Blockchain-empowered Data-driven Networks: A Survey and Outlook
Xi Li, Zehua Wang, Victor C.M. Leung, Hong Ji, Yiming Liu, Heli Zhang
aa r X i v : . [ c s . N I] J a n B LOC KCHAIN - EMPOWERED D ATA - DRIVEN N ETWORKS : AS
URVEY AND O UTLOOK
A P
REPRINT
Xi Li [email protected]
Zehua Wang [email protected]
Victor C.M. Leung [email protected]
Hong Ji [email protected]
Yiming Liu [email protected]
Heli Zhang [email protected]
February 1, 2021 A BSTRACT
The paths leading to future networks are pointing towards a data-driven paradigm to better cater tothe explosive growth of mobile services as well as the increasing heterogeneity of mobile devices,many of which generate and consume large volumes and variety of data. These paths are alsohampered by significant challenges in terms of security, privacy, services provisioning, and networkmanagement. Blockchain, which is a technology for building distributed ledgers that provide animmutable log of transactions recorded in a distributed network, has become prominent recently asthe underlying technology of cryptocurrencies and is revolutionizing data storage and processing incomputer network systems. For future data-driven networks (DDNs), blockchain is considered asa promising solution to enable the secure storage, sharing, and analytics of data, privacy protectionfor users, robust, trustworthy network control, and decentralized routing and resource managements.However, many important challenges and open issues remain to be addressed before blockchaincan be deployed widely to enable future DDNs. In this article, we present a survey on the existingresearch works on the application of blockchain technologies in computer networks, and identifychallenges and potential solutions in the applications of blockchains in future DDNs. We identifyapplication scenarios in which future blockchain-empowered DDNs could improve the efficiencyand security, and generally the effectiveness of network services.
Keywords data-driven networks, blockchain, networking technologies, blockchain-empowered data-driven networks
With the explosive increase of network users and their myriads of network-connected devices generating and consum-ing massive amounts of digital contents, future networks will encounter significant challenges to provide intelligentservices in an efficient and secured manner to users due to the large volumes of data generated, transmitted and pro-cessed in the networks. According to the 2019 Cisco report [1], global Internet data will increase nearly three-foldsin the next five years, and global data traffic will reach 4.8 ZB/year by 2022. Obviously, the paths towards futurenetworks are pointing towards a data-driven paradigm, i.e. , using data-driven networks (DDNs) [2] [3], to enhanceexisting services and enable new services to both humans and devices. However, there are obstacles along these pathsarising from challenges including security, privacy, services provisioning, and network management. Among the possi-ble solutions for future DDNs, blockchain, which gains prominence as the technology for cryptocurrencies, has drawnmuch attention from both academia and industry and is now considered as a promising direction to address manychallenges. Compared with traditional data tools such as database, blockchain gets rid of the centralized authority,and establishes integrity and transparency in distributed nodes. Therefore, future blockchain-empowered data-drivennetworks (BDNs) can provide intelligent services with better quality of experience (QoE) for users on the premiseof more secure data storage, sharing, and analytics, better protection of users’ privacy, more robust and trustworthy rXiv
Template
A P
REPRINT
Figure 1: Skeleton of the survey.network control, and decentralized routing and resource managements. Therefore, although research on BDNs is stillin the early stage, it is timely and significant to survey the underlying techniques and technologies and shine the lighton critical open issues.Blockchain is a distributed and practically immutable ledger of transactions. Each block in a blockchain contains oneor more transactions and points to the prior one [4]. As the primary purpose of a blockchain, transaction records can besecurely saved and validated in an untrusted peer-to-peer system by applying a consensus mechanism in a decentralizedmanner. The consensus mechanism is used to guarantee that the majority of peers can reach an agreement on thetransaction records disseminated over the network with gossip protocols. In addition, transactions can implement andexecute the operational codes saved in a blockchain, enabling software services among untrusted users. Since first usedin cryptocurrency [5], blockchain has attracted increasing interests with lots of use cases and applications. Particularly,blockchain has been envisioned as a promising solution to many problems when new technologies, including cloudand edge computing [6, 7], big data [8], and Internet of Things (IoT) [9, 10], are integrated with networks.By leveraging blockchain, current DDNs can evolve from many perspectives in the near future. Future DDNs shouldbe able to provide feasible solutions for dynamic access control, guarantee the integrity and validity of the exchangeddata, and preserve the privacy of mobile users. Blockchains may also be integrated in future DDNs for maintainingtight synchronization among different network elements equipped with storage, computing, and networking resources.While the advantages are quite inspiring, several significant research challenges need to be well investigated beforethe widespread implementation and deployment of BDNs will be possible, such as scalability in transaction through-put, decentralized intelligent network control, and resource management and allocation. However, to the best of ourknowledge, these challenges have not been well addressed in previous work. In this survey, we explore the relatedpioneer research works and investigate the applications of blockchain technology in computer networks to gleam theirpotential benefits when applied to future DDNs. A number of research challenges are also identified for BDNs whenconsidering the unique characteristics of blockchain technology. Our work takes the first steps toward a better under-standing of how blockchains may be embedded in future DDNs to improve the performance of data-driven applicationsand network management, and opens up pathways to the development and deployment of future BDNs.To better present our work, we summarize the structure of this article in Fig. 1. Section 2 presents the related works andmotivations. Section 3 presents the background of blockchain technology and describes the future DDNs. Section 4discusses the framework of future BDNs. Section 5 surveys in more details the related works on promising servicesenabled by future BDNs. Section 6 summarizes the existing works on management and control for future BDNs.Technical challenges and future directions of BDNs are discussed in Section 7. Section 8 concludes this article. Please2 rXiv
Template
A P
REPRINT also refer to the online supplemental file for additional materials including the characteristics and operation types ofblockchain, a summary of references in Section 7, and a list of acronyms.
The emergence of big data technologies has sped up the emergence of data-driven Internet to support new applicationsand efficient allocation of resources [11]. The DDNs can extract valuable information of the network through analysisof the massive amount of data collected over network nodes and end-points, based on which optimal strategies arelearned (most likely using machine intelligence that can improve its learning outcome also through big data analytics)and fed back to the network nodes and end-points to enhance network operation and management [12]. In this sub-section, we review some existing works on DDNs, as summarized in Table 1. These works have fueled the growinginterest in DDNs and continued evolution of the concepts of DDNs to leverage growing types of data. Challengesarising from the development toward future DDNs are discussed with more details in Section 2.2.1.Table 1: Existing research works on Data-driven Networks.Focus Ref. ContributionsNetwork architecture [13] Propose a data-based network architecture to enhance personal-ized QoE in 5G networks.[11] Propose a network architecture that collects data and optimizesnetwork from bottom up.[14] Propose a mobile network framework enabled by big data to pro-vide efficient resource allocation, content delivery, and RAN op-timization services.[15] Discuss the benefits of introducing big data to SDNs.Resource allocation [16] Introduce a big data-driven mobile network framework to pro-vide best QoE and minimize the cost.[2] Combine ICNs, SDNs and big data to achieve optimal resourceallocation and intelligent content delivery.[17] For computing and caching scenario, design a data-driven archi-tecture to achieve ultra-low latency in 5G networks.[18] Use DDN to improve the performance of VANETs.Network security [19] Combine artificial intelligence with DDN to monitor maliciousbehavior.[20] Propose a data-driven automated security monitoring architec-ture.Network optimization [21] Introduce the data-driven information plane to provide flexibleand intelligent services.[22] Propose a data-driven intelligent radio access network to pro-mote network optimization.Traffic management [12] For minimizing network traffic by allocating caching resource ofcontent routers.[23] Evaluate traffic prediction methods based on machine learningin DDN.Energy efficiency [24] Develop a comprehensive solution to improve energy efficiencyin IoT.
1) Network architecture.
Targeting emerging applications enabled by fifth-generation (5G) networks, Wang et al. in [13] propose a data-based network architecture for collecting users feedback and objective data that can describea user’s subjective experience to enhance personalized QoE in 5G networks. A DDN framework consisting of thedata plane, control plane, information plane, and market plane is considered in [11]. The information plane for dataacquisition collects topology information, quality of service (QoS) and QoE data, and provides these data to thecontrol plane, which utilizes these data to optimize the network architecture, protocols, resource management and taskscheduling. A mobile network framework enabled by big data analytics is proposed in [14], which provides efficientresource allocation, content delivery, and radio access network (RAN) optimization services. Cui et al. in [15] discuss3 rXiv
Template
A P
REPRINT the benefits of introducing big data to software-defined networks (SDNs), with respect to traffic engineering, cross-layer design, and security assurance.
2) Resource allocation.
In [16], a big data-driven mobile network is discussed, in which users’ and operators’ dataare collected and analyzed to provide best QoE through appropriate resource allocation while minimizing the cost ofthe infrastructure. Fang et al. in [2] combine the advantages of information-centric networks (ICNs), SDNs, as well asbig data technology to achieve optimal resources allocation, separation of data control and forwarding, and intelligentcontent delivery. To achieve ultra-low latency for computing and caching scenarios, a data-driven model incorporatinga data cognitive engine and a resource cognitive engine is proposed in [17]. The data cognitive engine analyzes cachingand computing data, and the resource cognitive engine performs resource awareness, such as dynamic storage andcomputing resources in small cell clouds. In vehicular networks, Cheng et al. in [18] show that the vehicle mobilitytracking data and the measurement data can be used to evaluate vehicular ad-hoc network (VANET) performance,optimize resource allocation, and design new protocols with big data intelligence.
3) Network security.
Sammarco et al. combine artificial intelligence methods with DDN and use unsupervisedclustering-based procedure to analyze the generated data, thus enabling malicious behaviors to be monitored in thenetwork [19]. In the field of IoT security, Astaras et al. introduce a data-driven automated security monitoringarchitecture [20], which is based on reusable security templates, and perform advanced data analysis through deeplearning to detect anomalies in each layer of an IoT system.
4) Network optimization.
Huang et al. in [21] introduce the data-driven information plane into the traditional SDNarchitecture to provide more flexible and intelligent network services. In [22], Lin et al. propose a data-driven intelli-gent RAN by applying big data in wireless network and machine learning methods to all layers of the communicationsystem to promote intelligent network optimization.
5) Traffic management.
In order to deal with the rapid growth of traffic in the network, Yao et al. in [12] suggestthat the SDN and content-centric network architectures should be combined to minimize network traffic by solvingthe problem of caching resource allocation on content caching routers. In [23], Ma et al. focus on the important roleof traffic prediction in active network optimization, and evaluate several existing traffic prediction methods based onmachine learning in DDN to implement better traffic management.
6) Energy efficiency.
Zhang et al. utilize the data-driven mechanism to develop a comprehensive solution for large-scale and energy-efficient IoT network [24], which uses reinforcement learning to analyze the data collected by sensornetworks in order to improve energy efficiency in IoT.
Recently, researchers have investigated and proposed a series of solutions that optimize blockchain technology andapply it to computer networks for different applications and services. There are several surveys, as summarizedin Table 2, that comprehensively review these proposed solutions from multiple angles to highlight the benefits ofadopting blockchain technology in computer networks.
1) Combined with cloud and edge computing.
The current states of blockchain for the cloud market, which is dom-inated by a few providers currently, is investigated in [25]. The authors provide some solutions based on blockchainand smart contracts, and then discuss the pending problems for applications in the cloud market. Yang et al. in [26]summarize the current research works of adopting blockchain in edge computing, and survey the enabling techniquesin terms of networking, storage, and computing.
2) Combined with artificial intelligence (AI).
Several pioneering works have shown the potential benefits ofblockchain in AI, such as enhancing data security, making collective decisions, and realizing distributed intelligence.Salah et al. in [27] present a broader perspective of blockchain-enabled AI applications, including decentralizedblockchain-based data storage and management, decentralized infrastructure, and decentralized AI applications.
3) Combined with security management.
Blockchain is considered in [28, 29] to build a distributed, trustworthy,and secure architecture for IoT applications . Khan et al. in [30] review and categorize common security issues relatedto the layered IoT architecture. Decentralized networks supported by digital ledger technology (DLT), which providesa method to promote trusted interactions between IoT devices, have the potential to achieve security control, identitymanagement, and traceable machine-to-machine transactions. Zhu et al. in [31] focus on identity management systemsin IoT, with higher requirements for scalability, mobility, security, and privacy. They investigate the recent identitysolutions based on blockchain, and point out that blockchain can transfer access control and identity management tothe edge computing devices close to the identity owners. Pohrmen et al. in [32] investigate heterogeneous networksincorporating IoT, SDN, fog architecture as well as blockchain technology. The distributed structure of blockchain canmake the entire system more resilient to attacks and single points of failure, thereby reducing the impacts of attacks4 rXiv
Template
A P
REPRINT in SDN-based IoT and cloud-based communications. Yang et al. in [33] survey the fundamental issues in the securityservice architecture, and the usage of blockchain in the data, control, network, and application planes. Salman et al. in [34] investigate some blockchain solutions for typical security services.
4) Application in IoT networks.
Christidis et al. in [35] study how to adopt blockchain to promote the sharingeconomy, while using smart contracts to automate multi-step processes in IoT networks. Focusing on the central-ized client/server model in the current IoT, Atlam et al. in [36] evaluate the benefits and challenges of combiningblockchain with IoT in details. For the typical IoT scenarios, the advantages of blockchain ( e.g. , distribution, trust-free, transparency, decentralization, and automation) can help improve the services of smart cities, industrial domains,smart grids, etc . Xie et al. in [37] consider the applications of blockchain in smart cities, such as smart citizens, smarthealthcare, smart transportation, supply chain management and other issues. Al-Jaroodi et al. in [38] present the re-quirements, opportunities, and challenges of introducing blockchain into different industrial fields, where blockchainledgers can introduce digital identities, security records, smart contracts, and distributed storage to the industrial ap-plications. Siano et al. in [39] introduce blockchain into energy systems to improve the traditional centralized powersystem. This work reviews the distributed transactive energy systems (TESs) and proposes an interactive energy man-agement architecture for peer-to-peer energy exchange based on distributed ledgers which uses proof of energy (PoE)as a consensus mechanism.
Many researchers have investigated the integration of blockchain technique into various network architectures. A nextgeneration blockchain network (NGBN) model based on peer-to-peer interactions is proposed in [40]. The physicallayer and application layer are converged into a single networking layer called the blockchain network layer (BNL),which incorporates encryption, storage, traffic balance, token control, and consensus to ensure secure data transfer withlow latency. In addition, several research works have considered how to combine blockchian with existing frameworks,Table 2: Existing works of blockchain-enabled applications and services in communication networks.Focus Ref. ContributionsCombined with cloudand edge computing [25] Compare three solutions based on blockchain in cloud comput-ing, with emphasis of cloud market standards.[26] Summarize the current research on the integration of blockchaintechnologies and edge computing.Combined with artifi-cial intelligence [27] Show a broader perspective of blockchain-enabled AI applica-tions.Combined withsecurity management [28] Discuss a distributed and untrusted IoT architecture based onblockchain.[29] Summarize the use of DLT in the IoT and the problems that DLTcan solve.[30] Investigate the blockchain solution for the security of the IoTlayered architecture.[31] Focus on the identity management systems based on blockchainin IoT.[32] Investigate heterogeneous networks which take IoT, SDN, fogarchitecture, and blockchain into one paradigm.[33] Investigate the application of blockchain in security of multi-layer network.[34] Investigate some blockchain solutions for several types of secu-rity services.Applied in IoTnetworks [35] Study on how to use blockchain to promote the sharing of ser-vices and resources in IoT.[36] Describe the benefits and challenges of combining blockchainwith the IoT.[37] Present a survey on some blockchain application in smart cities.[38] Focus on the requirements, opportunities, and challenges of in-troducing blockchain into different industrial fields.[39] Introduce blockchain into energy system to improve the tradi-tional centralized power system.5 rXiv
Template
A P
REPRINT such as SDN and cloud computing, to optimize the architecture and functionalities of the network in order to providesecure and decentralized services in typical communication scenarios.
1) Blockchain-based SDN architecture.
In order to deal with security challenges in SDN, Weng et al. in [41]propose a secure blockchain-based SDN network architecture, which consists of the data plane, blockchain plane,control plane, and application plane. The blockchain plane provides functionalities of resource-recording and resource-sharing among multiple controllers in the control plane. All the application flows and network events associatedwith the respective network conditions are recorded on the blockchain as transactions. In the control plane, multiplecontrollers participating in the underlying blockchain are responsible for recording network data from the applicationplane and the data plane as transactions into the blockchain. A kind of Byzantine fault tolerance (BFT) protocols, likethe Ripple network [42], are adopted to guarantee low latency and avoid the temporary forks.For optical networks, a distributed blockchain-based network architecture is described in [43] along with two dis-tributed multi-controller credible routing (MCR) schemes for software-defined data center optical networks. Datacenters connect to multi-domain elastic optical networks that implement the computing, storage, and optical spec-trum resources allocation, respectively. These domains are software-defined and manipulated by collaborating SDNcontrollers, and each data center can accommodate trustful cross-domain lightpaths based on the blockchain network.
2) Blockchain-based cloud computing architecture.
Sharma et al. in [44] propose a distributed blockchain-basedcloud architecture at the edge of the network, which consists of the device layer, fog layer, and cloud layer. Thedevice layer transmits the filtered raw data to the SDN-enabled fog layer, where all SDN controllers are connectedin a distributed manner using a blockchain, and each SDN controller is responsible for network management. Thefog nodes transmit the processed data to the distributed cloud and device layers, which enables them to access andoffload computing tasks to the cloud when computation resources are insufficient. To improve the performance ofcomputation as well as the data transfer and storage in the blockchain, a consensus protocol is proposed to combinethe advantages in both Proof of Work (PoW) and Proof of Stake (PoS) consensus mechanisms by 2-hop blockchaintechnique.
3) Blockchain-based IoT architecture.
In [45], a blockchain-based multi-layer mode is proposed for IoT, whichconsists of the edge layer and high-level layer. The edge layer is considered as a local area network, where a number ofnodes and a leading central node are deployed for providing interfaces to the high-level layer for addressing, allowingbidirectional data transfer, and participating in high-level layer activities. In the high-level layer, all the nodes aredata-independent with full replica records exchanged between each other across the blockchain. They operate with acommon interface to the edge layer, independent of the operation of the edge nodes.
4) Blockchain-based VANET architecture.
For vehicular communication systems, Zhang et al. in [46] presenta security blockchain-based architecture of VANET with mobile edge computing. The architecture consists of theperception layer, edge computing layer, and service layer. The perception layer, where vehicles and roadside units(RSUs) are connected through blockchain, guarantees high security for the transmitted data. Computation resourcesand cloud services are provided by the edge computing layer as well as the service layer. The edge computing layeris responsible for handling a large number of transactions and other computation intensive tasks for the perceptionlayer, while the service layer ensures the security recording of data, including traffic violation history, traffic accidentinformation, etc . As the types of data and services keep growing, the concepts of data-driven services will continue to evolve, whichresults in increasing scale and performance requirements of future DDNs, as well as other challenges, as illustrated inFig. 2.
1) Privacy:
For data-driven services, the analytics of user’s data is conducive to support personalized applications.With the increasing variety of data collected and stored in the future DDNs by Internet service providers and data-mining companies, how to protect users’ privacy while providing intelligent services becomes a great challenge. Ac-cording to the research in [47, 48], there are many issues that affect users’ privacy. Massive user data, which arecollected, transmitted, and analyzed in the network, may reveal informative results and carry immense values. How-ever, users may not be informed about what their data are used for, nor how to protect their privacy or secure theirpersonal information. There have been several disclosed cases in which illegal operations have been performed with-out authorization or users’ knowledge, resulting in severe personal information leakage and punishable offence topublic safety. On the other hand, most data mining uses linking, reconstruction, and inference operations, whichmay potentially breach users’ privacy. Conventional privacy protection approaches ( e.g. , encryption, anonymity, client6 rXiv
Template
A P
REPRINT personalization, etc. ) focus on enterprise networks by addressing the vulnerabilities at network gateways and accessportals [49, 50]. These operations are performed by the centralized storage, computing, and data processing. However,these methods cannot prevent the leakage of user privacy caused by internal malicious attacks. With mobile usersusing online applications that generate/consume data in a decentralized manner, previous privacy protection mecha-nisms may no longer be adequate for personal data and privacy protection. Future DDNs should solve this problemin a complete, consolidated and secure paradigm instead of the current patchwork and isolated solutions deployed byindividual Internet service providers.
2) Security:
Security has been a matter of great concern in computer networks due to the proliferation of manyforms of malicious attacks [51, 52], e.g., de-anonymization attacks, sniffer attacks, distributed denial-of-service attacks(DDoS), and Domain Name System (DNS) attacks. In future DDNs, the variety and complexity of data resources andformats, as well as their respective transmission, storage and analytics requirements, elevate the security issues to a newheight with new challenges. Data-driven applications are designed on the premise of effectively utilizing various datato provide intelligent and customized services. False information and malicious attacks can destroy the credibility ofthe data, resulting to unacceptable QoS or even hamper the normal operating of the devices and terminals. Furthermore,unlawful access to sensitive and private data by hostile hackers may bring unimaginable losses for the data owners andthe public. Existing security solutions for computer networks are designed to counteract specific malicious attacks ina certain network environment [53, 54]. The current approach of reacting to the ever-emerging novel malicious attackswith tailored solutions leads to delays in the introduction of effective actions and tedious patching on the existingnetworks. New security paradigms need to be devised for future DDNs to provide powerful protection in terms of theentire life cycle of the data.
3) Authentication:
For DDNs with multiple service providers, infrastructures, and users, authentication is very im-portant for mutual identification and data confidentiality. However, there are still several problems in authentications[55, 56]. Currently, the most widely used form of user authentication in modern computer systems is password, whichis vulnerable to password leakage as well as the password-cracking and social-engineering attacks. In addition, thetraditional threats to Internet users include message replay, dependence on trusted third parties which might be compro-mised, and man-in-the-middle (MITM) attacks. All these threats can prevent legitimate users from being successfullyauthenticated or allow illegitimate users to be authenticated. As to the existing solutions, in some network scenar-ios, anonymous authentication may provide secure authentication and access control [57]. However, anonymity alsobrings some problems. Many studies have focused on the implementation of pseudonymization technology (PT) byintroducing trusted authorities [58], which are easily affected by the single point of failure. In addition, consideringthat anonymous users are often indistinguishable, it is difficult for the network service provider to punish misbehavinganonymous users. When individual authentication is required, digital certificates are often used for this purpose [59],which may consume a significant amount of network overhead for computation and certificate validation. Therefore,the efficient implementation of effective authentication methods becomes very important.
DDNs require massive storage and processing of data in the networks, for which traditional solutions such asdatabase technologies are applicable. Compared with representative database technologies such as Oracle and MySQL,blockchain has obvious merits. As a distributed ledger, blockchain can effectively avoid data leakage and single pointof failure caused by authorized administrators, while its anonymity can protect the privacy of data owners to a certainextent. The blockchain only retains the two database operations of data reading and adding, and the process of addingdata must be verified by the consensus protocol, which ensures the transparency and integrity of the entire database.Figure 2: Challenges of future data-driven networks.7 rXiv
Template
A P
REPRINT
Therefore, applying blockchain to the future DDNs can uniformly solve the above challenges in terms of privacy, se-curity and authentication. Specifically, blockchain technology provides a promising solution that can address many ofthe challenges in future DDNs while bringing several potential benefits as follows:
1) Security and privacy:
Existing works have discussed about how to apply blockchain to improve security andprivacy in the networks from different points of view[60, 61]. First, blockchain nodes are decentralized and supportive to network robustness. Even if some nodes in thenetwork are compromised by various attacks, other nodes can work normally and the data will not be lost. This featureof blockchain increases the overall network robustness compared to existing centralized and distributed data systems.Second, the transparency of blockchain can be utilized to make the data flow in the network completely open to users.This provides traceability to users’ personal data usage, thus informing the users how their data is used. Third, theimmutability of data in the blockchain increases the reliability of activities in the network and enhance the mutual trustbetween users and service providers. Finally, pseudonym of blockchain helps network users to keep their real-worldidentity hidden for privacy protection.
2) Data and model sharing:
In order to support massive data and application requests for sharing data and informationin a secure and effective manner, applying blockchain in future DDNs can protect network data from tampering, andeffectively preserve the privacy for network users. Moreover, blockchain is more transparent and secure in datasharing. By better protecting the privacy for users’ data and the security of users’ application, BDNs can providebetter personalized services with more add-on values.
3) Credibility and malicious operation tracing:
With the immutable and distributed advantages of blockchain, futureDDNs can increase the credibility of the network. Malicious operations and mendacious messages can be traced backby all the participants that can access the transaction records saved in the blockchain. Cloud-based services can storethe credit value information of the service providers, which enables both the users and service providers to correctlyverify the legitimacy of the operations.
4) Enhanced decentralized solutions:
As a distributed ledger of transactions, blockchain technology is quite suit-able for peer-to-peer interaction and decentralized intelligent services, such as federated learning-based applications.As long as the nodes of the system are running compatible consensus mechanism or protocol, they can access thetransaction records while not being able to change any of them without being noticed. Thus, it can provide a moresecure and convenient method compared with current decentralized solutions for data sharing and cooperation amongindependent nodes in a large scale network deployment.
A blockchain employs decentralized digital ledgers that are maintained by peer nodes. Generally speaking, the archi-tecture of blockchain involves six layers, i.e. , the data layer, network layer, consensus layer, incentive layer, contractlayer, and application layer, as illustrated in Fig. 3 [62]. Each layer has specific core functions and key technologiesas described below. As for more background of blochchain including the characteristics and operation types, pleaserefer to the online supplemental file.•
Data layer : The data layer is the lowest layer in the blockchain architecture. Its key technologies includeMerkle tree, asymmetric encryption, timestamp, digital signature, and hash function. The data layer con-structs the basic data structure of a blockchain for organizing and storing data [63]. A typical blockchaindata structure consists of a header part and a body part. The header part contains meta-information such asversion information, hash value of the previous block header, timestamp, nonce, Merkle root of the containedtransactions, and target difficulty used to calculate the next block. The body part stores a Merkle tree of theverified hash data, which enables the blockchain to verify the existence and integrity of data efficiently andsecurely.•
Network layer : Blockchains operate in a peer-to-peer paradigm. The main task of the network layer is to sup-port the exchange of information between peer nodes in the network, while ensuring the security and privacyof the data. In a peer-to-peer network, the node that generates a transaction broadcasts the transaction to itsneighboring nodes. Each node that receives the transaction then verifies it according to the correspondingcheck list. Only authenticated transactions are forwarded by the node.•
Consensus layer : The consensus layer encapsulates the consensus algorithm for the distributed nodes. Itguarantees the data consistency and fault-tolerance of the shared ledger. There are some well-known con-sensus algorithms such as PoW, PoS, Delegated PoS (DPoS), Practical Byzantine Fault Tolerance (PBFT),8 rXiv
Template
A P
REPRINT
Raft, etc . PoW requires nodes to compete for opportunities to append blocks on the ledger through mathe-matically difficult calculations [63]. It relies on the computing power of nodes. In a PoS system, the creatorof the next block is determined by the number of assets a peer holds and the duration since the peer createda block last time. Compared to PoW, PoS requires far less resources to run, but decreases the liquidity ofcryptocurrency. DPoS is a variant of PoS, which generates block producers in a round-robin order. It sacri-fices the complete decentralization characteristic to provide high throughput and scalability. PBFT uses thedemocratic mechanism of majority-rule to select leaders and keep accounts. It is a fault-resistant, fast, andlong-lived mechanism. However, a large number of nodes will consume a heavy communication overhead.Raft is an easy-to-understand general consensus protocol that uses the “leader and follower” model. As apluggable consensus module, Raft is currently supported by some popular blockchain architectures such asHyperledger Fabric.•
Incentive layer : The incentive layer, which integrates economic rewards into a blockchain system, mainlyappears in public blockchains. In such a decentralized system, nodes are actually self-interested and theirfundamental purpose of participating in data validation and book-keeping is to maximize their own bene-fits. Using tokens to reward nodes that follow the rules is the most extensive incentive mechanism in theblockchain system. The incentive layer formulates the token issuing and distribution mechanism, whichdetermines the total amount and circulation of tokens.•
Contract layer : The contract layer encapsulates various Opcodes, Chaincodes, and smart contracts. Inblockchain systems, smart contracts are predefined commitments and rules [64], which are executed auto-matically when some predetermined conditions are met. Smart contracts enable trusted transactions withouta third party, and these transactions are traceable and irreversible. Opcodes and Chaincodes specify the detailsof trading and processes, thereby increasing the autonomy and programmability of the blockchain network.•
Application layer : As a basic technology, blockchain has been applied in multiple fields, such as financialmarket, healthcare systems, smart city, and energy market. The highest layer of blockchain includes variousapplications to provide secure, distributed, and customized services. In the long run, blockchains will bringmore solutions to future networks when combined with advanced data and communication technologies.Researchers have continued to contribute to the technologies and protocols associated with blockchain [35]. A compre-hensive survey on the security and privacy aspects of blockchain is presented in [65]. Neudecker et al. in [66] associatethe attacks and security requirements with network layer design, and investigate the requirements and adversary modelof the network layer design. The impacts of consensus and incentive mechanisms on the consensus participants areinvestigated in [67] from a game-theoretic perspective, which highlights how the consensus mechanisms affect theemerging applications of blockchain networks. A vademecum has been made in [68] to guide designers. The authorsnot only comprehensively introduce the existing blockchain platforms, but also propose the key requirements whenevolving from the permissionless blockchains to the permissioned blockchains.
Data LayerApplication Layer Programmable
Currency
Programmable
Financy
Programmable
Society
Contract Layer Opcode Smart Contract Chaincode
Incentive Layer
Token Issuing Mechanism Token Distribution MechanismConsensus Layer ...PoW PoS PBFTNetwork Layer P2P Network Proving
Mechanism
Transmission
Mechanism
Block-structuredTime Stamp
Asymmetric
EncryptionHash FunctionMerkle TreeDigital Signature RaftDPoS
Figure 3: The six-layer blockchain protocol stack.9 rXiv
Template
A P
REPRINT
The rapid development and widespread of data-driven applications have brought promising opportunities to reshapethe Internet architectures as well as operation and optimization solutions, thanks to emerging data analytic capabilitiesthat can uncover the knowledge and statistical patterns hidden in massive data [11]. Data-driven functionalities maybecome some of the most important features for future computer networks, particularly to efficiently handle the sky-rocketing user traffic while leveraging the huge amount of data generated and digested inside the networks to improvethe effectiveness of network management, resource allocation, and security control.
1) Layered architecture in future DDNs:
The architecture of future DDNs consists of three planes, i.e. , the data,management, and network planes, as shown in Fig. 4. Unlike the traditional vertical layered model, the proposed3-dimensional (3D) model also describes the interactions within and between the three planes.The data plane consists with four layers, including the data sensing layer, data repository layer, data processing andanalyzing layer, and data intelligence and service layer. The data sensing layer collects various data, including usertraffic, network performance indexes, as well as the network management signalling and operational directives, fromthe network entities. Such real-time information is critical to enable the use of advanced data analytic methods to revealthe true network status and problems. The data repository layer stores all the collected data. The cloud architecturecould be used for such purposes. Then, the data processing and analyzing layer processes the data and outputs thenecessary information to enable the data intelligence and service layer to solve network problems and support optimaldecision-making. This process enables effective resource allocation strategy for network self-optimization, as well asend-to-end network intelligence as discussed in the context of the other two planes.The network plane consists of the network infrastructure, controller, and application layers. The network infrastructurelayer includes all kinds of network entities in various typical communication scenarios. The network controller layerworks with function modules in distributed network nodes and centralized management servers and is responsible forpolicy making and dispatching. The application layer provides various network applications and services for the users,such as network reconstruct, secure transmission, and QoE.The management plane has three layers that are responsible for network operation, network administration, and net-work optimization, respectively. Taking advantage of data analytics and end-to-end network intelligence in the dataplane, a series of automatic operations including smart maintenance, troubleshooting, configuration, and optimizationfunctionalities can be implemented in the management plane. It can further provide elaborate fine-grained networkcontrol according to practical requirements and real-time feedback of network status to improve QoE for users.
2) Characteristics of Future DDNs:
With three planes orchestrating effectively, future DDNs are expected to havefollowing advantages in support of intelligent services with desirable QoE.•
Intelligent and autonomous control : Future DDNs with highly sophisticated data computing and analyticcapabilities can support autonomous network operation and maintenance. With the help of virtualization andever-increasing computing capacity, network elements become more intelligent and thus can react quickly tochanges in the network environment. Thus, future DDNs can update the network parameters autonomouslyin real time during network operation.
Data Intelligence and Service
Data Processing and Analyzing
Data RepositoryData SensingNetwork Optimization
Data Plane Management Plane
Network AdministrationNetwork Operation
Data Driven OAM
Data Driven Control and
Application
Figure 4: The layers in future DDNs.10 rXiv
Template
A P
REPRINT • Low cost and high efficiency : Data from all network entities could be stored and analyzed efficiently to opti-mize network operation, which can decrease the operating expense (OPEX) and capital expenditure (CAPEX).Intelligent self-maintaining scripts and algorithms could be designed for joint resource allocation and smartnetwork operation to free engineers from cumbersome manual tasks. Furthermore, due to the growing com-puting capacity and hence intelligence of network elements, many decisions can be made and executed in adistributed manner, which can potentially greatly decrease the OPEX.•
Low latency and real-time operation : In future DDNs, the data plane can perform the time-consuming dataprocessing and model training offline, thus providing the tuned model to the management plane for quickreaction to the changes of the the network or user requirements. The distributed and cooperative operation ofthe network entities in the network plane may also help to realize low latency for the end users.•
Heterogeneity and scalability : Future DDNs are flexible, extensible, and scalable for the ever-increasing net-work coverage and heterogeneous types of accessing devices. Moreover, network nodes and servers, whichmay be implemented across platforms, can collaborate effectively in data collecting, processing, and analyz-ing by highly intelligent information exchange on the premise of powerful security, privacy and authenticationprotections.
While future DDNs promise to provide efficient, flexible and intelligent services for an ever-increasing number ofusers and their terminal devices, there are many challenges that may hamper the fast evolution of present-day networkstowards DDNs, as we have addressed in Section 2.2.1. By leveraging blockchain technology, we believe that promisingBDN solutions can be found to address these challenges, as indicated by the discussions in Section 2.2.2. To facilitateresearch towards these solutions, we present in this section a general framework of BDNs, which provides a scalableand universal architecture for integrating blockchain into future DDNs.As shown in Fig. 5, our framework of BDNs is based on the future DDN described in Section 3.2. While the dataengine collects as well as processes massive and various data in different domains of the network, the blockchainengine interacts with the data engine by providing secure data storage, private data sharing, and decentralized networkoperation.
1) Network domains:
We divide the network into three domains, namely the access network domain, core networkdomain, and application domain.
The access network domain is where a myriad of terminals connect to the network by different communication tech-niques through wired or wireless links. There are also various edge network nodes, such as base stations and accessingpoints, edge switches and routers, gateways and edge data centers, cooperating together to perform access control, re-source allocation, and data processing, with huge amount of generated and digested data including resource utilizationstatus, operational node performance, and environment monitoring results.
The core network domain consists of large number of powerful routers and high-speed optical links, distributed cloudcomputing centers and data storage repositories, and facilitates the main functions of network management, resourceoptimization, and the signalling and data packets.
The application domain provides interfaces for various kinds of services and user applications, such as industrial IoT,medical system, vehicular network, and smart city. The application layer interacts with the core network and accessnetwork layers via application-control interfaces of the nodes to support different services flexibly and intelligently.
2) Data engine and blockchain engine:
The core modules of the proposed BDN framework are the data engine andblockchain engine. All the network domains interact with the data engine, which is responsible for various operationsrelated to all kinds of data. And then, the data engine and blockchain engine exchange data and parameters with eachother to coordinate operation and optimization of the network.
The data engine is composed by several core components. These modules are physically distributed in various nodesand devices across the network. The data collection module, which is equipped in all the users’ terminals and net-work nodes, collects various data including user behavior, resource utilization status, traffic data, device performance,management signalling and operational messages from the network in a timely manner. It also monitors and sensesexternal network environment, such as geographic location and electromagnetic interference. The collected data isstored in the data storage modules of edge and core data centers based on the cloud architecture. The data analyticsmodule, equipped with strong computing ability as well as a comprehensive library of data-mining and data analyticalgorithms, is called upon to digest these stored data in different dimensions according to the requirements from thedata optimization module. All the nodes with available computing and storage capabilities, including powerful servers11 rXiv
Template
A P
REPRINT in data centers, routers, edge nodes, and user terminals, can be equipped with a data analytics module, enabling them toexecute data analytics tasks independently or cooperatively. There are interfaces for sharing information among thesemodules and various nodes, which is an important standardization work for the BDNs. The close cooperation amongthese modules in the data engine guarantees integrity and efficiency of data-driven services and operations. TTheMerkle tree and block hash are used to secure verification of data in a large dataset, and help to verify the consistencyof the data. A set of nodes store all the blockchain data, which can be easily synchronized and maintained in the eventof an individual node failure.
The blockchain engine takes advantage of the decentralized, secure, and private features of blockchain, and is supportedby each node inside the blockchain network. It enables a series of network functions, including trust mechanisms, dataintegrity, virtual resource management, attack-resilience and fault-tolerance, which will be discussed in Section 5and 6 with more details. It can further support secure and private data sharing, storage, and processing as well asdecentralized network operation according to requirements and feedback, which can provide users with better QoE.In addition, the blockchain engine interacts with the data engine in a secure, effective, and trustful manner. Theblockchain engines in the network nodes collaboratively select a node according to the applicable consensus algorithmand authorize it to create a new block to encapsulate the transaction data with a specific timestamp generated over thenetwork. In this way, a myriad of data can be shared among completely decentralized nodes and entities.With the data and blockchain engines, the proposed framework of BDNs can form the basis for the design and imple-mentation of future BDNs to enable intelligent and secure distributed network operations. Considering the explosiveincrease of network data and their myriad of values, BDNs are capable of performing comprehensive data fusing andinformation extraction over the huge amount of collected raw data. By correlating various influencing factors andnetwork performance, the data engine can deduce the causality and logic behind these data with the help of advanceddata analytics and machine learning techniques by taking advantage of the ever-increasing computing capacity of thenodes. Then, it can optimize network operation intelligently and efficiently, especially when the network behaviorsare complicated and a massive number of network parameters need to be managed. Furthermore, with proper autho-rization, the output of the data engine can be made available to third-party application providers through standardizedinterfaces, thus supporting diverse and high-quality services to users. The blockchain provides decentralized controlthat is particularly fitting for the condition that organizations or individuals are situated in different geolocations. Theindividual nodes in future BDNs can delegate or vote their representative authority for decentralized control rather thanlimiting it to a few authorized nodes. Data integrity can also be guaranteed by a blockchain, which introduces a setof protocols to verify the participating nodes as well as new transactions. After the majority of nodes have reached averification consensus, the new transactions are recorded and saved in the new block within the distributed architectureFigure 5: Architecture of blockchain-empowered data-driven networks.12 rXiv
Template
A P
REPRINT to improve the data security and integrity. The blockchain engine operates harmoniously and effectively with the dataengine to protect the secure data storage and trustful network management.Under the proposed framework of BDNs, data from all the network entities at various levels could be stored and pro-cessed to optimize network management and operation, which can reduce both OPEX and CAPEX. Blockchain withdigital signature and hash functions are applied to guarantee the data integrity and immutability in the network. In thisway, the future BDNs can provide efficient solutions to achieve optimal network control and data-driven applications,especially when the network environments and scenarios are complicated and frequently changing with various usersrequirements.
In future networks, massive data generated by connected devices and network equipment provide promising possibil-ities for improving the QoS of the emerging data-driven applications through data sharing. However, these data arescattered in different systems, and controlled by the respective service providers. How to enable secure data sharingand analyzing in a decentralized manner is an important issue that may be addressed through the application of BDNs.Meanwhile, privacy is one of the key issues in data sharing. Conventional privacy technologies, such as encryption,authentication, and role-based access control, may not be sufficient to satisfy the efficiency and security requirementsin future computer networks. As BDNs may provide effective solutions to privacy issues, many researchers haveconducted pioneer works from different directions.A data sharing system based on blockchain can greatly simplify the data acquisition process, control access to thestored data, track the use of data, and facilitate users’ data ownership and privileges. A blockchain-empowered datasharing architecture is designed in [69] for distributed multiple parties. These parties agree to share their own data soas to implement a collaborative task together. Considering the common distributed data sharing framework involvingmultiple parties, all the IoT end devices are interconnected within a permissioned blockchain, which is maintainedby the super nodes implemented with computing and storage resources. Using the federated learning method, thepermissioned blockchain stores the federated data model learned over decentralized parties instead of all the raw data,which can trace the usage of data for further auditing while ensuring data security and privacy. How to improve themapping of raw data to federated data under the limited resource constraint of IoT devices is a key problem in datasharing.In order to maximize the data collection ratio as well as geographic fairness, a joint deep reinforcement learning(DRL) and blockchain-based secure data sharing framework is proposed by Liu et al. in [70]. A fully distributedDRL scheme is designed to help each mobile device to sense nearby environment to achieve maximum data collectionamount, geographic fairness, and minimum energy consumption. Ethereum blockchain is utilized to build a trustedplatform for mobile devices to share data without any third-party organization. Mobile devices in the blockchaincollect data through a multiagent DRL-based method, and transmit the encrypted data with respective private key anddigital signature to ensure data safety. Further improvement can be carried out from the aspect of improving the nodeintelligence using DRL.Chen et al. in [71] propose a blockchain-based data sharing incentive mechanism for the Internet of vehicles, anddesign a data quality-driven auction to perform the negotiation among data buyers and sellers for guaranteeing high-quality data and maximizing social welfare. A smart contract is designed for traffic data sharing among vehicles. Atamper-resistant consortium blockchain is introduced to securely store the on-chain data such as timestamp and sharedinformation after completing the data sharing ensuring the security and scalability of the algorithm.Many researchers utilize blockchain to realize the secure management and storage of distributed data with encryptionto facilitate sharing. A data storage and sharing framework is presented in [72], which consists of the InterPlanetaryFile System (IPFS), the Ethereum blockchain, and the attribute-based encryption (ABE) technology. Along with con-trolling the data that they own, users also encrypt the shared data through specified access control schemes. Moreover,encrypted keyword indexes are established for shared files, and the smart contract ensures that users need to pay a ser-vice fee only when the cloud server returns the correct result. However, it is challenging to reduce the costs associatedwith users’ attribute revocation and access policy updating. For secure blockchain-based online storage, Fukumitsu etal. propose a scheme without any third-party central server [73]. Data are secretly divided into several parts and sentto the nodes through secret sharing, which makes it difficult for attackers to obtain all the user data. The distributedstorage nodes remove the need for a centralized repository. A data sharing mechanism called
Meta-Key is proposedin [74], which shares encrypted data based on the distributed storage compatible architecture of blockchain. It allowsusers to encrypt data with their own public keys, and store the keys to the dedicated storage nodes in a blockchain13 rXiv
Template
A P
REPRINT network. The system can be improved by adding erasure codes to further enhance the security and reliability of thedata cipher-text. In [75], Raman et al. utilize a new combination of private key encryption, distributed storage, andShamirs secret sharing schemes to distribute transaction data, with data integrity ensured by specific encoding schemes.In addition, the Shamirs secret sharing scheme is used for hash value and dynamic region allocation to ensure securedata sharing under the premise of data integrity.In healthcare, the authors in [76] present a conceptual design employing cloud-assisted blockchain technology forusers to share their personal health data securely. They focus on continuous dynamic data, which account for most ofthe data generated by wearable devices and mobile devices, and integrate blockchain and cloud storage technologies tocollect and share the dynamic personal health data. Jin et al. investigate a novel secure and privacy-preserving medicaldata sharing mechanism in [77], which incorporate two types of blockchain techniques, namely, the permissionedand permissionless blockchains. In [78], different sharing requirements of medical data from medical institutionsand individuals are analyzed to drive the design of a healthcare information exchange platform composed of twoloosely-coupled blockchains. The transaction packing algorithm in the two blockchains can effectively increase thesystem throughput and the fairness of data sharing.
MeDShare is proposed as a blockchain-based medical data sharingsystem in [79]. A model based on blockchain for data sharing between cloud service providers is designed to ensureimmutability and non-tampering of data. Also, smart contracts and access control schemes are introduced to controland track the process of data sharing for detecting violation of permissions on data. This mechanism can safely achievedata provenance and auditing while ensuring user privacy.A decentralized personal data management platform is proposed in [80] to protect data privacy by utilizing theblockchain technology, targeting fine-grained access control management with out-of-chain storage. Two types oftransactions, namely transactions of access control and transactions of data storage and retrieval, are recorded byblockchain. Individual users can control their data, while service organizations can access the data to provide personal-ized services after legal authorization. One of the main contributions of this platform is to use protocols and encryptionto overcome the public nature of blockchain; however, it does not mitigate threats coming from malicious nodes in thenetwork, which can increase their reputation and then carry out attacks.Focusing on privacy requirements in IoT networks, an end-to-end data privacy-preserving framework named
PrivBlockchain is described in [81], which utilizes smart contracts to enforce privacy requirements and regulationsbetween data owners and data consumers. By deploying core components of
PrivBlockchain such as smart contracts,gateway nodes, etc. , with appropriate transaction protocols, IoT resources can be added, stored, and shared withoutviolating the end users’ privacy, thereby protecting privacy throughout the entire IoT data lifecycle.The authors in [82] propose a hierarchical blockchain-based data usage auditing architecture that relies on auditablecontracts in the blockchain to provide controllable and transparent data retrieving, sharing, and processing. Thisarchitecture consists of three entities: a data owner, a data controller, and a data processor. The authors make assump-tions about the actions of both data owners and service providers, and design an auditable contract and encryptionmechanism that not only protects user privacy, but also forbid unauthorized entities to process the data.Chen et al. in [83] propose a distributed partial ledger storage technique based on blockchain to protect user privacyin social networks. This technique combines blockchain technology with database file storage method, and thus canstore sensitive information securely. In order to save the computing power of user equipment, the proposed schemedoes not need to synchronize the entire ledger, and use proof of communications as a consensus mechanism. However,it does not perform well when the network is small ( e.g. , less than 1000 users) or has limited computational resources.In order to protect multimedia data privacy and provenance, Vishwa et al. design a decentralized data managementplatform to store, query, share, and audit multimedia data in [84]. The proposed platform combines blockchain encryp-tion with a cloud storage solution called data-lake to ensure that multimedia owners have knowledge of the collectionand utilization of their data without relying on a third party. The authors make full use of the trust-free nature ofblockchain, and use smart contracts to program the authenticity checks and authorization rules into the system toensure user rights; however, overall characteristics of the system have not yet been fully verified.Using searchable symmetric encryption, it is now possible to implement secure encrypted data search in a distributeddatabase. Jiang et al. in [85] propose a bloom filter-enabled search protocol to address data privacy issues in multi-keyword search. It can not only provide privacy-protected multi-keyword search in the blockchain but also improvethe efficiency of data search.Privacy protection in e-health systems has attracted wide attention with several research works. Al Omar et al. in [86]design a patient-centric medical data management system, which uses blockchain as the storage technology. Thesystem adopts a protocol called
MediBchain to ensure data privacy, accountability, authenticity, and integrity. Thepatients and doctors act as data senders, and their data are first encrypted in the
MediBchain . The private accessibleunit (PAU) is an intermediate unit between blockchain and users, which is used to authenticate the identity of the data14 rXiv
Template
A P
REPRINT senders and recipients. User data is stored in the blockchain. Each transaction in the blockchain returns a transactionidentifier that helps the user to access the data in the future. The authors in [87, 88, 89] also focus on the privacy ofelectronic health record (EHR). Specifically, Dagher et al. in [87] propose a framework named
Ancile , which is builton the Ethereum platform. Smart contracts and encryption technologies are used in this framework to enable datasecurity, access control, privacy and interoperability for EHRs. Magyar et al. in [88] consider how blockchain helpsto solve secure data storage problems while providing patient data in a standardized manner. The distributed EHRmanagement framework in [89] protects user privacy during data collection, management, and distribution, by usingblockchain for managing access control and EHR storage.
In future BDNs, various data will be stored inside the networks with diverse security requirements. As an immutabletransaction ledger, blockchain can provide secure and distributed storage while enforcing data integrity via proof-of-retrievability schemes. Data can be stored not only in the blockchain, but also off-chain, depending on the importanceof the data and practical requirements.
Storj is an open-source implementation for blockchain-based cloud storage to provide data security and integrity indistributed applications [90]. The data is broken up and stored in peer nodes across the network, while blockchainstores metadata information about where to find the data pieces.
Storj returns control of cloud data to users, andimproves security and privacy by running untrusted and fault-tolerant systems on trusted data providers. When a userneeds to access the data, the blockchain is queried and then returns the required metadata to retrieve the original data.However, it is not feasible to use the Bitcoin blockchain to store metadata currently, so the system uses a technologynamed Florincoin instead. A system named
BlockDS is designed in [91] to provide secure data storage and keywordsearch, which consists of three parts: distributed data storage, anonymous access control, and private keyword search.Similar to
Storj , data references, rather than encrypted data themselves, are stored in the permissioned blockchains, inwhich different clients have different data access authorities. The keyword searching component is modeled as a smartcontract in the blockchain, and data customers who have been authorized by the anonymous access control layer areallowed to search in the cloud storage without downloading the whole dataset. Therefore, the system can outsourcethe data storage to service providers to achieve a smooth distributed network.Considering that users often have a large amount of under-utilized storage resources in their devices, a secure decen-tralized storage framework named
BlockStore is proposed in [92]. Users can rent out their available storage resources,and the system collects and distributes them to the tenants. The difference between
BlockStore and traditional solutionsis that it provides powerful auditing capabilities through smart contracts to ensure data security and prohibit doublerentals. This solution is just a basic framework for blockchain storage and can be expanded in multiple directions,such as adding incentive mechanism or designing a more efficient wallet data structure.In order to improve the security of cloud storage and reduce transmission delay, a blockchain-based security architec-ture is presented in [93] for distributed peer-to-peer cloud storage. The uploaded files are divided into encrypted datapackages and then transmitted to the nodes in the peer-to-peer network randomly. A genetic algorithm is integratedinto the architecture to solve the problem of data packages placement between multiple users and data centers in adistributed environment. As a trading mechanism between storage service consumers and providers, blockchain storesfile locations such as file hash and file URL, and uses Merkle trees to ensure data integrity. Through performanceand security analysis in a multi-user network with multiple data centers and blockchain, this architecture has shown toachieve a lower file loss rate and transmission delay.
Mchain proposed in [94] has a two-layer structure to improve the security and latency of the virtual machine (VM) fordata storage in the infrastructure-as-a-service (IaaS) cloud architecture. The authors use the same trust assumptions asthe original blockchain network, that VM measurements data in the network may be attacked before or after storage.The first layer of the architecture constructs the semi-finished block after verifying the generated data packets, andthe second layer performs PoW tasks on the semi-finished block to generate tamper-resistant metadata to ensure dataintegrity. The system also includes an access control that can be updated to ensure that sensitive information cannotbe accessed by the public.
Mchain separates the consensus process from the original blockchain and executes time-consuming PoW tasks in the background. From users’ view, not only can it guarantee security, but also reduce thepacket waiting time. This scheme can be improved by incorporating a consensus protocol that is more efficient thanPoW.Han et al. apply blockchain to medical data storage to build a decentralized, secure, tamper-resistant health informa-tion storage system in [95]. In order to improve the latency of data validation, the system utilizes a hybrid healthblockchain, which combines the consortium and private blockchains. The encrypted medical records are stored in theprivate blockchain, which is used as a database in medical institutions. The consortium blockchain enables medical15 rXiv
Template
A P
REPRINT information sharing between nodes by storing medical data submitted from all participating medical institutions. Thenode that wants to share its medical record can add its block to the consortium blockchain from the current privateblockchain directly, thus achieving secure medical information storage, privacy protection, and medical data sharing.However, the cost of deploying the proposed model is not considered in this article.A blockchain-based data storage method is proposed to implement the secure domain name system in [96]. Thismethod can create multiple domain name service nodes in parallel, and the hash value of the file data is stored in theblockchain. To securely store the log files, a platform based on blockchain in the cloud is presented in [97] to achievethe integrity of the log files and logging process as well as the proof of non-repudiation.
The issues with trust in traditional data-driven networks can be efficiently addressed by blockchain technology inBDNs, which can build a trustable platform to share data and execute computations among different stakeholdersand organizations. In BDNs, users can feel safe to share transaction data with stranger nodes by leveraging theadvantages of blockchain, which provides distributed and immutable ledger with efficient consensus protocol to ensurethe trustfulness.Huang et al. in [98] propose a revocable chameleon hash (RCH) based on complexity assumptions and bilinear pairingfor deriving a self-redactable blockchain (SRB) to enable an intelligent trust-layer for IoT. Specifically, a collisioncould be found by RCH via ephemeral trapdoor and the SRB could enable the block content in the blockchain to bere-written and the redacted block hash to remain unchanged without suffering hard forks.
BlockTDM is a blockchain-based trusted data management scheme for mobile edge computing [99]. The core of
BlockTDM is a configurable blockchain architecture, which consists of the edge device layer, blockchain networklayer, edge nodes layer, and the cloud center layer. The data gathered from the edge devices as well as their hashvalues are transmitted to the blockchain network layer for storage.
BlockTDM supports matrix-based multi-channeldata segment and sensitive data isolation, by defining a channel matrix to perform data access, transfer and usagesafely and effectively in an untrusted environment. Moreover, users are capable of defining the data sensitivity, and
BlockTDM can encrypt the transaction body, ( i.e. , data payload) before saving the transaction in the blockchain system.Using smart contract, conditional access is also implemented with decryption algorithms for users who want to accessthe protected blockchain and transaction data.
BlockTDM provides a versatile blockchain-based paradigm for trustedand tamper-proof data sharing and process.Zhang et al. propose a blockchain-based trust scheme and elaborate a quality assurance application of blockchainin [100] to enhance data security and partners’ collaboration in smart manufacturing. Each block records differentkinds of digital data and information via asymmetric encryption, including task information, service information,identity data, transaction data, asset data, contract data, et al.
The block is verified and stored by all network nodesafter broadcast, and a consensus would be achieved on the latest blockchain. This scheme can build the trust ofparticipants in the system and the trust between participants. Its main challenges lie in the reliability of data source,the operating cost of the blockchain, and the deployment of smart contracts.A blockchain-based decentralized trust management system for vehicular networks is designed in [101]. The authorsconsider two types of adversary models including malicious users and damaged RSUs, and use blockchain to achievedecentralized trust management in the network. Vehicles can validate the messages leveraging Bayesian inferencemodel, and then generate and upload a rating result for each received message from other vehicles. The RSUs cal-culate the trust value offsets of these rating results, and then upload the block containing these data to blockchainusing a consensus mechanism combining PoW and PoS. The more total value of offsets (stake) is in the block, theeasier an RSU can find the nonce for the hash function. In this way, all RSUs collaboratively maintain a reliable andtrustful blockchain for the vehicular network. However, how to jointly realize privacy protection and effective trustmanagement is still a problem to be solved in such vehicular networks. Another privacy-preserving anonymous repu-tation system (BARS) is proposed in [102] to realize trustable VANETs. Under the trust model of privacy protectionmanaged by semi-trusted authorities, the blockchain enables transparency for the certification and revocation with theproofs of presence and absence. All broadcast messages are recorded in “Blockchain for Messages” as permanentevidence to evaluate vehicle reputation, thus achieving an effective trust model in VANET. Furthermore, a reputationevaluation algorithm is presented with both direct historical interactions and indirect opinions about vehicles to guar-antee the reliability of messages. Yang et al. propose a blockchain-based traffic event validation and trust verificationmechanism [103]. The authors assume that there is a Public Key Infrastructure (PKI)-based model in VANET andonly consider events that can be verified by the vehicles or RSUs. A proof-of-event (PoE) consensus algorithm is uti-lized by RSUs to detect passing vehicles within certain adjacent area in an incident when the collected data meets thecorresponding threshold. Meanwhile, the related information of incidents are recorded in the blockchain permanently.16 rXiv
Template
A P
REPRINT
The transactions on blockchain include two consecutive stages, namely, synchronizing the local blockchain and syn-chronizing the global blockchain. In this way, the warning messages could be broadcast in an appropriate region andtime periods. The proposed mechanism can effectively record the correctness of traffic events, and provide traceableevents with trust verification.
In this section, we discuss DDN management and control supported by blockchain from the perspectives of accesscontrol, routing mechanisms, virtual resource management, resilient network and fault-tolerant mechanisms, as sum-marized in Table 3. Table 3: Management and Control of BDNs.Focus Ref. ContributionsAccess Control [104] Use blockchain to manage access authority of largedatasets.[105, 106] Propose an access control framework with integratedmessage encryption.[107, 108] Present an attribute-based access control scheme for IoT.[109, 110] Achieve access control in a cross-layer and cross-domainnetwork.[111, 112,113] Improve security of access control through blockchain-assisted intrusion detection.Routing Mechanisms [114] Propose a Blockchain-based secure BGP routing system.[115, 116,117] Use blockchain to enhance the security and efficiency ofrouting in IoT and WSN.[118, 119,120] Design routing scheme to support cross-chain communi-cation and routing.[121] Use blockchain to combine incentive mechanism withsmart routing.Virtual ResourceManagement [122, 123,124] Propose virtual resource placement scheme supported byblockchain at the edge of the network.[125, 126] Design resource management and sharing mechanism inIoT based on blockchain.Resilient andFault-tolerantMechanisms [127, 128] Study on the network fault-tolerance mechanism sup-ported by blockchain.[129] Present a blockchain-based botnet detection architecturefor IoT.[130, 131] Utilize blockchain to build distributed resilient networkarchitecture.
There have been many proposed approaches of blockchain-based access control designed for supporting differentkinds of data-driven services. In BDNs, there are many factors that should be considered in access control, includingprivacy protection, authentication, and resource allocation, with the target of sending and deriving data in a secure,efficient, and decentralized manner.The blockchain-based access control system presented in [104] aims to effectively manage access authority of largedatasets and protect against data breaches. The Hyperledger fabric blockchain [132] is designed to deploy blockchainidentity-based access control (BIBAC) and blockchain role-based access control (BRBAC). BIBAC supports accesson a user-by-user basis, while BRBAC assigns users’ roles to access specific assets on the blockchain. Hyperledgermodeling tool is adopted for the smart contract and transaction processing functions. Though Hyperledger modelingraises questions about stability, the analysis indicates that ecosystem ensures data transparency and traceability forsecure data sharing, auditabilty and data self-sovereignty for the owner.Wang et al. in [105] propose a secure cloud storage framework with access control by Ethereum blockchain andciphertext-policy attribute-based encryption (CP-ABE). The blockchain is responsible for storing the publicly availableinformation, achieving supervision function, and tracking the behavior of the data access. The cipher-text data is17 rXiv
Template
A P
REPRINT stored by the data owner, and could be decrypted and used within the valid data-processing period via leveragingsmart contracts. When the attributes of data user satisfy the corresponding requirement and threshold, the data usercould decrypt correctly and access the data before the valid data-processing period expires. By using Ethereum smartcontract, this framework can be combined with most CP-ABE algorithm to achieve decentralized fine-grained accesscontrol.An access control architecture based on private blockchain is proposed in [106] for information-centric networks(ICN), where the content is divided into n original blocks, and encoded into m ( m > n ) blocks by xor-codingalgorithm. The xor-based encoding/decoding scheme could encrypt and decrypt messages efficiently. In order to meetthe requirement that only authorized users can read and decrypt the protected content cached in ICN nodes, the authorsmake some security assumptions for the ICN environment. Specifically, the content provider (CP) constructs a blockcontaining the decryption information in a given sequence, and signs this block with its private key. Then the CPpublishes the blockchain and adds new blocks for updating. The blocks are verified by users with the CP’s publickey via downloading the whole blockchain from ICN. The users find their own information and decrypt it using theirprivate key. Unlike other strategies, there is no need to request decryption information from the CP, which makes itunnecessary for the CP to be always online and thus improves the robustness of the ICN. Further security analysis isneeded to prove the security of the framework in the presences of malicious nodes and attacks.An attribute-based access control (ABAC) scheme for IoT is presented in [107]. It abstracts the roles or the identitiesinto a set of attributes issued by the attribute authorities, which manage each of the attributes and distribute themto proper users. The authors use consortium blockchain to record the attributes in a distributed manner to addressthe single point of failure and unauthorized data tampering problems. The attribute authorities act as the servingnodes in a consortium blockchain and as the key generation center (KGC) when new IoT devices register with thesystem. The devices could collect, transmit, process, and share the data in the system, and associate with the accesspolicies according to their functional and secure requirements. Note that the devices must prove their ownership ofthe corresponding attributes that satisfy the policy. Thus, the devices are not responsible for transaction verificationbut have only read permission with the blockchain according to their attributes. By reducing the overall computingand communication overheads and enhancing the system flexibility, this design is suitable for IoT scenarios. Anotherblockchain-based ABAC approach is designed in [108] to keep track of users’ attributes, and manage the trust levelwith the attribute issuing entities (AIE) by administrators. If the calculated trust level is higher than the correspondingthreshold set by the devices, the attribute could be considered valid. When all the required attributes are valid, theaccess request is granted. Otherwise, the access request is denied. This decentralized access control system solves theproblem of lacking trust, and further research work need to consider the issues of information leakage and latency.In [109], the authors propose a privacy-oriented blockchain-based distributed hierarchical access control framework,which consists of the cloud layer, fog layer, and device layer. The edge network provides the access function forthe devices. In the fog layer, the security access manager is responsible for recording and verifying transactions thatinclude key management information. The cloud layer with multi-blockchains helps to achieve interconnection andtraceability among blockchains. The cloud layer also stores the encrypted data generated by all kinds of mobile devices,and the data can be accessed directly using the encryption key. The network is divided into different side-blockchains tospeed up the verification and save the storage space. This framework achieves decentralized, fine-grained auditabilityand scalability requirements. However, blockchain technology is not fully utilized.Paillisse et al. implement and evaluate a three-layer architecture in [110], which supports distributed access controlbased on permissioned blockchain in cross-domain communications. Private blockchains and BFT protocols are ap-plied to this structure, so that moderate storage space can be used to store thousands of access strategies. Authorizeduser could access network resources, and routers could determine the authority of the users. The Locator/ID Separa-tion Protocol (LISP) is adopted to support the communications of control and data planes. The control plane storesspecific access policies and updates them via the Hyperledger, which brings secure and tamper-proof features.Besides, blockchain-assisted intrusion detection plays an important role in identifying possible threats and achievingsecure access control in DDNs [111]. To enhance the detection capabilities of the intrusion detection system (IDS),Ujjan et al. in [112] combines blockchain with the collaborative intrusion detection network (CIDN) in SDN toachieve trust-based communication among IDS nodes. Not only can each node monitor and identify malicious trafficin the network, but also share the Snort signature rule set with neighboring nodes. This CIDN model provides pro-tective measures against internal attacks, enabling accurate detection of some typical common attacks such as DDoS.Unsupervised deep learning methods can further improve the performance of such frameworks. A deep blockchainframework (DBF) is proposed in [113], which uses a Bidirectional Long Short-Term Memory (BiLSTM) algorithm todiscover network attacks from network data migration in cloud systems. This intrusion detection algorithm was eval-uated on the UNSW-NB15 and BoT-IoT data sets. The results show that the proposed framework can achieve simple,secure, and transparent intrusion detection between clouds, supporting users and cloud providers for secure access and18 rXiv Template
A P
REPRINT data migration. Note that although blockchain-based IDS can achieve security and privacy in access control, in termsof system latency and scalability, blockchain-based IDSs should be further tested and improved in the DDNs.
BDNs can provide services to collect, process, and store data and information through networks with single or multiplehop relays. By taking advantage of blockchain as a trusted, decentralized, self-organizing ledger system, effectiverouting mechanisms have been proposed by many researchers.A blockchain-based secure Border Gateway Protocol (BGP) routing system named
RouteChain is proposed in [114],to prevent BGP hijacking and maintain a consistent view of the Internet routing paths. Autonomous systems (ASs) aregrouped according to the geographical proximity, and then a bi-hierachical blockchain-based model is built to detectmisbehavior over the Internet. Each transaction contains BGP announcements that are exchanged among peers, anda PoA-based consensus protocol called
Clique is used to achieve consensus among these ASs. Since the propagationand verification of transactions between ASs takes some time,
RouteChain cannot prevent all ASs from being attackedcompletely. In addition, grouping ASs according to the ideal situation may not work well in reality. Despite the abovedefects, this architecture can still reduce BGP attacks in the network effectively.A blockchain-based distributed reputation management system is presented in [115] to protect secure routing inIoT networks. Considering the behavior of malicious routers refusing to provide routing services by discarding datapackets, the system rates the reputation of each router and stores it in the blockchain in a decentralized and immutablemanner. The reputation results are used to evaluate the trustworthiness of each router to protect the routing mechanismfrom misbehavior. To reduce the complexity of blockchain in IoT networks, a group mining scheme rather thanPoW-based consensus protocol is designed to achieve consensus among IoT nodes. The proposed scheme can realizereputation management of the IoT routing process against selfish nodes, and has stable convergence under differentnetwork scales. Further analysis needs to be done to evaluate the security performance of this routing system. Targetingan untrusted IoT environment, Ramezan et al. propose a blockchain-based contractual routing (BCR) protocol in [116].Each source IoT device implements a smart contract on the blockchain to update topology. The routing algorithmcoded in the smart contract finds the proper route from a source IoT device to a destination device or gateway, and theeffectiveness of the route could be guaranteed by any intermediary device. In order to enhance routing security andefficiency for wireless sensor networks (WSNs), reinforcement learning is adopted in [117] to propose a blockchain-based trusted routing scheme. The authors assume that the blockchain network is trusted and the nodes in WSNcan be static or dynamic, which use the proposed blockchain platform to provide dynamic and trusted routing. Eachtransaction contains routing information. The routing packets, confirmed by the verification nodes, are encapsulatedinto blocks and stored in the blockchain, which guarantees that these routing packets are traceable and tamper-proof.Then, a reinforcement learning model is utilized to adaptively select the best route to destination nodes. The system caneffectively suppress attacks by malicious nodes while guaranteeing lower latency and better throughput performance.Apart from these blockchain-based routing solutions, some researchers consider adding routing function to makeselected blockchain nodes transmit requests between different blockchain networks.
Interchain is proposed with ahandshaking method to complete asset transfer for cross communications between blockchains [118]. However, itdoes not consider any consensus algorithm in the framework. Chen et al. in [119] introduce a private token-basedinter-blockchain communication scheme to provide crossover communications between different blockchains withoutany intermediary. The authors utilize a routing algorithm and PBFT as the consensus algorithm. The main limitationof this work is that the proposed method negatively affects the system throughput.
Anlink blockchain [120] is an enterprise blockchain architecture that uses an inter blockchain communication protocolto connect multiple blockchains and enable cross-chain communications. The proposed architecture is composed of
Ann-Router , AnnChain , as well as other blockchain-based systems. These components are divided into four partici-pants, namely, the validator, surveillant, nominator, and connector. Ann-Router is a blockchain router to dynamicallymaintain all the related information registered on sub-chains and link sub-chains, and enable multiple blockchainsto communicate with each other in the network. Delegated stake-PBFT is used as the consensus protocol. In addi-tion to meeting the communication needs between blockchains, enterprise blockchain systems also need to meet theregulatory requirements, provide privacy protection, and support heavy transaction loads.Ersoy et al. in [121] combine an incentive mechanism with smart routing to reduce the communication and storage cost.Each participating node in the propagation of a transaction could receive a share of the transaction fee. They analyzethe sufficient and necessary conditions that encourage the spread of messages as well as to discourage the networkfrom introducing Sybil nodes, and design a routing mechanism considering a first-leader-then-block type consensusprotocols where the round leader who creates the block is known in advance. The proposed routing mechanism reducesthe propagation of redundant communication while encouraging nodes to propagate messages.19 rXiv
Template
A P
REPRINT
Wireless network virtualization is considered as a promising technology to enable sharing of physical infrastructureand networking slices for enhancing network capacity, coverage, and wireless security. Some pioneer work have beencarried out in BDNs.An architecture leveraging SDN, edge computing, and blockchain is proposed in [122] to enable wireless networkoperators to utilize their resources efficiently and securely. Implemented as an overlay architecture on existing net-works, the key components include SDN controller, network aggregator, primary wireless resource owners (PWROs),virtual wireless network operators (VWNOs), edge computing component, and blockchain platform with its manager.Specifically, the sublease and release of networking slices between PWROs and VWNOs could apply blockchain tooffer traceability and auditability so as to prevent double-spending of the same networking slice in the same periodand location for wireless virtualization. SDN controllers, which act as blockchain managers, keep their own privatekeys, and use public keys while subleasing (or releasing) networking slices to VWNOs (or from VWNOs to PWROs).Therefore, the sharing framework in [122] allows users to move/switch between virtual networks while maintaininga secure connection. The blockchain guarantees that it is impossible to generate malicious transactions that subleasewireless resources with authorization; however, open issues including large storage space and long consensus delaystill need to be addressed.Samaniego et al. in [123] introduce a virtualization of IoT components (virtual resources) at the edge of IoT networks.Permission-based blockchain protocols are introduced including blockchain as a service (BaaS) in the cloud and aprivate multi-chain in a fog network, to enable provisioning of virtual resources directly on edge devices. Only regis-tered users can access the blocks in the chain to write and read configurations, thereby managing the virtual resourceprovisioning and multi-tenant access in a secure manner. Considering the latency and bandwidth consumption, the foglayer implements a multi-chain for hosting virtual resources and only transfers useful information to the cloud. Theevaluation of data is also performed in the fog layer to provide a time-effective decision-making process.The authors in [125] present the concept of using blockchain as the ledger for network slice leasing, and analyzesome use cases for future industrial IoT networks. A 5G Network Slice Broker is introduced into blockchain in thiswork, enabling the factories to obtain the required slices automatically and dynamically to achieve efficient operationon-demand. Blockchain is applied to manage virtual network slice trading for secure network operation, and enablenew functionalities, such as spectrum management, data processing, and network infrastructure as a service.A blockchain-based workflow management system (BCWMS) is proposed in [126] to share heterogeneous logisticsresources among different customers to satisfy different operation logic. To simplify the process of systems and re-source association for IoT devices, the virtual resource gateway (VRG) is introduced to enable resources managementwith different granularity via specific gateways and virtual links. Also, a resource blockchain is designed to guaranteedata reliability and the accuracy of front-line resource usage data for enabling customer decisions.
EdgeChain is designed in [124] to build a decentralized platform for mobile edge application without any trustedthird party. The authors use stochastic programming for multiple service providers to make mobile edge applicationplacement decisions. The cost is modeled by jointly considering the edge hosts, latency, and service chaining. Theblockchain is used to store all placement transactions, including global resource availability, allocation, and consump-tion information, which are traceable by every mobile edge service provider and application vendor who consumesvirtual resources. However, this work does not take into consideration about the users’ behavior, which may have animpact on system cost and resource deployment.
In BDNs, fault or malicious activities should be readily identified and recovered to enable a decentralized, resilientand fault-tolerant system.
BeeKeeper is proposed in [127] as a fault-tolerant blockchain-based IoT service system, which implements a thresholdsecure multi-party computing (TSMPC) protocol. Servers perform homomorphic computations on shares and generateresponses to help users processing encrypted data. Since shares and responses are verifiable by leveraging blockchain,malicious nodes can be easily detected. However, the main limitation of
BeeKeeper is that the number of active andhonest servers should be more than a threshold to keep the protocol work effectively.The authors in [128] propose a fault-tolerant incentivisation mechanism based on payment strategies conditional ontask execution results for distributed mobile peer-to-peer crowd services (MPCS) systems. Moreover, smart contract
MPCSToken is designed to facilitate service auction, task execution and payment settlement process.
MPCSToken contract is implemented on Ethereum blockchain to build a micropayment mechanism based on blockchain payment20 rXiv
Template
A P
REPRINT channels to avoid bulky payments and mitigate payment settlement risks. The system can improve the utility ofparticipants effectively, and can also run cost-effectively on the Ethereum blockchain.A blockchain-based botnet detection architecture for IoT, called
AutoBotCatcher , is presented by Sagirlar et al. in [129]to analyze communities of IoT devices and detect botnets. A permissioned BFT blockchain is utilized to store thetraffic flows of the IoT devices. A set of pre-identified parties collect and audit these traffic flows as blockchaintransactions to perform botnet detection collaboratively without any trusted third party. Specifically, there are two mainactors, namely agents and block generators. Agents could monitor IoT network traffic flows and send collected trafficinformation as blockchain transactions. Block generators ( i.e. , the powerful trusted full node in the IoT domain) aimat modeling mutual contact information of IoT devices and generating mutual-contact graph, which is then exploitedto detect the logical communities. Based on these,
AutoBotCatcher can perform dynamic and collaborative botnetdetection on large networks, but its robustness and the ability to deal with internal threats still require further evaluation.Mylrea et al. in [130] investigate the application of blockchain and smart contract, which provides anatomicallyverifiable cryptographic signed distributed ledger to build decentralized, resilient, and smart energy grids as wellas energy trading platforms. The authors focus on the technical characteristics of blockchain (security, scalabilityand speed), thus apply the blockchain to provide an innovative trading platform where producers and consumerscan exchange their residual energy or demand flexibly. Blockchain is utilized to verify time, user, and transactiondata, and protect these data with an immutable cryptographic signed ledger to improve the trustworthiness, integrity,and resilience of energy delivery systems. Moreover, energy customers can also verify data from other entities byleveraging blockchain to creat a distributed trust mechanism.A trusted and resilient blockchain-based architecture for IoT is designed in [131] to support data integrity auditing andenable better resilience to improve the system scalability. It considers a system in which drones are deployed to providebetter connectivity among IoT devices. Specifically, the hashed data records collected from drones are stored in theblockchain network, and each blockchain receipt for each data record is generated and stored in the cloud consideringthe limitations of battery and processing capability of drones while guaranteeing data security. To improve robustness,a full copy of the entire distributed ledger is stored on every distributed node in the blockchain network. The systemcan improve the network reliability and security, as well as collect data in real time to provide data guarantee for dronecontrol.
In this section, we discuss the challenges and future directions for BDNs in terms of security and privacy, scalability,network management, and resource management. A summary of the references for these issues can be found in theonline supplemental file. Furthermore, standardization activities related to BDNs have been gaining momentum, someof which are also presented in this section.
Although blockchain technology could provide a promising solution for enhancing security and privacy in a varietyof applications, BDNs, which integrates blockchain with DDNs, may encounter new security challenges, such asselfish mining attacks [133], balance attacks [134], BGP hijacking attacks [135], and eclipse attacks [136]. In [90],blockchain is used as a common record for all user encryption metadata. An attacker can pretend to be a legitimatenode to read encrypted metadata, and then attempt to decode it through an offline brute-force attack. Therefore,appropriate and effective security measures must be developed for BDNs. One of the main advantages of blockchainis the provision of pseudo-user anonymity, which is critical to safety. For most of the methods discussed above, thetransaction data records can reveal the user’s identity by looking into the public key, the ACL, or the provenancedata. Therefore, by analyzing the data stored in a blockchain, the users’ behavioral activities can be tracked, and theirtrue identities can be exposed. To solve this problem, further research is needed to provide a completely anonymousapproach that meets the security requirements. Another issue is the possibility of quantum attacks. The encryption ofblockchain will no longer be secure when practical quantum computers are realized [137]. Therefore, new “quantum-safe” encryption mechanisms will need to be developed for future BDNs.Machine learning can be adopted to support BDNs, because ubiquitous blockchain nodes with machine learningalgorithm can react to network status and users’ requirements in a decentralized and timely manner, which not onlyimproves robustness but also reduces latency. The application of ML in BDNs can detect malicious behavior onthe blockchain by deploying the trained models and algorithms. In this way, ML could assist blockchain systems inidentifying and preventing theft, fraud, and illicit transactions on the chain. Furthermore, a blockchain can store large-21 rXiv
Template
A P
REPRINT scale data in a secure and tamper-resistant manner, ensuring data confidentiality and auditability of the collaborativetraining process and the trained ML model via cryptographic techniques.The world is experiencing a data explosion in both amount and diversity. Networking data is continuously generatedby users and network entities. The approaches to store, organize, and process these data are being actively explored.The combination of blockchain and big data in the network can reduce operating costs and data storage risks inBDNs. Towards this direction, Yue et al. in [138] propose a blockchain-based big data sharing model to improve datacredibility and ensure secure data circulation. Also, to enhance the security of big data platforms, an access controlframework is presented in [139] by relying on blockchain to ensure proper implementation of optimal access in adecentralized environment without central authority and administrators. At the same time, the historical data of thenetwork can quickly identify and prevent malicious authentication and authorization behaviors through data miningand real-time analyzing, thereby ensuring the cybersecurity for future BDNs. However, it may lead to huge dataredundancy. Therefore, how to perform data off-chain operation on the premise of ensuring privacy and integrity is aproblem that should be further investigated.
When blockchain technology is applied to future DDNs, scalability is another important problem that needs to beaddressed. For example, the Bitcoin blockchain requires more than 100 GB of storage and can handle only 7 transac-tions per second on average. If we increase the volume of each block ( i.e. , more transactions are packed together), itsthroughput could increase, but generating and propagating blocks would take a longer time. On the other hand, reduc-ing the block interval time can increase throughput, but this can also lead to stale blocks that have no contribution tothe main chain and reduce the security. Thus, designing blockchain for BDNs requires a trade-off between scalabilityand security [140].To enhance scalability of a BDN, the designed blockchain must also ensure its security. In order to solve the problemof storage, Bruce et al. in [141] proposed a novel blockchain structure, which uses a database called account tree anddeletes the old transaction records instead of holding all addresses. Then, the nodes only need the latest part of theblockchain to synchronize with the network, saving storage space in the blockchain. The low storage nodes proposedin [142] store linearly combined fragments of each block by using erasure codes. These nodes can recover data bydownloading fragments from other nodes and performing inverse linear operations. This technology ensures that anynode in the blockchain can be easily reconstructed from such nodes with the data integrity guaranteed. However,compatibility of this technique with other networks still needs more investigation.In terms of redesigning blockchain, the work in [143] proposes a theoretical lightweight blockchain structure, whichcan handle transactions without incurring additional delays. However, without a comprehensive analysis of consensusand security, it is unclear whether it can withstand attacks from corrupted nodes in the network. Poon et al. in [144]propose a solution for scalable computing on blockchains named Plasma, which is scalable to a significant amount ofstate updates per second. Plasma divides all blockchain calculations into a set of MapReduce functions. Each sidechainis implemented with a smart contract, and the root chain only needs to handle a small amount of commitment from thesidechain. The scalability of the root chain can thus be enhanced. However, Plasma can only guarantee security andscalability at the same time when the withdrawal delays is fixed; otherwise its security level may reduce.To address the scalability issues, ML techniques can benefit the blockchain in making better decisions, such as offeringmore efficient data sharding or pruning solutions. Leveraging ML techniques, blockchain applications can supportpredictive analytics to ensure the requirements for resources to be accurately met and to improve the efficiency ofblockchain operation. Specifically, leveraging the training data, ML-based mining algorithms may manage tasksin a more intelligent manner rather than adopting the brute-force approach. Since ML algorithms can predict andspeedily calculate data, it would also provide a feasible way for miners to select more important transactions toperform. ML techniques can also enable more efficient off-chain solutions or adjust the block size dynamically tomake the blockchain-based system more scalable.
Intelligent network management enables the introduction of autonomous mechanisms into the network to performtasks such as planning, configuration, management, optimization, and self-repair. Applying blockchain to the DDNand correlating functions by using smart contracts that can be executed automatically is a good way to implementautonomous functions in intelligent management. A new intelligent network architecture is proposed in [145], whichuses smart contracts to handle the relationship between operators and users, thereby simplifying spectrum and infras-tructure sharing. At the same time, the consensus protocol can also make the network highly resilient and resistantto DoS flooding. ADEPT [146] enables the devices connected to it to communicate in a secure and efficient man-22 rXiv
Template
A P
REPRINT ner in the IoT network, while supporting complex business logic. It also integrates multiple protocols to achieve anautonomously coordinated, secure, and decentralized network.However, because of distributed data storage used in blockchain, each node maintains a complete copy of the data.Once a new transaction is added to the block, all copies of the data should be updated synchronously. This frequentdata updates may increase the difficulty of blockchain-based network management. Neudecker et al. in [66] identifyfive requirements for the blockchain network layer: performance, low cost of participation, anonymity, DoS resistance,and topology information hiding. The network layer needs to be weighed and optimized in these five areas. However,most of these requirements are not quantitatively described. Therefore, the design of the blockchain network shouldbe further explored, and the data in the network should be used to improve the performance of blockchain.Smart contracts in the blockchain can enable consensus-based coordination and verification in the network in a mannerbeneficial to network management, but there are some issues that require further investigations. One of the problemsis that in the basic construction of decentralized autonomous management, it is difficult to fix bugs once the systemis in operation [147]. Another problem is the participant management problem in the decentralized autonomous orga-nization. This involves how to manage participants of the consensus mechanism, and how to develop incentives andaccountability mechanisms. Important details and new capabilities need to be considered to promote an autonomous,secure, transparent and more efficient network management.Applying edge computing to BDNs can achieve reliable access and control for edge-distributed networks, thereby pro-viding large-scale network control and management services. The storage resources on edge computing devices canbe used for content caching, with BDNs maintaining a credit system to support the business model. One example ofthis approach is the modular consortium architecture based on blockchain and IPFS software stack presented in [148],which contains private side-chains and a consortium blockchain. The private side-chains maintain logs of data oper-ations that occur within a private IoT network and the consortium blockchain connects edge servers, thus achievingsecurity and privacy of IoT data. In addition, the application of big data can improve the automatic optimization of thefuture networks, enabling BDNs to realize self-adjustment and intelligent networking.
Application of blockchain in DDN could make the resource management problem more critical. A decentralized cloudresource management framework based on blockchain techniques is presented in [149] to save energy significantly.This framework schedules requests without relying on a centralized scheduler in the cloud. The decentralization of theblockchain ensures that the failure of a data center does not affect the system’s resource management, which gives theframework excellent robustness.Another issue is the resources consumed by a blockchain. In order to reach decentralized consensus among nodes withpoor synchronization, some existing consensus protocols in the blockchain networks ( e.g. , PoW protocols) will con-sume a huge amount of computing and energy resources. To reduce the resource consumption of the blockchain, someother alternative PoX consensus protocols have been proposed in permissionless blockchain systems [150, 151, 152].These PoX solutions address two main goals, namely, incentivizing the provision of useful resources and improvingperformance [67]. However, the resources required by these consensus protocols have not been extensively analyzedand quantified.Edge computing provides a low-latency and distributed computational task offloading platform for end nodes withlimited resources. Edge computing decomposes the large-scale services conventionally handled by the data centers(i.e., in the cloud), dividing them into smaller parts and distributing them to the edge nodes for processing, therebyspeeding up data processing and reducing latency. This aligns very well with the decentralized machine learning solu-tion that we mentioned above for future BDNs. An example is the blockchain-based computation offloading schemein [153] that protects data integrity in task offloading while optimizing resource allocation using a non-dominatedsorting genetic algorithm. A deep learning method to optimize an auction-based resource allocation scheme for edgeservers is presented in [154]. However, it only considers an individual edge server. A feasible approach for interac-tions between multiple edge servers to optimize resource utilization while satisfying latency and other performanceconstraints should be developed to support BDNs.
In order to speed up the development, implementation and adoption of BDNs, standardization efforts are quite crucialto ascertain interoperability among network entities, user terminals, and various applications. While DDNs have at-tracted wide attention from researchers, their standardization work is still in the early stage. ITU-T has recently issued23 rXiv
Template
A P
REPRINT the Y.3650 series to discuss the use-cases and application scenarios for big-data driven networking [3]. The marketand related industry are quite fragmented, and a widely acknowledged and commonly accepted framework coveringdifferent technologies and platforms is still lacking. Meanwhile, standards for blockchains and DLT have started toemerge since late 2017, and several organizations have intensified their efforts to develop various standards [155].The IEEE P2418 series, which focuses on generic architectures, interoperability, and vertical-industry standards forblockchain/DLT is under development. ISO/TC307 is an international technical committee that has been very activein setting global standards in wide areas including use cases, smart contracts, security and privacy, governance, andinteroperability between blockchain applications. ITU-T Focus Group on Application of Distributed Ledger Technol-ogy (FG DLT) has been created to analyze the standardization demands of applications and services based on DLT.The Enterprise Ethereum Alliance (EEA) has more than 500 members worldwide and is considered as one of themost active industry alliances. It focuses on development of technical specifications and certification of Ethereum forenterprises. The W3C blockchain community group has been working on new technologies and use cases related toblockchain.Although there are many global standardization initiatives ongoing for DDNs and blockchain technologies, standard-ization efforts for BDNs need to be launched to support the vigorous development of an industry ecosystem. Industryhas an urgent need to standardize some aspects of BDNs. The application programming interface (API) formats shouldbe unified among different manufacturers and service providers, so that data-driven services may bloom and flourish.The APIs between data engine and block engine need to be standardized to support the information exchange amongvarious modules. Meanwhile, in order to support efficient data transfer across network operators, the processingmethod of the data ( i.e. the label for sensitive data, authentication and authorization policies) calls for correspondingstandards as well. The joint efforts from interest groups, industry-driven forums, standards-development organizations,together with academic experts, will be needed to push the standardization and widespread adoption of BDNs.
The current generation computing networks are rapidly evolving towards future DDNs connecting a massive numberof heterogeneous users and devices, generating and consuming a huge amount of data that require processing atstrategic locations within the networks, all for the purpose of enabling ubiquitous intelligent services. In this article,we have reviewed the fundamentals of the cutting-edge blockchain technologies and how they are proposed to beapplied in computer networks. We have further analyzed the challenges of future DDNs and discussed how blockchaincan help address many of these challenges. We have systematically surveyed and categorized the pioneer researchworks of using blockchain in networks to motivate blockchain as a solution to empower future DDNs. We havehighlighted the specific advantages of BDNs, made possible by the decentralization of blockchain that is reshaping theconventional paradigms of running complex systems. We have presented a framework for BDN, and discussed howdecentralization can be beneficially applied to different layers ( i.e. , application layer, network layer, and link layer) invarious dimensions ( i.e. , data plan and control plan) for multiple purposes ( i.e. , security control, privacy protection,robustness enhancement) with different kinds of resources ( i.e. , computational resource, bandwidth/communicationresource, and storage resource). We have also identified a number of research challenges, and presented the broaderperspectives on how BDN might evolve in the future to incorporate AI, big data and edge computing. We hope thisarticle will provide impetus for a substantial increase in research towards the design and implementation of BDNs, aswell as the enhancement of existing services and development of new services based on BDNs.
References [1] VNI Cisco. Cisco visual networking index: Forecast and trends, 2017–2022.
White Paper , 1, 2019.[2] Chao Fang, Song Guo, Zhuwei Wang, Huawei Huang, Haipeng Yao, and Yunjie Liu. Data-driven intelligentfuture network: Architecture, use cases, and challenges.
IEEE Communications Magazine , 57(7):34–40, 2019.[3] ITU-T Y.3650. Framework of big-data-driven networking, 2018.[4] Weichao Gao, William G Hatcher, and Wei Yu. A survey of blockchain: Techniques, applications, and chal-lenges. , pages 1–11, Jul. 2018.[5] Satoshi Nakamoto. Bitcoin: A peer-to-peer electronic cash system. http://bitcoin.org/bitcoin.pdf .Accessed May 4, 2019.[6] Weisong Shi, Jie Cao, Quan Zhang, Youhuizi Li, and Lanyu Xu. Edge computing: Vision and challenges.
IEEEInternet of Things Journal , 3(5):637–646, 2016. 24 rXiv
Template
A P
REPRINT [7] Wei Yu, Fan Liang, Xiaofei He, William Grant Hatcher, Chao Lu, Jie Lin, and Xinyu Yang. A survey on theedge computing for the internet of things.
IEEE access , 6:6900–6919, 2017.[8] Fan Liang, Wei Yu, Dou An, Qingyu Yang, Xinwen Fu, and Wei Zhao. A survey on big data market: Pricing,trading and protection.
IEEE Access , 6:15132–15154, 2018.[9] John A Stankovic. Research directions for the internet of things.
IEEE Internet of Things Journal , 1(1):3–9,2014.[10] Jie Lin, Wei Yu, Nan Zhang, Xinyu Yang, Hanlin Zhang, and Wei Zhao. A survey on internet of things:Architecture, enabling technologies, security and privacy, and applications.
IEEE Internet of Things Journal ,4(5):1125–1142, 2017.[11] Hao Yin, Yong Jiang, Chuang Lin, Yan Luo, and Yunjie Liu. Big data: Transforming the design philosophy offuture internet.
IEEE Network , 28(4):14–19, 2014.[12] Haipeng Yao, Chao Qiu, Chao Fang, Xu Chen, and F Richard Yu. A novel framework of data-driven networking.
IEEE Access , 4:9066–9072, 2016.[13] Ying Wang, Peilong Li, Lei Jiao, Zhou Su, Nan Cheng, Xuemin Sherman Shen, and Ping Zhang. A data-driven architecture for personalized qoe management in 5g wireless networks.
IEEE Wireless Communications ,24(1):102–110, 2016.[14] Shuangfeng Han, I Chih-Lin, Gang Li, Sen Wang, and Qi Sun. Big data enabled mobile network design for 5gand beyond.
IEEE Communications Magazine , 55(9):150–157, 2017.[15] Laizhong Cui, F Richard Yu, and Qiao Yan. When big data meets software-defined networking: Sdn for bigdata and big data for sdn.
IEEE network , 30(1):58–65, 2016.[16] Kan Zheng, Zhe Yang, Kuan Zhang, Periklis Chatzimisios, Kan Yang, and Wei Xiang. Big data-driven opti-mization for mobile networks toward 5g.
IEEE network , 30(1):44–51, 2016.[17] Min Chen, Yongfeng Qian, Yixue Hao, Yong Li, and Jeungeun Song. Data-driven computing and caching in 5gnetworks: Architecture and delay analysis.
IEEE Wireless Communications , 25(1):70–75, 2018.[18] Nan Cheng, Feng Lyu, Jiayin Chen, Wenchao Xu, Haibo Zhou, Shan Zhang, and Xuemin Sherman Shen. Bigdata driven vehicular networks.
IEEE Network , 32(6):160–167, 2018.[19] Matteo Sammarco, Miguel Elias Mitre Campista, Marcin Detyniecki, Tahiry Razafindralambo, andMarcelo Dias de Amorim. Unsupervised detection of adversarial collaboration in data-driven networking. In , pages 1–8. IEEE, 2019.[20] Stefanos Astaras, Sofoklis Efremidis, Angela-Maria Despotopoulou, John Soldatos, and Nikos Kefalakis. Deeplearning analytics for iot security over a configurable bigdata platform: Data-driven iot systems. In , pages 1–6. IEEE, 2019.[21] Haojun Huang, Hao Yin, Geyong Min, Hongbo Jiang, Junbao Zhang, and Yulei Wu. Data-driven informationplane in software-defined networking.
IEEE Communications Magazine , 55(6):218–224, 2017.[22] I Chih-Lin, Qi Sun, Zhiming Liu, Siming Zhang, and Shuangfeng Han. The big-data-driven intelligent wirelessnetwork: Architecture, use cases, solutions, and future trends.
IEEE vehicular technology magazine , 12(4):20–29, 2017.[23] Bo Ma, Weisi Guo, and Jie Zhang. A survey of online data-driven proactive 5g network optimisation usingmachine learning.
IEEE Access , 8:35606–35637, 2020.[24] Chaofeng Zhang, Mianxiong Dong, and Kaoru Ota. Enabling computational intelligence for green internet ofthings: Data-driven adaptation in lpwa networking.
IEEE Computational Intelligence Magazine , 15(1):32–43,2020.[25] Rafael Brundo Uriarte and Rocco De Nicola. Blockchain-based decentralized cloud/fog solutions: Challenges,opportunities, and standards.
IEEE Communications Standards Magazine , 2(3):22–28, 2018.25 rXiv
Template
A P
REPRINT [26] Ruizhe Yang, F Richard Yu, Pengbo Si, Zhaoxin Yang, and Yanhua Zhang. Integrated blockchain and edgecomputing systems: A survey, some research issues and challenges.
IEEE Communications Surveys & Tutorials ,21(2):1508–1532, 2019.[27] Khaled Salah, M Habib Ur Rehman, Nishara Nizamuddin, and Ala Al-Fuqaha. Blockchain for ai: Review andopen research challenges.
IEEE Access , 7:10127–10149, 2019.[28] Muhammad Salek Ali, Massimo Vecchio, Miguel Pincheira, Koustabh Dolui, Fabio Antonelli, andMubashir Husain Rehmani. Applications of blockchains in the internet of things: A comprehensive survey.
IEEE Communications Surveys & Tutorials , 21(2):1676–1717, 2018.[29] Qingyi Zhu, Seng W. Loke, Rolando Trujillo-Rasua, Frank Jiang, and Yong Xiang. Applications of distributedledger technologies to the internet of things: A survey.
ACM Comput. Surv. , 52(6), November 2019.[30] Minhaj Ahmad Khan and Khaled Salah. Iot security: Review, blockchain solutions, and open challenges.
FutureGeneration Computer Systems , 82:395–411, 2018.[31] Xiaoyang Zhu and Youakim Badr. Identity management systems for the internet of things: A survey towardsblockchain solutions.
Sensors , 18(12):4215, 2018.[32] Fabiola Hazel Pohrmen, Rohit Kumar Das, and Goutam Saha. Blockchain-based security aspects in hetero-geneous internet-of-things networks: A survey.
Transactions on Emerging Telecommunications Technologies ,30(10):e3741, 2019.[33] Wenli Yang, Erfan Aghasian, Saurabh Garg, David Herbert, Leandro Disiuta, and Byeong Kang. A surveyon blockchain-based internet service architecture: Requirements, challenges, trends and future.
IEEE Access ,2019.[34] Tara Salman, Maede Zolanvari, Aiman Erbad, Raj Jain, and Mohammed Samaka. Security services usingblockchains: A state of the art survey.
IEEE Communications Surveys & Tutorials , 21(1):858–880, 2018.[35] Konstantinos Christidis and Michael Devetsikiotis. Blockchains and smart contracts for the internet of things.
IEEE Access , 4:2292–2303, 2016.[36] Hany F Atlam, Ahmed Alenezi, Madini O Alassafi, and Gary Wills. Blockchain with internet of things: Benefits,challenges, and future directions.
International Journal of Intelligent Systems and Applications , 10(6):40–48,2018.[37] Junfeng Xie, Helen Tang, Tao Huang, F Richard Yu, Renchao Xie, Jiang Liu, and Yunjie Liu. A survey ofblockchain technology applied to smart cities: Research issues and challenges.
IEEE Communications Surveys& Tutorials , 2019.[38] Jameela Al-Jaroodi and Nader Mohamed. Blockchain in industries: A survey.
IEEE Access , 7:36500–36515,2019.[39] Pierluigi Siano, Giuseppe De Marco, Alejandro Rolán, and Vincenzo Loia. A survey and evaluation of thepotentials of distributed ledger technology for peer-to-peer transactive energy exchanges in local energy markets.
IEEE Systems Journal , 2019.[40] Zhibin Lei, Chao Feng, Yang Liu, Dennis SF Lee, Tony Tsang, Jun Liang, Zhijun Xiong, Yuquan Liu, and GangChen. Next generation blockchain network (ngbn). In , pages 452–456. IEEE, 2019.[41] Jiasi Weng, Jian Weng, Jia-Nan Liu, and Yue Zhang. Secure software-defined networking based on blockchain,2019.[42] David Schwartz, Noah Youngs, and Arthur Britto. The ripple protocol consensus algorithm, 2014.[43] Hui Yang, Yajie Li, Shaoyong Guo, Jian Ding, Young Lee, and Jie Zhang. Distributed blockchain-based trustedcontrol with multi-controller collaboration for software defined data center optical networks in 5g and beyond.In
Optical Fiber Communication Conference , pages Th1G–2. Optical Society of America, 2019.[44] Pradip Kumar Sharma, Mu-Yen Chen, and Jong Hyuk Park. A software defined fog node based distributedblockchain cloud architecture for iot.
IEEE Access , 6:115–124, 2017.26 rXiv
Template
A P
REPRINT [45] Cheng Li and Liang-Jie Zhang. A blockchain based new secure multi-layer network model for internet of things.In , pages 33–41. IEEE, 2017.[46] Xiaodong Zhang, Ru Li, and Bo Cui. A security architecture of vanet based on blockchain and mobile edgecomputing. In
Proc. IEEE HotICN’18 , pages 258–259, Aug 2018.[47] Yong Yu, Yannan Li, Junfeng Tian, and Jianwei Liu. Blockchain-based solutions to security and privacy issuesin the internet of things.
IEEE Wireless Communications , 25(6):12–18, Dec. 2018.[48] Keke Gai, Yulu Wu, Liehuang Zhu, Lei Xu, and Yan Zhang. Permissioned blockchain and edge computingempowered privacy-preserving smart grid networks.
IEEE Internet of Things Journal , pages 1–1, Mar. 2019.[49] Alfred Kobsa. Privacy-enhanced personalization.
Commun. ACM , 50(8):24–33, August 2007.[50] Jiafeng Hua, Hui Zhu, Fengwei Wang, Ximeng Liu, Rongxing Lu, Hao Li, and Yeping Zhang. Cinema: Efficientand privacy-preserving online medical primary diagnosis with skyline query.
IEEE Internet of Things Journal ,6(2):1450–1461, 2019.[51] Pradip Kumar Sharma, Saurabh Singh, Young-Sik Jeong, and Jong Hyuk Park. Distblocknet: A distributedblockchains-based secure sdn architecture for iot networks.
IEEE Communications Magazine , 55(9):78–85,Sep. 2017.[52] Libo Feng, Hui Zhang, Liqi Lou, and Yong Chen. A blockchain-based collocation storage architecture fordata security process platform of wsn. , pages 75–80, May. 2018.[53] Yuhua Xu, Houtao Sun, Feng Xiang, and Zhixin Sun. Efficient ddos detection based on k-fknn in softwaredefined networks.
IEEE Access , 7:160536–160545, 2019.[54] Xudong He, Jian Wang, Jiqiang Liu, Zhen Han, Zhuo Lv, and Wei Wang. Dns rebindingdetection for localinternet of things devices. In
Frontiers in Cyber Security , pages 19–29, Singapore, 2020. Springer Singapore.[55] Hui Yang, Haowei Zheng, Jie Zhang, Yizhen Wu, Young Lee, and Yuefeng Ji. Blockchain-based trusted authen-tication in cloud radio over fiber network for 5g. , pages 1–3, Aug. 2017.[56] Beini Zhou, Hui Li, and Li Xu. An authentication scheme using identity-based encryption blockchain.
IEEESymposium on Computers and Communications (ISCC) , pages 556–561, Jun. 2018.[57] Chris Y.T. Ma, David K.Y. Yau, Nung Kwan Yip, and Nageswara S.V. Rao. Privacy vulnerability of publishedanonymous mobility traces. In
Proceedings of the Sixteenth Annual International Conference on Mobile Com-puting and Networking , MobiCom ’10, page 185–196, New York, NY, USA, 2010. Association for ComputingMachinery.[58] Rongxing Lu, Xiaodong Lin, Tom H. Luan, Xiaohui Liang, and Xuemin Shen. Pseudonym changing at so-cial spots: An effective strategy for location privacy in vanets.
IEEE Transactions on Vehicular Technology ,61(1):86–96, 2012.[59] Petra Wohlmacher. Digital certificates: A survey of revocation methods. In
Proceedings of the 2000 ACMWorkshops on Multimedia , MULTIMEDIA ’00, page 111–114, New York, NY, USA, 2000. Association forComputing Machinery.[60] Shixiong Yao, Jing Chen, Kun He, Ruiying Du, Tianqing Zhu, and Xin Chen. Pbcert: Privacy-preservingblockchain-based certificate status validation toward mass storage management.
IEEE Access , 7:6117–6128,Dec. 2019.[61] Xueping Liang, Sachin Shetty, Deepak Tosh, Charles Kamhoua, Kevin Kwiat, and Laurent Njilla. Provchain:A blockchain-based data provenance architecture in cloud environment with enhanced privacy and availability.In
Proceedings of the 17th IEEE/ACM international symposium on cluster, cloud and grid computing , pages468–477. IEEE Press, 2017.[62] Yong Yuan and Fei-Yue Wang. Blockchain and cryptocurrencies: Model, techniques, and applications.
IEEETransactions on Systems, Man, and Cybernetics: Systems , 48(9):1421–1428, 2018.27 rXiv
Template
A P
REPRINT [63] Mingli Wu, Kun Wang, Xiaoqin Cai, Song Guo, Minyi Guo, and Chunming Rong. A comprehensive survey ofblockchain: From theory to iot applications and beyond.
IEEE Internet of Things Journal , 2019.[64] Hiroki Watanabe, Shigeru Fujimura, Atsushi Nakadaira, Yasuhiko Miyazaki, Akihito Akutsu, and Jay JunichiKishigami. Blockchain contract: A complete consensus using blockchain. In , pages 577–578. IEEE, 2015.[65] Rui Zhang, Rui Xue, and Ling Liu. Security and privacy on blockchain.
ACM Comput. Surv. , 52(3):1–34, July2019.[66] Till Neudecker and Hannes Hartenstein. Network layer aspects of permissionless blockchains.
IEEE Commu-nications Surveys & Tutorials , 21(1):838–857, 2018.[67] Wenbo Wang, Dinh Thai Hoang, Peizhao Hu, Zehui Xiong, Dusit Niyato, Ping Wang, Yonggang Wen, andDong In Kim. A survey on consensus mechanisms and mining strategy management in blockchain networks.
IEEE Access , 7:22328–22370, 2019.[68] Marianna Belotti, Nikola Boži´c, Guy Pujolle, and Stefano Secci. A vademecum on blockchain technologies:When, which and how.
IEEE Communications Surveys & Tutorials , 2019.[69] Yunlong Lu, Xiaohong Huang, Yueyue Dai, Sabita Maharjan, and Yan Zhang. Blockchain and federated learn-ing for privacy-preserved data sharing in industrial iot.
IEEE Transactions on Industrial Informatics , 2019.[70] Chi Harold Liu, Qiuxia Lin, and Shilin Wen. Blockchain-enabled data collection and sharing for industrial iotwith deep reinforcement learning.
IEEE Transactions on Industrial Informatics , 2018.[71] Wuhui Chen, Yufei Chen, Xu Chen, and Zibin Zheng. Toward secure data sharing for the iov: A quality-drivenincentive mechanism with on-chain and off-chain guarantees.
IEEE Internet of Things Journal , 2019.[72] Shangping Wang, Yinglong Zhang, and Yaling Zhang. A blockchain-based framework for data sharing withfine-grained access control in decentralized storage systems.
IEEE Access , 6:38437–38450, 2018.[73] Masayuki Fukumitsu, Shingo Hasegawa, Junya Iwazaki, Masao Sakai, and Daiki Takahashi. A proposal of asecure p2p-type storage scheme by using the secret sharing and the blockchain. In , pages 803–810. IEEE, 2017.[74] Dagang Li, Rong Du, Yue Fu, and Man Ho Au. Meta-key: A secure data-sharing protocol under blockchain-based decentralized storage architecture.
IEEE Networking Letters , 1(1):30–33, Mar. 2019.[75] Ravi Kiran Raman and Lav R Varshney. Distributed storage meets secret sharing on the blockchain. In , pages 1–6. IEEE, 2018.[76] Xiaochen Zheng, Raghava Rao Mukkamala, Ravi Vatrapu, and Joaqun Ordieres-Mere. Blockchain-based per-sonal health data sharing system using cloud storage. In , pages 1–6. IEEE, 2018.[77] Hao Jin, Yan Luo, Peilong Li, and Jomol Mathew. A review of secure and privacy-preserving medical datasharing.
IEEE Access , 7:61656–61669, 2019.[78] Shan Jiang, Jiannong Cao, Hanqing Wu, Yanni Yang, Mingyu Ma, and Jianfei He. Blochie: A blockchain-basedplatform for healthcare information exchange. In , pages 49–56. IEEE, 2018.[79] QI Xia, Emmanuel Boateng Sifah, Kwame Omono Asamoah, Jianbin Gao, Xiaojiang Du, and Mohsen Guizani.Medshare: Trust-less medical data sharing among cloud service providers via blockchain.
IEEE Access ,5:14757–14767, 2017.[80] Guy Zyskind, Oz Nathan, et al. Decentralizing privacy: Using blockchain to protect personal data. In , pages 180–184. IEEE, 2015.[81] Faiza Loukil, Chirine Ghedira-Guegan, Khouloud Boukadi, and Aïcha Nabila Benharkat. Towards an end-to-end iot data privacy-preserving framework using blockchain technology. In
International Conference on WebInformation Systems Engineering , pages 68–78. Springer, 2018.28 rXiv
Template
A P
REPRINT [82] Nesrine Kaaniche and Maryline Laurent. A blockchain-based data usage auditing architecture with enhancedprivacy and availability. In , pages 1–5. IEEE, 2017.[83] Yun Chen, Hui Xie, Kun Lv, Shengjun Wei, and Changzhen Hu. Deplest: A blockchain-based privacy-preserving distributed database toward user behaviors in social networks.
Information Sciences , 2019.[84] Alka Vishwa and Farookh Khadeer Hussain. A blockchain based approach for multimedia privacy protectionand provenance. In , pages 1941–1945.IEEE, 2018.[85] Shan Jiang, Jiannong Cao, Julie A. McCann, Yanni Yang, Yang Liu, Xiaoqing Wang, and Yuming Deng.Privacy-preserving and efficient multi-keyword search over encrypted data on blockchain. In , pages 405–410. IEEE, 2019.[86] Abdullah Al Omar, Md Zakirul Alam Bhuiyan, Anirban Basu, Shinsaku Kiyomoto, and Mohammad ShahriarRahman. Privacy-friendly platform for healthcare data in cloud based on blockchain environment.
FutureGeneration Computer Systems , 95:511–521, 2019.[87] Gaby G Dagher, Jordan Mohler, Matea Milojkovic, and Praneeth Babu Marella. Ancile: Privacy-preservingframework for access control and interoperability of electronic health records using blockchain technology.
Sustainable Cities and Society , 39:283–297, 2018.[88] Gábor Magyar. Blockchain: Solving the privacy and research availability tradeoff for ehr data: A new disruptivetechnology in health data management. In , pages 000135–000140.IEEE, 2017.[89] Richard Nuetey Nortey, Li Yue, Promise Ricardo Agdedanu, and Michael Adjeisah. Privacy module for dis-tributed electronic health records (ehrs) using the blockchain. In , pages 369–374. IEEE, 2019.[90] Shawn Wilkinson, Tome Boshevski, Josh Brandoff, and Vitalik Buterin. Storj a peer-to-peer cloud storagenetwork. 2014.[91] Hoang Giang Do and Wee Keong Ng. Blockchain-based system for secure data storage with private keywordsearch. In , pages 90–93. IEEE, 2017.[92] Sushmita Ruj, Mohammad Shahriar Rahman, Anirban Basu, and Shinsaku Kiyomoto. Blockstore: A securedecentralized storage framework on blockchain. In , pages 1096–1103. IEEE, 2018.[93] Jiaxing Li, Jigang Wu, and Long Chen. Block-secure: Blockchain based scheme for secure p2p cloud storage.
Information Sciences , 465:219–231, 2018.[94] Bo Zhao, Peiru Fan, and Mingtao Ni. Mchain: a blockchain-based vm measurements secure storage approachin iaas cloud with enhanced integrity and controllability.
IEEE Access , 6:43758–43769, 2018.[95] Huirui Han, Mengxing Huang, Yu Zhang, and Uzair Aslam Bhatti. An architecture of secure health informationstorage system based on blockchain technology. In
International Conference on Cloud Computing and Security ,pages 578–588. Springer, 2018.[96] Jingqiang Liu, Bin Li, Lizhang Chen, Meng Hou, Feiran Xiang, and Peijun Wang. A data storage methodbased on blockchain for decentralization dns. In , pages 189–196. IEEE, 2018.[97] Manish Kumar, Ashish Kumar Singh, and TV Suresh Kumar. Secure log storage using blockchain and cloudinfrastructure. In , pages 1–4. IEEE, 2018.[98] Ke Huang, Xiaosong Zhang, Yi Mu, Fatemeh Rezaeibagha, Xiaojiang Du, and Nadra Guizani. Achievingintelligent trust-layer for iot via self-redactable blockchain.
IEEE Transactions on Industrial Informatics , 2019.[99] Zhaofeng Ma, Xiaochang Wang, Jain Deepak Kumar, Khan Hanees, Zhen Wang, and Hongmin Gao. Ablockchain-based trusted data management scheme in edge computing.
IEEE Transactions on Industrial In-formatics , 2019. 29 rXiv
Template
A P
REPRINT [100] Yongping Zhang, Xiwei Xu, Ang Liu, Qinghua Lu, Lida Xu, and Fei Tao. Blockchain-based trust mechanismfor iot-based smart manufacturing system.
IEEE Transactions on Computational Social Systems , 2019.[101] Zhe Yang, Kan Yang, Lei Lei, Kan Zheng, and Victor CM Leung. Blockchain-based decentralized trust man-agement in vehicular networks.
IEEE Internet of Things Journal , 6(2):1495–1505, 2018.[102] Zhaojun Lu, Wenchao Liu, Qian Wang, Gang Qu, and Zhenglin Liu. A privacy-preserving trust model basedon blockchain for vanets.
IEEE Access , 6:45655–45664, Aug. 2018.[103] Yao-Tsung Yang, Li-Der Chou, Chia-Wei Tseng, Fan-Hsun Tseng, and Chien-Chang Liu. Blockchain-basedtraffic event validation and trust verification for vanets.
IEEE Access , 7:30868–30877, 2019.[104] Uchi Ugobame Uchibeke, Kevin A Schneider, Sara Hosseinzadeh Kassani, and Ralph Deters. Blockchainaccess control ecosystem for big data security. In , pages 1373–1378. IEEE, 2018.[105] Shangping Wang, Xu Wang, and Yaling Zhang. A secure cloud storage framework with access control basedon blockchain.
IEEE Access , 7:112713–112725, 2019.[106] Xiaobin Tan, Chaoming Huang, and Liguo Ji. Access control scheme based on combination of blockchainand xor-coding for icn. In , pages160–165. IEEE, 2018.[107] Sheng Ding, Jin Cao, Chen Li, Kai Fan, and Hui Li. A novel attribute-based access control scheme usingblockchain for iot.
IEEE Access , 7:38431–38441, 2019.[108] Sophie Dramé-Maigné, Maryline Laurent, and Laurent Castillo. Distributed access control solution for the iotbased on multi-endorsed attributes and smart contracts. In , pages 1582–1587. IEEE, 2019.[109] Mingxin Ma, Guozhen Shi, and Fenghua Li. Privacy-oriented blockchain-based distributed key managementarchitecture for hierarchical access control in the iot scenario.
IEEE Access , 7:34045–34059, 2019.[110] Jordi Paillisse, Jordi Subira, Albert Lopez-Bresco, Alberto Rodríguez-Natal, Vina Ermagan, Fabio Maino, andAlbert Cabellos. Distributed access control with blockchain. In
ICC 2019 - 2019 IEEE International Conferenceon Communications (ICC) , pages 1–6, May 2019.[111] Weizhi Meng, Elmar Wolfgang Tischhauser, Qingju Wang, Yu Wang, and Jinguang Han. When intrusiondetection meets blockchain technology: A review.
IEEE Access , 6:10179–10188, 2018.[112] Raja Majid Ali Ujjan, Zeeshan Pervez, and Keshav Dahal. Snort based collaborative intrusion detection sys-tem using blockchain in sdn. In , pages 1–8. IEEE, 2019.[113] Osama Alkadi, Nour Moustafa, Benjamin Turnbull, and Kim-Kwang Raymond Choo. A deep blockchainframework-enabled collaborative intrusion detection for protecting iot and cloud networks.
IEEE Internet ofThings Journal , 2020.[114] Muhammad Saad, Afsah Anwar, Ashar Ahmad, Hisham Alasmary, Murat Yuksel, and Aziz Mohaisen.Routechain: Towards blockchain-based secure and efficient bgp routing. In , pages 210–218. IEEE, 2019.[115] Min Li, Helen Tang, and Xianbin Wang. Mitigating routing misbehavior using blockchain-based distributedreputation management system for iot networks. In , pages 1–6. IEEE, 2019.[116] Gholamreza Ramezan and Cyril Leung. A blockchain-based contractual routing protocol for the internet ofthings using smart contracts.
Wireless Communications and Mobile Computing , 2018, 2018.[117] Jidian Yang, Shiwen He, Yang Xu, Linweiya Chen, and Ju Ren. A trusted routing scheme using blockchain andreinforcement learning for wireless sensor networks.
Sensors , 19(4):970, 2019.30 rXiv
Template
A P
REPRINT [118] Donghui Ding, Tiantian Duan, Linpeng Jia, Kang Li, Zhongcheng Li, and Yi Sun. Interchain: A framework tosupport blockchain interoperability. 2018.[119] Zhi-dong Chen, YU Zhuo, Zhang-bo Duan, and HU Kai. Inter-blockchain communication.
DEStech Transac-tions on Computer Science and Engineering , (cst), 2017.[120] ZhongAn Tech. Anlink blockchain network whitepaper, 2017.[121] O˘guzhan Ersoy, Zhijie Ren, Zekeriya Erkin, and Reginald L Lagendijk. Transaction propagation on permis-sionless blockchains: Incentive and routing mechanisms. In , pages 20–30. IEEE, 2018.[122] Danda B Rawat. Fusion of software defined networking, edge computing, and blockchain technology forwireless network virtualization.
IEEE Communications Magazine , 57(10):50–55, 2019.[123] Mayra Samaniego and Ralph Deters. Virtual resources & blockchain for configuration management in iot.
Journal of Ubiquitous Systems & Pervasive Networks , 9(2):01–13, 2017.[124] He Zhu, Changcheng Huang, and Jiayu Zhou. Edgechain: Blockchain-based multi-vendor mobile edge appli-cation placement. In , pages222–226. IEEE, 2018.[125] Jere Backman, Seppo Yrjölä, Kristiina Valtanen, and Olli Mämmelä. Blockchain network slice broker in 5g:Slice leasing in factory of the future use case. In ,pages 1–8. IEEE, 2017.[126] Ming Li and GQ Huang. Blockchain-enabled workflow management system for fine-grained resource sharingin e-commerce logistics. In , pages 751–755. IEEE, 2019.[127] Lijing Zhou, Licheng Wang, Yiru Sun, and Pin Lv. Beekeeper: A blockchain-based iot system with securestorage and homomorphic computation.
IEEE Access , 6:43472–43488, 2018.[128] Fengrui Shi, Zhijin Qin, Di Wu, and Julie McCann. Mpcstoken: Smart contract enabled fault-tolerant incentivi-sation for mobile p2p crowd services. In , pages 961–971. IEEE, 2018.[129] Gokhan Sagirlar, Barbara Carminati, and Elena Ferrari. Autobotcatcher: Blockchain-based p2p botnet detectionfor the internet of things. In , pages 1–8. IEEE, 2018.[130] Michael Mylrea and Sri Nikhil Gupta Gourisetti. Blockchain for smart grid resilience: Exchanging distributedenergy at speed, scale and security. In , pages 18–23. IEEE, 2017.[131] Xueping Liang, Juan Zhao, Sachin Shetty, and Danyi Li. Towards data assurance and resilience in iot usingblockchain. In
MILCOM 2017-2017 IEEE Military Communications Conference (MILCOM) , pages 261–266.IEEE, 2017.[132] Christian Cachin. Architecture of the hyperledger blockchain fabric. In
Workshop on distributed cryptocurren-cies and consensus ledgers , volume 310, page 4, 2016.[133] Ittay Eyal and Emin Gün Sirer. Majority is not enough: Bitcoin mining is vulnerable.
Communications of theACM , 61(7):95–102, 2018.[134] Christopher Natoli and Vincent Gramoli. The balance attack against proof-of-work blockchains: The r3 testbedas an example. arXiv preprint arXiv:1612.09426 , 2016.[135] Maria Apostolaki, Aviv Zohar, and Laurent Vanbever. Hijacking bitcoin: Routing attacks on cryptocurrencies.In , pages 375–392. IEEE, 2017.[136] Atul Singh et al. Eclipse attacks on overlay networks: Threats and defenses. In
In IEEE INFOCOM . Citeseer,2006.[137] Divesh Aggarwal, Gavin K. Brennen, Troy Lee, Miklos Santha, and Marco Tomamichel. Quantum attacks onbitcoin, and how to protect against them.
Papers , 2017.31 rXiv
Template
A P
REPRINT [138] Yue Li, Junqin Huang, Shengzhi Qin, and Ruijin Wang. Big data model of security sharing based on blockchain.In , pages 117–121.IEEE, 2017.[139] Hamza Es-Samaali, Aissam Outchakoucht, and Jean Philippe Leroy. A blockchain-based access control for bigdata.
International Journal of Computer Networks and Communications Security , 5(7):137, 2017.[140] Vitalik Buterin. On sharding blockchains. 2017.[141] Bruce J D. The mini-blockchain scheme.
White paper , 2014.[142] Doriane Perard, Jerome Lacan, Yann Bachy, and Jonathan Detchart. Erasure code-based low storage blockchainnode. In , 2018.[143] Ali Dorri, Salil S Kanhere, and Raja Jurdak. Towards an optimized blockchain for iot. In
The second IEEE/ACMconference on Internet of Things Design and Implementation, IoTDI 2017 , 2017.[144] Buterin V Poon J. Plasma: Scalable autonomous smart contracts.
White paper , pages 1–47, 2017.[145] Taras Maksymyuk, Juraj Gazda, Longzhe Han, and Minho Jo. Blockchain-based intelligent network manage-ment for 5g and beyond. In , pages 36–39, July 2019.[146] Sanjay Panikkar, Sumabala Nair, Paul Brody, and Veena Pureswaran. Adept: An iot practitioner perspective.
Draft Copy for Advance Review, IBM , 2015.[147] Usman W Chohan. The decentralized autonomous organization and governance issues.
Available at SSRN3082055 , 2017.[148] Muhammad Salek Ali, Koustabh Dolui, and Fabio Antonelli. Iot data privacy via blockchains and ipfs. In
Proceedings of the Seventh International Conference on the Internet of Things , page 14. ACM, 2017.[149] Chenhan Xu, Kun Wang, and Mingyi Guo. Intelligent resource management in blockchain-based cloud data-centers.
IEEE Cloud Computing , 4(6):50–59, 2017.[150] Marshall Ball, Alon Rosen, Manuel Sabin, and Prashant Nalini Vasudevan. Proofs of useful work.
IACRCryptology ePrint Archive , 2017:203, 2017.[151] Fan Zhang, Ittay Eyal, Robert Escriva, Ari Juels, and Robbert Van Renesse. { REM } : Resource-efficient miningfor blockchains. In { USENIX } Security Symposium ( { USENIX } Security 17) , pages 1427–1444, 2017.[152] Sunoo Park, Krzysztof Pietrzak, Joël Alwen, Georg Fuchsbauer, and Peter Gazi. Spacecoin: A cryptocurrencybased on proofs of space. In
IACR Cryptology ePrint Archive . 2015.[153] Xiaolong Xu, Xuyun Zhang, Honghao Gao, Yuan Xue, Lianyong Qi, and Wanchun Dou. Become: Blockchain-enabled computation offloading for iot in mobile edge computing.