Intrusion Detection Systems for Networked Unmanned Aerial Vehicles: A Survey
Gaurav Choudhary, Vishal Sharma, Ilsun You, Kangbin Yim, Ing-Ray Chen, Jin-Hee Cho
IIntrusion Detection Systems for NetworkedUnmanned Aerial Vehicles: A Survey
Gaurav Choudhary, Vishal Sharma,Ilsun You, Kangbin Yim
Department of Information Security EngineeringSoonchunhyang University, ROK
Email: { gauravchoudhary7777, ilsunu } @gmail.com,vishal [email protected], [email protected] Ing-Ray Chen
Department of Computer ScienceVirginia TechVA, USA
Email: [email protected]
Jin-Hee Cho
U.S. Army Research LaboratoryMD, USA
Email: [email protected]
Abstract —Unmanned Aerial Vehicles (UAV)-based civilian ormilitary applications become more critical to serving civilianand/or military missions. The significantly increased attention onUAV applications also has led to security concerns particularly inthe context of networked UAVs. Networked UAVs are vulnerableto malicious attacks over open-air radio space and accordinglyintrusion detection systems (IDSs) have been naturally derivedto deal with the vulnerabilities and/or attacks. In this paper, webriefly survey the state-of-the-art IDS mechanisms that deal withvulnerabilities and attacks under networked UAV environments.In particular, we classify the existing IDS mechanisms accordingto information gathering sources, deployment strategies, detectionmethods, detection states, IDS acknowledgment, and intrusiontypes. We conclude this paper with research challenges, insights,and future research directions to propose a networked UAV-IDS system which meets required standards of effectiveness andefficiency in terms of the goals of both security and performance.
Index Terms —Unmanned aerial vehicle, intrusion detectionsystem, security, attack, vulnerability.
I. I
NTRODUCTION
The proliferation of unmanned aerial vehicles (UAVs) andtheir diverse applications in many different domains havebeen realized due to their merit of dynamic reconfigurability,fast response, and ease of deployment. In particular, theapplications of networked UAVs have attracted major industryplayers such as Google, Facebook, Boeing, and Amazon. Inaddition, their applications in serving military and civilianmissions have been explored in diverse domains to providepublic safety, surveillance, medical services, and/or militarymission support [13]. In Table I, we discuss the key applicationdomains where UAVs can be applied to assist given missionsin different domain context.The key merit of UAVs is known as its high reconfigurabilityand mobility. However, its mobility also exposes an issueof controllability towards the aerial vehicles and causes linkdistortion in UAV networking. Despite these concerns, UAV-assisted networks have been recognized for the benefit of easy deployment of wireless connectivity that does not require anyphysical infrastructure [19, 20].UAVs provide high benefits to assist the goals of manydifferent applications, as summarized in Table I. However,they also introduce the following challenges [21]: (1) thearchitectural design of drone communication lacks a standardor unification; (2) UAV-assisted communication networks suf-fer from an issue of dedicated spectrum sharing; (3) UAVdeployment and path planning should be considered duringspectrum allocations due to its potential impact on energyefficiency; and (4) UAV communications introduce additionaloverhead to architectural design, deployment, and consistencywith large and reliable networks along with their security. Inthis work, we particularly focus on security challenges.This paper provides the following key contributions : • We survey the key state-of-the-art UAV-IDS approachesand associated taxonomies which can provide a goodoverview to answer what a UAV-IDS system is, what thekey components need to be considered in the UAV-IDS,and what the key security concerns should be considered,associated with the key components of the UAV-IDS. • We discuss the main research challenges and hurdlesto build a cyber-physical hardened UAV-IDS systemunder highly resource-constrained, hostile, dynamic, anddistributed environments reflecting the key characteristicsof military tactical characteristics. • We suggest future research directions to move towardsbased on the discussed research challenges and learnedlessons / insights.The remainder of the paper is organized as follows. SectionII provides the background and goal of UAV-IDS. Section IIIdiscusses the taxonomies used in the structure and classifica-tion of UAV-IDS. Section IV discusses the evaluation tech-niques of the state-of-the-art UAV-IDS approaches. Section Vdiscusses research challenges derived from the inherent char-acteristics of UAV-IDS environments. Section VI concludesthe paper and suggests future work directions. a r X i v : . [ c s . CR ] J u l ABLE ID
OMAINS AND APPLICATIONS OF
UAV S Domain Key example applications Achieved roles by UAVsLaw enforcement surveillance Search and rescue Equipped with cameraPublic safety communications Voice communications in case of disaster Aerial base stationsEnvironmental applications Climate change Equipped with sensorsLogistics Goods shipping / delivery in urban areas Drone as a transportation mediumMilitary applications Searches for lost or injured soldiers Armed with live video remote communications to groundtroops, essential gear, or weaponsMedical field applications Delivering aid packages, medicines, vaccines, bloodand other medical supplies to remote areas Drone as a transportation mediumVideo and photography Events (e.g., social gatherings, sports games, orcompetitions) Equipped with cameraAgriculture Crop monitoring and soil and field analysis Equipped with sensors
II. O
VERALL D ESCRIPTION OF
UAV-IDS E
NVIRONMENTS
An unmanned aerial vehicle based intrusion detection sys-tem (UAV-IDS) is developed to detect anomaly behaviouror illegal activities in a network by automatically analyzingthe behaviors or activities based on given hypothesis and/orpolicies, which are governed by the security rules of the givennetwork [2]. The UAV-IDS monitors system configuration,data files, and/or network transmission to check whether thereexists an attack. Hence, the UAV-IDS is to mitigate the effectof the attacks aiming to prevent any covert / overt operationsfrom exposed vulnerabilities of the system. In addition, UAV-IDSs aim to detect the misuse of UAVs. Misuse can be definedas any undesirable activity which can cause any harmful effectin terms of either performance or security to an entire swarm ofthe UAVs. Attacks explore the vulnerabilities of UAV systems,where the vulnerabilities can be the result of misconfigurationsof UAV networks, an implementation fault, flawed designsand/or protocols [4].Fig. 1 shows an example scenario for an UAV-IDS. UAV-IDSs monitor signals, command traffic, control instructions,working behavior, energy consumption, and/or operations ofUAV components. In addition, it analyzes the data flowand gather information from different components of UAVsduring their operations as a network node. The UAV-IDSsare capable of enhancing reliability and/or security of UAVcommunications in an efficient and effective manner.A UAV-IDS can be placed on a UAV or a ground controlsystem and maintains the security and reliability of the UAVand the ground control system. The placement of a UAV-IDScan be determined based on the level of required security,such as required security levels in terms of confidentiality,integrity, availability, and authorization. In addition, a UAV-IDS is responsible for ensuring guarding the UAV systemagainst unlawful activities or attacks. The incorporated secu-rity policies for UAVs provide low-complex rules to detectanomalies or potential threats. These policies can be designedthrough different approaches or algorithms based on the re-quirements of the UAV system. Most existing IDSs for UAVsuse behaviour-based detection mechanisms [23].III. T
AXONOMIES OF
UAV-IDS S
YSTEM C OMPONENTS
We summarize the key component taxonomies of the UAV-IDS system in Fig. 2 which discusses its key components,
Fig. 1. An operational overview of UAV-IDS. including information gathering sources, deployment strate-gies, detection methods, detection states, acknowledgment, andintrusion types. We give the detail of each component as below.
A. Information Gathering Sources
A UAV is embedded with a cyber-physical system con-sisting of sensors and/or actuators. Sensors provide data (orinformation) to an actuator that can control the UAV. Thecollected data are used for analysis to make mission-relatedcritical decisions. The information gathering sources can beclassified as follows: [10] • Sensors : Sensors collect information in terms of signalsand/or behavior through sensors like inertial sensors,location sensors, and/or threat sensors. The sensor maybe implicitly tied with a UAV or explicitly tied to aspecific task object, like weather capturing sensors. Thenformation retrieval by a malicious node from any ofonboard sensors in a critical situation can impact theperformance of a UAV in a networked scenario. • Communications links : Communication links supporttransmissions directly to UAVs in mission areas and/orallow simultaneous sharing of information among multi-ple UAVs and the ground system. They also secure datatransfers by monitoring the traffic between a source anda destination. • Ground control system (GCS) : The GCS has a signif-icant component in UAVs and is charged of conductingintelligent surveillance and reconnaissance based on datagenerated by the unmanned aircraft’s payload. • UAV components : The components within a UAV in-clude a power supply unit, antennas, transceiver units,navigation systems, and an inbuilt UAV control system.All inter- and intra-communications take place throughthese components, in which the information is exchangedamong these components for an effective control andmaneuvering of UAVs. This information should be ex-amined for security purposes and timely patches shouldbe available upon the detection of potential threats. • Deployment strategies : The deployment of an IDS inUAVs is critical because effective optimization is requiredfor balancing the trade-off between IDS operations andUAV transmissions. The system should be maintainedto enhance performance of the UAV with their effectiveoperations and environmental conditions along with con-trollable activities of the deployed IDS. The IDS can bedeployed based on two methods:1)
Ground-coordinated or network-initiated basis :In the ground coordinated IDS, all the gatheredinformation is analyzed on the ground station andappropriate decisions are made on the basis ofanalyzed data; and2)
Autonomous or host basis : With an autonomousdeployment of IDSs, UAVs acting as hosts to deployIDSs should conduct data analysis and control otherUAVs, along with coordinating between these two.In this deployment type, the IDS is placed withinthe system control of UAVs in the form of hardwareor software.
B. Intrusion Detection Systems (IDSs)
The key mechanism of IDSs can be classified as follows: • Specification-based [26] : A UAV-IDS is incorporatedwith respective rules specified based on the expectedbehaviors of UAVs. These specified rules are applied tomonitor successful executions of the UAV system. • Signature-based [27] : This method aims to detect knownattacks based on pre-defined, known signatures. Upondetecting anomaly activities, a detection operation istriggered to identify a matched signature to ensure thedetection of an intrusion. • Anomaly-based [14] : Anomaly behavior is detectedbased on a failure or an illegal activity observed in a sys- tem. With the goal of detecting known and/or unknownattacks, this method uses learning or a filtering mecha-nism, which can significantly enhance the detection ofunknown attacks in the absence of pre-defined signaturesof the unknown attacks. • Hybrid-based [1] : This method is a hybrid approachby integrating two or more detection methods, such asspecification plus anomaly, in order to provide a strongdetection policy that can catch known and/or unknownattacks.
Fig. 2. Taxonomy of UAV-IDSs.
C. Detection States
Based on the source of information, we can categorize twomajor detection states: • On-site (or dynamic) : The detection state is evaluatedbased on data collected and monitored from real-timeoperations; the detection analysis and decisions are madeat on-site UAVs. • Off-site (or static) : The detection state is evaluated basedon data collected by an IDS from all information sources;the detection is made based on the analysis of all collecteddata received in the IDS.
D. IDS Acknowledgment
Based on the result of data analysis, a UAV-IDS makesdecision on whether there exists an attack via an IDS acknowl-edgment. This IDS acknowledgment has two forms:
ABLE IIS
URVEY ON THE S TATE - OF -T HE -A RT ON E XISTING
UAV IDS A
PPROACHES .Reference Proposed scheme Key method(s) / metricsBlasch et al. [3] War planning situation awareness tool ROC for visualizationLauf et al. [10] Hybrid IDS Maxima and cross-correlation detectionShen et al. [23] Markov game theoretic approach Deployment of IDS, configuration of email-filtering, firewallsettings, and shutdown or reset policies for servers.Muniraj and Farhood [12] Framework for detection of cyber-physical attacks onthe sensors Anomaly-based detectors based on knowledge of the physicalsystem and statistical analysis.Sedjelmaci et al. [17] Hierarchical IDS Threat classification & behavior monitoringMitchell and Chen [11] Behavior rule-based evaluations Minimizing false positives & false negativesKwon et al. [8] Safety analysis under stealthy cyber attacks Real-time safety assessmentLauf and Robinson [9] Distributed resource system Intrusion-tolerance strategyShen et al. [24] Game theoretic approach Based on the three levels: object, situation and threat.Sedjelmaci et al. [15] Security game framework Optimal setting identification based on intrusion detection ratewith minimum overhead using Bayesian gameTrafton and Pizzi [25] Network service suite Framework for information assurance of UAVs • Instant acknowledgement : In this acknowledgment, anIDS monitoring is performed at real-time and decisionsor alarms are generated in the form of instant acknowl-edgments. • Periodic acknowledgement : In this acknowledgement,an IDS continuously gathers the data but the decisionsare based on the periodical analysis of received data.
E. Intrusion Types
A UAV-IDS should be able to detect the following intrusiontypes: • Virus, worms, and/or malware; • Modification of signals, modification of sharing informa-tion; • Routing attacks, UAV capturing, and/or path alteration;and • Message forgery, UAV spoofing, and/or GPS spoofing.IV. S
URVEY OF E XISTING
UAV-IDS A
PPROACHES
UAV networks are highly sensitive over which criticalinformation will be exchanged between UAVs and the groundstation. Table 2 summarizes the state-of-the-art UAV-IDSapproaches, aiming to enhance security and performance withthe end goal to build a cyber-physical hardened system beingprotected against inside and outside attackers on networkedUAVs. Below we survey these existing approaches using ourproposed taxonomies discussed in Section III.Blasch et al. [3] proposed a war planning situation aware-ness tool by leveraging the Receiver Operating Characteristic(ROC) plots to visualize the effectiveness of their classifi-cations. The effective classification is developed based onmatrices where situation assessment is used to derive relationsbetween a given classification and a location. Lauf et al. [10]developed a decentralized anomaly-based detection technique,which uses maxima and cross-correlation detection methods.The Maxima Detection System (MDS) allows the characteri-zation of either one or zero suspicious nodes. Cross-correlationdetection methods are capable of detecting multiple intrusions.However, this work does not capture the quality of the IDSbased on detection errors including false positives and falsenegatives. Shen et al. [23] took a game theoretic approach by con-sidering three levels of states: object, situation, and threat.This approach projects attack activities while focusing on thestates of the network. Shen et al. [24] further developed acooperative surveillance strategy to improve the performancethrough adaptive Markov game based on the cooperativejamming strategies. These are performed on the basis of fourdefensive parameters, including IDS deployment, configura-tions of email-filtering, firewall settings, and shut down orreset policies for servers.Muniraj and Farhood [12] focused on the attacks over smallUAVs by identifying malicious activities over their sensors. Inthe proposed framework that detects cyber-physical attacks,sensors are designed based on the knowledge of physicalsystem and statistical analysis techniques. However, the pro-posed scheme was not capable of detecting combination ofpiece-wise constant attacks of smaller magnitude. Sedjelmaciet al. [17] proposed an hierarchical IDS and intrusion responsemechanism by classifying threats and monitoring UAV behav-ior to detect malicious activities.Mitchell and Chen [11] proposed a specification-based de-tection technique to guard a UAV system against cyber-attacks.This work used a behavior rule-based UAV-IDS, in which thebehavior rules are constructed based on defined attack models,considering reckless, random, and opportunistic attacks. Thiswork minimized detection errors (i.e., false positives and falsenegatives) based on the critical tradeoff between security andperformance of UAVs. Kwon et al. [8] developed a real-timesafety assessment algorithm based on reachability analysis todeal with cyber attacks.Lauf and Robinson [9] developed a distributed sensingmechanism to build a fault-tolerant resource managementsystem. The proposed mechanisms uses a service discoveryprotocol (SDP) where SDP flooding can introduce a burst ofcommunications leading to traffic congestion or bottleneckissues. Sedjelmaci et al. [16] proposed a threat estimationmodel based on estimated beliefs towards whether a threatexists in the system. In addition, this work incorporated spe-cific detection policies to maintain data integrity and networkavailability. Sedjelmaci et al. [15] took one step further topropose a robust UAV assisted network against lethal attack-ers, namely a Security Game Framework (SGF), which isormulated based on Bayesian game among the suspectednodes. This approach formulates two attack-defense problemsbetween IDS and the attacker, and between intrusion ejectionsystem and the suspected nodes.Trafton and Pizzi [25] proposed the so called Joint AirborneNetwork Services Suite, which aims to integrate an airbornemilitary network by allowing the implementation of variouspossible hardware and software solutions. In this work, an IDSis considered as an integral part of their assurance strategy.V. R
ESEARCH C HALLENGES
UAVs are operated remotely while receiving control andcommand messages from ground stations. These commandand control messages are transmitted over different channelsand variable transmission rate. Security vulnerabilities can beexploited to compromise confidentiality, integrity, availability,and authorization of networked UAVs [6, 7]. Message securityand control signal protections are achieved by cryptographicmechanisms. However, security issues, like unauthorized ac-cess, malicious control, illegal connection, or other maliciousattacks, require strategic solutions without compromising per-formance. Identifying and mitigating threats in UAV networksefficiently and effectively is a first step to secure UAV net-works [18, 22].The significant increase of threats and/or attacks in UAVnetworks brought our attention on the issue of the deploymentof IDS which will play a key role to achieve the effectivenessand efficiency of the UAV-IDS [5]. In a UAV environment, anIDS is being operated based on specific rules and/or policiesto determine whether an observed activity is malicious or not.The results of the IDSs can be used to develop strategies tomitigate the identified risks. However, the design and develop-ment meeting these two requirements (i.e., effectiveness andefficiency of the developed IDS) is not a trivial goal because itoften requires a time-consuming, high-overhead process whichcan often exceed the benefit of introducing high security inpractice.To achieve the UAV-IDS system that meets required lev-els of effectiveness (i.e., minimizing detection errors withminimum service interruptions) and efficiency (i.e., reducingcomputational and communication overhead), we identify thefollowing challenges on the table to pave a way to build acyber-physical hardened UAV-IDS system: • Detection latency : The detection latency can be used as ameasure of agility of an IDS. However, there is a criticaltradeoff in that triggering the IDS more often leadsto incurring more communication/computation overhead,which naturally results in low efficiency, and vice-versa.Hence, we need to make a good balance to achieveboth efficiency and effectiveness in order to build anaffordable, secure UAV-IDS systems in practice. • IDS computational cost : The computational cost as-sociated with IDSs is closely related to how much wewant to achieve the accuracy of the IDS and securityvulnerabilities we allow in a given system. Again, this issue is not trivial because more cost not only incurs highoverhead, but brings more benefit in enhancing security. • Implementation overhead : The high implementationoverhead of IDSs causes power consumption and de-grades the performance of UAVs resulting in a networkshutdown. • Threat & behavior modeling : IDS detection techniquesincorporate behaviors of UAVs. The rules are designedby reflecting the UAVs’ behavior and/or possible threats.However, accurate observations of threat/attack behaviorsand accordingly their correct modeling is not a trivial taskalthough achieving it can provide an enormous benefit toexpedite the development of better defense strategies. • Effective threat assessment : An effective threat assess-ment is critical to mitigating vulnerabilities and risksassociated with threats occurred. In particular, developingeffective threat assessment policies is the key to enhanceboth security and performance of UAV-IDS systems. • Maximum network throughput with minimum cost :This is a typical tradeoff issue any network can face asthese two goals are conflicting to each other. However,based on dynamic monitoring to capture an accuratesystem state, both goals that are dynamically set can beachievable. • Lightweight IDS with minimum resource consump-tion : As UAVs are battery-operated and resource-constrained, the development of lightweight IDS mech-anisms is highly challenging but a must to achieve innetworked UAVs. • Effective monitoring and attack response : The re-sponse against an attack is naturally related to howquickly the attack is detected by a given IDS. Thisimplies that the effectiveness of the IDS is closelyrelated to how quickly the system can respond to thedetected attack. This is indeed the issue of the agilityof a system which should take appropriate actions inorder to minimize damages or vulnerabilities caused bythe intrusion which exploits system vulnerabilities. Thecontested nature of UAV environments, characterized byresource-constraints, high hostility, high dynamics, anddistributed nature, also adds more challenges to achievethis goal.VI. C
ONCLUSION & F
UTURE R ESEARCH D IRECTIONS
In this work, we provided a brief overview of the state-of-the-art UAV-IDS mechanisms. In addition, we discussedrelated design challenging issues to develop effective andefficient UAV-IDS mechanisms, considering high resource-constraints, high hostility characterized by sophisticated at-tack/threat behaviors, and distributed nature causing high secu-rity vulnerabilities. We also defined the taxonomies to describethe key components of UAV-IDS systems based on the state-of-the-art existing works. Lastly, we discussed key researchchallenges that should be considered for future research plans,aiming to build an affordable, secure cyber-physical UAV-IDSsystem.s future work directions, we plan to conduct the following: • Define an attack model that captures key attackbehaviors targeting for UAV-IDS systems . We willderive an attack graph and build a set of correspondingcountermeasures to deal with those attacks. • Develop a behavior rule specification-based UAV-IDS that uses minimum memory while maximizing detectionaccuracy by checking the anomaly of an observed behav-ior. Formal verification and Bayesian estimation basedground truth check for anomaly behaviors can be used tovalidate the developed set of specification rules. • Measure the effectiveness and efficiency of the devel-oped lightweight UAV-IDS using the metrics of agilityor resilience. A
CKNOWLEDGMENT
This work was supported by the Institute for Information &communications Technology Promotion (IITP) grant fundedby the Korea government (MSIP) (2015-0-00508, Develop-ment of Operating System Security Core Technology for theSmart Lightweight IoT Devices). This work is also supportedin part by the U.S. AFOSR under grant number FA2386-17-1-4076. Dr. Ilsun You is the corresponding author.R
EFERENCES [1] M. A. Aydın, A. H. Zaim, and K. G. Ceylan, “A hybrid intrusiondetection system design for computer network security,”
Com-puters & Electrical Engineering , vol. 35, no. 3, pp. 517–526,2009.[2] E. Biermann, E. Cloete, and L. M. Venter, “A comparison ofintrusion detection systems,”
Computers & Security , vol. 20,no. 8, pp. 676–683, 2001.[3] E. P. Blasch, J. J. Salerno, and G. P. Tadda, “Measuringthe worthiness of situation assessment,” in
Proceedings of the2011 IEEE National Aerospace and Electronics Conference(NAECON) , 2011, pp. 87–94.[4] H. Debar, M. Dacier, and A. Wespi, “Towards a taxonomy ofintrusion-detection systems,”
Computer Networks , vol. 31, no. 8,pp. 805–822, 1999.[5] P. Garcia-Teodoro, J. Diaz-Verdejo, G. Maci´a-Fern´andez, andE. V´azquez, “Anomaly-based network intrusion detection: Tech-niques, systems and challenges,”
Computers & Security , vol. 28,no. 1-2, pp. 18–28, 2009.[6] D. He, S. Chan, and M. Guizani, “Communication securityof unmanned aerial vehicles,”
IEEE Wireless Communications ,vol. 24, no. 4, pp. 134–139, 2017.[7] A. Y. Javaid, W. Sun, V. K. Devabhaktuni, and M. Alam, “Cybersecurity threat analysis and modeling of an unmanned aerialvehicle system,” in , 2012, pp. 585–590.[8] C. Kwon, S. Yantek, and I. Hwang, “Real-time safety as-sessment of unmanned aircraft systems against stealthy cyberattacks,”
Journal of Aerospace Information Systems , vol. 13,no. 1, pp. 27–45, 2015.[9] A. P. Lauf and W. H. Robinson, “Fault-tolerant distributedreconnaissance,” in
IEEE Military Communications Conference(MILCOM’2010) , 2010, pp. 1812–1817.[10] A. P. Lauf, R. A. Peters, and W. H. Robinson, “A distributedintrusion detection system for resource-constrained devices inad-hoc networks,”
Ad Hoc Networks , vol. 8, no. 3, pp. 253–266,2010. [11] R. Mitchell and I. R. Chen, “Adaptive intrusion detection ofmalicious unmanned air vehicles using behavior rule specifica-tions,”
IEEE Transactions on Systems, Man, and Cybernetics:Systems , vol. 44, no. 5, pp. 593–604, 2014.[12] D. Muniraj and M. Farhood, “A framework for detection ofsensor attacks on small unmanned aircraft systems,” in , 2017, pp. 1189–1198.[13] G. Pajares, “Overview and current status of remote sensingapplications based on unmanned aerial vehicles (uavs),”
Pho-togrammetric Engineering & Remote Sensing , vol. 81, no. 4,pp. 281–329, 2015.[14] A. Patcha and J.-M. Park, “An overview of anomaly detectiontechniques: Existing solutions and latest technological trends,”
Computer networks , vol. 51, no. 12, pp. 3448–3470, 2007.[15] H. Sedjelmaci, S. M. Senouci, and N. Ansari, “Intrusion de-tection and ejection framework against lethal attacks in UAV-aided networks: A bayesian game-theoretic methodology,”
IEEETransactions on Intelligent Transportation Systems , vol. 18,no. 5, pp. 1143–1153, 2017.[16] H. Sedjelmaci, S. M. Senouci, and M.-A. Messous, “How todetect cyber-attacks in unmanned aerial vehicles network?”in
IEEE Global Communications Conference (GLOBECOM) ,2016, pp. 1–6.[17] H. Sedjelmaci, S. M. Senouci, and N. Ansari, “A hierarchicaldetection and response system to enhance security against lethalcyber-attacks in uav networks,”
IEEE Transactions on Systems,Man, and Cybernetics: Systems , 2017.[18] V. Sharma and R. Kumar, “Teredo tunneling-based securetransmission between UAVs and ground ad hoc networks,”
International Journal of Communication Systems , vol. 30, no. 7,2017.[19] V. Sharma, M. Bennis, and R. Kumar, “UAV-assisted heteroge-neous networks for capacity enhancement,”
IEEE Communica-tions Letters , vol. 20, no. 6, pp. 1207–1210, 2016.[20] V. Sharma, R. Sabatini, and S. Ramasamy, “UAVs assisteddelay optimization in heterogeneous wireless networks,”
IEEECommunications Letters , vol. 20, no. 12, pp. 2526–2529, 2016.[21] V. Sharma, R. Kumar, and R. Kumar, “QUAT-DEM:Quaternion-DEMATEL based neural model for mutual coor-dination between UAVs,”
Information Sciences , vol. 418, pp.74–90, 2017.[22] V. Sharma, R. Kumar, K. Srinivasan, and D. N. K. Jayakody,“Coagulation attacks over networked UAVs: concept, chal-lenges, and research aspects,” in
International Conference onCommunication, Management and Information Technology (IC-CMIT) . Warsaw, Poland: IEEE, 2017, pp. 1–5.[23] D. Shen, G. Chen, E. Blasch, and G. Tadda, “Adaptive markovgame theoretic data fusion approach for cyber network de-fense,” in
IEEE Military Communications Conference (MIL-COM 2007) , 2007, pp. 1–7.[24] D. Shen, G. Chen, J. B. Cruz, and E. Blasch, “A game theoreticdata fusion aided path planning approach for cooperative UAVISR,” in , 2008, pp. 1–9.[25] R. Trafton and S. V. Pizzi, “The joint airborne network servicessuite,” in
IEEE Military Communications Conference (MIL-COM’2006) , 2006, pp. 1–5.[26] C.-Y. Tseng, P. Balasubramanyam, C. Ko, R. Limprasittiporn,J. Rowe, and K. Levitt, “A specification-based intrusion de-tection system for AODV,” in