Kimihiro Mizutani

Rethinking Packet Classification for Global Network View of Software-Defined Networking

In software-defined networking, applications are allowed to access a global view of the network so as to provide sophisticated functionalities, such as quality-oriented service delivery, automatic fault localization, and network verification. All of these functionalities commonly rely on a well-studied technology, packet classification. Unlike the conventional classification problem to search for the action taken at a single switch, the global network view requires to identify the network-wide behavior of the packet, which is defined as a combination of switch actions. Conventional classification methods, however, fail to well support network-wide behaviors, since the search space is complicatedly partitioned due to the combinations. This paper proposes a novel packet classification method that efficiently supports network-wide packet behaviors. Our method utilizes a compressed data structure named the multi-valued decision diagram, allowing it to manipulate the complex search space with several algorithms. Through detailed analysis, we optimize the classification performance as well as the construction of decision diagrams. Experiments with real network datasets show that our method identifies the packet behavior at 20.1 Mpps on a single CPU core with only 8.4 MB memory, by contrast, conventional methods failed to work even with 16 GB memory. We believe that our method is essential for realizing advanced applications that can fully leverage the potential of software defined networking.

State-of-the-Art Deep Learning: Evolving Machine Intelligence Toward Tomorrow’s Intelligent Network Traffic Control Systems

Currently, the network traffic control systems are mainly composed of the Internet core and wired/wireless heterogeneous backbone networks. Recently, these packet-switched systems are experiencing an explosive network traffic growth due to the rapid development of communication technologies. The existing network policies are not sophisticated enough to cope with the continually varying network conditions arising from the tremendous traffic growth. Deep learning, with the recent breakthrough in the machine learning/intelligence area, appears to be a viable approach for the network operators to configure and manage their networks in a more intelligent and autonomous fashion. While deep learning has received a significant research attention in a number of other domains such as computer vision, speech recognition, robotics, and so forth, its applications in network traffic control systems are relatively recent and garnered rather little attention. In this paper, we address this point and indicate the necessity of surveying the scattered works on deep learning applications for various network traffic control aspects. In this vein, we provide an overview of the state-of-the-art deep learning architectures and algorithms relevant to the network traffic control systems. Also, we discuss the deep learning enablers for network systems. In addition, we discuss, in detail, a new use case, i.e., deep learning based intelligent routing. We demonstrate the effectiveness of the deep learning-based routing approach in contrast with the conventional routing strategy. Furthermore, we discuss a number of open research issues, which researchers may find useful in the future.

Efficient Virtual Network Optimization Across Multiple Domains Without Revealing Private Information

Building optimal virtual networks across multiple domains is an essential technology for offering flexible network services. However, existing research is founded on an unrealistic assumption: providers will share their private information including resource costs. Providers, as well known, never actually do that so as to remain competitive. Secure multi-party computation, a computational technique based on cryptography, can be used to secure optimization, but it is too time consuming. This paper presents a novel method that can optimize virtual networks built over multiple domains efficiently without revealing any private information. Our method employs secure multi-party computation only for masking sensitive values; it can optimize virtual networks under limited information without applying any time-consuming techniques. It is solidly based on the theory of optimality and is assured of finding reasonably optimal solutions. Experiments show that our method is fast and optimal in practice, even though it conceals private information; it finds near optimal solutions in just a few minutes for large virtual networks with tens of nodes. This is the first work that can be implemented in practice for building optimal virtual networks across multiple domains.

An Overlay-Based Data Mining Architecture Tolerant to Physical Network Disruptions

Management scheme for highly scalable big data mining has not been well studied in spite of the fact that big data mining provides many valuable and important information for us. An overlay-based parallel data mining architecture, which executes fully distributed data management and processing by employing the overlay network, can achieve high scalability. However, the overlay-based parallel mining architecture is not capable of providing data mining services in case of the physical network disruption that is caused by router/communication line breakdowns because numerous nodes are removed from the overlay network. To cope with this issue, this paper proposes an overlay network construction scheme based on node location in physical network, and a distributed task allocation scheme using overlay network technology. The numerical analysis indicates that the proposed schemes considerably outperform the conventional schemes in terms of service availability against physical network disruption.

Routing or Computing? The Paradigm Shift Towards Intelligent Computer Network Packet Transmission Based on Deep Learning

Recent years, Software Defined Routers (SDRs) (programmable routers) have emerged as a viable solution to provide a cost-effective packet processing platform with easy extensibility and programmability. Multi-core platforms significantly promote SDRs’ parallel computing capacities, enabling them to adopt artificial intelligent techniques, i.e., deep learning, to manage routing paths. In this paper, we explore new opportunities in packet processing with deep learning to inexpensively shift the computing needs from rule-based route computation to deep learning based route estimation for high-throughput packet processing. Even though deep learning techniques have been extensively exploited in various computing areas, researchers have, to date, not been able to effectively utilize deep learning based route computation for high-speed core networks. We envision a supervised deep learning system to construct the routing tables and show how the proposed method can be integrated with programmable routers using both Central Processing Units (CPUs) and Graphics Processing Units (GPUs). We demonstrate how our uniquely characterized input and output traffic patterns can enhance the route computation of the deep learning based SDRs through both analysis and extensive computer simulations. In particular, the simulation results demonstrate that our proposal outperforms the benchmark method in terms of delay, throughput, and signaling overhead.

The Deep Learning Vision for Heterogeneous Network Traffic Control: Proposal, Challenges, and Future Perspective

Recently, deep learning, an emerging machine learning technique, is garnering a lot of research attention in several computer science areas. However, to the best of our knowledge, its application to improve heterogeneous network traffic control (which is an important and challenging area by its own merit) has yet to appear because of the difficult challenge in characterizing the appropriate input and output patterns for a deep learning system to correctly reflect the highly dynamic nature of large-scale heterogeneous networks. In this vein, in this article, we propose appropriate input and output characterizations of heterogeneous network traffic and propose a supervised deep neural network system. We describe how our proposed system works and how it differs from traditional neural networks. Also, preliminary results are reported that demonstrate the encouraging performance of our proposed deep learning system compared to a benchmark routing strategy (Open Shortest Path First (OSPF)) in terms of significantly better signaling overhead, throughput, and delay.

On the fast-convergence of delay-based load balancing over multipaths for dynamic traffic environments

Multipath routing enables the source exploit multiple available paths to transfer data to destination. This technique has drawn much attention by efficiently utilizing the bandwidths, preserving packets order and so on. However, these load balancing schemes are not for the delay-related issue and thus unsuited for the real-time applications. To deal with the delay-sensitive features, a load balancing scheme named Effective Delay-Controlled Load Distribution (E-DCLD) has been proposed to lower the end-to-end delay and the associating packet reordering possibility. Nevertheless, to compute the optimal load for each path, this scheme uses gradually approaching method that needs extra convergence rounds, and performs unsatisfactory especially when path status is unstable. In this paper, we propose a Convex optimization-Based Method (CBM) to effectively figure out the best load ratio for each path based on the model of E-DCLD. The proposed method could count out the result at once and overcome the low convergence rate problem of the original solution. Experimental results demonstrate that our solution could significantly decrease the end-to-end packet delay and total packet delay.

An efficient framework for data-plane verification with geometric windowing queries

Modern networks have complex configurations to provide advanced functions, but the complexity also makes them error-prone. Network verification is attracting attention as a key technology to detect inconsistencies between a configuration and a policy before deployment. Existing verifiers, however, either generally verify various properties over the policy at the cost of efficiency, or efficiently perform configuration analysis without paying much attention to the policy. This paper presents a novel framework of data-plane verification, which flexibly checks the inconsistency with great efficiency. For the purpose of generality, our framework formalizes a verification process with three abstract steps: each step is related to 1) packet behaviors defined by a configuration, 2) operator intentions described in a policy, and 3) the inspection of their relation. These steps work efficiently with each other on the simple quotient set of packet headers. This paper also reveals how the second step can be regarded as the windowing query problem in computational geometry. Two novel windowing algorithms are proposed with solid theoretical analyses. Experiments on real network datasets show that our framework with the windowing algorithms is surprisingly fast even when verifying the policy compliance; e.g., in a medium-scale network with thousands of switches, our framework reduces the verification time of all-pairs reachability from ten hours to ten minutes.

Efficient virtual network optimization across multiple domains without revealing private information

Building optimal virtual networks across multiple domains is an essential technology to offer flexible network services. However, existing research is founded on an unrealistic assumption; providers will share their private information including resource costs. Providers, as is well known, never actually do that to remain competitive. Technically, secure multiparty computation, which is a computational technique based on the cryptography, can be used to secure optimization, but it is too time-consuming. This paper presents a novel method to optimize virtual networks built over multiple domains, with great efficiency but without revealing any private information. Our method employs secure multi-party computation but only for masking sensitive values; it can optimize virtual networks under limited information without any time-consuming technique. It is solidly based on the theory of optimality, and is assured of finding reasonably optimal solutions. Experiments show that our method is fast and optimal in practice even concealing private information; it finds nearly optimal solutions in just a few minutes for large virtual networks with tens of nodes. This is the first work that can be implemented in practice for building optimal virtual networks across multiple domains.

Towards a Low-Delay Edge Cloud Computing through a Combined Communication and Computation Approach

There are many applications which cannot be executed by mobile devices due to their limitations in memory, processing, battery, among others. One solution to this would be offloading heavy tasks to cloud servers in the edge of the network, in a service model called Edge Cloud Computing. The main Quality of Service requirement of this model is a low Service Delay, which can be achieved by lowering Transmission Delay and Processing Delay. Works in literature focus on either one of those two types of delay. This paper, however, argues that an approach which combines transmission and processing technologies to lower Service Delay would be more efficient. This idea is defended by an analysis of the service model and existing stochastic modeling of the Edge Cloud Computing system. We conclude that a dual focus approach would be the only way of truly minimizing the Service Delay, therefore being the desired method to improve Quality of Service. We conclude by laying the foundation for a future model that follows such concept.