DIVANet'18- Proceedings of the 8th ACM Symposium on Design and Analysis of Intelligent Vehicular Networks and Applications

Full Citation in the ACM Digital Library

SESSION: Software-Defined Vehicular Networks

A Deep Reinforcement Learning-based Trust Management Scheme for Software-defined Vehicular Networks

Vehicular ad hoc networks (VANETs) have become a promising technology in intelligent transportation systems (ITS) with rising interest of expedient, safe, and high-efficient transportation. VANETs are vulnerable to malicious nodes and result in performance degradation because of dynamicity and infrastructure-less. In this paper, we propose a trust based dueling deep reinforcement learning approach (T-DDRL) for communication of connected vehicles, we deploy a dueling network architecture into a logically centralized controller of software-defined networking (SDN). Specifically, the SDN controller is used as an agent to learn the most trusted routing path by deep neural network (DNN) in VANETs, where the trust model is designed to evaluate neighbors' behaviour of forwarding routing information. Simulation results are presented to show the effectiveness of the proposed T-DDRL framework.

Blockchain-Based Distributed Software-Defined Vehicular Networks via Deep Q-Learning

Nowadays, in order to support flexibility, agility, and ubiquitous accessibility among vehicles, software defined networking has been proposed to integrate with vehicular networks, known as software defined vehicular network (SDVN). Due to a variety of data, flows, and vehicles in SDVN, a distributed SDVN is necessary. However, how to reach consensus in distributed SDVN efficiently and safely is an intractable problem. In this paper, we use a permissioned blockchain approach to reach consensus in distributed SDVN. The existing permissioned blockchain has a number of drawbacks, such as low throughput. We virtualize the underlying resources (e.g., computing resources and networking resources), jointly considering the trust features of blockchain nodes to improve the throughput. Accordingly, we formulate view change, computing resources allocation, and networking resources allocation as a joint optimization problem. In order to solve this joint problem, we use a novel deep Q-learning approach. Simulation results show the effectiveness of our proposed scheme.

Distributed SDN Controller Placement Using Betweenness Centrality & Hierarchical Clustering

Software-defined networking (SDN) separates the control plane from the data plane. This simplifies network management and provides flexibility to the network administrator. Such an architecture can be implemented in various types of networks including wide-area networks and vehicular networks. Two different approaches are possible for the control plane, namely the centralized controller approach and the distributed multiple controllers approach. For efficient network management in a large distributed network, the location and number of controllers should be optimized to provide the service providers desired system performance. This paper proposes a new framework that solves the SDN controller placement problem by combining both the hierarchical clustering and betweenness centrality concepts (denoted as HC-BC). The performance of the proposed framework is evaluated using a real-world network and compared to three other algorithms in terms of worst-case switch-to-controller latency and domain imbalance. The simulation results show that the proposed HC-BC framework achieves the best compromise between the latency and the domain imbalance between different clusters, with the added advantage of having lower computational complexity.

SESSION: Routing and Content Delivery

Information-Centric Strategies for Content Delivery in Intelligent Vehicular Networks

Efficient content delivery in vehicular networks will play a fundamental role in empowering envisioned vehicular and smart transportation applications. However, the peculiar characteristics of vehicular networks (e.g., high mobility, dynamic network topologies, and intermittent connectivity), as well as the string QoS requirements of the applications, challenge the design of efficient solutions for content delivery in these environments. In this paper, we discuss the recent trend of applying the information-centric networking (ICN) paradigm for content delivery in vehicular networks. By doing so, we highlight the potentials of this novel paradigm and how it can deal with the severe challenges of vehicular networks. Besides, we point out the limitations of current ICN implementations when applied in vehicular networks. Furthermore, we discuss the new challenges that need be tackled in ICN-based vehicular networks, the proposed solutions encountered in the literature and, based on that, we provide some guidelines for the design of new solutions and some future research directions.

Adaptive Forwarding Control using Network Coding for Efficient Multicasting in Mobile Ad-hoc Networks

In mobile ad-hoc networks (MANETs) composed of mobile devices, network resources such as battery capacity and bandwidth are limited. Therefore, in such networks, it is particularly important to reduce unnecessary packet transfers and the use of network resources and power to ensure efficient multicast communications. In this study, we apply network coding to conventional flooding in order to achieve efficient and reliable multicasting in MANETs. Network coding makes it possible to reduce the number of packets transmitted over the network and to use devices and network resources effectively. The packet-reduction benefits obtained using network coding depend greatly on the network topology. Therefore, we propose a novel method called adaptive forwarding control, which uses network coding and flooding for efficient multicasting. Our proposed scheme switches between conventional flooding and network coding, depending on the topology, to derive the benefits from both flooding and network coding.

An Energy-efficient UAV-based Data Aggregation Protocol in Wireless Sensor Networks

Energy efficiency is an important issue in Wireless Sensor Networks (WSNs). The energy is mainly consumed by data sensing, data transmission and movement of sensors. The energy consumed by data transmission is much larger than data sensing. A potential solution for energy saving is applying the external devices to collect data to prolong the network lifetime. In this paper, we adopt an unmanned aerial vehicle (UAV) serving as the data mule and propose a novel energy-efficient UAV-based data aggregation protocol in WSNs to reduce the energy consumption of sensors. By considering a clustered WSN, our approach computes an optimal path for data mule through all cluster heads (CHs) while achieving a relatively high system-wide energy efficiency. Moreover, we introduce a genetic algorithm to derive a near-optimal solution for a large-scale WSN while reducing the computing time. We compare our protocol with state-of-the-art approaches, and the simulation results demonstrate that the proposed algorithm improves the network performance on energy efficiency.

SESSION: Quality of Service

Multi-objective Approaches to Improve QoS in Vehicular Ad-hoc Networks

n view of the need for new optimization techniques capable ofincreasing the reliability of the Inter-vehicular CommunicationSystems (IVC), indispensable for the development of IntelligentTransportation Systems (ITS), this work proposes the use of a meta-heuristic based on evolutionary computation, to optimize parame-ters in the MAC-layer in search of the best throughput, latency andpacket loss values, in an urban scenario. These approaches showedup to 91% of reduction in the time needed to find the best MAC-layer configurations for multi-objective optimization of throughput,latency and packet loss, when compared to the use of exhaustivesearch.

Congestion Mitigation in Densely Crowded Environments for Augmenting QoS in Vehicular Clouds

Parking lots in densely crowded environments such as stadiums, theaters, and hospitals provide great opportunities for vehicular cloud services. A cloud environment formed by individual vehicles, where each vehicle offers its resources as a service has shown feasible practices in 5G network scenarios. Moreover, resource management in 5G must be achieved in accordance with user-centric QoS requirements. In alignment with this, a key enabler of the user-centric service scheme is Network Slicing. The formation of multiple slices in such a dense environment, the congestion between sender and receiver, and resource management and allocation are topics of current research. This paper has the following contribution: First, a framework of Vehicular Clouds being restricted to individual slices in 5G cellular networks is proposed. Second, a queuing strategy for congestion control in a densely crowded environment such as parking lots is designed. Finally, a resource allocation algorithm that enables maximum matching between the tasks to be executed and the candidate slices is developed. The novelty of this approach comes from the fact that congestion control is performed at the Access Points (AP). We do this by introducing a control module that makes queuing decisions at the time of request arrival. By incorporating control module in AP, our aim is to provide AP resources in terms of transmission period to different slices, thereby, allowing WiFi resources to be shared along with the 5G radio resources. The performance benefits of the proposed solution has been investigated through simulation tests.

An Adaptive Power Level Control Algorithm for DSRC Congestion Control

Vehicular industries and researchers have invested efforts to reduce avoidable accidents through the means of Vehicle to Vehicle (V2V) wireless communication using Vehicular Ad Hoc Networks (VANETs). Up-to-date information on the location, speed and other important parameters for each vehicle is shared with neighboring vehicles through the periodic exchange of Basic Safety Messages (BSMs). With a high vehicular density, network congestion can quickly arise in the 5.9GHz spectrum, rendering the system as unreliable because safety messages are not delivered and received on time. To alleviate this problem, there has been considerable research, in recent years, on distributed congestion control algorithms for VANETs. These approaches are generally based on rate control or power control of the transmitted packets. In this paper, we propose a novel, adaptive power control algorithm to reduce packet congestion in VANETs. Instead of requiring all vehicles to use the same power level, our approach allows each vehicle to dynamically adjust the transmit power of BSM packets, depending on its current speed. The goal is to prioritize which other vehicles will receive BSM packets from a given vehicle. Our simulation results demonstrate the advantages of the proposed algorithm regarding commonly used metrics such as packet loss, Beacon Error Rates(BER), Channel Busy Time(CBT) and Inter-Packet Delay(IPD).

SESSION: Security and Applications

Optimized Anonymity Updating in VANET Based on Information and Privacy Joint Metrics

With the continuous development of the vehicular ad hoc network (VANET), many challenges related to network security have come one after another, among which privacy issues are particularly prominent. To help each network user decide when and where to protect their privacy, we suggest creating a user-centric privacy computing system in VANET. A risk assessment function and a set of decision weights are proposed to simulate the driver's decision-making intent in the vehicle network. Besides, proposed information and privacy joint metrics are used as the key indicators for dynamic selection of Mix-zone. Finally, by considering three influencing factors: maximum road capacity, user-centric quantitative privacy and attacker information measurement, defined mixzone creation mechanism to achieve privacy protection in VANET.

Energy Trade-offs in Millimeter Wave Vehicular Networks by exploiting a Hyperfractal Model

We present results on the trade-offs between the end-to-end communication delay and the energy spent in millimeter wave vehicular communications in urban settings. This study exploits the self-similarities of the traffic and vehicle locations in cities captured in our innovative model called "hyperfractal''. We enrich the model by incorporating road-side infrastructure.

We use analytical tools to derive theoretical bounds for the end-to-end communication hop count under two different energy constraints: either total accumulated energy, or maximum energy per node. More precisely, we prove that the hop count is bounded by O(n^1-α/(d_m-1) ) where α<1 and d_m>2 is the precise hyperfractal dimension. This proves that for both constraints the energy decreases as we allow to chose among paths of larger length. In fact, the asymptotic limit of the energy becomes significantly small when the number of nodes becomes asymptotically large. A lower bound on the network throughput capacity with constraints on path energy is also given. The results are confirmed through simulations using different fractal dimensions and path loss coefficients.

PlaSA - Platooning Service Architecture

Platooning is a solution that allows automated vehicles to travel very close to each other, enabling enhancements in terms of safety, traffic flow and highway capacities. Furthermore, it provides a more convenient and comfortable driving experience to the drivers. The research on Platooning covers a very large spectrum of areas, ranging from communication strategies to controller solutions. However, most of the proposed works tend to present very specific solutions, treating platooning almost as a "black-box". Thus, it is imperative to find solutions that allow a proper assembly of every component that is crucial for a proper behavior of the system as a whole. This paper presents an open and modular architecture for platooning as a service, discussing its general assumptions and the logical and functional models. Additionally, some related preliminary results are also presented.