Trust management in a decentralized vehicular network, such as vehicular ad hoc network, is particularly challenging due to the lack of centralized communication infrastructure, and a fast varying feature of vehicular environment. In this paper, we propose a decentralized trust management scheme for vehicular networks. The proposed protocol uses a fuzzy logicbased trust calculation scheme to evaluate the direct trust where trustee nodes are located within the transmission range of a trustor node. A reinforcement learning-based approach is also employed to estimate the indirect trust where the behaviors of trustee cannot be observed directly. The extensive simulations are conducted to show the advantage of the proposed scheme over other baseline approaches.

Space-assisted vehicular networks (SAVN) provide seamless coverage and on-demand data processing services for user vehicles (UVs). However, ultra-reliable and low-latency communication (URLLC) demands imposed by emerging vehicular applications are hard to be satisfied in SAVN by existing computation offloading techniques. Traditional deep reinforcement learning algorithms are unsuitable for highly dynamic SAVN due to the underutilization of environment observations. An AsynchronouS federaTed deep Q-learning (DQN)-basEd and URLLC-aware cOmputatIon offloaDing algorithm (ASTEROID) is presented in this paper to achieve throughput maximization considering the long-term URLLC constraints. Specifically, we first establish an extreme value theory-based URLLC constraint model. Second, the task offloading and computation resource allocation are decomposed by employing Lyapunov optimization. Finally, an asynchronous federated DQN-based (AF-DQN) algorithm is presented to address the UV-side task offloading problem. The server-side computation resource allocation is settled by an queue backlog-aware algorithm. Simulation results verify that ASTEROID achieves superior throughput and URLLC performances.

Existing task offloading mechanisms are developed on some single and rigid quality of service (QoS) performance metrics, which is widely apart from satisfying the true intent of a user vehicle (UV), thereby resulting in low quality of experience (QoE), large queuing latency, and poor reliability. There is an unprecedented demand for an intent-aware task offloading strategy that provides improved QoE and guarantees reliability. In this paper, we develop a novel task offloading framework for air-ground integrated vehicular edge computing (AGI-VEC), which is called the learning-based Intent-aware Upper Confidence Bound (IUCB) algorithm. IUCB enables a UV to learn the long-term optimal task offloading strategy while satisfying the long-term ultra-reliable low-latency communication (URLLC) constraints in a best effort way under information uncertainty. IUCB can achieve three-dimension intent awareness including QoE awareness, URLLC awareness, and trajectory similarity awareness. Simulation results demonstrate that IUCB significantly outperforms existing EMM, sleeping-UCB, and UCB mechanisms in terms of QoE, end-to-end delay, queuing delay, throughput, and times of task offloading failure.

By running applications and services closer to the user, edge processing provides many advantages, such as short response time and reduced network traffic. Deep-learning based algorithms provide significantly better performances than traditional algorithms in many fields but demand more resources, such as higher computational power and more memory. Hence, designing deep learning algorithms that are more suitable for resource-constrained mobile devices is vital. In this paper, we build a lightweight neural network, termed LiteNet which uses a deep learning algorithm design to diagnose arrhythmias, as an example to show how we design deep learning schemes for resource-constrained mobile devices. Compare to other deep learning models with an equivalent accuracy, LiteNet has several advantages. It requires less memory, incurs lower computational cost, and is more feasible for deployment on resource-constrained mobile devices. It can be trained faster than other neural network algorithms and requires less communication across different processing units during distributed training. It uses filters of heterogeneous size in a convolutional layer, which contributes to the generation of various feature maps. The algorithm was tested using the MIT-BIH electrocardiogram (ECG) arrhythmia database; the results showed that LiteNet outperforms comparable schemes in diagnosing arrhythmias, and in its feasibility for use at the mobile devices.

Integrating edge computing and non-orthogonal multiple access (NOMA) into the power Internet of Things (PIoT) provides a promising way to meet the stringent demands of communication delay, energy consumption, and massive connectivity. The service tasks can be partitioned into independent subtasks, and can be processed locally by PIoT devices or offloaded to the edge servers through reusing radio resource blocks (RBs). In this paper, we propose a Multi-timescale multi-dimEnsion Resource allocatIon and Task Splitting algorithm, namely MERITS, for NOMA-edge computing-based PIoT. First, Lyapunov optimization is exploited to decompose the long-term stochastic optimization problem into three short-term deterministic subproblems, i.e., RB allocation in a large timescale, task splitting and computation resource allocation in a small timescale. The first subproblem is modeled as a one-to-many matching problem with externalities, and solved by the proposed swap matching-based RB allocation algorithm. The second subproblem is relaxed into a continuous convex problem which is easily to be solved, and the third subproblem is solved by Lagrange dual decomposition. Simulation results demonstrate that MERITS outperforms the existing resource allocation and task splitting algorithms in terms of energy consumption, queuing delay, queue backlog, and connection success ratio in the massive connectivity scenario.

In recent years, Vehicle-to-Vehicle (V2V) and Vehicle-to-Infrastructure (V2I) communication brings more and more attention from industry (e.g., Google and Uber) and government (e.g., United States Department of Transportation). These Vehicle-to-Everything (V2X) technologies are widely adopted in future autonomous vehicles. However, security issues have not been fully addressed in V2V and V2I systems, especially in key distribution and key management. The physical layer key generation, which exploits wireless channel reciprocity and randomness to generate secure keys, provides a feasible solution for secure V2V/V2I communication. It is lightweight, flexible, and dynamic. In this paper, the physical layer key generation is brought to the V2I and V2V scenarios. A LoRa-based physical key generation scheme is designed for securing V2V/V2I communications. The communication is based on Long Range (LoRa) protocol, which is able to measure Received Signal Strength Indicator (RSSI) in long-distance as consensus information to generate secure keys. The multi-bit quantization algorithm, with an improved Cascade key agreement protocol, generates secure binary bit keys. The proposed schemes improved the key generation rate, as well as to avoid information leakage during transmission. The proposed physical layer key generation scheme was implemented in a V2V/V2I network system prototype. The extensive experiments in V2I and V2V environments evaluate the efficiency of the proposed key generation scheme. The experiments in real outdoor environments have been conducted. Its key generation rate could exceed 10 bit/s on our V2V/V2I network system prototype and achieve 20 bit/s in some of our experiments. For binary key sequences, all of them pass the suite of statistical tests from National Institute of Standards and Technology (NIST).