Machine Learning to Improve Multi-Hop Searching and Extended Wireless Reachability in V2X

Abstract

Multi-hop relay selection is a critical issue in vehicleto-everything networks. In previous works, the optimal hopping strategy is assumed to be based on the shortest distance. This study proposes a hopping strategy based on the lowest propagation loss, considering the effect of the environment. We use a twostep machine learning routine: improved deep encoder-decoder architecture to generate environmental maps and Q-learning to search for the multi-hopping path with the lowest propagation loss. Simulation results show that our proposed method can improve environmental recognition and extend the reachability of multi-hop communications by up to 66.7%, compared with a shortest-distance selection.

A Blockchain Based Federated Learning for Message Dissemination in Vehicular Networks

“…However, fuzzy logic is also dependent on thresholds and weights to be set in the rule base for making inferences. In [5], satellite images are used to detect buildings and obstacles to enable machine learning driven channel characterization. The path with lowest propagation loss is used for message dissemination in [5].…”

“…Optimal relay selection mechanisms result in better coverage, more reliable connectivity and less communication overhead [3]. Various intelligent relay selection schemes depending on a vehicle's distance from predecessor, moving direction, speed and propagation loss in environment have been proposed using fuzzy logic [4] or machine learning algorithms [5]. Existing literature shows improved packet delivery ratio by machine learning algorithms in multi-hop V2V communications [6].…”

Lateral confinement of high‐impedance surface‐waves through reinforcement learning

“…Therefore, the optimal SWG design policy , can be obtained following the Bellman's optimality criterion as follows: where the transition probability from s to when action a is taken, is represented as . Our proposal estimates this value that changes the patch sizes as a Q ‐learning task [9 ]. For the SWG design policy , the Q ‐value that maps the dimension‐state of each row of patches to the action of increasing or reducing its size ( s,a ), is defined as the expected discounted reward of taking the action a in the per‐row dimension state s , according to the design policy (5 ).…”

“…where the transition probability from s to s ′ when action a is taken, is represented as P s, a (s ′ ). Our proposal estimates this value P s, a (s ′ ) that changes the patch sizes as a Q-learning task [9]. For the SWG design policy p, the Q-value that maps the dimension-state of each row of patches to the action of increasing or reducing its size (s,a), is defined as the expected discounted reward of taking the action a in the per-row dimension state s, according to the design policy p (5).…”

Accelerating wireless channel autoencoders for short coherence-time communications

“…These smart communication systems rely on various detection, classification, and prediction tasks, such as signal detection and signal type identification for spectrum sensing. To address these tasks, DL provides powerful automated means for communication systems to learn from spectrum data and adapt to its dynamics [5]- [7]. Wireless communications data come in large volumes at high rates, and are subject to interference and security threats due to the shared nature of the medium [8], [9].…”