Spatial Intelligence toward Trustworthy Vehicular IoT

TLDR

Spatial challenges for the vehicular Internet of Things arise from mobility, high density, sparse connectivity, and heterogeneity. The study proposes two techniques—decentralized moving edge and multi‑tier multi‑access edge clustering—to address these spatial challenges. The authors implement a decentralized moving‑edge architecture where vehicles act as edge nodes performing distributed communication, caching, and computing, and a multi‑tier multi‑access edge clustering scheme that integrates licensed and unlicensed links, using fuzzy logic and Q‑learning to balance conflicting metrics and enable self‑evolving operation. Simulations demonstrate that the proposed protocols outperform existing alternatives, while highlighting several open research problems.

Abstract

Spatial challenges for the vehicular Internet of Things come from mobility, high density, sparse connectivity, and heterogeneity. In this article, we propose two techniques, namely decentralized moving edge and multi-tier multi-access edge clustering, to handle these challenges. The "vehicle as an edge" concept of the decentralized moving edge provides a more suitable solution to meet the throughput and latency performance requirements by conducting distributed communication, data caching, and computing tasks at vehicles. Multi-tier multi-access edge clustering generates different levels of clusters for more efficient integration of different types of access technologies including licensed/unlicensed long-range low-throughput communications and unlicensed short-range high-throughput communications. We employ fuzzy logic to jointly consider multiple inherently contradictory metrics and use Q-learning to achieve a self-evolving capability. Realistic computer simulations are conducted to show the advantage of the proposed protocols over alternatives, and several open research problems are discussed.