Survey and Analysis of Digital Twin Integration for Network and Service Optimization in Vehicular Edge Computing

Abstract

In recent years, Vehicular Edge Computing (VEC) has gained attention as a key approach to address the complexity caused by the fusion of diverse vehicular applications. It has emerged as a promising paradigm for future intelligent transportation systems. VEC facilitates computation-intensive and latency-sensitive vehicular applications by providing computing and caching capabilities near vehicles. This improves transmission efficiency and lowers congestion. VEC is nonetheless susceptible to implementation challenges due to the highly dynamic nature of vehicular networks, which have characteristics of high mobility, opportunistic connectivity, and heterogeneous user demand. This complicates network and service management. In this line of reasoning, Digital Twin (DT) technology, which provides virtual models of objects, processes, and attributes, enables intelligent decision-making in management. The paper conducts a systematized review and comparative analysis of up-to-date literature that combines DT with VEC systems for network and service optimization. Our review highlights key methodologies, including DRL-assisted DT architectures, multi-agent offloading scenarios, and edge-cloud collaboration protocols. We summarize pioneer studies, indicate prevalent resource control and predictive modeling trends, and note common weaknesses such as scalability and data synchronization. The study explores the concept of DT, its applicability in various industries, and its potential for vehicular network modeling and simulation. Apart from that, we also discuss existing research trends, identify challenges such as scalability and real-time data acquisition, and introduce potential avenues for future research.