Task Offloading for CAVs Edge Computing Environment: Taxonomy, Critical Review, and Future Road Map

The rapid advancement of Intelligent Transportation Systems (ITS) has led to a paradigm shift toward the adoption of Connected Autonomous Vehicles (CAVs). In recent years, CAVs have emerged as a prominent research focus due to their potential to reduce road traffic accidents caused by human error, optimize traffic flow, create new economic opportunities, and enhance travel convenience. However, the increasing demand for compute and delay-sensitive applications, such as real-time navigation and sensor data processing, exceeds the capabilities of current onboard vehicle resources. Consequently, task offloading has gained significant attention, allowing certain computational tasks generated by CAVs operations to be offloaded to external cloud or edge servers. The existing review literature has been limited in its focus on task offloading techniques specifically for CAVs architecture. Therefore, this study aims at presenting a comprehensive survey on task offloading in CAVs through a systematic review guided by key research questions. We first provide a technical background and then propose a broad coverage taxonomy of existing literature, analyzing promising solutions such as Machine Learning (ML) and heuristic-based techniques. In addition, we present a taxonomy of execution environments, metrics, and datasets. Finally, we highlight key research challenges and future trends, providing valuable insights for advancing task offloading in CAVs architecture.