Cooperative Perception in Connected Vehicles: Integrating Sensing and Communication for IoT-Based Autonomous Driving

Cooperative perception (CP) is poised to revolutionize autonomous driving by creating a collective sensory system. However, its real-world viability is critically dependent on the performance of the underlying V2X communication network. A persistent gap in research is the tendency to evaluate perception and communication in isolation, leading to overly optimistic performance estimates that ignore the feedback loop where network congestion degrades perception, and vice-versa. This paper addresses this gap by providing a systematic, empirical quantification of a critical phenomenon: the performance collapse cliff in CP systems. Using B5GCyberTestV2X, a framework that ensures tight bidirectional synchronization between traffic (CARLA) and 5G network (OMNeT++/SIMU5G) simulators, this work demonstrates that the theoretical benefits of sharing rich sensor data are not just diminished but completely nullified by the network degradation they induce. The results show that while a lightweight fused object strategy achieves a 2.1× perception gain with sub-20 ms latency, heavier strategies trigger a collapse where the Age of Information (AoI) exceeds 300 ms, rendering the cooperative data useless. ... mehrThis work’s primary contribution is the empirical characterization of this tipping point, proving that the communication channel is the dominant bottleneck and arguing for a paradigm shift towards communication-aware adaptive perception algorithms.