ASSGO

The need for enhanced safety and sustainability of mobility as a service is also addressing the promise of connected vehicles, edge computing, and AI. Providing all three of latency, energy, and security at scale in a heavily connected vehicular environment, remains a loose thread. This concern is addressed in this paper through the formulation of the ASSGO framework\u2014Adaptive Secure Slice and Green Offload, which primarily focuses on the integration of software-defined networking (SDN) and lightweight edge intelligence for optimal energy, latency, and security in Edge-AI vehicular surroundings. Defense strategies in the controller use a 1D Temporal Convolution Network (TCN) with a greedy multi-objective optimizer for green offloading and real-time anomaly detection, and adaptive SDN slicing to sustain service during adversarial events. In the testbed which is a combination of reproducible SUMO\u2013Mininet\u2013WiFi\u2013Ryu\u2013Docker, the application of ASSGO yields 38\\% lower latency, 27% lower energy, and 97% attack detection accuracy as compared to more recent SDN-based baselines validated across 10 independent runs with p<0.05. This shows the adequacy of composable SDN for the orchestration of Edge-AI transport systems in a manner that is secure and sustainably resilient.