DOC: Delay-Optimized Caching for Cold Start Mitigation in Vehicular Edge Serverless Computing

This paper addresses cold start latency in serverless edge computing for Vehicle-Infrastructure Collaboration (VIC) environments. We formulate the container caching problem by introducing Request Function Chains (RFCs) and proving its NP-hardness. We propose the Delay-Optimized Caching (DOC) algorithm, modeling the problem as an Integer Nonlinear Programming formulation, linearizing it to ILP, and providing theoretical performance guarantees through competitive ratio analysis. Experiments using both synthetic and real trajectory-driven scenarios demonstrate that DOC reduces total system cost by 25.92% compared to baselines. Our algorithm effectively balances efficiency and resource utilization, significantly improving performance for delay-sensitive vehicular applications.