SMPPO: A Shared Memory Scheduling Approach for Real-Time Heterogeneous Serverless Edge Environment

Serverless edge computing is transforming distributed systems, especially for IoT, by merging serverless architecture with edge computing. This evolution demands efficient scheduling methods, as traditional approaches struggle with dynamic and heterogeneous environments. Deep Reinforcement Learning (DRL), particularly the Proximal Policy Optimization (PPO) algorithm, offers a promising alternative. Our proposed method, SMPPO, introduces a global trajectory memory where agents learn from both their own and others' experiences. Simulations show SMPPO consistently outperforms others, improving total performance by 15% and response time by 25%. This robust, scalable approach advances scheduling in serverless edge computing.