Graph Reinforcement Learning-based CNN Inference Offloading in Dynamic Edge Computing
It addresses efficient and accurate inference offloading for edge computing systems, but is incremental as it builds on existing early-exit and reinforcement learning approaches.
This paper tackles the problem of offloading CNN inference tasks in dynamic edge computing networks by proposing a graph reinforcement learning-based early-exit mechanism (GRLE) to handle uncertainties in communication and server capacity, achieving up to 3.41x higher average accuracy compared to baseline methods.
This paper studies the computational offloading of CNN inference in dynamic multi-access edge computing (MEC) networks. To address the uncertainties in communication time and Edge servers' available capacity, we use early-exit mechanism to terminate the computation earlier to meet the deadline of inference tasks. We design a reward function to trade off the communication, computation and inference accuracy, and formulate the offloading problem of CNN inference as a maximization problem with the goal of maximizing the average inference accuracy and throughput in long term. To solve the maximization problem, we propose a graph reinforcement learning-based early-exit mechanism (GRLE), which outperforms the state-of-the-art work, deep reinforcement learning-based online offloading (DROO) and its enhanced method, DROO with early-exit mechanism (DROOE), under different dynamic scenarios. The experimental results show that GRLE achieves the average accuracy up to 3.41x over graph reinforcement learning (GRL) and 1.45x over DROOE, which shows the advantages of GRLE for offloading decision-making in dynamic MEC.