Indoor Heterogeneous Multi-Access Edge Computing Systems: Online Learning for Channel Variation-Aware Task Offloading

We investigate task offloading in indoor heterogeneous multi-access edge computing (MEC) systems with cellular and WiFi networks. Due to unpredictable mobile device mobility and spatially varying multipath fading, MEC systems face time-varying wireless channel conditions, making it challenging to make deterministic task offloading decisions. We propose an online learning-based task offloading decision algorithm that enables mobile devices to learn spatially varying channel conditions and optimize task offloading policy over time. Our algorithm minimizes the energy consumption of each mobile device while ensuring maximum task offloading delay guarantees. Numerical simulation results demonstrate the effectiveness of our algorithm.