Joint Resource Allocation and Cache Placement for Location-Aware Multi-User Mobile-Edge Computing

With the growing demand for latency-critical and computation-intensive Internet of Things (IoT) services, IoT-oriented network architecture, mobile-edge computing (MEC), has emerged as a promising technique to reinforce computation capability resource-constrained IoT devices. To exploit cloud-like functions at edge, service caching been implemented reuse task input/output data, thus effectively reducing delay incurred by data retransmissions repeated execution same task. In multiuser cache-assisted MEC system, users’ preferences different types possibly dependent on their locations, play an important role in joint design communication, computation, caching. this article, we consider multiple representative where users location share preference profile given set services. Specifically, exploiting location-aware profiles, propose optimization binary cache placement, edge resource, bandwidth (BW) allocation minimize expected sum-energy consumption, subject BW limitations well latency constraints. solve mixed-integer nonconvex problem, deep learning (DL)-based offline placement scheme using novel stochastic quantization-based discrete-action generation method. The proposed hybrid framework advocates both benefits from model-free DL approach model-based optimization. simulations verify that DL-based saves roughly 33% 6.69% energy consumption compared with greedy popular caching, respectively, while achieving up 99.01% optimal performance.