While mobile edge computing (MEC) alleviates the computation and powerlimitations of mobile devices, additional latency is incurred when offloadingtasks to remote MEC servers. In this work, the power-delay tradeoff in thecontext of task offloading is studied in a multi-user MEC scenario. In contrastwith current system designs relying on average metrics (e.g., the average queuelength and average latency), a novel network design is proposed in whichlatency and reliability constraints are taken into account. This is done byimposing a probabilistic constraint on users' task queue lengths and invokingresults from extreme value theory to characterize the occurrence oflow-probability events in terms of queue length (or queuing delay) violation.The problem is formulated as a computation and transmit power minimizationsubject to latency and reliability constraints, and solved using tools fromLyapunov stochastic optimization. Simulation results demonstrate theeffectiveness of the proposed approach, while examining the power-delaytradeoff and required computational resources for various computationintensities.
展开▼