Energy Optimization in Statistical AoI-Aware MEC Systems

Abstract

Mobile edge computing (MEC) -based computational offloading can help age-sensitive devices handle their tasks and reduce the age of information (AoI) of tasks. However, the inherent randomness of wireless channels makes it challenging to realize AoI provisioning for age-sensitive services in MEC systems. To address this issue, we propose a statistical AoI-aware MEC system that incorporates a stochastic network calculus (SNC)-based statistical AoI provisioning theoretical framework to support the tail distribution analysis of AoI. Particularly, we derive the closed-form expression of upper-bounded statistical AoI violation probability. Based on our analytical work, we formulate an energy consumption minimization problem by jointly optimizing offloading strategy, power, and bandwidth allocation in the AoI-aware MEC system. To solve the intractable problem, we propose a dynamic joint optimization algorithm based on block coordinate descent. Extensive simulations show the proposed algorithm achieves at least 13.2% energy consumption reduction compared to the RLTBB, GCGH, and PA-fixedB algorithms.