Binary vs partial offloading in wireless powered mobile edge computing systems with fairness guarantees

Abstract

Mobile Edge Computing (MEC) has recently emerged as a new communications/computing concept that amends the limited computing of IoT devices by completely or partially offloading the computational tasks to the MEC servers at the network edge (typically co-located with the base stations). Because IoT devices are typically power limited, the potential of the MEC is further enhanced by its integration with wireless power transfer technology, especially for those IoT devices with a high duty cycle that requires frequent battery replacement. This paper develops fairness-aware resource allocation schemes for a WPT-assisted MEC system whose Energy Harvesting Users (EHUs) employ either binary or partial offloading. Specifically, the proposed schemes optimize the computational speeds and the energy harvesting and offloading durations of the EHUs with the aim to maximize the minimum of their computed bits (sum of locally and remotely processed bits of each EHU), subject to the RF energy harvested from the base station. When EHUs are concentrated closer to the base station, remote processing is preferred over local processing, as local processing consumes more energy than the Radio Frequency (RF) power for offloading data to the MEC server, but this effect diminishes for lower values of the computational effort needed for the processing of a single bit. Interestingly, in terms of the sum computation rate, the partial offloading scheme only slightly outperforms the binary offloading scheme, but only when the EHUs are moderately away from the base station.