Utility Maximization in Multi-Functional UAV- and IRS-Aided SWIPT-MEC Networks: A Dynamic Feature Extraction-Based Optimization Approach

Abstract

UAV (unmanned aerial vehicle) and IRS (intelligent reflecting surface) assisted mobile edge computing (MEC) faces low quality of service (QoS) due to the single functionality of UAV and IRS. In response, we propose a simultaneous wireless information and power transfer (SWIPT)-MEC network aided by multi-functional UAV and IRS, where a SWIPT Base Station (SBS) is employed to charge UAVs and assist them in processing tasks. In addition, the multi-functional UAVs can compress task before forwarding it, while the multi-functional IRS enhances both communication and charging channels. To maximize long-term system utility, we design a dynamic feature extraction (DFE)-aided Alternative Optimization Framework (DAOF) to jointly optimize IRS phase-shift, task compression, collaborative offloading, resource allocation and charging duration of UAVs. Numerical results demonstrate that DAOF achieves the highest system utility over other benchmarks.