SSDNN: A Self-Supervised DNN for Energy Efficiency Optimization in UAV CF-mMIMO Under URLLC

With the rapid advancement of unmanned aerial vehicle (UAV) technology, UAV cell free massive multiple-input multiple-output (CF-mMIMO) systems demonstrate significant potential for enhancing wireless communication performance. However, optimizing energy efficiency (EE) to meet the growing demands of communication has become a critical research challenge. This paper explores resource allocation in the UAV CF-mMIMO uplink under ultra reliable and low latency communication (URLLC) constraints to improve EE. We first conduct a systematic analysis in the system model to identify key factors impacting EE. Accounting for these factors, an optimization problem is then formulated. To solve this problem, we employ a self-supervised deep neural network (SSDNN) composed of two submodules: One is to extract channel characteristics of all users equipment (UEs), and the other is to optimally allocate power for both UAVs and ground user equipments (GUEs) via a resource allocation layer. Numerical results show that our proposed method achieves higher EE compared to approaches that do not consider GUE interference.