Optimal Load Allocation of On-Board Detection for Multi-UAV Cooperative NOMA under Payload Constraints

Unmanned aerial vehicle (UAV) swarm can provide ubiquitous coverage if innovative non-orthogonal multiple access (NOMA) is deployed and the widespread propagation-induced interference is effectively alleviated through on-board multi-user detection (MUD). While centralized implementation of on-board MUD brings unaffordable complexity for a single UAV, this paper proposes to share and allocate the detection load among multiple UAVs for finest exploitation of cooperative processing power. Observing the inherent disparities among users under NOMA transmission, this paper first excavates the closed-form trade-offs between achievable sum-rate and the MUD load corresponding to the UAV-user matchings, which leads to a sum-rate maximization problem under payload constraints on processing complexity. To address the non-trivial integer matching, this paper develops a quadratic transformation to the original problem, and proves it an equivalent conversion through Karush-Kuhn-Tucker (KKT) conditions. The problem is further simplified into a series of subproblems employing successive lower bound approximation which is proved to obtain polynomial-time complexity with good convergence. Numerical results show remarkable complexity reduction compared with centralized processing, as well as more than 10% sum-rate gain compared with other allocation methods.