Optimal Power Allocation for the Downlink of Cache-Aided NOMA Systems

In this work, we investigate the optimal power allocation for the downlink of cache-Aided NOMA systems, in which the user with weaker link condition caches the intended message of user with stronger channel condition within each cell, resulting in different signal decoding at receiver side from that of conventional NOMA. The goal is to minimize the total needed power of all users subject to the least data rate constraint of each user. To solve this non-convex optimization problem, we first transform the required power of users in each cell to that of their serving BS. With the power constraint transformation, the original power allocation problem can be transformed into a power optimization problem with reduced dimensions. The optimality of the newly formulated problem is analyzed. In addition, we show that the optimal solutions to the two optimization problems are unique and one-To-one mapping, which motivates us to get the optimal power control for the original problem by solving the newly formulated problem. An iterative power control algorithm is then designed, which is guarantee to converge to the unique optimal point given that the feasible power region is non-empty. Simulation results demonstrate the power-saving capability of our proposed power allocation scheme for cache-Aided NOMA.