Analysis of user pairing non-orthogonal multiple access network using deep Q-network algorithm for defense applications

Non-orthogonal multiple access (NOMA) networks play an important role in defense communication scenarios. Deep learning (DL)-based solutions are being considered as viable ways to solve the issues in fifth-generation (5G) and beyond 5G (B5G) wireless networks, since they can provide a more realistic solution in the real-world wireless environment. In this work, we consider the deep Q-Network (DQN) algorithm-based user pairing downlink (D/L) NOMA network. We have applied the convex optimization (CO) technique and optimized the sum rate of all the wireless users. First, the near-far (N-F) pairing and near-near and far-far (N-N and F-F) pairing strategies are investigated for the multiple numbers of users, and a closed-form (CF) expression of the achievable rate is derived. After that, the optimal power allocation (OPA) factors are derived using the CO technique. Through simulations, it has been demonstrated that the DQN algorithms perform much better than the deep reinforcement learning (DRL) and conventional orthogonal frequency-division multiple access (OFDMA) schemes. The sum-rate performance significantly increases with OPA factors. The simulation results are in close agreement with the analytical results.