A Reinforcement Learning-Assisted OFDM-IM Communication System against Reactive Jammers

An innovative orthogonal frequency division multiplexing with index modulation (OFDM-IM) transmitter design is proposed in this paper to enable high-speed communication against reactive jammers. The proposed model can dynamically adjust its index modulation (IM) parameters and modulation types, including a novel multi-carrier noise modulation capability that enhances robustness under heavy jamming conditions. Moreover, a reinforcement learning (RL) mechanism is implemented to find the optimal defense strategy without needing any information about the jammer. To validate our approach, we conducted extensive computer simulations to evaluate the system's performance against various jammer types. Our simulation results revealed that subcarrier adaptation (adjusting IM parameters) enhances system performance towards higher throughput, while noise modulation improves bit error rate (BER) performance. Moreover, the results verify the model's ability to maintain robust communication in the presence of sophisticated reactive jamming attacks, outperforming several benchmark models.