Meta learning for NOMA-MISO systems based full-duplex transmission

The fast growth in wireless communication capacity demands adaptable systems that can control wireless resources efficiently to ensure a good and fair user experience. Machine learning (ML) and deep learning (DL) offer the opportunity to achieve modern communication goals. In this paper, wireless service is provided to several clients using a combination of non-orthogonal multiple access (NOMA) and multiple-input single-output (MISO) over a full-duplex (FD) channel. After using a pairing technique, the BS divides the coverage area into clusters is one suggested approach. Each cluster has only two users. Once everyone in a cluster is treated equally and their circumstances are considered, a sufficient amount of power is assigned to each of them using the NOMA technique. Feature engineering techniques are applied over the generated datasets to help the models find the pattern between the input datasets that leads to robust models. An examination and evaluation of the signal-to-interference-pulse-noise ratio (SINR), and the bit error rate (BER) attained for each user in the cluster in an acceptable value. Moreover, the potential throughput for strong user is maximized. To train different DL models controlled by one ML model to predict the optimal power allocation factor (PAF) that controlled transmitted power.