Power Allocation with Non-Orthogonal Multiple Access for 5G Heterogeneous Network

The fifth generation of mobile network technology (5G) through heterogeneous networks (HetNet) demands increasing the capacity of cellular wireless communications networks. In addition to the HetNet, a Non-Orthogonal Multiple Access (NOMA) can enhance its performance techniques to support a large number of users. Coordinated multi-point communication (CoMP) has also been identified as a desirable technology for reducing interference in HetNet. However, existing techniques cannot fulfil the high demands of the next 5G, which require direct interference between users in such an orthogonal architecture and good system-level performance. This study aimed to determine the optimal resource allocation for the downlink HetNet 5G NOMA system. The system models were derived to analyze the power performance based on the power allocation factor. An algorithm is constructed based on the deep learning coordinated multi-point to simulate the performance of the downlink 5G. Deep learning can be implemented to improve resource allocation in the context of CoMP by instructing a neural network to predict the optimal power allocation for a specific set of users to resemble the NOMA with HetNet system. Simulation validates the proposed algorithm in terms of the power allocation factor and the impact of the SNR. The results indicate that the power allocation factor ratio for various user signal strengths is capable of allocating dynamic resources to meet the need for efficient resource management.