Deep Learning Based Hybrid Multiple Access Consisting of SCMA and OFDMA Using User Position Information

摘要

This paper proposes a deep-learning-based uplink hybrid multiple access scheme consisting of both sparse code multiple access (SCMA) and orthogonal frequency-division multiple access (OFDMA). SCMA improves the system throughput when the carrier-to-noise ratio (CNR) is high. However, SCMA performance is significantly degraded, compared to OFDMA, when the CNR is low. To overcome this problem, the proposed scheme introduces a combination of SCMA and OFDMA as a novel multiple access pattern. The scheme determines the appropriate pattern among SCMA-only, OFDMA-only, or their combination, by utilizing user position information through deep learning. The effectiveness of the proposed scheme is demonstrated in terms of system throughput under different user distributions via computer simulations.