A Novel GNN-based Decoding Scheme for Sparse Code Multiple Access (SCMA)

Sparse Code Multiple Access (SCMA) is a promising code-based multiple access technique for achieving higher spectral efficiency and massive connectivity that is crucially required in the B5G applications such as massive machine-type communication (mMTC). Traditional SCMA receivers use Maximum Likelihood (ML) and Message Passing Algorithm (MPA) for signal decoding, which nonetheless suffer from extremely high computational complexity. To resolve the issue, we propose to use Graph Neural Networks (GNN) to replace MPA for decoding, aiming at reducing the decoding complexity while maintaining satisfactory Bit Error Rate (BER) performance. Simulation results show that our proposed solution can achieve much higher decoding accuracy and faster decoding speed.