A Novel Receiver Technique to enable NOMA Without User Selection

Abstract

NOMA (Non Orthogonal Multiple Access)/MUST (Multi-User Superposition Transmission) is one of the major candidates for next generation wireless access technologies as they provide better user fairness and improved spectral efficiency. Successive Interference Cancellation (SIC) receivers enable multiple users to be scheduled in the same time-frequency resources while differentiated in power domain. Two user Power Domain NOMA (PD-NOMA) analysis clearly shows that the SNR difference between the users is very important in order to achieve inter-user interference cancellation. Hence user selection or pairing becomes crucial and can lead to heavy co-channel interference when users are paired wrongly. In order to overcome this problem, a new method of NOMA is proposed where the user pairing can be arbitrary, but an advanced receiver technique is used which enables successful decoding of the desired user’s data. A novel Gaussian Mixture based receiver is proposed. It is shown that the error rate performance of this detector is very close to that of a joint detector and the proposed technique can be used both in uplink and downlink directions. Another key contribution of this work is to show the over-modeling technique which enables the decoder to operate without having to know any prior information about the interferers. This enables a practical approach to build Gaussian Mixture Model (GMM) based receivers for a variety of interference scenarios.