RSMA Architectures for Downlink Communication Systems

Abstract

Demand for the internet is increasing day by day, and the number of internet users is also growing drastically. Multiple access (MA) plays an important role in the fifth-generation (5G) and upcoming sixth-generation (6G) era. Space division multiple access (SDMA) and non-orthogonal multiple access (NOMA) are some of the existing MA schemes used in the 5G networks.

Recently, rate-splitting multiple access (RSMA) has gained much attention as a can didate MA scheme for future radio access, which uses the rate-splitting (RS) concept at the transmitter. It splits each user message into private and common/public parts, and outperforms these existing MA schemes in any network load, user deployments, and channel state information at the transmitter (CSIT) inaccuracy. It works based on the principle of “partially decoding the interference and partially treating the interference as noise”. However, RSMA suffers from large encoding complexity, receiver complex ity, and signaling burden (signaling burden means that each user needs to know how to split and combine the decoded common and private messages in order to recover the intended message at the receiver, and furthermore, signaling burden increases if split and combining patterns are complicated). In this thesis, to overcome these limitations in the existing schemes, we tried to develop four new MA schemes or architectures, i.e., NOMA assisted RSMA schemes and another one without using SIC in any users, for the downlink cases with the base station (BS) serving single antenna users and compared their sum rate performance and bit error rate (BER) performance with existing work through simulations.

First, we propose three NOMA-assisted RSMA schemes by using the concepts of RSMA and NOMA, and do analysis of their sum-rates and BER. We derive optimiza tion frameworks for these three proposed schemes for sum-rate maximization using the weighted minimum mean square error (WMMSE) approach. Finally, we compare their complexity and performance with the existing ones and show that they perform well with less complexity, and conclude with their performance observations in different scenarios, like in perfect and imperfect CSIT, and at different parameter values/settings, i.e., for different channel strength differences and angles between the users.

Next, we tried proposing another MA technique for the multiple input single output (MISO) systems by allocating equal power to all transmit messages at the BS and with out using any successive interference cancellation (SIC) at the receiver side. We derive BER and sum-rate equations for this proposed scheme and compare its performance with the existing state of the art. Also, we solve the weighted sum rate (WSR) optimization problem for this scheme using WMMSE algorithm. Finally, we conclude that our pro posed scheme performs better than the existing schemes with less receiver complexity, less encoding complexity, and reducing the signaling burden.

Also, to the best of our knowledge, there were no works found on BER simulation of different architectures of RSMA (except 1-layer RS), but we have simulated the BER for all the linearly precoded RSMA schemes. All simulations have been carried out in MATLAB environment and used the CVX toolbox for optimization purposes.