Abstract
The non-orthogonal multiple access (NOMA) has the potential to improve the spectrum efficiency and the user connectivity compared to the orthogonal schemes. The codebook design is crucial for the the performance of the NOMA system. In this paper, we comprehensively investigate the NOMA codebook design involving the characteristics from multiple signal domains. The minimizing of the pairwise error probability is considered as the target of the optimization. The neural network framework is explored for the optimization, and the mapping functions on the edges are considered as weights. The method of batch gradient descent is applied for optimizing the weights and correspondingly the codebook. The simulation results reveal that with the optimized codebook the error performance is significantly improved compared to the schemes in the literature.
Highlights
The evolution of wireless communications has triggered the rapid development of Internet of things and mobile Internet
By allowing multiple signals from different user equipments (UEs) superimposed on the same resource element (RE), non-orthogonal multiple access (NOMA) improves the spectrum efficiency and the user connectivity compared to the traditional orthogonal multiple access (OMA)
The method of batch gradient descent is applied for optimizing the weights and correspondingly the codebook
Summary
In the existing work in the literature, the research object is the entire NOMA transmission including channel encoding, modulation and mapping, detection and decoding. We investigate the NOMA codebook design involving the characteristics from multiple signal domains but excluding the channel encoding and decoding. Our optimization is based on the given factor graph, which has the similar form to the single-layer neural network. THE PERFORMANCE METRIC In general, the performance analysis of the NOMA transmission can be carried out from two perspectives: sum-rate analysis based on information theory and error-probability analysis based on modulation theory. The error performance can be evaluated by the pairwise error probability which is an upper bound for the average symbol error rate This bound is tight in the high SNR regions for both AWGN channels and Rayleigh fading channels.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have