Performance Analysis of Rate Splitting in K -User Interference Channel Under Imperfect CSIT: Average Sum Rate, Outage Probability and SER

Interference alignment (IA) and rate splitting (RS) are two promising interference processing techniques to handle the interference problems in the multi-input multi-output (MIMO) multi-cell cooperation scenario. However, under the imperfect channel state information (CSIT) in the practical systems, the poor robustness of IA has become a bottleneck of its performance gain. On the other hand, RS can provide a robust solution, yet its outage performance is limited by its SIC decoding process. To make up these two schemes' shortcomings and achieve the robust and reliable multi-cell cooperation, in this paper, we propose a novel transmission scheme, signal and interference alignment based rate splitting (SIA-RS), and provide comprehensively analyze the performance under CSIT quantization error. Closed-from expressions are derived for the average sum rate, outage probability and symbol error rate (SER) and their asymptotic versions in high signal-to-noise ratio (SNR) region. Due to signal alignment and common message, there exists certain correlation between the key variables. The nested finite weighted sum of independent and correlated Erlang random variables is used to approximate the exact expressions of performance. The relationship between "splitting" and "alignment" is revealed via the analytical derivation and numerical simulation. The simulation results show that the proposed SIA-RS achieves the best average sum rate compared with conventional IA and RS schemes, indicate that alignment would further reduce the outage probability, and suggest to the separate modulation schemes for common and private messages in terms of SER.