Rate Splitting Multiple Access Enhanced Visible Light Communications under Fairness Constraint

Rate splitting multiple access (RSMA) has been recently proposed for visible light communications (VLC) systems to further increase the spectral efficiency (SE) compared with traditional non-orthogonal multiple access (NOMA) approaches. However, state-of-the-art RSMA-VLC systems usually utilize optimization algorithms to determine the channel precoding matrix, leading to high complexities in achieving optimality. Despite the computational complexities, another major concern is that these schemes only focused on the weighted sum rate (WSR) without considering the fairness issue among multiple users, leading to poor BER performances for certain users. This is detrimental for VLC where light is adopted to let each user achieve a high data rate. Neglecting fairness problems may result in inadequate information signal rates for some important users within the system, so that the quality of services (QoS) metric cannot be satisfied. To address the QoS issue, this paper introduces a fairness criterion by considering both public and data rates between VLC users. To cope with the increased system computational complexity issue introduced by RSMA-VLC fairness constraint, we proposed a Genetic Algorithm (GA) based optimization method, named GAF-RSMA, to find the optimal precoding matrix. To thoroughly study the WSR performance of GAF-RSMA under signal-to-interference-plus-noise ratio (SINR) constraints, we conducted numerical simulations. Simulations results show that GAF-RSMA outperforms conventional NOMA in terms of SE, while ensuring fairness between VLC users. For different scenarios, RSMA brings up to 15.23% improvements in SE. It is worth pointing out that as the SINR requirement becomes more stringent, i.e., with the increment of minimum SINR constraint, the WSR of the GAF-RSMA system dropped. Our results indicate that the GA algorithm can be applied to achieve an optimal precoding matrix for RSMA VLC systems while maintaining system’s WSR and user’s SINR values.