Joint transmitter and receiver spatial modulation for reconfigurable intelligent surface enabled VLC

In visible light communications (VLC), besides assisting the data transmission while the direct link is blocked, reconfigurable intelligent surface (RIS) can adopt spatial modulations (SM) to superimpose its own modulated data on the reflected light signals, leading to an increased overall system throughput. A straight-forward method to perform SM using RIS is named RIS enabled SM (RIS-SM), where the RIS elements are divided into groups and only one of them are activated during each transmission process based on the data to be transmitted. However, the performance of RIS-SM is greatly affected by channel correlation, leading to higher bit error rates (BER). State-of-the-art approach, named RIS enabled receiver SM (RIS-RSM), allows all RIS elements to reflect signals, but only one of the multiple PD receivers is activated due to the spatial modulation design. Although RIS-RSM reduces the impact of channel correlation, its data rate in the spatial domain is limited due to the number of receivers used, lacking the flexibility in using best suitable RIS elements as in RIS-SM. In this paper, we propose a joint transmitter-and-receiver SM architecture for the RIS enabled VLC systems, named as RIS-JtSM, where spatial modulation takes place at both the RIS elements and the PD receivers. The reflected light signals received by the PD receivers contain three data streams that can be decoded in a joint manner: one distinct stream is from the light source, and two other coupled steams that are spatially modulated at the RIS elements and the PD receiver simultaneously. These streams are jointly decoded by the receiver using maximum-likelihood detections (MLD). We derive an upper bound for the BER of RIS-JtSM to evaluate its performance, and simulation results verify the tightness of our analytical upper bound in additive white Gaussian noise (AWGN) channels. In addition, simulation results show the superior BER performance of RIS-JtSM compared with RIS-SM and RIS-RSM under the same overall data rate and mean emitted light power. To reduce the complexity of MLD, we propose a selection based MLD (S-MLD) method, which decodes the spatially modulated data first, and then performs the detection of the other two streams in successively. Compared with MLD, S-MLD can reduce the detection complexity to 1/N²_r (with N_r the number of PDs) while maintaining near-optimal BER performance. This is the first attempt in studying a general joint transmitter-and-receiver SM design for RIS enchanted VLC systems, and in the future, to ensure the real impact of RIS-JtSM thorough evaluations using Monte-Carlo simulations under practical indoor VLC fading channel models with moving receivers should be carried on.