RIS Enabled Spatial Modulation in VLC with Maximized Euclidean Distance

In visible light communications (VLC) systems, re-configurable intelligent surface (RIS) is capable of transmitting the light source's signals to the receiver even when the line-of-sight (LoS) link is blocked. RIS is also capable of realizing spatial modulation (SM), so as to modulate its own messages on top of the reflected light signals, and thereby further increase overall system throughput. However, the straight-forward combination of RIS with SM in VLC, like RIS-enabled SM (RIS-SM) and generalized SM (RIS-GSM) in the literature, may encounter a huge bit error rate (BER) degradation as large as 21dB compared with direct RIS links without SM, due to high channel correlations between the RIS elements. That is, the reflection links' channel coefficients' similarity may ruin the minimum Euclidean distance among received symbol vectors. To tackle this BER degradation challenge, we propose an algorithm which can maximize the minimum Euclidean distance at the receiver, called MaxED, by tuning the RIS reflection directions and grouping scheme. In MaxED, we jointly select proper RIS-SM signal groups and optimize the reflection directions using exhaustive search. To reduce the computational complexity of MaxED while maintaining good BER performance, we apply a particle swarm optimization (PSO) algorithm. Simulation results show that, compared with traditional RIS-SM and RIS-GSM, MaxED can have transmitted power gains of approximately 18dB and 17dB, respectively, when BER=1E-6. Our findings reveal that MaxED can effectively reduce the impact of channel correlation in RIS enabled spatial modulation systems, making RIS-SM and RIS-GSM more feasible for real applications.