Accurately Decoding MIMO Streams in VLC

Among the efforts to overcome the problem of rapidly saturating RF bands, visible light communication (VLC) is garnering a renewed interest as it can be enabled using commodity LED lamps. As indoor spaces are typically illuminated by multiple lamps comprising multiple LEDs each, a natural approach to efficiently utilize the bandwidth of all the LEDs is to use MIMO communication. The state of the art approach to decode MIMO streams is to use channel matrix. Although channel matrix based decoding method (CMDM) works very well in conventional RF technologies, when used in VLC, it suffers from several limitations, such as high sensitivity to environmental conditions, and need for sophisticated receivers. To overcome these limitations, we propose PCDM, a novel parallelogram - clustering based decoding method, which is fundamentally different from CMDM and achieves an order of magnitude lower bit error rate compared to CMDM. We implement and extensively evaluate these two methods using a real VLC MIMO testbed. Our results show that PCDM outperformed CMDM in all scenarios.