Adaptive Channel Estimation in VLC for Dynamic Indoor Environment

The channel of indoor VLC system is usually considered static for ease in most of the cases however in a real-world scenario, the changing effect due to people density, shadowing, dimming, background lights and interiors create dynamism in the VLC channel albeit with slow variation. Thus, the channel time-varying effect cannot be ignored entirely in modeling the VLC system. The impact of the dynamic channel can't be mitigated just by increasing LED transmission power. Fortunately, a possible alternative way is to estimate the channel state information (CSI) in dynamic VLC environment. This paper considers a dynamic VLC environment where a decrease in the normalized received power follows Rayleigh distribution. In this paper, we propose the estimation of the channel coefficients using variants of least mean square (LMS) algorithm such as normalized (NLMS), zero attracting (ZA-LMS), block (BLMS) and fast block (FBLMS). This paper tests the suitability of adaptive algorithms in terms of mean square error (MSE) and tap-weights convergence, computational complexity and the number of pilot symbols required in VLC dynamic channel. FBLMS and NLMS perform better in comparison to other algorithms as observed from results obtained. At lower people density in FBLMS and NLMS the difference in the number of MSE convergence samples is less compared to the higher people density as FBLMS converges faster to ideal zero MSE than NLMS.