| Visible light communication(VLC) is a new wireless access technology developed in last decades. The technology has a number of advantages such as: environment friendly, occupying no radio frequency spectrum, high-speed data rate and so on. Thus, it has been a hotspot issue all over the world. However, for an indoor transmission environment, the received signal contains both direct light and multiple reflection light, which will bring multi-path delay, causing Inter Symbol Interference(ISI) problem when the system transmission rate is very high. This paper does a series of research on equalization method to overcome multipath interference, so as to improve the VLC system performance.Firstly, this paper introduces the basic principles and structure of the VLC system, then the channel model of VLC system is studied. The multiple-bounce response and Root Mean Square(RMS) delay performance based on ray tracing model show that ISI has serious impact on system performance. Then, a variety of equalization techniques are discussed in detail, including Zero Forcing(ZF) equalization, Least Mean Square(LMS) equalization, Recursive Least Square(RLS) equalization and LMS algorithm based on Decision Feedback Equalizer(DFE) structure. Computer simulation results show that ZF equalization needs known channel information which is not available for the time-varying channel; LMS and RLS equalization can update the equalizer weight coefficients adaptively, but more training sequences need to be sent in LMS algorithm, which will result in the lower efficiency of information transmission while RLS method has instability problems. Considering all the problems described above, the paper introduces a new equalization technique adapted to VLC system, namely artificial neural network(ANN) adaptive equalization. Equalization problem is changed into a classification issue of “0†and “1†in binary, which can be solved by artificial neural network. The paper studies the neural network equalization based on linear equalizer and decision feedback equalization. Theoretical analysis and simulation results show that, in order to achieve the same performance, ANN adaptive equalization requires less training sequences than LMS equalization. In particularly, when the data rate reaches up to 700Mbit/s or higher, ANN method exhibits an excellent Bit Error Rate(BER) performance. The new equalization method can adapt to the complexity of the VLC channel better, improving the overall performance of V LC system. |
