Research On Channel Estimation Technology In LiFi

After entering the era of network, the interaction of information communication and the supply and demand of services are becoming more and more frequent and convenient. Wi Fi technology has been matured and widely used, making the Internet industry growth blowout. However, many problems are also increasingly exposed, such as the unstable Wi Fi signal, limited application,narrow spectrum range and unreasonable layout of the hot point. The communication technology using high speed light visible light to transmit information has double functions of lighting and communication. The communication style using LED as the media is called "Li Fi".Due to meteorological factors or relative displacement, Optical wireless channel environment has such characteristics: randomness and fluctuation. In the high speed transmission system, using high order OFDM modulation is time-varying multipath effect, effective means to improve the efficiency of utilization of bandwidth. To ensure strict orthogonality of the receiving end of OFDM subcarriers, we must fully and accurately estimate the transmission channel. It can be said that the channel estimation accuracy directly affects the performance of Li Fi system.At present, LS, MMSE and DFT are three basic channel estimation algorithms, such as the base of tower; EM and SVD are routine applied algorithms, such as the body of tower. Aiming at the poor traceability and bad match degree, this paper proposes an improved algorithm, such as the top of tower. The iterating and finding optimal characteristics of EM algorithm and the reducing order characteristics of SVD algorithm are integrated deeply. The problem of the dense matrix which iterative converged slowly is overcome by using the gradient descent method. The problem of the low-rank matrix recovery is transformed into the problem of the matrix the nuclear norm minimization smoothly.In the simulation, the running ability of improved algorithm is tested and evaluated by the BER/MSE curve comparison string, visual differences and data analysis of the picture. Through theoretical analysis and simulation tests show that the algorithm model can maintain good performance consistently in both the AWGN and the RMF channel.