Research And Application On Fast Likelihood Detection For Outdoor Visible Light Communications

Recently, the technology of VLC(Visible Light Communication) has developed rapidly, and it effectively alleviates the contradiction of spectrum resources shortage, furthermore it has the advantages of energy conservation, environment protection, safety and health, and fast installation. In addition to being used in indoor high speed communication, VLC is also widely used in outdoor scenario such as the communication among cars, the communication between street lamps and human beings and the communication between lighthouses and boats. Nevertheless OVLC(Outdoor Visible Light Communication) is influenced by the factors including of turbulence and attenuation in atmosphere channel, these factors lead to the random change of the optical intensity, which exerts a severe impact on signal detection. This thesis focuses on the research of signal design and detection algorithm of OVLC. The main research contents and results are summarized as follows:1. Aiming at the complexity of OVLC channel which is affected by many factors without uniform model, the joint fast blind likelihood detection algorithm of the channel and the signal is proposed. The channel characteristics are analyzed, and the classic channel models in different turbulence states are introduced. Making the full use of characteristics of OVLC channel and signal, the joint fast blind likelihood detection algorithm of channel and signal is proposed. The algorithm transforms the problem of exhaustively searching signal into the problem of linearly searching signal. The complexity of detecting signal is decreased into the complexity of ordering signal. To decrease the algorithm complexity, the modified algorithm of joint fast blind likelihood is put forward to halve the complexity by using the convex characteristic of decision function. To solve the error floor in join blind likelihood algorithm, the methods of one nonzero bit training and sequence code are proposed to eliminate error floor. Simulations demonstrate that joint fast blind likelihood detection performs better than mean value algorithm in the channel evaluation and signal detection, both algorithms can eliminate the floors.2. Aiming at the lower complexity and less processing delay in practical system, the method of one bit training is researched, and the maximum likelihood detection with two thresholds is proposed. The one bit training system is modeled to analyze the noise in the system. The ZF-ZF and ZF-ML detection algorithm are given, and by analyzing the probability density function of the signal received conditioned on the signal sent, the maximum likelihood detection algorithm with two thresholds is obtained; The BER(bit error rate) of repetition coded MISO system with ZF-ML receiver is derived, the full diversity of ZF-ML is proved. To make BER least, it is proved that the BER is a convex function of training symbol power and information symbol power in the limit of total optical power, and the method of optimal power allocation is given; Simulations demonstrate that the performance of ZF-ML is better than ZF-ZF, and it can achieve full diversity, the performance can be improved by power allocation.3. The OVLC system with the rate of 10 Mbit/s at the distance of 50 m is designed and implemented. The joint fast blind likelihood detection algorithm of channel and signal, and the fast blind likelihood detection algorithm based on one bit training are tested in practice. The system structure, electric circuit and optical front end are designed, and the frequency-amplitude response of LED is measured. The pre-equalization method is used to compensate the low-pass factor of LED and achieved in FPGA. At last the mean-value algorithm, the joint fast blind likelihood detection algorithm and the fast blind likelihood detection algorithm are tested and compared in experiment to test the algorithm.