Research And Implementation Of Channel Estimation And Symbol Sychronization In Visible Light Communications

Visible Light Communications(VLC) has been one of the most popular research areas in wireless communications due to many strengths such as ease of implementation, no radiant harm to human bodies, low cost, dual functions of illumination and data communication. This is because VLC uses visible light to convey signal, while the traditional radio communication has none of the strengths. This work studies the channel estimation, design of long training sequence and symbol synchronization problems in VLC. Apart from that, the hardware implementation of a testbed VLC system is also presented. Details of the work are as follows:1. Study of channel estimation algorithms in VLC systems. OFDM channel estimation algorithms in VLC can be borrowed from traditional radio OFDM systems. These are the LS and LMMSE algorithms in frequency domain, and also the time domain approaches such as time domain LS(ML), time domain LMMSE. The circulant matrix based LS approach and other algorithms are proposed. Detailed analysis and comparison are presented for these algorithms. Also, the influence of Symbol Timing Offset(STO) and Sampling Clock Offset(SCO) on the performance of channel estimation are studied.2. Study of the design of long training sequence in OFDM VLC systems. The performance of channel estimation not only depends on the estimation algorithms used, but also the training sequence. This problem is deduced to be an optimization problem. The Cross Entropy(CE) algorithm is proposed to guide the design of the long training sequence that has the minimal(or near minimal) Peak-to-Average-Power-Ratio(PAPR) under the same power constraint.3. Study of the symbol synchronization problem in OFDM VLC systems. STO and SCO and Carrier Frequency Offset(CFO) can have negative influences on the performance of OFDM systems. As with VLC systems, CFO doesn’t exist at all. Methods of estimating and compen-sating the phase rotation of QAM symbols are proposed. These are the single symbol based estimation and compensation method, multiple symbol based method and whole frame based method. Among them, the whole frame based method is optimal.4. The design and hardware implementation of some modules in a testbed VLC system are presented. The fundamental principle of fixed-point design is presented first. Then the system performance difference between floating-point algorithms and fixed-point algorithms is presented to serve as a benchmark to evaluate the effectiveness of the fixed-point algorithms. On the basis of this, hardware implementation based on Xilinx Kintex 7 series FPGA devel-opment board is discussed in detail. The module test methodology, transmitter integration and debugging, hardware resource utilization are presented at the end of this part.