Half Of Blind Faith In Ofdm System Design And Realization Of The Estimator

Orthogonal frequency division multiplexing (OFDM) is one of the key technologies of the fourth generation mobile communication standard LTE, offering high-quality and high-rate mobile multimedia transmission for the future communication. It has been used in many communication networks, such as Digital Video Broadcasting (DVB), Wireless Local Area Network (WLAN). Channel estimation is the fundament of equalization and coherent demodulation, playing a very important part role in the OFDM system.There are three different classifications in channel estimation:pilot channel estimation, blind channel estimation, and semi-blind channel estimation. Firstly, the model of wireless channels and the principle of OFDM system are introduced; the simulation models of OFDM system and wireless channel are constructed. Then the semi-blind channel estimator is designed and implemented in DSP C64x based on HIPERLAN/2standard, the goal of the estimator is to improve the performance of the low-speed mobile HIPERLAN/2terminals.1. Three algorithms, finite-alphabet based blind channel estimation algorithms, precoding technique-based blind channel estimation algorithms and subspace-based blind channel estimation algorithms are analyzed in the paper. By comparing the performance of these methods in the HIPERLAN/2model, the paper gets the conclusions:the subspace-based algorithm which can have zero subcarrier is better than others.2. In order to overcome the disadvantages of slow convergence rate, the paper uses pilot symbols to change the blind algorithm to semi-blind algorithm, thus the computational complexity is greatly reduced.The simulation result shows that the performance of this novel semi-blind algorithm is greatly improved in the time-varying channel estimation.3. According to the algorithm principle, the algorithm is implemented with floating-point C code in VC, and the fixed-point code is designed in order to achieve in fixed-point DSP. By comparing the simulation results, the performance of the fixed-point approach is slightly worse than the floating, because the censored processing causes some loss of precision. But it still has a very good performance in channel estimation.4. The C language code is transplanted to the DSP. In order to enhance the execute speed of the code; the paper optimizes C code by deploying software pipeline, intrinsic function, et al. The code execution speed is greatly improved at last.