| Pilot patterns’ selection of CS-based channel estimation and DDLMMSE channel estimation algorithm in OFDM systems are studied. Firstly, the model of CS channel estimation is established, and a novel method on optimal pilot patterns’ selection is proposed based current studies, simulations verify its advantages. Secondly, the problem exsited in DDLMMSE algorithm have been analysed, and an improved method is proposed, simulations verify its effectiveness. The main contents include:(1) A novel method on pilot patterns’ selection of sparse channel estimation is proposed. Drawbacks of pilot patterns’ selection by using the coherence parameter minimization have been proved through theoretical analysis and simulation experiments, and cumulative coherence function is proposed instead of coherence parameter, it will remedy drawbacks. Channel estimation will carry out with optimal pilot patterns selected by novel method, and channel impulse responses are obtained by OMP and LS. The simulations with 100 times Monte Carlo demonstrate that channel estimation performance with pilot patterns selected by cumulative coherence function is 30% better than pilot patterns selected by coherence, 68% is the same and the remaining is 2%. In addition, cumulative coherence function tends to convergence while the total iteratioin number at 300 times. Therefore, it is necessary to reduce the operation time by setting the iteration number at 300 times around.(2)An improved DDLMMSE channel estimation algorithm has been proposed. DDLMMSE channel estimation algorithm is obtained from modified LS algorithm. The diagonal matrix in DDLMMSE algorithm will be augmented to width K by using correlation between subcarriers, and estimate accuracy would be enhanced without unobvious increment of complexity. The SNR will decrease with the increasing width while bit error rate is fixed. There is about 0.5d B reduction of SNR in K ?3 compared with K ?1. We can select appropriate K according to the performance. |
PDF Full Text Request