Research Of Pilot-interpolation Algorithm In Channel Estimation Of OFDM Systems

With social progress and development, the wireless broadband digital communication system and cause broad researchers interests. Pilot interpolation algorithm research of OFDM technology caused extensive concern at home and abroad, the first thing is to put the known pilot signals to insert each sub-carrier, using least-square estimation, least mean-square error estimation and singular value decomposition estimation to estimate the pilot points channel response, then by linear interpolation, gaussian interpolation, DFT interpolation, Cubic-spline interpolation etc interpolation algorithms estimate pilot sub-carrier channel response, in order to further reduce the BER, improve the performance of the system, interpolation algorithm has become OFDM technology focus.This paper in order to reduce BER and improve the performance of the system, puts forward a new interpolation algorithm. main research work of this paper are as follows: (1) Aiming at OFDM channel characteristic, establishes mathematical model of wireless channels. Based on which this paper is in-depth discussions of block and comb two pilot. Discusses LS, MMSE and SVD algorithm, and compare linear interpolation with gaussian interpolation, DFT interpolation, Cubic-spline interpolation performance, and analyzes the reasons of high BER algorithm, in order to put forward the improved algorithm to lay the foundation.(2) Aiming at the factor of noise isn't considered in the LS algorithm, affects estimated accuracy, this paper proposes an based on the threshold judgment LS algorithm, this algorithm will limit time channel response of LS to less than cycle prefix length, and set up a threshold, set zero when the noise components within cycle the prefix length and invalid diameter response , which improves the estimated accuracy.(3) Aiming at high BER problem in the existing interpolation algorithm, this paper proposes cubic hermite interpolation algorithm, cubic hermite interpolation uses two nearby pilot point channel response and channel response first order derivative, to estimate data sub-carrier channel response between two pilot points, and through simulation, cubic hermite interpolation has lower BER compared with linear interpolation, gaussian interpolation and DFT interpolation, but little higher than cubic spine interpolation, howeve, cubic hermite interpolation'computational complexity is more simple.