Estimation Based On OFDM Underwater Acoustic Communication System Offset And Channel

Underwater acoustic channel, with large relative delay and high Doppler spread, has been considered one of the most challenging wireless channels. Orthogonal frequency division multiplexing (OFDM) will be one of the most potential key technologies in the future underwater acoustic communication system due to low-complexity frequency-domain equalizer at receiver formed to combat high time-dispersive multipath channels. However, it is very sensitive to carrier frequency offset (CFO), CFO will yield a seeable inter-carrier interference (ICI) in the case normalized CFO is larger than0.05, this severely degrades the system performance. In addition, an accurate estimation of underwater acoustic channel is also important to demodulate and restore the transmission signals at receiver. In view of this, this paper focuses on the CFO estimation and channel estimation algorithms for OFDM-based underwater acoustic communication system. The main contents are arranged as follows:1) In order to estimate the fractional CFO, we first research the MUSIC algorithm based on null subcarriers (NS). Then, the four blind CFO estimation algorithms without requiring any auxiliary information, including O-M algorithm, Y-G algorithm, Least Square (LS) algorithm and J-J-J algorithm, are investigated. From simulation and theoretical analysis, it follows that the MUSIC method based on NS is very stable regardless of the adopted signal constellation, while the remaining methods are affected by the adopted signal constellation.2) To estimate the integer CFO, firstly, we present three pilot-assisted estimation algorithms:Schmidl algorithm, Morelli algorithm and Maximum Likelihood (ML) algorithm. Subsequently, we investigate the minimum power sum (MPS) algorithm. The result of simulations shows that among the four algorithms, ML is optimal; at low Signal-to-noise Ratio (SNR), the performance of MPS is the worst one, however, its performance grows dramatically with the SNR increase, when the SNR reaches a certain value, its performance is only worse than ML; in middle and high SNR regions, the estimation performance of Schmidl algorithm is the worst, and Morelli followed. When there exit a residual fractional CFO at system, all methods show different loss in performance, but there is no change on their performance order.3) To accurately estimate the parameters of underwater acoustic channel, we introduce five channel estimation algorithms based on Compressed Sensing (CS):Matching Pursuit (MP), Orthogonal Matching Pursuit (OMP), Gradient Pursuit (GP), Parallel Coordinate Descent (PCD) based on iterative-shrinkage, and Dantzig-Selector (DS) algorithm; finally, a kind of random distributed pilot pattern is proposed. The results of simulation show:given pilot pattern, these CS-based algorithms are superior to LS algorithm and these channel estimation methods have an decreasing order in BER performance as follows: DS>PCD>OMP-GP>MP>LS; given a channel estimator, the proposed random pilot pattern performs much better than uniform and consecutive ones in the BER performance.