Research On Wide-band HF-OFDM Channel Estimation Based On Compressive Sensing

Short wave(High-Frequency) communication is widely used in the field of remote communication owing to the advantages of strong anti-destruction performance, low cost of network and far communication distance. With the increasing demands of communication services, the existing narrowband HF channel can't meet the requirements of high speed HF communication. However, the conditions of HF channel are relatively harsh, orthogonal frequency division multiplexing(OFDM) is applied to wideband HF communication system in order to improve the spectrum efficiency and oppose multipath delays in HF channel. At the same time, in order to ensure the good transmission performance of wideband HF OFDM system, it is necessary to carry out accurate channel estimation at the receiver.Firstly, in view of the actual situation that the OMP algorithm requires to predict the sparse degree of HF channel and the HF channel sparsity is difficult to predict, an improved algorithm mOMP is proposed, which doesn't need to predict sparse degree of the channel. In order to avoid the effects of the strong single-frequency-interference on the channel estimation, firstly, the interference detection and suppression in pilot sub-carriers is needed, then HF channel estimation. The updated residual value is used as the decision condition to control the circulation and termination of the proposed algorithm, so as to adapt the sparsity of HF channel. Simulation results show that compared to the LS algorithm, the compressed sensing based channel estimation algorithms proposed in this thesis has not only the lower bit error rate(BER), but also highter system spectral efficiency. In addition, the performance of mOMP algorithm is nearly the same with OMP algorithm, but the inputted HF channel sparsity has serious influence on OMP algorithm, while the mOMP algorithm is not affected and the performance of the channel estimation is stability.Secondly, in order to further improve the reconstruction accuracy of HF channel, a pilot optimization scheme based on genetic algorithm is proposed, and it takes the minimization of the cross correlation value of the measurement matrix as the optimization goal.Through the study of the method of judging whether the measurement matrix is meet the property of RIP, related thesises found that the smaller the cross-correlation value of measurement matrix is, the higher probability of signal reconstruction becomes. Thus, the thesis can get the optimal pilot sequences by minimizing the cross correlation value of the measurement matrix as the goal of the optimization. The analogy with the traditional traveling salesman problem(TSP), genetic algorithm is used to optimize the pilot sequences in the thesis. Compared with the existing pilot optimization schemes, simulation results prove the proposed feasibility and effectiveness of proposed scheme.