Research On Channel Estimation Algorithm Based On Sparse Coprime Sensing

Sampling,which is the key module at the head of digital signal processing system,convert the simulative signal in nature to the digital sequence.According to the Nyquist Theorem,the sampling rate has to be larger than two times of the highest frequency of the band-limit signal.People have been finding a more sparse sampling methods,the coprime sampling,as one of effective sparse sampling methods.has,in the past,been used in some signal processing applications such as range improvement in radar,and for identifying sinuoids in noise.Recently,many Theory about the sparse coprime sensing have arisen,such as the DOA(Direction Of Arrival)estimate based on the sparse coprime sensing,system identification based on the sparse coprime sensing,autocorrelation estimation with coprime samplers and arrays and power spectrum estimation with coprime DFT filter banks.Here we raise two methods for spectrum selective channel estimation based on sparse coprime sensing.We first introduce the concept of coprime sampling and the theory of LTI system identification based on sparse coprime sensing,then we use this algorithm to identify the FIR filter,analyze the algorithm in detail,and compare it with the traditional LMS channel estimating algorithm.For the weakness of system identification based on sparse coprime sensing on the type of channel,we raise the algorithm of channel estimation based on coprime sampling for input stream and output stream.Through coprime sampling for the input stream which contains multi-frequency at any arbitrarily rate,We can get the high radio solution autocorrection of the input stream,and do the FFT(Fast Fourier Transmission)for the autocorrection to get the power spectrum,then do the soft threshold filter for power spectrum to discard the false peaks,similarly,the output stream power spectrum can also be made out,final y we can get the channel spectrum amplitude with the computation of input stream and output stream power spectrum.Theoretical calculation and simulation are made for the method.The theory of power spectrum estimation based on coprime DFT filter banks shows that we can use a N-order DFT filter bank and a M-order filter bank to estimate the signal power spectrum directly,instead of doing the FFT for autocorrection.Compare with traditional DFT analyze filter bank which needs MN filters.This theory greatly reduce the number of filters.Channel estimation based on coprime DFT filter banks arise in this paper,this algorithm can analyze the regular input stream to get the MN point power spectrum based on coprime DFT filter banks,similarly,we can get the MN point power spectrum of output stream,then,the channel spectrum amplitude can be computed out.Considering the extra false value,the regular input stream should be designed appropriately.Theoretical calculation and simulation are made for the method.