Research On Key Technology Of Sparse Channel Estimation In Ground-air OFDM System

The ground air communication system is very important to the Unmanned Aerial Vehicle(UAV),which implements the information interaction between the mission payload and the ground control station.The broadband is an inevitable trend.The Orthogonal Frequency Division Multiplexing(OFDM)is one of the key technologies of the new generation wireless mobile communication system,which can cope with the frequency selective fading of wireless channel effectively,with high frequency utilization rate.It can meet the requirement of large capacity downlink data transmission in UAV communication system.However,the UAV channel not only has frequency selective fading characteristics,but also has a time-varying characteristic with the change of the flight state and environment.The time-frequency double selective fading channel affects signal transmission seriously.So the channel estimation is necessary to complete the coherent demodulation and ensure the signal transmission.At present,there have been some research achievements on channel estimation in high speed mobile environment of wireless OFDM system.However,the traditional channel estimation method in frequency domain needs large number of pilots in large delay scenario,and reduces the transmission rate seriously.In recent years,the channel estimation method based on compressive sensing(CS)can save the number of pilots,but there are still some issues to be researched further,including the pilot design,the elimination of inter-carrier interference(ICI),and the improvement of sparse recovery algorithm efficiency.Aimed at the Ground-Air communication system of UAV,several key problems of the channel estimation based on the CS are researched in this paper,including the pilot design,the sparse recovery method based on linear time-varying(LTV)model,the ICI cancellation and the improvement of the sparse recovery algorithm,specially as following:First,the model and sparse characteristics of the UAV ground-air channel are studied.On the basis of the existing measurement results of UAV ground-air channel,the channel model and parameter generation method are improved,which provide a verification model for the following methods of the paper.Meanwhile,the sparsity of the UAV ground-air channel under different time delay and different system sampling period are analyzed,and the result that the taps of channel impulse response have obvious sparsity in the time domain,which provides the basis for channel estimation based on CS.Second,using double criteria,a new pilot optimization algorithm based on cyclic parallel tree is proposed.After studying the current pilot design criteria under different definitions of the measurement matrix mutual correlation,it is found that all the existing criteria have their limitations.For the new pilot optimization algorithm proposed here,in the iteration process,the traditional pilot design criteria with less computation is used to optimize one location by one,which is based on the minimization of the maximum value of the inner product between different columns in the measurement matrix.After each iteration,the new criterion is used to determine the initial node of each branch,taking the mutual coherence of all columns of the measurement matrix into consideration.If either index is no longer smaller,the iteration is over.Third,a new sparse channel recovery method based on the LTV model is proposed.The UAV ground-air channel belongs to the LTV channel model,the gain of each tap varies linearly with time during several continuous OFDM symbol.Based on this,the OFDM time-varying channel estimation is equivalent to a CS model,and the average of the tap gain during one symbol is constructed,or based on the DCS(Distributed Compressed Sensing),the averages of the tap gain during several successive symbol periods are reconstructed at the same time.Then,the slope of tap gain during the current symbol period can be gotten by the average of the two adjacent symbols.Finally,according to the transmission relationship of OFDM system under LTV model,or the traditional ICI iterative cancellation method,the sent data can be gotten.Fourth,based on LTV model,the new ICI elimination method is proposed.The expression of ICI in frequency domain under LTV model is derived.Based on the analysis of ICI characteristics,a sparse reconstruction model based on pilot coding and the grouped pilot optimization algorithm are proposed,which uses the special structure of pilot-pair,to eliminate the influence of the un-known data sub-carries on the pilot sub-carriers.Meanwhile,based on the characteristics of the ICI in LTV channel model,the complex matrix operation in traditional ICI iterative elimination method is changed to look-up and several multiplication operations.While keeping performance nearly unchanged,the computation is reduced significantly.Finally,an improved OMP algorithm based on search space pre-processing is proposed.Base on the characteristics of non-zero tap position during adjacent symbol period,the probability distribution of non-zero tap position of current symbol is calculated by the channel information of the previous symbol,then the search space of the OMP algorithm is divided into priority set and supplement set.Due to the strong correlation of non-zero tap positions between adjacent symbol periods,for most iteration,only the priority set is used,so the computation of OMP algorithm is reduced.At the same time,the variation characteristic of the residual value of two successive iterations is analyzed,and the threshold method is used to judge the search set switching and iteration termination.The improved OMP algorithm realizes the adaptation to the new burst tap and the sparsity of channel.