Research On Channel Estimation Techniques For OFDM Systems In Doubly Selective Fading Channels

The future wireless communication system should be able to meet the needs of broadband wireless communication in high-speed mobile scenarios.For this kind of time-frequency dual selective fading channel,single carrier communication system needs complex adaptive equalization algorithm to resist the inter symbol interference in fading channel.As a multi carrier transmission technology,orthogonal frequency division multiplexing(OFDM)technology has been widely used because of its high spectrum efficiency and strong anti-multipath ability.It can effectively suppress the frequency selective fading caused by multipath effect by inserting cyclic prefix,but it is very sensitive to the time selective fading caused by Doppler effect.A little Doppler frequency shift will destroy the orthogonality between the subcarriers of OFDM system,resulting in serious inter carrier interference(ICI).When traditional channel estimation techniques are applied to OFDM systems in double selective fading channels,they need to occupy a large number of pilot resources,and the performance is poor,especially the system time-varying is strong,the performance deteriorates seriously.To solve the above problems,this paper first studies the channel estimation method based on compressed sensing(CS),analyzes the statistical characteristics of inter carrier interference caused by Doppler spread and its distribution among subcarriers,and points out that the ICI received by a certain subcarrier mainly comes from a few neighboring subcarriers.In order to suppress ICI,protective pilots are inserted on both sides of the effective pilot;the basis expansion model(BEM)is used to model the double selective fading channel to reduce the parameters to be estimated;the distributed compressed sensing theory is used to recover the joint sparse base coefficient vector,and the discrete long ellipsoid sequence is used to smooth it.The simulation results show the superiority of the algorithm.Secondly,the channel estimation method based on deep learning is studied.Using deep neural network,the prior channel data is used to train the learning network off-line through supervised learning,and the whale optimization algorithm is used to search the optimal learning rate of the network,then the trained network is used for online channel estimation.Through off-line training and on-line estimation,the estimation error caused by linear interpolation without considering channel time-varying is corrected,and the channel estimation accuracy and system error performance are further improved.Finally,some top-level design and implementation of the above channel estimation algorithm are carried out on FPGA to further verify the effectiveness of the algorithm.