Research On Multipath Delay-spread Estimation Algorithm In OFDM Communication System

With the development of China mobile Internet technology,LTE technology has become a research hotspot in current stage.As the core technology of the LTE,Orthogonal Frequency Division Multiplexing(OFDM),which has a high data transmission rate,can effectively overcome the frequency selection in multipath fading channels and has been applied to many aspects of people's life.However,adjacent symbols may be overlapping because of the multipath effect.In this case,OFDM symbol timing synchronization error would lead to the wrong position of the FFT processing window.One window would contain the values of two consecutive OFDM symbols,which brings the InterSymbol Interference(ISI)and InterCarrier Interference(ICI)and largely influences the performance of OFDM systems.As a solution to the problem,multipath delay-spread estimation in OFDM system can be applied to the symbol timing recovery,interference elimination and so on.This paper proposes a low complexity delay-spread estimation algorithm in OFDM system under multipath fading channels.Firstly,the paper presents the basic principle of OFDM system and the mathematical model of the multipath fading channels.Based on the channel model,we reviews several typical blind delay-spread estimation algorithms.Generally those algorithms calculate the correlation of each sample in the received signal symbols with its N-th later sample to estimate the delay-spread.For example,Karthik's algorithm uses the cosine similarity as the correlator and the complexity is high.Then,we propose a new blind delay-spread estimation algorithm with low computational complexity,which won the best student paper award in HMWC 2014.It uses the Manhattan distance to calculate the correlation of samples to estimate the multipath delay-spread.The computational complexity for multiplication is reduced from polynomial time to zero,so the whole complexity is greatly reduced.Compared with the cosine similarity,which is used for calculating the angle between two vectors,Manhattan distance obtains the correlation by calculating the difference between two vectors,hence it has a higher precision.The simulation results verify that our algorithm has better performance in accuracy than Karthik's.In the end,this paper establishes a mathematical model of the threshold's effect on estimation error,theoretically illustrates the difficulty to obtain the optimal threshold and proposes an algorithm to obtain the suboptimal threshold based on L1-norm.It effectively improves the accuracy of our algorithm.