Research Of Transmission Technology Based On Filter Bank Multi-Carrier In Wireless Communications

With the rapid development of big data and Internet of things,the amount of information will grow tremendously in modern communication systems,and various new application scenarios are emerging,thus the age of 5G needs to answer these challenges.Faced with the requirements of the ultra-low latency,high spectral efficiency,high transmission rate and business diversity in the 5G,traditional communication systems based on orthogonal frequency division multiplexing(OFDM)cannot satisfy the new application requirements because of out-of-band leakage and low spectral efficiency.Filter bank multicarrier(FBMC)has attracted considerable attention due to its low out-of-band leakage and high spectral efficiency.FBMC has the following characteristics: good out-of-band suppression,no cyclic prefix,non-orthogonal between sub-carriers,flexible control between sub-carriers,suitable for fragmented spectrum utilization.Filtering on each subcarrier,it is a multicarrier transmission scheme with low complexity.Due to application of the filter,it affects the complexity and performance of the FBMC system.At the same time,due to the overlapping structure of time-domain and the inherent interference of the FBMC system,the peak-to-average power ratio(PAPR)suppression,channel estimation and MIMO signal detection techniques for the traditional OFDM system cannot be directly applied to the FBMC system.To solve these problems,this thesis has carried out the following researches and innovations:1.The characteristics of the prototype filter will affect the performance of the FBMC system,and its design must satisfy either perfect reconstruction(PR)or near perfect reconstruction(NPR)conditions.After analyzing the shortcomings of unconstrained condition optimization,this thesis proposes a new method of constrained optimization,which uses the side lobes and ISI/ICI as the constraints to directly optimize the coefficients of the prototype filter.This method simultaneously reduces the stop band energy and the side lobes under NPR conditions.The simulation results indicate that the proposed prototype filter reduces more stopband energy and side lobes compared with the optimized prototype filter of the window function.Using fast orthogonal unitary transform to calculate discrete Fourier transform(DFT)and Walsh-Hadamard transform(WHT),this thesis proposes a low complexity cyclic prefix coding FBMC(CC-FBMC)system.Compared with traditional FBMC systems,this method remarkably reduces the overall complexity of transform and convolution operations.2.The FBMC system has very high PAPR value.Due to the overlapping structure of time domain for FBMC system,the related technology of PAPR suppression for traditional OFDM system cannot be directly applied to the FBMC system.In order to solve this problem,some new methods of PAPR suppression are proposed for the FBMC system.Firstly,a random allocation prototype filter(RAPF)method is proposed to effectively suppress PAPR.Secondly,using the advantages of RAPF algorithm,two hybrid algorithms,the RAPF algorithm with Hadamard transform,the RAPF algorithm with DFT spread spectrum are proposed.The PAPR of the FBMC system is reduced furtherly.Thirdly,the idea of hierarchical search and PAPR estimation algorithm is used.This thesis proposes a low complexity hybrid partial transmit sequence(H-PTS)algorithm for FBMC system.Under certain circumstances,the complexity of C-PTS algorithm is reduced by two orders of magnitude.3.Channel estimation of the FBMC system is a very challenging task.The channel estimation of the FBMC system is more complicated when wireless channel exhibits time selectivity or frequency selectivity.In order to improve the accuracy of channel estimation of the FBMC system,this thesis uses clustering algorithm to solve the DFT channel estimation problem of FBMC system.The clustering algorithm accurately separates the noise inside maximum delay from channel coefficients,thereby improving the channel estimation accuracy of the FBMC system.In this thesis,the AP-NCSVR algorithm is proposed by using support vector regression.It solves the problem that interpolation method of the traditional channel estimation is not accurate.The channel estimation accuracy based on clustering and support vector machine is better than the method of traditional channel estimation at the cost of increasing complexity.4.The channel estimation of the FBMC system is studied by compressed sensing since the wireless channel is sparse.This thesis proposes a sparse adaptive subspace tracking(SASP)algorithm for channel estimation of FBMC systems.For the SASP algorithm,this thesis developed new AP-SASP and coded SASP algorithms to estimate the channel frequency response(CFR)values of the FBMC system.The SASP algorithm can accurately estimate the CFR values in the doubly selective channels,and the computational complexity of the SASP method is reduced by the low complexity of the SP algorithm.Compared with the traditional methods of channel estimation,like AP-OMP,AP-ROMP,coded orthogonal matching pursuit(OMP)and coded ROMP,the BER performances of AP-SASP and coded SASP methods are better than those of AP-OMP,AP-ROMP,coding-OMP and coding-ROMP.The complexity of the SASP algorithm is only 90% of the complexity of the traditional OMP,ROMP,and CoSaMP algorithms.5.In order to solve the problem of FBMC combined with MIMO signal detection,this thesis analyzes the signal detection problems of low-frequency selective and mildfrequency selective scenarios based on the Hadamard coding MIMO-FBMC system.Finally,this thesis proposes a sparse signal detection algorithm of regularized SASP(RSASP)combined with zero-forcing(ZF).Simulation results show that the BER performance of RSASP and ROMP algorithm is better than CoSaMP algorithm.In addition,the RSASP and ROMP algorithms have almost the same BER performance,but the complexity of the RSASP algorithm is lower than that of the ROMP algorithm.