Nonlinear Distortion Recovery Based On Compressed Sensing In OFDM System

Orthogonal Frequency Division Multiplexing(OFDM)signal has the problem of high peak to average power ratio(PAPR),which makes the High Power Amplifier(HPA)enter the saturated region.It introduces nonlinear distortion and deteriorate the Bit Error Rate(BER)performance of the system.Therefore,how to suppress PAPR at the sending end is particularly important.PAPR suppression technology can be divided into three categories: signal distortion,multi-signal and probability,and coding.Among them,the amplitude clipping technology of signal distortion technology is widely used in PAPR suppression due to its low complexity.However,clipping is a nonlinear process,which leads to in-band distortion and out-of-band leakage.The former reduces the BER performance of the system,while the latter interferes with the adjacent channel spectrum.Although the filtering can filter out of band leakage,the nonlinear distortion caused by clipping still exists.With the extensive research of compressed sensing(CS)in sparse signal processing,CS algorithm can be used to reconstruct the sparse signal in time domain.Based on this,this thesis uses CS algorithm to reconstruct the clipping noise.The specific research contents are as follows:1.The improved scheme for clipping noise cancellation in compressed sensing.Existing CS reconstruction scheme can improve the BER performance of clipping noise,however,when reliable compressed observation values are selected,they are easily disturbed by noise folding phenomenon,which leads to lower reconstruction accuracy of sparse signals and lower BER performance.In order to make up for the loss of BER performance,the number of compressed sensing observations needs to be increased,but it increases the computational complexity of the system,in view of the above problem,this thesis first of all,once the basic characteristic of the clipping noise was deduced,and then,analyzing the characteristics of the clipping noise distribution,finally,to estimate position of clipping noise,the score as compression perception index of refactoring.The simulation results show that the improved scheme can effectively reconstruct the clipping noise and improve the BER performance of the system with fewer compressive sensing observation vectors.2.The ?-law-based compressed sensing iterative clipping noise elimination scheme.When filtering out-of-band leakage through filtering,it causes signal peak regeneration.In order to achieve the ideal PAPR,iterative cilpping and filtering(ICF)is often required,and each iteration introduces a different degree of clipping noise,degrading system BER performance.To solve the problem of clipping noise introduced by iteration,this thesis first analyzes the basic characteristics of clipping noise generated in each iteration,and then uses the noise enhancement factor to quantify the iterative introduction of clipping noise based on the clipping noise generated in the first iteration.The ?-law companding algorithm sets the compression threshold and compresses the quantized clipping noise,so that the compressed clipping noise still has approximately sparse characteristics,and satisfies the condition of CS reconstruction.The simulation results show that the proposed scheme not only effectively eliminates the clipping noise introduced by iteration,improves the BER performance of the system,but also obtains a better PAPR suppression effect than traditional ICF.