Research On Data Compression Algorithms For Railway Wireless Communication Air Interface Monitoring

When monitoring the air interface of railway communications,a large amount of air interface data needs to be stored in order to reproduce interference,which will inevitably bring huge storage resource consumption.Therefore,compressing air interface data helps reduce storage pressure.However,the compression ratio of some common compression algorithms is not high when applied to railway air interface data.It is extremely necessary to design a compression algorithm suitable for railway air interface data.This thesis first verifies and analyzes the performance of common compression algorithms such as Huffman coding and LZMA,and the LEC algorithm in railway air interface data compression,and discusses the impact of data accuracy on the compression ratio of each algorithm.Based on this,the thesis designs and implements waveform dictionary compression algorithm,simulates and analyzes the applicability of waveform dictionary compression algorithm in railway air interface data compression.Finally,the thesis explores sparse representation for air interface data compression,and improves the sparse reconstruction algorithm and dictionary generation algorithm.The performance is verified through simulation analysis.The thesis simulates common compression algorithms such as Huffman coding,LZMA and LEC algorithms under different parameter,and the results show that the compression ratio does not exceed 15%.The compression ratio of each compression algorithm has been improved by reducing the data accuracy.At this time,the algorithm with the highest compression ratio is the PPmd algorithm,which achieved a compression ratio of 41%.Because the LEC algorithm does not fully utilize the data information,the thesis modifies the group number encoding and output data structure of the LEC.Combining modified LEC and LZMA2 algorithms obtains a compression ratio of about 48%.Due to the low compression ratio of lossless compression algorithms such as Huffman coding,LZMA,and LEC algorithms,this thesis proposes a lossy compression algorithm based on pattern matching——waveform dictionary algorithm.This method segments the input data and matchs these segments to dictionary waveform,and then use the dictionary index corresponding to the dictionary waveform that approximately matches the data segment to replace the data to achieve compression.Simulation analysis shows that the compression ratio is about 84% when the matching threshold is0.06 times the maximum amplitude.The comparison of the spectrum and the demodulated data before and after compression verify the consistency of the reconstructed data with the original data.The thesis analyzes the influence of threshold on reconstruction error.Inappropriate thresholds may cause low compression ratio or excessive reconstruction errors.Secondly,the dictionary capacity will also affect the compression ratio,and inappropriate dictionary capacity will cause the compression ratio to decrease.Although the waveform dictionary algorithm can obtain a high compression ratio,it does not have the capability of random decompression.Therefore,this paper researches sparse representations and explores two main problems of sparse representations: dictionary design and model solving algorithms.In terms of dictionary design,in view of the weak anti-noise ability of the dictionary produced by KSVD algorithm,the thesis proposes to use a hybrid dictionary to improve the anti-noise and anti-interference ability of the dictionary;in the model solving algorithm,in order to solve the problems existing in the OMP algorithm and the SP algorithm,an improvement on the SP algorithm is proposed.Simulation results show that the improved algorithm reduces the reconstruction error of the algorithm and improves the anti-noise and anti-interference ability.Because compressed sensing can also achieve the purpose of random decompression,and can be sampled and compressed at the same time,the thesis also studies compressed sensing,and compares the compressed sensing reconstruction error of the KSVD basis,FFT basis,and DCT basis at different sampling rates.It is obtained that the reconstruction error of the dictionary produced by KSVD algorithm is the smallest.The thesis analyzes the reconstruction error of compressed sensing and sparse representation at the same compression ratio.The performance of sparse representation is better than compressed sensing.Therefore,it is recommended to use the sparse representation method for railway air interface data compression.This method can obtain a compression ratio of about 77%,similar to the compression ratio of the waveform dictionary,and has a random decompression capability.