Research On Perturbed Compressed Sensing Based Signal Processing In Wireless Communication

Compressed sensing breaks through the limitations of the traditional Nyquist sampling theory and there are large numbers of related achieve-ments emerging in academia and industry.However,research on the de-sign and optimization of wireless communication systems based on the compression sensing theory is still lacking a well-established framework.Besides,the research considering the influence of wireless fading channel-s,the non-ideal characteristics of simulation equipment,and other factors is still far from enough and has a long way to go.On this account,this thesis is based on the perturbed compressed sensing theory,which consid-ers the influence of non-ideal factors and mainly studies the algorithm of compressed sensing in the wireless communication system.Insight of the time-varying channels scenario,the influence of error of channel estimation is modeled as a perturbation in the measurement matrix,thereby a novel system framework is established by the guidance of perturbed compressed sensing.This thesis starts with the general model of perturbed compressed sensing reconstruction and design a reconstruction algorithm based on the prior information.Then,the proposed algorithm is applied on the single-user scenarios to improve the performance of compressed data transmis-sion system by using perturbation compressive sensing.Further,extend-ed to multi-user scenarios,the thesis focuses on the compressed sensing based activation detection problem in the grant-free multi-user uplink ac-cess system.Finally,this thesis also discusses the impact of multi-antenna combined with 1bit quantization on the grant-free multi-user uplink access system based on compressed sensing and the corresponding solutions.The research above not only promotes the application of compressed sensing theory but also significantly improves the performance of the wireless com-munication system.Specifically,this thesis mainly includes the following four aspects of innovation work:1)The thesis proposes a perturbed compressed sensing reconstruc-tion algorithm based on block-sparsity structure.Existing perturbed com-pressed sensing reconstruction algorithms do not make full use of the struc-ture of the original signal to improve reconstruction performance.There-fore,beginning with the block-sparsity signal,this thesis analyzes the in-fluence of unknown perturbation in the measurement matrix during the re-construction process and designs a perturbation offset mechanism that is suitable for the block structure,to eliminate the impact of the perturbation.Using the perturbation offset mechanism,a perturbed compressed sens-ing reconstruction algorithm for block-sparsity signals is proposed,and the corresponding theoretical analysis is given.By utilizing the block-sparsity structure,the proposed algorithm no only improves the tolerance to per-turbation,but also enhances the reconstruction performance.Simulation results indicate that compared with the existing perturbation reconstruction algorithm,the proposed algorithm obtains excellent performance in various scenarios.2)This thesis proposes a compressive data transmission system un-der the slow time-varying channel.In practice,due to the impact of time-varying channels,there are part of the impact remaining in the mea-surements after channel equalization,which seriously damages the perfor-mance of reconstruction.In this thesis,we model the remaining impact of channel or channel equalization errors as a perturbation in the measuremen-t matrix and design a more robust data reconstruction algorithm,by using the perturbed compressed sensing reconstruction algorithm and the semi-blind channel estimation structure.The proposed algorithm can maintain the accuracy of data transmission under the poor channel condition.The proposed algorithm also effectively reduces the need for pilot resources and improves transmission efficiency.Simulation results show that by using the perturbed compressed sensing reconstruction algorithm,the proposed algo-rithm works accurately and robustly with a high transmission efficiency,in the time-varying channel scenario.3)This thesis proposes a multi-user detection and data reconstruction algorithm for the grant-free multi-user uplink access system based on the compression sensing theory.Existing active user detection algorithms in the grant-free multi-user uplink access system,neither fully utilize the tem-poral correlation in user data,nor consider the impact of time-varying chan-nels.In this thesis,we design a new multi-slot user data model,which is more in line with the actual situation and is more conducive to exploit po-tential auxiliary information in the scenario.Based on this model,a novel multi-user detection and data reconstruction algorithm for the grant-free multi-user uplink scenario is proposed.In the proposed algorithm,the channel estimation error caused by the time-varying channel is modeled as the perturbation in the measurement matrix,and the temporal correla-tion between each time slot is exploited to enhance the load capacity of the system and achieve an obvious performance gain.Simulation results show that the proposed algorithms outperform the other active user detection al-gorithms and show robustness in the time-varying channel.4)The thesis proposes a multi-user detection and data reconstruction algorithm for the one-bit quantization grant-free multi-user uplink access system.One-bit quantization can significantly reduce the hardware require-ments and energy consumption,so it has become the future trend of the wireless communication system.However,the active user detection algo-rithms,which are based on the traditional compressed sensing theory,are not suitable for the one-bit quantization case.To deal with this problem,this thesis uses the one-bit compressed sensing theory to transform the ac-tive user detection model into a one-bit compressed sensing reconstruction model.Besides,this thesis also models the channel estimation error in the system as the perturbation in the measurement matrix and proposes a one-bit perturbed compressed sensing reconstruction model to solve the prob-lem of active user detection and data reconstruction problem.The simula-tion results show that,compared with the existing single-bit reconstruction algorithm,the proposed algorithm exhibits a significant performance gain in the single-bit multi-user unlicensed uplink access scenario.