Research Of Frequency-Hopping Signal Detection Technology Based On Sparsity

Frequency-Hopping communication is a common method of communication in the field of modern wireless communication countermeasure.As the Frequency-Hopping has good anti-jamming,anti-interception and strong networking ability,making itself very popular in military radio communication and civilian mobile communication.Therefore,the research on the technology of Frequency-Hopping signal analysis and processing becomes particularly important.On the basis of reading a large amount of literature,this paper focuses on Time-Frequency analysis and parameter estimation of Frequency-Hopping signals based on sparse principles.Frist of all,the principle and main parameters of the Frequency-Hopping system are briefly introduced to provide a theoretical basis for the follow-up research.At the same time,several traditional Time-Frequency analysis techniques,such as Short-Time Fourier Transform distribution,Wigner-Ville Distribution and its improved distribution are discussed in-depth.Combined with simulation the factors which affect the Time-Frequency analysis are discussed,meanwhile the performance of these Time-Frequency analysis algorithm is compared.Secondly,in the side of Time-Frequency analysis,this paper focuses on a Time-Frequency analysis algorithm based on sparse reconstruction for Frequency-Hopping signal.According to the dual sparsity characteristic of Frequency-Hopping signal,this algorithm establishes a sparse reconstruction problem model of Time-Frequency matrix and then the steepest descent method is used to solve the model.In order to solve the l₀ norm problem which is non-convex and difficult to solve directly,this paper proposed a function-T-Sigmoid function,which can approximate the l₀ norm.Through theoretical analysis and algorithm simulation,the Time-Frequency analysis effect of approximating l₀ norm of T-Sigmoid function and traditional Time-Frequency analysis algorithm are compared and analysis.Then further study the difference between the T-Sigmoid function and the existing approximation l₀ norm function-Gaussian function in approximate accuracy,Time-Frequency matrix estimation accuracy and computational complexity.At the same time,the performance of sparse reconstruction algorithm in parameter estimation is analyzed.The results indicate that the approximation l₀ norm of T-Sigmoid function has good Time-Frequency estimation effect and strong real-time performance.Finally,aiming at the parameter estimation of Frequency-Hopping signal,this paper studies a parameter estimation method of Frequency-Hopping signal based on sparse linear regression.In view of the original algorithm is easily disturbed by noise,and it can only estimate the amount of time and can not estimate the frequency directly.In this paper,an optimized and improved algorithm is proposed.Compared with the original algorithm,by reconstructing the Time-Frequency matrix and then maximizing the Time-Frequency matrix by column,the most of the noise is eliminated.The simulation results show that the robustness of the optimization algorithm is better,which not only improves the estimation accuracy of the frequency hopping period and the timing offset,but also accurately estimates the frequency of the Frequency-Hopping signal and then restores the Frequency-Hopping pattern.However,the disadvantage of the optimization algorithm is that it can only estimate the parameters of a single Frequency-Hopping signal,and it is no longer suitable for the multi-component signals.