Research On Signal Detection Technology Under Low SNR

As the communication environment becomes more and more complex and changeable,the signal at the receiving end is often mixed with strong background noise,making the received SNR often very low.In such an environment with low SNR,the performance of detection algorithms declines seriously or even fails.Based on the existing detection algorithms,this paper is devoted to research a new signal detection algorithm with good detection performance at low SNR.Firstly,based on the analysis and research of the detection algorithm based on power spectrum density,an adaptive signal decomposition method,which called variational mode decomposition,was introduced into the field of signal detection for the first time,and a signal detection algorithm based on variational mode decomposition was proposed.For the detection algorithm based on power spectrum density,the division of power spectrum density is often made by artificial and which is very subjective.With the introduction of the variational mode decomposition method,the power spectrum density is divided adaptively from the view of signal processing to make it more precise and accurate.The power spectrum density is divided in this way can also obtain certain denoising effect.The signal to be measured is first decomposed by the variational mode decomposition,and then the obtained intrinsic mode function is estimated by power spectrum,and an appropriate test statistic is constructed.The simulation results show that the proposed algorithm can still achieve a high detection probability at low SNR.Secondly,a signal detection algorithm based on power spectrum entropy is proposed,which based on the difference of information entropy between the signal and noise.In order to gain a better estimate of the power spectrum entropy,Bartlett periodic diagram method is used to estimate the power spectrum density,which can significantly reduce the estimated variance and make the power spectrum more smoother.However,the resolution of power spectrum density is also reduced at the same time,and it is found that there is a compromise between the two conditions,that is,when the number of segments K is equal to the number of data points of each segment N,the best estimated power spectrum density can be obtained.Then the histogram method is used to calculate the power spectrum entropy,and it is used as the test statistic of the algorithm.The simulation experiment shows that the algorithm can achieve the detection probability of more than 98% at the signal-to-noise ratio of-15 d B,and the detection effect does well in the low SNR scenario.Furthermore,due to the limited performance improvement of single-node detection algorithm,multi-user collaboration is considered to further improve the detection performance.A power spectrum entropy signal detection algorithm based on multi-user collaboration is proposed.Firstly,in order to improve the robustness of the algorithm,a two-threshold decision strategy is adopted for the local detection algorithm based on power spectrum entropy.Hard decision fusion is adopted for the sensor nodes that can make a decision directly in the local place,and soft decision fusion is adopted for the sensor nodes that cannot make a direct decision.Then the final statistical decision is made according to the soft and hard merging strategy proposed in this paper.Simulation results show that compared with single node detection,the detection performance of multi-user cooperative detection algorithm can be significantly improved.The detection performance increases with the number of users participating in the collaboration.However,considering the practical application,the performance improvement cannot be achieved only by increasing the number of cooperative users.Therefore,it is necessary to select an appropriate number of cooperative users,which can not only improve the reliability and accuracy of detection algorithm under low SNR,but also make the whole detection system not consume too much communication overhead.