Research On Weak Signal Detection Theory Based On Nonlinear Stochastic Resonance

With the increasing number of wireless communication equipments, coupled with more diverse modulation modes and ever-increasing crowded wireless spectrum, the noise and interference in wireless communication systems have increased dramatically, and the received signal takes on the characteristics of low signal-to-noise ratio (SNR). These factors will have negtive impact on the application of the wireless communication systems both in military and civilian fields, and the detection of weak signal in wireless communication systems is facing severe challenges. Therefore, how to improve the detection performance of weak signals under low SNR conditions is the key problem to be solved.Nowadays, the traditional weak signal detection methods, in the framework of linear system, improve the detection performance by suppressing noise. However, there are some problems with these methods:1. When the noise is suppressed, the useful signal is suppressed simultaneously. The received signal processed by linear methods can not meet the receiving sensitivity requirements, and it is difficult to extract useful signal from strong noise;2. Linear system is the source of noise itself, and more additional noise will be produced when more systems are involved in the procedure of signal processing;3. A linear system does not have the ability to improve the SNR, and the increasing of noise will lead to the decreasing of the output SNR.To ensure the transmission performance of the wireless communication systems, a novel weak signal detection technology based on nonlinear stochastic resonance (SR) is proposed in this paper. SR is a curious nonlinear physical phenomenon which describes that when the nonlinear system, the weak signal forcing and a moderate amount of random noise forcing reach the matching state, there is a cooperative phenomenon taking place:incoherent noise power is feeding into the coherent signal, then the input signal receives assistance from noise to trigger a stronger response from a nonlinear SR system. It shows that the "non-coherent" noise can be used to enhance the transmission of a "coherent" signal of a known form in a nonlinear SR system, revealing a possibility of turning the noise from a nuisance into a benefit, which shows the particular advantage of SR technique in weak signal detection under low SNR.Aiming at the new requirements and new challenges in weak signal detection under low SNR conditions, the full research contents of this paper are mainly divided into the following four aspects:1. Under Gaussian noise, quantitative studies on the mechnism and perfomance of the bistable stochastic resonance (BSR) system response to weak periodic signals and aperiodic binary modulated signals are investigated. For weak periodic signal inputs, through a comprehensive analysis of the signal frequency and the noise intensity’s effect on BSR ststem, an analytical expression of the bistable system parameters is derived, and the SR phenomenon is assured to occur by tuning system parameters. On this basis, a quantitative analysis on the SNR gain of the bistable system is given. For aperiodic binary modulated signal inputs, the response mechanism of the bistable system outputs is analyzed. Through analyses on the quantitative relationship between the bistable system response speed and the symbol interval or modulated frequency, an analytical expression of the bistable system parameters is deduced, and the aperiodic stochastic resonance (ASR) phenomenon is assured to occur by tuning system parameters. On this basis, the output SNR and transmission error rate performance of the BSR-based aperiodic binary modulated signal processing are analyzed.2. Under generalized Gaussian noise, a novel signal detection scheme based on nonlinear threshold system (NTS) is proposed. The algorithm firstly processes the received signal using the NTS, then the characteristics of NTS output signal are analyzed, and finally the bit error ratio (BER) expression of the proposed detection algorithm is derived under the minimum error probability criterion. Simulation results show that under assumption of Gaussian noise, the BER performance of the linear optimal detection algorithm is better than that of the proposed detection algorithm; Under Laplacian noise (non-Gaussian noise) conditions, the BER performance of the proposed algorithm is significantly improved when compared with the linear optimal detection algorithm.3. To improve the performance of non-zero mean signal detection using energy detection at low signal-to-noise ratio (SNR), an improved energy detection (IED) algorithm using generalized stochastic resonance (GSR) is proposed. Firstly, to generate GSR with the (direct current) dc in the signal, a dc component whose optimal amplitude is determined by the offset coefficient is added to the received signal. Secondly, a decision result is given from a comparison of the test statistic calculated by the cumulative energy of the sampled resonant signal and the optimal detection threshold derived under the minimum average error probability criterion. Finally, the performance analyses of the proposed algorithm are given. The theoretical analyses and simulation results show that the error probability performance can be improved under low SNR conditions, and under the same error probability conditions, lower sample numbers are needed for the IED algorithm, when compared with the conventional ED algorithm.4. Aming at the visual and performance requirements on binary image processing under low peak signal-to-noise ratio (PSNR) conditions, a novel binary image enhancement algorithm based on bistable stochastic resonance (BSR) is proposed. Firstly, the two-dimensional binary image pixels are converted into one-dimensional aperiodic binary PAM signal by scanning the image pixels along the row or column direction. Secondly, the aperiodic binary modulated signal is enhanced by the BSR system. Finally, the enhanced one-dimensional aperiodic binary PAM signal is converted into two-dimensional binary image. Simulation results show that the visual effect improvement of the proposed algorithm is superior to that of the traditional binary image enhancement methods such as median filtering, wiener filtering and mathematical morphology processing. For binary image with PSNR to be 7.31 dB, 11.14 dB PSNR improvement is got using the proposed algorithm, and 4.96 dB,2.96 dB and 4.96 dB PSNR improvement can be got using median filtering, wiener filtering and the mathematical morphology processing methods.