基于压缩采样值的跳频信号检测和参数估计

发布时间：2019-02-12 12:29

【摘要】：跳频通信具有抗干扰、低截获和易组网等优点,在民用和军事通信中有广泛的应用。近年来为了提高抗干扰能力,跳频通信有向宽频带、高跳速发展的趋势。这给基于奈奎斯特采样架构的跳频捕获系统带来了诸多问题,最突出的就是前端采样数据量大、后续传输和处理困难。在不损失信息的前提下,压缩感知技术能以极低的速率采集宽带稀疏信号,为解决跳频信号的非协作接收和处理提供了新的思路。本文主要研究基于压缩信号处理(Compressive Signal Processing, CSP)的跳频信号检测和参数估计算法。相对于基于奈奎斯特采样值的传统处理方式,基于少量压缩采样值的压缩跳频信号处理方式能有效的降低运算量,简化信号处理流程,从而提高系统工作的时效性。现将本文主要研究内容和创新点总结如下：1.在噪声水平已知时,针对高斯白噪声中的未知信号检测问题,提出一种压缩能量检测算法(Compressive Energy Detection, CS-ED)。根据单个压缩采样值在不同假设条件下其数字特征不同的特点,该算法将压缩采样值的方差作为判决依据,完成检测任务。实验结果表明,该算法相对于传统的能量检测算法,CS-ED算法用少量检测性能的损失换取了算法时效性较大的提高。2.在噪声水平未知时,提出一种基于压缩信号处理的压缩自相关检测算法(Compressive Auto-Correlative Detection, CS-ACD)。该算法充分利用了信号的稀疏性和传感矩阵的严格等距特性,由稀疏系数自相关向量的不同统计分布进行检测判决。仿真结果表明,在相同的压缩采样次数下,相对于重构原信号后再做检测的算法,CS-ACD算法拥有更低的错误概率；通过和现有压缩检测算法的对比,在信噪比大于-2dB时,CS-ACD算法可在保证检测性能的前提下降低运算量。3.针对仅存在单个跳频信号的情况,提出一种基于压缩信号处理的跳频信号跳变时刻估计算法(Compressive Hopping Transition time Estimation, CS-HTE)。该算法仅需重构单采样周期内.,跳频信号在傅里叶正交基上两个权值最大的稀疏系数,并根据这两个系数的相对大小判定前后两跳持续时间,通过不断的滑动压缩采样即可完成对单个跳频信号的跳变时刻估计。CS-HTE能克服时频不确定性带来的不利影响,可有效提高跳频信号参数估计的精度和时效性。4.针对跳频电台组网工作时的参数估计问题,在现有压缩域波达方向估计算法基础上,给出了一种基于压缩信号处理的跳频信号空时频联合估计算法(Compressive Spatial Time-Frequency Estimation, CS-STFE)。该算法利用压缩阵列信号在空域和频域的稀疏性,可对跳频信号的波达方向和语图进行联合估计。5.在宽带跳频信号压缩采样处理系统中,给出了FPGA和ARM数据共享和交互协议,设计产生了基于FPGA控制的宽带压缩感知测量波形,经测试符合跳频信号压缩采样处理系统的要求。
[Abstract]:FH communication is widely used in civil and military communication because of its advantages of anti-jamming, low interception and easy networking. In recent years, in order to improve the ability of anti-jamming, frequency-hopping communication has a trend of wide band and high speed hopping. This brings many problems to the frequency hopping acquisition system based on Nyquist sampling architecture. The most outstanding problem is that the front-end sampling data is large and the subsequent transmission and processing are difficult. On the premise of no loss of information, compressed sensing technology can collect wideband sparse signals at very low rate, which provides a new way to solve the problem of non-cooperative receiving and processing of frequency-hopping signals. This paper mainly studies the frequency hopping signal detection and parameter estimation algorithm based on compressed signal processing (Compressive Signal Processing, CSP). Compared with the traditional processing method based on Nyquist sampling value, the compressed frequency hopping signal processing method based on a small amount of compressed sampling value can effectively reduce the computational complexity, simplify the signal processing flow, and improve the timeliness of the system work. The main contents and innovations of this paper are summarized as follows: 1. When the noise level is known, a compression energy detection algorithm (Compressive Energy Detection, CS-ED) is proposed to detect unknown signals in Gao Si white noise. According to the different digital characteristics of a single compressed sampling value under different assumptions, the variance of the compressed sampling value is taken as the decision basis to complete the detection task. The experimental results show that compared with the traditional energy detection algorithm, the CS-ED algorithm gains a small loss of detection performance in exchange for a significant increase in the time-efficiency of the algorithm. 2. When the noise level is unknown, a compression autocorrelation detection algorithm (Compressive Auto-Correlative Detection, CS-ACD) based on compression signal processing is proposed. The algorithm makes full use of the sparsity of the signal and the strict equidistance of the sensor matrix, and detects and decides by the different statistical distribution of the sparse coefficient autocorrelation vector. The simulation results show that the CS-ACD algorithm has lower error probability than the original signal detection algorithm under the same compression sampling times. By comparing with the existing compression detection algorithm, when the SNR is greater than-2dB, the CS-ACD algorithm can reduce the computation cost under the premise of ensuring the detection performance. This paper presents a hopping time estimation algorithm (Compressive Hopping Transition time Estimation, CS-HTE) for frequency hopping signals based on compressed signal processing, which only has a single frequency hopping signal. The algorithm only needs to reconstruct the maximum sparse coefficients of the two weights of the frequency hopping signal on the Fourier orthogonal basis, and determine the duration of the two hops according to the relative size of the two coefficients. The jump time estimation of a single frequency hopping signal can be completed by continuous sliding compression sampling. CS-HTE can overcome the adverse effect of time-frequency uncertainty and improve the precision and timeliness of the parameter estimation of frequency hopping signal. 4. Aiming at the parameter estimation problem of frequency hopping station (FH) network, a joint space-time-frequency estimation algorithm (Compressive Spatial Time-Frequency Estimation, CS-STFE) based on compressed signal processing is presented based on the existing DOA estimation algorithm in compressed domain. Based on the sparsity of compressed array signals in spatial and frequency domain, the proposed algorithm can estimate the DOA and speech patterns of FH signals. In the broadband frequency hopping signal compression sampling processing system, the data sharing and interactive protocols between FPGA and ARM are given, and the waveform of broadband compression sensing measurement based on FPGA control is designed. The test meets the requirements of frequency hopping signal compression sampling processing system.
【学位授予单位】：解放军信息工程大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN914.41

【参考文献】