Detection And Parameter Estimation Of Frequency Hopping Signal Based On Compressive Measurements

With advantages such as anti-jam, low probability of interception and easy networking, frequency-hopping communication has been widely used in many civil and military areas. In recent years, in order to enhance the ability of anti-jam, the frequency-hopping communication has a trend towards wideband and high-speed hopping, bringing many problems to hopping interception system which is based on Nyquist sampling architecture. The most serious problems are huge amounts of sampled data in front-end, and low efficiency of transmitting and processing in back-end. The compressed sensing technology can sample the wideband sparse signal at a very low sampling rate without losing information, which can offer new ideas of solving the problem of frequency-hopping signal non-cooperation receiving and processing.This thesis mainly aims at researching the detection and estimation of the frequency-hopping signal parameter based on compressive signal processing (CSP). Compared with traditional methods based on Nyquist sampling processing, algorithms based on compressive signal processing can reduce the computation burden effectively, simplify the signal processing procedure, and improve the system’s timeliness as a consequence. The key work and innovations are summarized as follows:1. When the noise level is known, for the detection of unknown signal in Gaussian white noise, an algorithm called compressive energy detection (CS-ED) is proposed. This algorithm adopts the variance of single compressed sampling to accomplish the detection task based on that the numeral characteristics of the compressive samplings is different under different hypothesis. Simulation results show that CS-ED algorithm can improve the timeliness greatly with a little loss in detection performance compared with traditional energy detection algorithm.2. Proposed an algorithm called compressive auto-correlative detection (CS-ACD) based on CSP when the noise level is unknown. This algorithm makes full use of the sparsity of signal and the restricted isometry property of sensing matrix. The detection task is accomplished based on the different statistical distribution of the auto-correlation vector composed by the sparse coefficients. Simulations show that CS-ACD can have a lower error rate than the detection algorithm after fully reconstruction; compared with existing compressive detection algorithm, CS-ACD can reduce the computation burden by ensuring the detection performance.3. For the presence of only a single frequency-hopping signal, an algorithm that estimates the hopping transition time (CS-HTE) of frequency-hopping signal is proposed based on CSP. This algorithm only needs to reconstruct the first two coefficients which have the maximum weight and judge the duration time of the two hops before and after by the relative size of these two coefficients, and can estimate the hopping transition time through successively compressive sampling. CS-HTE can overcome the disadvantage brought by time-frequency uncertainty principle, and can effectively improve the frequency-hopping parameter estimation accuracy and timeliness.4. For the parameter estimation of frequency-hopping radio networking, a method called compressive spatial time-frequency joint estimation (CS-STFE) based on CSP is analyzed on the foundation of existing direction of arrival algorithm in compressive domain. This algorithm makes use of the sparsity in spatial and frequency domain of the compressive array signal, and can joint estimate multi frequency-hopping signals’ directions of arrival and spectrograms.5. In wideband frequency-hopping signal compressive sampling and processing system, the data sharing and interaction protocols between FPGA and ARM has been given, meanwhile the wideband measurement waveform for compressive sampling controlled by FPGA has been designed and produced, meeting the requirement of frequency-hopping signal compressive sampling and processing system.