Research On Separation And Parameter Estimation Of Signals From Radar Target With Micro-motions

When a radar target is moving, the target or structures on the target usually have micro-motions. The target has small vibrations, rotations or other high-order motions in addition to its translation, this is called mico-motions. Micro-motions on the target will induce a micro-Doppler shift in frequency domain with respect to the sent signal; this micro-Doppler contains the unique micro characteristics of the target, which can be potentially used as an important tool in radar detection and recognition. Recently, researches on micro-Doppler signal detection, separation, parameter estimation and feature extraction from micro-Doppler signals have attracted a lot of interest and become a hot topic in both radar engineering community and radar academia.The dissertation investigates the technologies of separation and parameter estimation of micro-Doppler signals reflected from both weak and strong micro-motions. The work contains two parts. In the first part, taking the chest vibration of human as a typical example, we design a continuous wave Doppler for the detection of human respiration and heartbeat, and investigate the algorithms for separation and parameter estimation of micro-Doppler signals reflected from weak micro-motions of a target. In the second part, based on the point scatter model and time frequency analysis, the algorithms for separation and parameter estimation of micro-Doppler signals from general strong micro-motions of target are studied.In detail, the work of this dissertation can be summarized as follows:1Separation and parameter estimation of micro-Doppler signals caused by weak micro-motions based on spectal analysis and harmonics model(1) A2.4GHz continuous wave Doppler radar is designed to study the weak vibration of human chest. Experiments indicate that after the respiratory signals are separated from heatbeat signals by band-pass filters, the Doppler radar can estimate both respiration and heartbeat rates of a human subject at a short distance about1meter.(2) When several weak point scatterers are simultaneously present, the harmoics of stronger point scatterers may overwhelm the micro-Dppler signals of weaker point scatterers. Therefore, taking human chest movement as an example, two methods for removing the harmonics are proposed. The first algorithm firstly estimates the respiration rate, and produces the harmonics of respiration adaptively to cancel the harmonics mixed with heartbeat signal. The second algorithm decomposes the harmonics into different intrinsic mode functions based on empirical mode decomposition, then, the harmonics are removed by auto-correlation creteria. Experimets show that both two methods can effectively extract the heartbeat signal from harmonics of respiration, and more accurate heartbeat rates can be obtained compared with the traditional band-pass filters.(3) Since the conventional time frequency distribution lacks enough time frequency resolution to analyze the non-stationary signal from weak micro-motion, an improved Hilbert-Huang transform is presented by combination of Hilbert transform with the algorithm of harmonics cancellation based on empirical mode decomposition. The experiments on the non-stationary respiratory signal show that the algorithm can characterize the time varying frequency of respiration more accurately.(4) Based on the relationship between the amplitude of the harmonics component and the amplitude of vibration, an extended algorithm for estimating the amplitude of vibration using ratios between amplitudes of spectral peaks of harmonics is proposed. Both experiments on human chest and simulations show that the algorithm can estimate the amplitude of weak vibration accurately in the case of high signal-noise-ratio.2Separation and parameter estimation of micro-Doppler signals caused by strong micro-motions based on point scatter model and time frequency analysis(1) Since the components of micro Doppler signals are overlapped in the time frequency domain, an algorithm for separation of micro-Doppler signals composed of linear frequency modulation (LFM) components and sinusoidal frequency modulation (SFM) components is proposed based on time frequency filter (TFF). To determine the masking function in TFF, Viterbi algorithm is adopted to estimate the instantaneous frequency at first, and then the components are filtered out one after another. Simulation results indicate that this algorithm can effectively separate the micro-Doppler signals even in a high noise environment, and compared with the conventional empirical mode decomposition and wavelet decomposition method, the algorithm produces more accurate results without significantly more computational complexity.(2) When the Hough-TFD technique is applied to parameter estimation of micro-Doppler signals, it needs several dimensional searches, resulting in much more computational burden. Therefore, this dissertation proposes a novel Hough-CPF method based on cubic phase function (CPF). The technique needs only one dimensional search to determine the parameters of the LFM signal, and2-dimensional search for parameter estimation of the SFM signal. Simulation results indicate that the method has almost the same SNR performance as Hough-WVD, but is computationally efficient.