Multistatic Low Frequency Ground-based Radar Moving Target Detection In Foliage Environment

The low frequency band and multistatic radar system is the future development trend of ground moving target reconnaissance radar system. On one hand, the low frequency band can improve the battlefield transparency in one-way situation. On the other hand, the multistatic radar is of high maneuverability, concealment and survival ability. Accordingly, the combination of the low frequency band and multistatic radar system will provide strong support to obtain the battlefield information and to master the battlefield situation. Considering the advantages of the combination of the low frequency band and multistatic radar system, weak moving target detection in foliage environment is researched in this dissertation based on a multistatic low frequency ground-based radar. In particular, a series of approaches are put forward, which are verified to be feasible and effective with the real data and simulation experiments. The research achievements of this dissertation have important theoretical significance and engineering value to improve the moving target detection performance of multistatic low frequency ground-based radar. The main contents of this dissertation are organized as follows.1. With the low frequency wide band ground-based radar, the amplitude of the foliage clutter in three typical foliage environments of different densities are characterized and parameterized. Firstly, several statistical models along with their k-th order moments are considered which may be designated as “curve fit” models and adapted to the foliage environments clutter. At the same time, the parameter estimation methods and goodness-of-fit(Go F) tests are developed for all the considered comparative models, which provide a theoretical basis for the following research of real data. Secondly, the aforementioned state-of-the-art clutter models are applied to these three data sets which were acquired in the three typical different foliage environments, including line-of-sight(LOS) forest environment, non line-of-sight(NLOS) forest-penetration environment, and NLOS bush-penetration environment. Then an appropriate model is chosen and tested as the most suitable for the real clutter data sets, respectively.(1) For LOS forest environment, the clutter reflected by the dominated tree trunks, causes high kurtosis and long tails, which is suitably characterized by the log-logistic model.(2) The NLOS forest-penetration environment clutter contains mainly the shrubs clutter and a little echo reflected by the trunks behind the shrubs, which can be seen as a moderately heterogeneous area. As a result, both log-logistic model and G0 model perform very well for the NLOS forest-penetration environment.(3) The bushes clutter is the main source of the NLOS bush-penetration environment clutter, which can be regarded as extremely heterogeneous scenes. Thus G0 and G models are demonstrated to be the best choice to model the clutter in the NLOS bush-penetration environment. Thirdly, the target detection performance in the foliage environment is investigated. On the basis of the real data, the results of Monte Carlo indicate that the more precise knowledge of the clutter statistical characteristics, the better radar target detection performance.2. To detect weak ground targets embedded in foliage environments, two different research directions are undertaken. One direction is to pursue the foliage clutter modeling and analysis in order to gain better understanding of the clutter and improve the detection performance. Based on the clutter models that have been investigated in the previous chapter and goodness-of-fit(Go F) tests, a new detector is proposed. Moreover, the detector is proved to have the constant false alarm rate(CFAR) property under log-logistic distribution. It is shown that in foliage environment clutter the proposed detector outperforms some other well-known CFAR detectors. The other direction focuses on advanced signal processing approaches to support target detection. In particular, a clutter suppression technique based on range alignment is firstly applied to suppress the time-varying clutter and the instable antenna coupling. Then an entropy weighted noncoherent integration(EWNI) algorithm is adopted to mitigate the multipath effects. In consequence, the proposed method effectively reduces the clutter considerably. Based on the high visual quality image, the target trajectory is detected robustly and the radial velocity is estimated accurately with the Hough transform(HT).3. In order to provide a comparison between conventional monostatic phased-array radar and multistatic radar, spatial diversity gain, geometry gain and data fusion rules are fully exploited and discussed. Particularly, a signal model for the multistatic radar in homogeneous colored Gaussian clutter is developed. Besides, considering the data fusion rules that play an important role in the radar detection performance, the target detectors for centralized multistatic radar and distributed multistatic radar are derived, respectively. Some remarkable conclusions can be summarized as follows from the simulation results.(1) The monostatic radar is prone to severe target fading and blind/low radial velocity, and hence it may suffer considerable performance degradation. In contrast, with spatial diversity gain and geometry gain the multistatic radar can overcome the aforementioned problems and then obtain significant gains of the target detection performance.(2) The centralized multistatic radar performs better than the distributed one, but with higher computational complexity.(3) Compared to the equal weighting coefficients, with the signal clutter noise ratio(SCNR) weighting coefficients the multistatic radar has better target detection performance.4. The analysis and optimization of the multistatic radar are studied. Firstly, the impacts of waveform selection, weighting selection strategy and multistatic geometry on the multistatic radar performance are characterized. In addition, based on the monostatic/bistatic ambiguity function(AF), multistatic AF is developed and formulated, which provides the basis for a complete description of a given multistatic radar system. Secondly, with the well-known relationship between the Cramér-Rao lower bound(CRLB) and the AF, that is, the CRLBs on estimation accuracy in delay and Doppler are proportional to the inverse of the second derivatives of the AF with respect to delay and Doppler, the multistatic CRLBs on the estimation accuracy of the target range and velocity for multistatic radar are derived, which serves as a guideline for multistatic radar analysis. At last, using genetic algorithm(GA) an approach for sensor placement of multistatic radar is put forward. Exactly, the multistatic CRLBs for range and velocity estimation are derived as the fitness functions. Then GA is used to perform the optimized sensor placement as one of global optimization, suggesting optimal sensor placement strategies to meet required radar performance goals.