Radars on multiple distributed airborne or ground based moving platforms are of increasing interest, since they can be deployed in close proximity to the event under investigation and thus offer remarkable sensing opportunities. Ground moving target indicator (GMTI) detects and localizes moving targets in the presence of ground clutter and other interference sources. Space-time adaptive processing (STAP) implemented with antenna arrays has been a classical approach to clutter cancellation in airborne radar. One of the challenges with STAP is that the minimum detectable velocity (MDV) of targets is a function of the baseline of the antenna array: the larger the baseline (i.e., the narrower the beam), the lower the MDV. Unfortunately, increasing the baseline of a uniform linear array (ULA) entails a commensurate increase in the number of elements. An alternative approach to increasing the resolution of a radar, is to use a large, but sparse, random array. The proliferation of relatively inexpensive autonomous sensing vehicles, such as unmanned airborne systems, raises the question whether is it possible to carry out GMTI by distributed airborne platforms. A major obstacle to implementing distributed GMTI is the synchronization of autonomous moving sensors. For range processing, GMTI processing relies on synchronized sampling of the signals received at the array, while STAP processing requires time, frequency and phase synchronization for beamforming and interference cancellation. Distributed sensors have independent oscillators, which are naturally not synchronized and are each subject to different stochastic phase drift. Each sensor has its own local oscillator, unlike a traditional array in which all sensors are connected to the same local oscillator. Even when tuned to the same frequency, phase errors between the sensors will develop over time, due to phase instabilities. These phase errors affect a distributed STAP system. In this dissertation, a distributed STAP application in which sensors are moving autonomously is envisioned. The problems of tracking, detection for our proposed architecture are of important.
The first part focuses on developing a direct tracking approach to multiple targets by distributed radar sensors. A challenging scenario of a distributed multi-input multi-output (MIMO) radar system (as shown above), in which relatively simple moving sensors send observations to a fusion center where most of the baseband processing is performed, is presented. The sensors are assumed to maintain time synchronization, but are not phase synchronized. The conventional approach to localization by distributed sensors is to estimate intermediate parameters from the received signals, for example time delay or the angle of arrival. Subsequently, these parameters are used to deduce the location and velocity of the target(s). These classical localization techniques are referred to as indirect localization. Recently, new techniques have been developed capable of estimating target location directly from signal measurements, without an intermediate estimation step. The objective is to develop a direct tracking algorithm for multiple moving targets. It is aimed to develop a direct tracking algorithm of targets state parameters using widely distributed moving sensors for multiple moving targets. Potential candidate for the tracker include Extended Kalman Filter.
In the second part of the dissertation,the effect of phase noise on space-time adaptive processing in general, and spatial processing in particular is studied. A power law model is assumed for the phase noise. It is shown that a composite model with several terms is required to properly model the phase noise. It is further shown that the phase noise has almost linear trajectories. The effect of phase noise on spatial processing is analyzed. Simulation results illustrate the effect of phase noise on degrading the performance in terms of beam pattern and receiver operating characteristics. A STAP application, in which spatial processing is performed (together with Doppler processing) over a coherent processing interval, is envisioned.
|
![[NJIT logo]](http://archives.njit.edu/vhlib/images/njit_logo.gif){kind=logo}
![[Van Houten Library logo]](images/vhl-top-image.jpg)
