White, Langford BartonGray, Douglas AndrewBalzan, Luke Anthony2013-05-202013-05-202012http://hdl.handle.net/2440/77835This thesis presents a background to and a series of interesting and novel results for a particular proposed model for narrowband MIMO radar systems. The proposed model is both novel and unique, comprising closely-spaced antenna arrays that transmit orthogonally-coded waveforms, and can be seen as a logical extension to existing models for conventional single-input, single-output radar systems. Signal processing and optimisation is performed on the proposed system with a view to yield enhanced estimation and tracking performance. The proposed signal and likelihood estimation models have been thoroughly defined, with a number of important approximations and simplifications to the models gained through the use and exploitation of orthogonally-coded waveforms. All approximations and assumptions have been justified through the use of simulated examples. The Cramer-Rao bound for the models is derived and verified as correct through the use of simulated data. Through comparison of the Cramer-Rao bound to statistical estimation variances obtained through extensive simulations, the proposed models are shown to be efficient, thereby demonstrating the validity of the bound to be used as performance metric for optimisation. With the knowledge that the proposed MIMO radar system is efficient, the Cramer-Rao bound is used as a measure for estimation performance optimisation. The bound is seen to be dependent on the choice of orthogonally-coded waveforms used in the MIMO radar system, and by framing the selection of codes as a convex optimisation problem, codes can be chosen to minimise the Cramer-Rao bound, and since the system has been shown to be efficient, this also reduces estimate variance, thus improving the estimation performance of the system. This optimisation problem has been examined and simulated extensively, with simulated data substantiating the claims of performance improvement. Finally, this thesis explores the idea of tracking for MIMO radar. A Kalman filter based tracker is proposed and simulated for the MIMO radar system. Extending the convex optimisation scheme discussed above, a similar optimisation problem is formed for the case of MIMO radar tracking. The optimisation problem has been simulated to select orthogonal codes for transmitting based on the predicted target motion, obtained from the Kalman filter tracker. By basing the optimisation on the predicted tracker outputs, an action-perception cycle for MIMO radar is established, where the system is able to adapt to its surroundings based on it’s current and predicted view of the environment. Simulations have been used to observe the performance improvements of implementing the optimisation scheme, and thereby showing the action-perception cycle for MIMO radar at work.radar; signal processing; detection and estimation; optimisation; trackingThesis20130410185307