In this paper, we tackle the problem of selecting the radar node positions to provide an estimate of the target state vector with a prescribed accuracy. The topology optimization problem is formulated as selection of a fixed number of radar node positions from a set of available ones, where the radar observations are modeled by a general non-linear model. We further propose a topology optimization framework for the simultaneous estimation of the multi-modal parameter vector. In particular, the task of joint position and velocity estimation is considered. The feasibility of the proposed approach is demonstrated for several cost functions, namely the frame potential as well as the log-determinant and maximum eigenvalue of the error covariance matrix.

The ultimate goal of any sensing system is to build situation awareness. Existing solutions for a single radar node that have to assure extended areas of coverage with high resolution measurements (in range, cross-range, and Doppler) are physically cumbersome (large antenna size) and typically require large operational resources (high transmit power, wide bandwidth and long integration time). Combining data from multiple spatially separated nodes located at several locations offers a possibility to use radars with low-cost omnidirectional antennas to cover wide areas and overcome operational limitations such as sector blockage due to landscape or high-rise buildings. Thus, performance of the complete system becomes dependent not only on the parameters of a single radar node, but on the number of nodes and their location (system topology) as well. A proper selection of both node-related (transmit power, operational frequency and bandwidth, integration time, etc.) and system-related (node location, node cooperation) resources is an important design task, which forms the major focus of this thesis. The first part of this dissertation is dedicated to the development of the radar network performance assessment tool,while the second part provides the framework for radar network topology optimization. The potential accuracy of the target parameters estimation has been used for radar network performance assessment. The developed tool incorporates parameters of a single radar node as well as system parameters (positions of the nodes and their cooperation), evaluated using Cramér-Rao lower bound. Using the tools developed, performance of different types of radar networks have been studied and compared in this thesis. For the radar network topology optimization several convex and greedy algorithms have been used, making the optimization approach versatile. Validation and performance comparison of the optimization algorithms have been performed in this thesis. The results obtained in this research can be used to evaluate the potential performance of radar networks for different applications and provide a solution to key problems of their topology design.

The bistatic radar network has significant advantage over the network of monostatic radars in terms of the system cost. Moreover, the separation of transmit and receive radar units lead to better system performance in contrast to monostatic radars. However, the number of transmit nodes is often limited by the system costs and available resources. To address this issue, we present the approach of topology optimization of bistatic radar network with limited number of transmit nodes. This problem is formulated as convex optimization problem, where the topology of bistatic radar network with fixed number of transmit nodes and minimal number of receive nodes is selected in order to meet prescribed localization accuracy requirements.