| In this thesis we study crucial problems within complex, large scale, networked control systems and mobile sensor networks. The first one is the problem of decomposition of a large-scale system into several interconnected subsystems, based on the imposed information structure constraints. After associating an intelligent agent with each subsystem, we face with a problem of formulating their local estimation and control laws and designing inter-agent communication strategies which ensure stability, desired performance, scalability and robustness of the overall system. Another problem addressed in this thesis, which is critical in mobile sensor networks paradigm, is the problem of searching positions for mobile nodes in order to achieve optimal overall sensing capabilities.;Novel, overlapping decentralized state and parameter estimation schemes based on the consensus strategy have been proposed, in both continuous-time and discrete-time. The algorithms are proposed in the form of a multi-agent network based on a combination of local estimators and a dynamic consensus strategy, assuming possible intermittent observations and communication faults. Under general conditions concerning the agent resources and the network topology, conditions are derived for the stability and convergence of the algorithms. For the state estimation schemes, a strategy based on minimization of the steady-state mean-square estimation error is proposed for selection of the consensus gains; these gains can also be adjusted by local adaptation schemes. It is also demonstrated that there exists a connection between the network complexity and efficiency of denoising, i.e., of suppression of the measurement noise influence. Several numerical examples serve to illustrate characteristic properties of the proposed algorithm and to demonstrate its applicability to real problems.;Furthermore, several structures and algorithms for multi-agent control based on a dynamic consensus strategy have been proposed. Two novel classes of structured, overlapping decentralized control algorithms are presented. For the first class, an agreement between the agents is implemented at the level of control inputs, while the second class is based on the agreement at the state estimation level. The proposed control algorithms have been illustrated by several examples. Also, the second class of the proposed consensus based control scheme has been applied to decentralized overlapping tracking control of planar formations of UAVs. A comparison is given with the proposed novel design methodology based on the expansion/contraction paradigm and the inclusion principle.;Motivated by the applications to the optimal mobile sensor positioning within mobile sensor networks, the perturbation-based extremum seeking algorithm has been modified and extended. It has been assumed that the integrator gain and the perturbation amplitude are time varying (decreasing in time with a proper rate) and that the output is corrupted with measurement noise. The proposed basic, one dimensional, algorithm has been extended to two dimensional, hybrid schemes and directly applied to the planar optimal mobile sensor positioning, where the vehicles can be modeled as velocity actuated point masses, force actuated point masses, or nonholonomic unicycles. The convergence of all the proposed algorithms, with probability one and in the mean square sense, has been proved. Also, the problem of target assignment in multi-agent systems using multi-variable extremum seeking algorithm has been addressed. An algorithm which effectively solves the problem has been proposed, based on the local extremum seeking of the specially designed global utility functions which capture the dependance among different, possibly conflicting objectives of the agents. It has been demonstrated how the utility function parameters and agents' initial conditions impact the trajectories and destinations of the agents. All the proposed extremum seeking based algorithms have been illustrated with several simulations. |
