| Motivated by the increasing size of complex systems and wide applications of control over networks, this dissertation considers a set of important open research problems related to the architecture, design, and analysis of distributed controllers over heterogeneous networks. In particular, we consider the effects of communication constraints among the subcontrollers and subcontroller failures on the overall system performance. First, we consider the design of distributed controllers over constrained and noisy links and provide linear matrix inequality (LMI) based procedures. The two forms of constraints considered are the case of limited number of communication links and the case of limited communication power with communication noise. Second, we consider the problem of cooperative control of l dynamic agents to optimize the system performance based on the Youla-Kucera (Y-K) parameterization. Convex conditions are given to account for the structured interconnection and random packet drops. Third, we develop two distributed architectures and propose necessary and sufficient conditions to provide pre-specified performance levels for different failure modes. Also, by using the Y-K parameterization, the effect of communication noise among subcontrollers on the system performance can be solved via convex programming. Last, we propose and analyze a prototypical model matching problem to address transient system performance due to time-varying topologies in communication networks and sensor/actuator failures. Examples and applications are provided to demonstrate the efficiency of the algorithms proposed in the dissertation. Future research directions are also provided. |
