Cyber physical complex networks, modeling, analysis, and control

This research scrutinize various attributes of complex networks; mainly, modeling, sensing, estimation, safety analysis, and control. In this study, formal languages and finite automata are used for modeling incident management processes. Safety properties are checked in order to verify the system. This method introduces a systematic approach to incident management protocols that are governed by mostly unsystematic algorithms. A portion of the used data in this study is collected by means of radar and loop detectors. A weighted t-statistics methodology is developed in order to validate these detectors. The detector data is then used to extract travel time information where travel time reliability is investigated. Classical reliability measures are examined and compared with the new entropy based reliability measure proposed in this study. The novel entropy based reliability measure introduces a more consistent measure with the classical definition of travel time reliability than traditional measures. Furthermore, it measures uncertainty directly using the full distribution of the examined random variable where previously developed reliability measures only use first and second moments. Various approaches of measuring network reliability are also investigated in this study. Finally, feedback linearization control scheme is developed for a ramp meter that is modeled using Godunov's conditions at the boundaries representing a switched system. This study demonstrates the advantages of implementing a feedback liberalized control scheme with recursive real time parameter estimation over the commonly practiced velocity based thresholds.