Experiments for Distributed and Networked Dynamic Resource Allocation, Scheduling, and Control

Introduction Advances in information technology such as object-oriented programming, real-time operating systems, and reliable computer communication networks, enable the deployment of effective control and automation systems [1]. Industry exploits information technology in distributed control system (DCS) products [2], whose components include proportional-integral-derivative (PID) and programmable logic (PLC) controllers. At the same time, DCSs can utilize distributed networked decision-making systems for factory-wide control solutions. There is an ongoing technological revolution in the application of networked embedded computing to control [3]. Advancements in this area are being driven by (i) low-cost and high-performance microcontrollers that have onboard data acquisition, and (ii) the controller area network (CAN) or wireless ethernet technologies that interconnect these microcontrollers. There are software tools for developing and implementing such embedded 1 distributed networked control systems [3]. The co-design of software and feedback controls for both DCS and networked embedded controls is a topic of current investigation for control and software engineers [1]. While industry continues to exploit information technology for distributed feedback control , universities face the challenge of revising their curricula to suit the educational needs of future engineers in this area. We can identify two research thrusts that support relevant educational experiences. First, there is the area of cooperative robotics where groups of ground, underwater, flying, or space robots are given tasks such as cooperative search, cooperative pushing of an object, or coordinated group motion. In these applications, there is typically a microcontroller on each vehicle, and a wireless ethernet supports vehicle-to-vehicle communications. University courses that provide educational experiences in this area are described in [4], [5]. Second, there is the research program on wireless sensor networks based on the University of California Berkeley motes, which are electronic modules that have a processor, sensors, and communication capabilities [6]. Networks of motes can be used for distributed networked control system problems, such as the tracking of an evader moving through a field of sensors. This article introduces a third approach to distributed networked feedback control based on an educational laboratory. This laboratory emphasizes: (i) low-cost experiments that retain key elements of realism; (ii) problems that are different from those studied in the cooperative robotics and sensor network areas; and (iii) new challenges from information 2 technology-enabled feedback control [1], [7], such as dynamic resource allocation, feedback scheduling of tasks, decision-making over networks, and nontraditional control objectives. We have developed the following experiments to study these challenges [8]: …