Co-Design of Arbitrated Network Control Systems With Overrun Strategies

Abstract

Abstract-This paper addresses co-design of platform and control of multiple control applications in a network control system. Limited and shared resources among control and noncontrol applications introduce delays in transmitted messages. These delays in turn can degrade system performance and cause instabilities. In this paper, we propose an overrun framework together with a co-design to achieve both optimal control performance and efficient resource utilization. The starting point for this framework is an Arbitrated Network Control System (ANCS) approach, where flexibility and transparency in the network are utilized to arbitrate control messages. Using a two-parameter model for delays experienced by control messages that classifies them as nominal, medium, and large, we propose a controller that switches between nominal, skip and abort strategies. An automata-theoretic technique is introduced to derive analytical bounds on the abort and skip rates. A co-design algorithm is proposed to optimize the selection of the overrun parameters. A case study is presented that demonstrates the ANCS approach, the overrun framework and the overall co-design. I. INTRODUCTION AND RELATED WORKEmbedded computing systems (ECS) are ubiquitous in a wide range of applications including transportation, energy, and healthcare. Design of ECS faces several challenges, especially in the context of high performance, due to strict requirements such as safety, real-time deadlines, and minimum power consumption. These systems typically consist of several control and non-control applications in which their components communicate via shared resources. The presence of several applications with different priorities and limitations on the processing elements introduces resource contention. Specifically, messages can be occasionally delayed, arriving too late to be useful. It is therefore highly desirable to address the design of ECS that directly accommodates the presence of imperfect message transmissions, provides efficient resource utilization, and meets stringent performance specifications. This paper presents a co-design of implementation platform and control so as to result in efficient resource utilization and desired real-time control performance in the presence of overruns in messages when they do not meet their deadlines.Existing research in co-design of control and implementation platform in the presence of non-ideal message transmissions can be categorized in two parts: i) Designing controllers to achieve desired performance and ii) Designing communication protocols to achieve efficient resource utilization. The former consists of procedures for the estimation of worst-case delays of messages and design of controllers that are robust to such delays. This approach however can be pessimistic and

Optimal Control and Stabilization for Networked Control Systems With Asymmetric Information

“…Recently, the control of autonomous unmanned underwater vehicle (AUUV) has gain increasing interests due to its extensive applications, such as deep-sea exploration, target tracking and precise striking [1], [2]. In this section, we shall investigate a simple AUUV system to illustrate the effectiveness of the proposed algorithm.…”

“…INTRODUCTIONOver the course of last few decades, advances in wireless communication have greatly boosted the development of networked control systems (NCSs). NCSs, containing the system, sensors, controllers and actuators where the operation is coordinated through a wireless communication, have attracted research interest due to its broad applications in electronic system, industrial manufacture and mobile communication [1], [2]. Comparing with the classical feedback control systems with wired point-to-point link, NCSs have been shown to be more cost-effective, provide higher flexibility and reduce the maintenance cost [3], [4].Recently, optimal control with asymmetric information (OCAI) has received increasing attention due to the urgent demand in applications, such as deep-sea research, co-ordination of supply and demand, unmanned aerial vehicles and automated highway systems [5]- [7].…”

Generalized Weakly Hard Schedulability Analysis for Real-Time Periodic Tasks

“…When a controller is implemented in a real-time system, studying how its response time behaves in overload conditions is of particular importance since it may impact the overall performance of the controlled system [39,40]. In control-scheduling co-design problems, the ( , ) model has generated great interest [14,31] and has been used in [10,11,35,36] as a constraint on the maximum number of jobs that can be dropped in a given time window, where those values are determined based on the minimum required level of quality of control. An integrated approach for controller synthesis is presented in [4], by finding task parameters that guarantee that the system is stable and provides the required control performance.…”

Weakly Hard Schedulability Analysis for Fixed Priority Scheduling of Periodic Real-Time Tasks

“…[2] extended the idea in [3] from uniprocessors to distributed cyber physical systems, and in [17] FlexRay is considered as the communication medium. The m-K model has also been investigated in the co-design of controls (with respect to their performance) and scheduling [16,26], and is used in [10,11,30,31] to de ne the maximum number of samples (jobs) that can be dropped over any number of consecutive samples (density of dropped samples), to guarantee a minimum level of quality to the controls. However, [11] and [10] do not provide methods for the schedulability analysis of a system and the density of dropped samples needs to be enforced by the operating system or smart sensors.…”