Trading computation for bandwidth: reducing communication in distributed control systems using state estimators

J.K. Yook, D.M. Tilbury, N.R. Soparkar
2002 IEEE Transactions on Control Systems Technology  
This paper describes a new framework for distributed control systems in which estimators are used at each node to estimate the values of the outputs at the other nodes. The estimated values are then used to compute the control algorithms at each node. When the estimated value deviates from the true value by more than a pre-specified tolerance, the actual value is broadcast to the rest of the system; all of the estimators are then updated to the current value. By using the estimated values
more » ... d of true value at every node, a significant savings in the required bandwidth is achieved, allowing large-scale distributed control systems to be implemented effectively. The stability, performance, and expected communication frequency of the reduced communication system are analyzed in detail. Simulation and experimental results validating the effectiveness and communication savings of the framework are also presented. * This research was supported in part by the NSF under the grants EEC95-92125 and CMS 99-77179. Related Work Although to the authors' knowledge, the proposed estimation framework has not been used previously in a distributed control framework, there are research results in related areas. We briefly summarize these here. Dead-Reckoning Dead reckoning is an approach to data replication in which every host models the state (position, velocity, and acceleration) of each remote entity by applying predictive extrapolation based on update information sent from each entity's local host. Dead reckoning typically reduces bandwidth because update packets can be transmitted at lower than frame rate frequencies. When remote sites are updated at a slower rate, receivers use extrapolation to provide seamlessly integrated remote and local entities on display. Dead reckoning protocols are implemented in distributed simulation games such as Amaze [3] , and virtual battlefield environments such as U.S. Army's Simulation Networking system [13]. Control Network Effects When a network is used to transmit data, the delay between the initiation time of the message and the delivery time is influenced not only by the speed of the network (bit-rate) and message size but also by the amount of other traffic on the network and the communication protocol used. The communication protocol also affects the reliability and fault-tolerance of the networked system through its priority assignment and error-correction schemes. In [9], communication protocol requirements for real-time systems were discussed in detail. Attempts to improve existing communication protocols include a new token-passing mechanism for real-time access, fault detection, and recovery of transmission error [8] and a new LAN architecture accounting for the time constrains of the message environment [1]. To decrease the network induced delay, communication scheduling [7] and dynamic routing [6] have also been considered. A study of the delay characteristics of popular control networks can be found in [11] . It is well-known that excessive delays in a feedback loop can destabilize a control system. Thus, in a distributed control system, the effect of the network-induced delays on the stability of the system should be
doi:10.1109/tcst.2002.1014671 fatcat:f5qbxiebuzbgpoeawh5ounpyeu