Using Kalman innovations for transmission control of location updates in a wireless network

Abstract

Distributed systems, such as cooperative unmanned aerial vehicles (UAVs) or unmanned ground vehicles (UGVs), typically depend on wireless communication networks to implement system control. The reliance on wireless communications within a networked control system (NCS) subjects their performance not only to system dynamic characteristics and feedback laws, but also the performance characteristics of the communications links. Fluctuations in communication system performance are reflected by corresponding changes in the control system performance. For instance, network protocols incorporate strategies that manage or limit the message traffic within the network to maintain stable operation of the network. However, these protocols are typically designed using the layered architecture approach, in which the network protocols are designed independently of the applications. In this paper, we present some initial steps toward designing cross-layer protocols to support networked control applications in wireless networks. We consider one portion of networked control in which the feedback of sensing information from a set of distributed agents is made to a control center over the network. We consider a multiple-access scenario in which the location updates must share the network resources with external traffic sources. When the network traffic is heavy, sending every location update over the network can result in increased congestion and decreased performance for both the control and regular network traffic. We propose schemes that use the innovations of a Kalman filter-based tracking system to limit which location updates are scheduled for transmission and thus improve the system performance.