Stochastic Diffusion, Adaptive Estimation, and Prediction Models for Wireless Networked Systems
Networked control systems (NCSs) are finding applications in a broad range of areas such as mobile sensor networks, unmanned vehicles, haptics collaboration, a host of security applications and industry quality control. For example, the economically efficient delivery of highly reliable electric power is increasingly dependent on networked Supervisory Control and Data Acquisition System. However, currently these networks are usually modeled as simple channels having limited transmission bandwidth, non-negligible communication delays and random packet dropouts with specific probability distributions. In fact, no description is given of the temporal variations. For wireless NCSs even for fixed placements of the transmitter and receiver, moving objects temporarily shadowing off multipath components can lead to temporal variations which considerably degrade their performance. This proposal focuses on developing and implementing new, generic, time varying random models for NCSs.
An integrated research plan is proposed that introduces a novel approach: modeling wireless networks, the physical links between transmitters and receivers as time varying processes identified and estimated in real-time. The research plan focuses on three themes. First, the development of models of communication networks as jump diffusions, which are solutions to stochastic differential equations (SDEs), driven by Levy processes which includes Brownian motion and pure jump processes. This modeling approach directly addresses issues such as the time varying multipath effects and time varying noise sources caused by the transmitter population. The models generalize to diverse propagation environments. Second, innovative approaches, based on sufficient statistics in conjunction with least square and maximum likelihood estimation methods, are proposed to identify finite dimensional channel parameter sets that characterize the behaviors of the SDEs with limited computational resources and in real time. Third, the effect of wireless networks on NCSs stability and performance is studied by considering the communication networks as subsystems modeled by the SDEs and connected to the plants to be controlled, actuators, and sensors.
This research provides technologies for NCSs in challenging time varying environments: high fidelity and robust time varying models and adaptive estimation algorithms. Particular benefits will accrue to wireless sensor networks (WSNs), where data rates are typically low, and the good ranging and geolocation capabilities of ultra-wide band (UWB), including military (battlefield sensors), homeland security (critical infrastructure protection), NASA (remote WSN), health monitoring, agricultural areas/farms, home networking, and wireless networks multimedia applications. The proposed models are dynamic and allow viewing wireless channels as such. This is particularly useful in the emerging areas of NCSs and control over wireless communication channels.