Stokastisk estimering och reglering över nätverk med kommunikations- och resursbivillkor
Tidsperiod: 2014-01-01 till 2016-12-31
Projektledare: Subhrakanti Dey
Budget: 2 910 000 SEK
Networked sensor and actuator systems have immense technological potential in a number of important social, environmental and modern industrial applications. This project will consider some key fundamental problems in networked estimation and control: (i) design and analysis of state estimation algorithms and quantizer design schemes using multiple sensors over rate-constrained channels with and without information loss, (ii) optimal and adaptive quantizer and sensor transmission scheme design for remote state estimation over rate-constrained, lossy and fading channels, (iii) dynamic quantizer design for stabilizing stochastic control over lossy error-prone channels, (iv) development and analysis of information theoretic notions for control design over random time-varying (RTV) channels, and (v) optimal and adaptive resource allocation for networked estimation and control over RTV fading channels, in particular optimal energy/power management for networked estimation and control algorithms using energy harvesting technology. We will use techniques such as novel dynamic quantization methods, high rate lattice vector quantization methods, nonlinear programming, and stochastic control algorithms using Markov decision processes and associated dynamic programming methods. The proposal will contribute to theoretical advances in networked estimation and control, and design, development and analysis of novel resource-efficient networked signal processing and control technology.