Optimal Real-Time Estimation Strategies for a Class of Cyber-Physical Systems Using Networked Mobile Sensors and Actuators

Abstract

The combination of physical systems and networks has brought to light a new generation of engineered systems: Cyber-Physical Systems (CPS) (CPS, 2008). CPS is defined in (Chen, 2008) in the following way: ``Computational thinking and integration of computation around the physical dynamic systems form Cyber-Physical Systems (CPS) where sensing, decision, actuation, computation, networking and physical processes are mixed". CPS is foreseen to become a highly researched area in the years to come with its own conferences (NSF, 2006; WCPS, 2008) and journals, e.g. (Gill et al, 2008). ``Applications of CPS arguably have the potential to dwarf the 20-th century IT revolution" (Lee, 2007). CPS applications can be found in medical devices and systems, patient monitoring devices, automotive and air traffic control, advanced automotive systems, process control, environmental monitoring, avionics, instrumentation, oil refineries, water usage control, cooperative robotics, manufacturing control, buildings, etc. The first step when considering a CPS is to determine the dynamics of its ``physical" part, i.e. the environment in which the sensors and actuators are going to operate. First by defining a matching mathematical model, and then by retrieving the values of the parameters of this model. In this paper, the parameter estimation process constitutes a CPS in itself as we are using a mobile actuator-sensor network for that purpose. The ``modeling-analysis-design (MAD) process in dynamic systems control is fundamental in control engineering practice. In both physical and mathematical modelling, the parameter estimation is essential in successful control designs. A precise parameter estimation depends not only on ``relevant measurements and observations, but also on ``rich excitation of the system. These are all known concepts in system identification for finite dimensional systems (Ljung, 2008). In control engineering practice, it is very common to estimate the parameters of a system given a mathematical model. Using observations or measurements, one can parameterize the model using different techniques. Sometimes, when the system to be modelled is spatially and temporally dynamic (i.e. the states depend on both time and space), common