Filtering And Control For Several Classes Of Discrete-time Stochastic Systems With Incomplete Channel

In wireless communications, when a signal travels from transmitter to receiverover the wireless network, there are some inevitable physical phenomena (such asrefection, refraction and difraction), which lead to a time-varying random changein the amplitude and phase of the transmitted signal. This kind of phenomenon isreferred to as incomplete channel. Based on the results obtained, a new measurementmodel modeling both sensor nonlinearities and fading channels, and a modifed Ricefading model whose channel coefcients have the time-varying statistical propertiesare proposed in this thesis in order to refect such phenomena more properly. Then,we research the control and fltering problems for several classes of discrete stochasticsystems with incomplete channels and develop some novel flter and controller designschemes. The compendious frame and description of the thesis are given as follows.The measurement output is subject to the sensor saturations described by sector-nonlinearities as well as the channel fadings caused typically in wireless communica-tion. At the same time, the distributed delays occur in a probabilistic way due to thetransmission network environment. By using the stochastic analysis combined withLyapunov functional approaches, a full-order dynamic output-feedback controller isdesigned such that, the closed-loop system is exponentially mean-square stable andsatisfes the prescribed H∞performance constraint.The H∞fuzzy fltering problem is considered for a class of discrete-time T-S fuzzy systems with the randomly occurring uncertainties，interval time-varyingdelays and channel fadings, and then the H∞output feedback control problem isinvestigated for T-S discrete-time fuzzy systems with infnite random distributedtime-delay and channel fadings. Some Bernoulli distributed white sequences areintroduced to account for the randomly occuring uncertainty, interval time-varyingdelay and infnite distributed time-delay. The fuzzy flter/output-feedback fuzzycontroller is designed, such that the fltering error system/closed-loop T-S fuzzycontrol system is exponentially mean-square stable, and the disturbance rejectionattenuation is constrained to a given level by means of the H∞performance index.The flter/controller parameters can be obtained by solving a convex optimizationproblem via the semidefnite program method.In particular, in a networked situation, the occurrence of the flter/controllergain variations may be subject to random changes due to environmental circum-stances. So, the phenomenon of the randomly occurring gain variations (ROGVs) is introduced into the system model so as to account for the parameter fuctuations oc-curring during the flter/controller implementation. Also, the improved Rice fadingmodel is utilized to describe the phenomena of channel fadings where the occurrenceprobabilities of the random channel coefcients are allowed to time-varying. Inten-sive analysis is carried out to obtain sufcient conditions under which the system isexponentially mean-square stable with a pre-specifed H∞performance. The mainresult contains all the information about the system parameters, the statistical in-formation of channel coefcients，the occurring probabilities of ROGVs, the sectorsof the sensor nonlinearities as well as the minimum and maximum delays, and itshows adequately these factors infuence on system performance and flter/controllerdesign.