Sensor network adjacency matrix set up based on anisotropic criterion

In this paper, a linear discrete time-varying system of sensors network is considered. Each sensor has appropriate dropout probability with Bernoulli distribution. The dropout occurs if measuremented output contains only a noise term. The external disturbance belongs to sequences of random vectors with bounded anisotropy of the extended vector. The estimation model of the system is given, for the model an adjacency matrix set up is suggested based on an anisotropic criterion. The input-to-estimation error system is derived, it has the multiplicative noise system type. The estimation problem is reduced to convex optimization one. The suggested method of optimization is based on applying bounded real lemma with the anisotropic norm boundedness sufficient condition. The solution of considered problem allows to decrease anisotropic norm of the input-to-estimation error system, this yields to better performance in the estimation problem.