Abstract
This article revisits the problem of H filtering for discrete-time systems with a time-varying delay in the state and parameter uncertainties residing in a polytope. By utilising the polynomially parameter-dependent idea, a new filter design procedure is proposed, which formulates the existence of admissible robust H filters into a set of linear matrix inequalities. These conditions are developed based on homogeneous polynomially parameter-dependent matrices of an arbitrary degree. As the degree grows, test of increasing precision is obtained providing less conservative filter designs. It is established that the results in the quadratic framework (that entail fixed matrices for the entire uncertainty domain), and the linearly parameter-dependent framework (that use linear convex combinations of matrices) are special cases of the proposed conditions for the zeroth degree and the first degree, respectively. Moreover, in addition to parameter dependence, the obtained conditions are also dependent on both the upper and lower bounds of the delay, which is obtained by using advanced techniques for achieving delay dependence. Several numerical examples are given to illustrate the effectiveness and advantage of the proposed filter design methods.