Stubborn state estimation for complex-valued neural networks with mixed time delays: the discrete time case

In this paper, the state estimation problem is investigated for a class of discrete-time complex-valued neural networks (CVNNs) with mixed time delays. We consider a scenario that the measurement output may contain the unexpected outliers. In order to attenuate the impact of measurement outliers on the state estimation performance, a stubborn state estimator is designed for discrete-time CVNNs. For the purpose of analysis and synthesis, the CVNNs under consideration are firstly transformed to an augmented system which includes the dynamics of the real and imaginary parts of original CVNNs. Then, by resorting to the Lyapunov functional approach, a sufficient condition is given to ensure that the estimation error system is asymptotically stable. Subsequently, the desired state estimator gain is determined by solving a set of matrix inequalities. Finally, two simulation examples are provided to demonstrate the effectiveness of the proposed stubborn state estimation scheme.