A Multiobjective Approach for Source Estimation in Fuzzy Networked Systems

In this paper, fuzzy networked systems with a randomly varying delay and quantization errors are considered to represent signal transmission systems of nonlinearly interactive sources. A source estimation scheme is proposed by using a multiobjective approach, addressing the concerns of estimation errors and transmission power consumption. A mixed
$H_{2}/H_{\infty }$
design is employed to enhance the estimation performance, while the number of quantized bits is optimized to reduce the power consumption. A Pareto front representation is adopted so that the proposed estimation scheme is designed from a broader perspective in contrast with the conventional single-objective approach. It turns out that the proposed source estimator parameters can serve as decision variables of a multiobjective optimization problem (MOP) with linear matrix inequality constraints. This MOP can be solved by using deterministic algorithms, such as interior-point methods, for solutions of internal variables and using stochastic algorithms, such as multiobjective evolutionary algorithms, for the global optimality. Numerical examples are provided to illustrate the proposed methodology.