Application of neural networks to spurious mode modeling in the design of multiplexing networks

A method of modeling spurious mode for microwave multiplexer design is presented. It is proved that the channel filter can be simplified to have the minimum structural features, yet still capable of modeling the effect of spurious mode accurately, which facilitates the application of neural networks in the multiplexer design. Model simplification and neural model development are detailed. The method enables fast and accurate simulation of the spurious mode over wide frequency range. Excellent agreements are observed between the results obtained from EM simulations and from the developed models.