A novel concept combining Neuro-computing and cellular neural networks for shortest path detection in complex and reconfigurable graphs

Abstract

This paper develops for the first time an analytical concept involving the Basic Differential Multiplier Method (BDMM) in a framework concept using Cellular neural networks (CNN) for finding shortest paths (SP) in reconfigurable graphs. The developed concept is modeled by coupled nonlinear ordinary differential equations (ODE). The resulting ODE parameters are the CNN templates that are, except from their dimension, independent of the different graph’s elements. The main advantage of the CNN concept is that both the costs of arcs and the selection of the origin-destination (s-t) pair are insured by external commands which are inputs of the CNN-processor model. This allows a high flexibility and an easy re-configurability of the developed concept, thereby without any re-training need. Further, the concept can handle even graphs with negative arc’s weights as well as graphs with nonlinear path’s costs.