Novel analogic algorithm for finding the shortest-path using simple CNN structure

In this paper we describe a new analogic algorithm for finding the shortest-path in planar graphs explicitly, based on the continuous-time cellular neural network (CNN) templates only. The constant-speed autowave propagation phenomena combined with elements of analog logic are encapsulated in two layers of basic CNNs, with a simple cell-circuit structure. Compared to the 2-D array of the coupled Chua´s circuits, widely used for modeling the autowave propagation, our realization is found to be simpler and, thus, more efficient for the VLSI implementation. Despite its simple structure, this system exhibits complex behavior and performs sophisticated processing resulting in the steady-state output which determines all shortest routes from origin to any point in a given graph. Then, the extraction of the shortest path itself is accomplished using the simple single-layer CNN.