Search

As networks expand in area and scale, there arise needs for high-speed, optimal path-design algorithms that can take into account a number of factors. This paper proposes an optimal path-design algorithm using a neural network. The proposed algorithm can be implemented by hardware digital logic circuits, and enables optimal path design to be performed 104 to 106 times faster than the conventional polynomial-time algorithms. The proposed method is also easily implemented in LSI. Another feature of the proposed method is the use of directed and weighted synaptic connections. © 1998 Scripta Technica, Electron Comm Jpn Pt 1, 81(11): 67–76, 1998 [ABSTRACT FROM AUTHOR]

Recently, application of neural networks using their parallel processing feature has progressed. This paper proposes a routing method using a neural network and based on dynamic programming. The proposed method finds the least-hop path between two points in a communication network. The method consists of the application of disturbance to the destination, propagation of the disturbance to the source, and search for the shortest path from the source to the destination, using the distribution of the influences of the disturbance at each node. © 1997 Scripta Technica, Inc. Electron Comm Jpn Pt 1, 80(8): 1–6, 1997 [ABSTRACT FROM AUTHOR]