Neuroutin': a novel high-speed adaptive-routing scheme using a neural network as a communications network simulator

The author discusses the hardware implementation of Neuroutin' and problems that arise with applying this scheme to large-scale communications networks. In every node of a communications network, neural networks are installed to simulate the traffic in the communications network and to determine the optimum output link. A neural unit in the simulator has delay circuits that correspond to the buffer memories in a communications network node. Since the simulator was implemented using analog hardware and operated in a parallel processing manner, this scheme is suitable for high-speed routing.<
>