A novel, to our knowledge, type of packet scheduler that could significantly outperform current state-of-the-art schedulers is presented. The operation and the design of such a scheduler are discussed, and a fully operational experimental implementation is described. The scheduler uses a neural network in a winner-take-all strategy to optimize decisions on the throughput of both a crossbar and a banyan switching fabric. The problems of high interconnection density are solved by use of a free-space optical interconnect that exploits diffractive optical techniques to generate the required interconnection patterns and weights.
© 2000 Optical Society of America
Original Manuscript: May 14, 1999
Revised Manuscript: September 24, 1999
Published: February 10, 2000
Roderick P. Webb, Andrew J. Waddie, Keith J. Symington, Mohammed R. Taghizadeh, and John F. Snowdon, "Optoelectronic neural-network scheduler for packet switches," Appl. Opt. 39, 788-795 (2000)