Expand this Topic clickable element to expand a topic
Skip to content
Optica Publishing Group

A Collisionless Wavelength-Division Multiple Access Protocol for Free-Space Cellular Hypercube Parallel Computer Systems

Not Accessible

Your library or personal account may give you access

Abstract

The performance of a MIMD parallel computer is critically impacted by the interconnection network performance, which in turn is determined by the network topology, implementation hardware, and communication protocol. Cellular hypercube (CH) interconnection networks, with emphasis on a symmetric cellular hypercube (SCH) network, were studied for the system discussed in this paper because they can exploit the communication locality observed in parallel applications [1], are reasonably scalable due to their O(logN) connectivities, and can be implemented with moderate requirements on the number of wavelength channels needed. While free-space optics can realize highly parallel CH networks [2], little progress has been made in designing an efficient protocol for optical data communication. In this paper a CH interconnection system based on a collisionless wavelength-division multiple access with reroute (WDMA-R) protocol is proposed. This system incorporates space-, time-, and wavelength-multiplexing to achieve dense communication, simple control, and multiple access. Analytic models based on semi-Markov processes were employed to analyze this protocol. The performance of the protocol in terms of network throughput and data packet delay is evaluated and compared to other protocols.

© 1995 Optical Society of America

PDF Article
More Like This
Free-space optical bus-based wavelength division multiple access (WDMA) networks

Yao Li, Adolf W. Lohmann, and Satish B. Rao
FC5 OSA Annual Meeting (FIO) 1992

Dispersionless optical cross-over bus interconnects using wavelength division multiple access

Yao Li, Tao Jiang, and Satish B. Rao
WV1 OSA Annual Meeting (FIO) 1992

Select as filters


Select Topics Cancel
© Copyright 2024 | Optica Publishing Group. All rights reserved, including rights for text and data mining and training of artificial technologies or similar technologies.