Performance of a MAC protocol for WDM networks with on- line scheduling

Authors:
Krishna M. Sivalingam;Jie Wang
Affiliations:
University of North Carolina at Greensboro, Greensboro, NC;University of North Carolina at Greensboro, Greensboro, NC
Venue:
INFOCOM'96 Proceedings of the Fifteenth annual joint conference of the IEEE computer and communications societies conference on The conference on computer communications - Volume 3
Year:
1996

Citing 6
Cited 0

Competitive algorithms for on-line problems

STOC '88 Proceedings of the twentieth annual ACM symposium on Theory of computing
Directory-Based Cache Coherence in Large-Scale Multiprocessors

Computer
Pre-allocation media access control protocols for multiple access WDM photonic networks

SIGCOMM '92 Conference proceedings on Communications architectures & protocols
LIGHTNING: a scalable dynamically reconfigurable hierarchical WDM network for high-performance clustering

HPDC '95 Proceedings of the 4th IEEE International Symposium on High Performance Distributed Computing
Efficient scheduling of nonuniform packet traffic in a WDM/TDM local lightwave network with arbitrary transceiver tuning latencies

INFOCOM '95 Proceedings of the Fourteenth Annual Joint Conference of the IEEE Computer and Communication Societies (Vol. 1)-Volume - Volume 1
SPLASH: Stanford parallel applications for shared-memory

SPLASH: Stanford parallel applications for shared-memory

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper investigates the performance of media access protocols based on on-line scheduling for an optically connected star-coupled system with Wavelength Division Multiple Access channels. Hybrid access protocols combining reservation and pre-allocation of receiver channels have been recently proposed. Transmission consisted of a reservation phase followed by a data phase. Each node was allowed to place a reservation for one WDM channel. This protocol is extended in this paper by allowing reservations on multiple channels and using simple on-line scheduling algorithms. Existing scheduling algorithms (all are off-line) for similar reservation problems tend to have high implementation and computational complexity. We require that scheduling algorithms be real-time and amenable to hardware implementation. The performance of the extended protocol incorporating two simple on-line scheduling algorithms is presented in this papel: We show that the slight increased computational overhead with scheduling is justified by reduced packet latency and higher utilization, especially for client-server traffic.