Elsevier

Computer Networks

Volume 31, Issue 20, 27 September 1999, Pages 2139-2152
Computer Networks

An efficient message scheduling algorithm for WDM lightwave networks

https://doi.org/10.1016/S1389-1286(99)00085-7Get rights and content

Abstract

Two important issues that need to be addressed when designing medium access control (MAC) protocols for Wavelength Division Multiplexing networks are message sequencing and channel assignment. Channel assignment addresses the problem of choosing an appropriate data channel via which a message is transmitted. This problem has been addressed extensively in the literature. On the other hand, message sequencing, which addresses the order in which messages are sent, has rarely been addressed. In this paper, we propose a new reservation-based message scheduling algorithm called RO-EATS that addresses both the channel assignment and message sequencing during its scheduling process. We formulate an analytical model and conduct extensive simulations to evaluate the performance of this algorithm. We compare the performance results of a well-known algorithm which only addresses the channel assignment issue with those of our new algorithm. The comparison shows that our new algorithm gives significant improvement over scheduling algorithms that do not consider message sequencing. As a result, we anticipate that these research results will lead to new approaches to message scheduling on WDM networks.

Introduction

With the proliferation of the World Wide Web (WWW) in all aspects of networking, current local and wide area networks can barely cope with the huge demand for network bandwidth. As a result, there is a world-wide effort in upgrading current networks with high-bandwidth fiber-optic links that can potentially deliver Tera-bits/s. Wavelength Division Multiplexing (WDM) is an effective technique for utilizing the large bandwidth of an optical fiber. By allowing multiple messages to be simultaneously transmitted on a number of channels, WDM has the potential to significantly improve the performance of optical networks. The nodes in such a network can transmit and receive messages on any of the available channels by using and tuning one or more tunable transmitter(s) and/or tunable receiver(s). Several topologies have been proposed for WDM networks 1, 2. Of particular interest to us in this paper is the single-hop topology where a WDM optical network is configured as a broadcast-and-select network in which all the inputs from the various nodes are combined in a passive star coupler, and the mixed optical information is broadcast to all destinations [3].

To unleash the potential of single-hop WDM passive star networks, efficient medium access control (MAC) protocols are needed to efficiently allocate and coordinate the system resources [1]. MAC protocols in a single-hop WDM passive star network environment can be divided into two main classes, namely pre-allocation-based protocols and reservation-based protocols. Pre-allocation-based techniques use all channels of a fiber to transmit messages. These techniques assign transmission rights to different nodes in a static and pre-determined manner. Examples of preallocation-based protocols can be found in 3, 4, 5, 6. On the other hand, reservation-based techniques allocate a channel as the control channel to transmit global information regarding messages to all nodes in the network. Once such information is received, all nodes invoke the same scheduling algorithm to determine when to transmit/receive a message and on which data channel. Examples of reservation-based protocols can be found in 7, 8, 9, 10, 11. Reservation-based techniques have a more dynamic nature and assign transmission rights based on the run-time requirements of the nodes. In this paper, we focus our attention on reservation-based techniques.

Two important issues that need to be addressed when designing (MAC) protocols for WDM networks are message sequencing and channel assignment. Channel assignment addresses the problem of choosing an appropriate data channel via which a message is transmitted. This problem has been addressed extensively in the literature. On the other hand, message sequencing, which addresses the order in which messages are sent, has rarely been addressed. In this paper, we propose a new algorithm that handles both the channel assignment and message sequencing issues pertaining to scheduling. Most of the existing reservation-based approaches schedule messages individually and independently of one another. They ignore that the way to choose the order of the message transmission may affect the performance of the network. To the best of our knowledge, only our previous paper [12]has addressed the issue of sequencing messages in WDM networks. Using that protocol, the order of transmission is determined by the message length. This algorithm has been shown to significantly improve the performance of a WDM network. However, if the difference of the message lengths are small, the algorithm will not be effective. Also it fails when messages are blocked due to avoiding receiver collisions.

In this paper, we propose a reservation-based protocol for scheduling variable-length messages in single-hop WDM passive star networks that overcomes the above deficiencies. The proposed technique addresses channel assignment and ordering message transmission. Our technique is more globally optimizing than existing approaches, since it not only shares global information among receiving and transmitting nodes, but it simultaneously considers multiple messages from different transmitting nodes. The scheduling algorithm is invoked when all nodes in the network have received the message control information. This approach not only provides more information, but it also reduces the number of times the scheduling algorithm need be invoked. This reduction results in lower scheduling overheads and permits more time for transmitter and receiver tuning.

Our algorithm is composed of two phases. The first phase decides the messages transmission order. The second phase is channel assignment. For the first phase, we use the global information on the receivers to impose a priority on the transmission ordering. This, as will be shown, reduces the average delay in the network. The second part of our algorithm is based on the Earliest Available Time Scheduling (EATS) algorithm, which has received a lot of attention as an effective algorithm for channel assignment in WDM networks [9]. We call our new algorithm Receiver Oriented-Earliest Available Time Scheduling (RO-EATS).

We formulate an analytical model and conduct extensive simulations for evaluating the performance of the RO-EATS algorithm. We then compare the performance results of the EATS algorithm with the performance results of our new algorithm. The comparison shows that our new algorithm significantly improves the performance of the EATS algorithm.

The difference between the EATS and the RO-EATS algorithms is that EATS only addresses channel assignment; whereas RO-EATS addresses both channel assignment and message sequencing. Message sequencing in RO-EATS algorithm fully employs the information of the states of the destination nodes, which gives a significant improvement. Furthermore, the analytical model of RO-EATS is another main contribution of this paper.

The remainder of this paper is organized as follows. Section 2specifies our system model and the problem to be addressed. Section 3presents our new scheduling algorithm and the techniques involved. Section 4provides our analytical performance model of RO-EATS. Section 5presents the performance results from simulation experiments and theoretical analysis of RO-EATS compared to EATS. Finally, Section 6concludes the paper with a summary of the results and a discussion of future work.

Section snippets

WDM system model

In this paper, we consider message transmission in a single-hop WDW optical network whose nodes are connected to a passive star coupler via two-way fibers. Each direction of the fiber supports C+1 WDM channels with the same capacity and there are N nodes in the network. The C channels, referred to as data channels, are used for message transmission. The remaining channel, referred to as the control channel, is used to exchange global information among nodes about the messages to be sent. The

Scheduling algorithm

Message transmission and reception in this system model works as follows: A node must transmit a control packet on the control channel in its assigned time slot before sending a message to its destination node. After one round-trip propagation delay, the destination node and the other nodes in the network receive the control packet. Then the distributed scheduling algorithm is invoked by each node to determine the data channel and transmission time slots for each message in the control frame.

A numerical example

In this section, we discuss the details of the proposed scheduling techniques in the context of an example. Fig. 2 presents a simplified system model of Fig. 1. Fig. 2 shows a network of 4 nodes and a set of 10 messages to be transmitted at the source nodes si through 3 channels Ci. In this figure, the boxes represent messages. We label each message in the queue by mi. Each message has basically three attributes. The first one is the source node of a message. If a message is attached to a node,

Analytical model

To compare the performance of the scheduling algorithms, we give an approximate mathematical model for a WDM network using the RO-EATS algorithm. The goal of this model is to study the performance of a WDM network using the new algorithm under the condition of a limited number of data channels in the network. In a practical WDM network, the number of channels is less than the number of nodes. The performance metric we use is the average message delay in the network.

In order to make the model

Simulation and analytical results

In this section, we first present the results of the above analytical model of a system using EATS and RO-EATS. We also present the design and simulation experiments of a WDM network using the different algorithms. We compare these results to verify the accuracy of the analytical model.

Fig. 6 shows the results of the analytical models of the system using RO-EATS and EATS. The mathematical equations that calculate the average message delay of the network using RO-EATS is presented in Section 5;

Conclusions

In this paper, we proposed a new reservation-based algorithm for scheduling variable-length messages in a single-hop WDM passive star network, denoted RO-EATS. Unlike many existing reservation-based techniques, the proposed algorithm addresses both message sequencing and channel assignment aspects of the scheduling problem. The message selection scheme which we adopt is used to impose an order on the message sequences according to the destinations of the messages and the states of the

Acknowledgements

This research work was supported in part by the Hong Kong Research Grant Council under the Grant RGC/HKUST 692/96E.

Maode Ma received the B.E degree in Computer Engineering from Tsinghua University, Beijing, China, in 1982, the M.E. degree in Computer Engineering from Tianjin University, Tianjin, China, in 1991. He is currently a Ph.D. candidate in the Department of Computer Science of the Hong Kong University of Science and Technology since 1994. From 1982 to 1985, he was a project engineer in computer industry in China. From 1986 to 1991, he was a system engineer for the Department of Computer Engineering

References (12)

  • B. Mukherjee, WDM-based local lightwave networks – Part I: single-hop systems, IEEE Network (May 1992)...
  • B. Mukherjee, WDM-based local lightwave networks – Part II: multi-hop systems, IEEE Network (July 1992)...
  • K Bogineni et al.

    Low-complexity multiple access protocols for wavelength-division multiplexed photonic networks

    IEEE J. Selec. Commun.

    (1993)
  • A Gantz et al.

    Time-wavelength assignment algorithms for high performance WDM star based systems

    IEEE Trans. Commun.

    (1994)
  • G.N. Rouskas, M.H. Ammar, Analysis and optimization of transmission schedules for single-hop WDM networks, Proceedings...
  • M.S. Borella, B. Mukherjee, Efficient scheduling of nonuniform packet traffic in a WDM/TDM local lightwave network with...
There are more references available in the full text version of this article.

Cited by (0)

  1. Download : Download full-size image
Maode Ma received the B.E degree in Computer Engineering from Tsinghua University, Beijing, China, in 1982, the M.E. degree in Computer Engineering from Tianjin University, Tianjin, China, in 1991. He is currently a Ph.D. candidate in the Department of Computer Science of the Hong Kong University of Science and Technology since 1994. From 1982 to 1985, he was a project engineer in computer industry in China. From 1986 to 1991, he was a system engineer for the Department of Computer Engineering of Tianjin University. From 1991 to 1994, he was a faculty member with the Department of Computer Engineering of Tianjin University. His current research interests include WDM optical computer networks, QOS of the computer communications, real-time system scheduling, system modelling and simulation, application of queuing theory, Markov decision process, and application of discrete event system algebra.
  1. Download : Download full-size image
Babak Hamidzadeh received M.S. and Ph.D. degrees in Computer Science and Engineering from The University of Minnesota in 1989 and 1993, respectively. In that period, he also worked as a research associate at The Systems and Research Center of Honeywell Inc., and as a research scientist at The Research and Technology Center of Alliant Techsystems Inc. for over 3 years. From 1993 to 1996 he was an Assistant Professor of Computer Science and Computer Engineering at The Hong Kong University of Science and Technology. Currently, he is an Assistant Professor of Electrical and Computer Engineering at The University of British Columbia. He is also a member of IEEE Computer Society. His areas of research include real-time computing, parallel and distributed processing, multimedia, and communication networks.
  1. Download : Download full-size image
Mounir Hamdi received the B.Sc. degree with distinction in Electrical Engineering (Computer Engineering) from the University of Southwestern Louisiana in 1985, and M.Sc. and Ph.D. degrees in Electrical Engineering from the University of Pittsburgh in 1987 and 1991, respectively. While at the University of Pittsburgh, he was a Research Fellow involved with various research projects on interconnection networks, high-speed communication, parallel algorithms, switching theory, and computer vision. In 1991 he joined the Computer Science Department at Hong Kong University of Science and Technology as an Assistant Professor. He is now an Associate Professor of Computer Science and the Director of the Computer Engineering Programme. His main areas of research are ATM/IP packet switching architectures, high-speed networks, wireless networking, and parallel computing. Dr. Hamdi has published over 100 papers on these areas in various journals, conference proceedings, and book chapters. He is on the editorial board of the IEEE Communications Magazine and Parallel Computing, and has been on the Program Committee of more than 30 international conferences and workshops. He was guest editor of a special issue of Informatica on “Optical Parallel Computing”. He co-founded and co-chaired the International Workshop on High-Speed Network Computing. Dr. Hamdi received the best paper award at the 12th International Conference on Information Networking. Dr. Hamdi received the “Best Ten Lecturers Award” and “Teaching Excellence Award” from the Hong Kong University of Science and Technology. Dr. Hamdi is a member of IEEE and ACM.

View full text