Shaping self-similar traffic at access of optical network

https://doi.org/10.1016/S0166-5316(03)00062-2Get rights and content

Abstract

The focus of this paper is shaping of self-similar traffic at the access of an optical node. Due to the lack of optical memories, we suggest a combined exploitation of electronic memories in the edge of the optical network by means of shaping and propose two novel shaping algorithms that dynamically shape the incoming traffic so as to meet Quality of Service (QoS) requirements while keeping with the optical core network constraints. The first algorithm follows a deterministic approach and is based on the service curves theory. It dynamically, on an interval basis, adjusts the shaping parameters to the token bucket described self-similar traffic. The second, alternative shaping algorithm follows a stochastic approach. It blindly shapes the traffic using a jumping observation window.

Introduction

Optics have been identified as a key technology able to provide a large capacity to the transport of massive IP flows while coping with different Quality of Service (QoS) requirements [3], [4]. This very IP traffic is known to present a self-similar nature [13] which, at the network level, implies mainly the need for large buffers. In the context of optical networks however, there is a lack of optical memories.

In our previous work [6], we have focused on end-to-end traffic management in multiservice optical network context in the presence of different traffic QoS requirement. To overcome the lack of optical memories, we have proposed to make a use, in an optimal way, of the electronic memories at the edge nodes of the optical network. We particularly studied the impact of traffic shaping at the periphery of optical network on the en-to-end performance in terms of loss and delay. We have simulated a scenario in which a multiservice optical network was set up in the presence and the absence of shaping. Our previous results have shown that the shaping of data traffic eases the contention resolution in the core nodes by reducing the level of burstiness of the incoming traffic and by enabling the use of optical resources, in term of time, space and spectral techniques, in an efficient manner. In the absence of shaping, we showed that the required fiber delay lines size was beyond feasible cases if QoS commitments were to be reached. In the presence of shaping however, we achieved the desired performance level with suitable values of fiber delay lines capacities. We thus propose, in this work, to offer QoS through a combined exploitation of electronic memories in the edges and optical resources in the core of the network by means of shaping.

There are basically two ways of solving the problem at hand. Due to the difficulty of dealing with self-similar traffic, one option is to use an envelope to upper bound traffic and hide its highly variable statistics. This can be achieved through the use of the service curves, the so-called (σ, ρ) method [5]. But because traffic statistics are precisely highly variable, this envelope should be adjusted to match the input traffic on an interval basis, wherein in each interval of a given length, a new envelope is recalculated [10].

Once input traffic is well described, the service curves theory, along with its bounds on delay and backlog, permits us to calculate the appropriate shaping parameters so as to meet the QoS requirements while obeying to the network constraints. In this context, we propose a novel shaping algorithm that dynamically changes the shaping parameters according to the input envelope, based on the service curves equations [8].

The deterministic approach comes however at the cost of rather conservative engineering and supposes the prior knowledge of incoming traffic, within a given interval. That is why, in this paper, we investigate alternatively classical Markovian modeling so as to incorporate the statistical nature of the input traffic and propose yet another shaping algorithm that is dynamic and adaptable to the incoming traffic variability and operates without prior knowledge of the traffic parameters [7].

The remainder of this paper is organized as follows. In Section 2, we investigate the deterministic approach. We set the end-to-end system and describe the modeling of both source and network in terms of arrival and service curves, respectively. We study the performance of the system in the absence of shaping and motivate the need for dynamically shaping the traffic in order to meet the network constraints. We then focus on dynamic shaping, we show the regions of shaping to meet both buffer and delay constraints and propose a novel algorithm for dynamic shaping based on the equations of the service curves. In Section 3, we adopt a statistical approach with a Markovian model for the input traffic and described our blind shaping algorithm. Numerical results are presented and discussed in Section 4. Concluding remarks are eventually given in Section 5.

Section snippets

Deterministic approach: equation-based dynamic shaping

The end-to-end system is shown in Fig. 1. It mainly consists of a collection of sources on a local area network (LAN) level, an optical network, represented in terms of buffer spaces and servers. A shaper, at the access of the network is used to make input traffic conformant to the network constraints.

Statistical approach: blind shaping

So far, we studied shaping through a deterministic viewpoint. This approach, rather straight-forward, may be quite conservative in upper-bounding input traffic. Things can get even worse when traffic is statistically complex, as is the self-similar case. We thus now study the end-to-end model of Fig. 1 stochastically.

Model

We consider the end-to-end system shown in Fig. 1. Let the self-similar traffic resulting from the LAN sources have the following characteristics: mean=100 Mbit/s, variance=108, and Hurst parameter H=0.7.

Let the packets be of maximal size M equal to 1540 bytes. At the network level, let Rn=227 Mbit/s be the rate of the server, with buffer capacity Bn equal to 100 packets. We assume without loss of generality that a fixed portion of the server at the network level, with service rate Rn and buffer

Conclusion

In this paper, we focused on the shaping of self-similar traffic at the access of an optical node. If this traffic is left as it is, it cannot satisfy the buffer and/or delay constraints at the network level, which may be very stringent in the optical case due to the lack of optical memories. In order to meet those requirements, shaping is essential. In this work, we proposed the use of electronic memories to carry out shaping in two different ways: a deterministic equation-based dynamic

Acknowledgements

This work was supported by the “RNRT ROM (Reseau Optique Multiservice)” under the decision number 99S 0201-0204 and the European IST Project DAVID (Data and Voice Integration over DWDM).

Halima Elbiaze received the BS degree in Applied Mathematics from the University of Rabat, Morocco, in 1996. She received the MS and Ph.D. degrees in Computer Science from the University of Versailles, France, in 1998 and 2002, respectively. She is currently an Assistant Professor in the Department of Computer Science, University of Quebec in Montreal, Canada. Her research interests are in the area of Quality of Service, performance evaluation and traffic engineering for high speed networks

References (14)

  • A.W. Berger, On the index of dispersion for counts for user demand modeling, presented at ITU, Madrid, Spain, June...
  • J.-Y. Le Boudec, Application of network calculus to guaranteed service networks, IEEE Trans. Inform. Theory 44 (3)...
  • N. Le Sauze, et al., Optical packet vs circuit switching in Metropolitan networks, in: Proceedings of the ECOC’99, vol....
  • P. Gravey, et al., Multiservice optical network: main concepts and first achievements of the ROM program, J. Lightwave...
  • R.L. Cruz, Quality of service guarantees in virtual circuit switched networks, IEEE J. Selected Areas Commun. 13 (6)...
  • H. Elbiaze, T. Atmaca, Traffic management in multi-service optical network, in: Proceedings of the IEEE International...
  • H. Elbiaze, T. Czachorski, T. Atmaca, Blind shaping strategy of self-similar traffic: Markovian model, Arch. Theor....
There are more references available in the full text version of this article.

Cited by (0)

Halima Elbiaze received the BS degree in Applied Mathematics from the University of Rabat, Morocco, in 1996. She received the MS and Ph.D. degrees in Computer Science from the University of Versailles, France, in 1998 and 2002, respectively. She is currently an Assistant Professor in the Department of Computer Science, University of Quebec in Montreal, Canada. Her research interests are in the area of Quality of Service, performance evaluation and traffic engineering for high speed networks (IP/WDM, TCP/IP, ATM, FR, etc.).

Tijani Chahed received the BS and MS degrees in Electrical and Electronics Engineering from Bilkent University, Turkey, in 1995 and 1997, respectively, and the Ph.D. degree in Computer Science from the University of Versailles, France, in 2000. He is currently an Assistant Professor in the Department of Telecommunication Networks and Services, Institut National des Télécommunications, France. His research interests are in the area of Quality of Service and traffic engineering for wired and wireless network services.

Tülin Atmaca received the BS degree in Mathematics from Istanbul University and a MS degree in Mathematical Engineering from Istanbul Technical University in Turkey. She received her Ph.D. in Computer Science from University of Paris-Sud, France in 1987. From 1986 to 1988 she has taught in the Department of Computer Science at the University of Paris-Sud (Paris XI). From 1989 to 1991 she was Assistant Professor in the Computer Science Department at the University of Paris VI. Since January 1992 she has been “Associate Professor” at National Institut of Telecommunications in France. Her research interests are in the areas of performance evaluation of telecommunication networks (ATM, FR, Optical Networks, TCP/IP, etc.) of traffic and congestion control, and Quality of Services aspects in these networks. She is involved in several national and international research projects in the field of the optical packet switching network and their performance.

Gérard Hébuterne holds a “Doctorat” (Ph.D.) and a “Habilitation à diriger les Recherches” from the University Paris-6. He has been with CNET (France Telecom Research Lab) from 1973 to 1994, where he was in charge of traffic studies in SPC switches. He participated also to the development of Broadband ISDN Standardisation (performance issues). He joined the Institut National des Télécommunications in July 1994, where he leads the Networks Department. His current research activity focuses on overload control in telecommunications systems and broadband networks (ATM, TCP/IP), and Quality of Service aspects in multiservices networks.

1

Tel.: +33-1-6076-4742; fax: +33-1-6076-4780.

View full text