Printer Friendly

Virtual path resource allocation schemes in ATM networks.


Today's networks are under ever-increasing competitive pressure to deliver more services to users while doing so at lower cost. Deregulation is occurring globally, resulting in a highly competitive industry with lowered entry barriers for new competitors as well as potential new markets for incumbents. At the same time, trends show a rapidly accelerating demand for bandwidth driven by advanced video and data services [3]. Currently, services are provided using multiple networks, often overlaying services on technologies not designed for today's range of applications. Asynchronous transfer mode (ATM) [5], by contrast, is a technology designed first and fore- most for flexibility to enable the carriage of multiple, and previously segregated, services over a single common backbone infrastructure. As multiple services share the same infra- structure, greater statistical gains can be achieved with improved utilization of backbone links.

The promise of ATM is in its capacity to provide a cost effective, flexible, and scalable network solution. Service providers are fundamentally in the business of selling bandwidth--a finite resource within their networks. The economics of ATM, therefore, revolves around maximizing revenue from the available bandwidth through:

* A consolidated backbone network with differentiated ser vices

* Maximal utilization of provisioned links

* Maximal statistical gains

* Reduced network management complexity

The challenge of ATM network design [1], [7], [17] is to provide these advantages in a scalable manner to accommodate large net -work sizes.

Congestion occurs when output rate of network is less than total input rate. Congestion in packet switching network is a state in which performance degrades due to the saturation of network resources such as bandwidth processor cycle and buffers. In ATM based B-ISDN, congestion control should support a set of ATM QoS. Congestion control procedures can be classified into preventive control and reactive control. In ATM networks CBR and rtVBR services uses preventive schemes and ABR and UBR services uses reactive schemes [11], [10].

Static bandwidth and Dynamic bandwidth allocation can be done under guidelines of the following procedures :

1. Connection admission control

2. Virtual path utilization

3. Usage parameter control

4. Traffic shaping (leaky bucket)

We will describe here how the objective of traffic control and congestion control can be done in ATM networks by bandwidth allocation using Virtual Path concept.

Virtual Path

In ATM Networks the VP (Virtual Path) concept for path layer has been standardized in ITU-T. A VP is a logical path introduced in ATM. Virtual path is established by the Virtual Path Identifier (VPI) and the VPI tables in cross connect nodes. The VPI is a number contained in the cell header that identifies the assigned path of the cell. VP bandwidth allocating is usually done by static schemes. This allocation is quite difficult to implement in ATM networks because of few reasons. The ATM networks supports different classes traffic with different bandwidth and QoS. The bandwidth of each VP is determined based on the forecast traffic, which is likely to be bursty and complicated. So reference model is not specified accurately. It makes it very difficult to estimate traffic characteristics because of different QoS of every user. This means static VP bandwidth allocation is not suitable for maximum utilization of resources.

The high quality VP networks offer the following benefits. A connection's end-to-end QoS is almost equal to the QoS yielded by the Virtual Circuit (VC) network. VP network control the delay and CLR at the access point. VP QoS does not interfere with the CAC procedure of accepting or rejecting of new connection.

To fulfill these demands and shortcoming of static VP allocation can be done by dynamic VP bandwidth allocation [15]. Self-healing virtual path, Virtual path core [8] [9] [12] extended the solutions in new directions and illustrated that the strategy based on virtual path is effective. Fault recovery is improved with faster connection re-route times of these critical high bandwidth links and performance management becomes less complex with greater consolidation of reported statistics and alarms. The design of ATM networks based on VPC (Virtual Path Connection) core specifically was given by [16].

CBR and rtVBR VP

Static bandwidth mechanism is applied for Virtual Path Core (VPC) for constant bit rate (CBR) or real-time variable bit rate (rt-VBR). That is allocated bandwidth for the VPC is constant, and the VPC shaping rate is constant for the duration of the VPC. VCCs can be carried over a real-time (CBR or rt-VBR) VPC in the core of the network. It is expected that bandwidth gains due to statistical multiplexing gains can be achieved for real-time applications such as voice with silence suppression and VBR real-time video. The QoS requirements of such applications dictate that a dedicated VPC category is needed. For a given cell loss ratio (CLR) objective, the shaping rate is determined based on a trade-off between the bandwidth savings expected and the delay experienced by the VCC's due to shaping. VCC's will be added to the VPC until the VPC bandwidth is exhausted.


A real-time VPC is not efficient for non-real-time VCC services because of the limited statistical multiplexing gains achieved in the VPC core. The bursty nature of ABR and UBR traffic is not well accommodated with this static VPC provisioning of bandwidth because it restricts access to excess bandwidth available in the core.

The ABR service category presents a unique set of characteristic which can be exploited to efficiently carry traffic dynamically of all non-real-time services :

* A minimum cell rate (MCR) guarantee that represents the static bandwidth required for the network to achieve the QoS of the constituent VCC's.

* Dynamic bandwidth allocation allows access to unused bandwidth in the network particularly for ABR and UBR VCC's to achieve and improved level of performance.

* A low-loss flow-controlled network provides efficient management of congestion and higher utilization of network resources. The cell loss in the ABR section of the network is minimized and is engineered to achieve the QoS objectives of the most stringent non-real-time VCC.

* Network fairness through explicit rate (ER) bandwidth allocation.

The VPC network design is based on dynamically shaped ABR VPC's to carry non-real-time services. It involves two main aspects:

* The use of ABR VPC's to carry non-real-time traffic through the network core.

* The implementation of per-VC fairness and isolation at the edge of the VPC core.

The functionality at the edge design is capable of carrying nrt-VBR, ABR, and UBR connections over the ABR VPC core.

By provisioning ABR VPCs, congestion is pushed to the network edge where the traffic management functionality maximizes performance and provides per-VC fairness and isolation. The network core becomes easier to manage, and core switches need only to support VP switching and explicit rate ABR.

The ABR VPC solution is able to minimize core buffer requirements while maintaining high performance over varying delays. Large buffers in the core switches will add queuing delay to the traffic during congestion. By pushing congestion to the network edge, ABR is able to utilize intelligent packet discard to ensure fairness at the VCC level.


The VP network based on instantaneous virtual path utilization, which is defined as the total cell rate of the active VC's normalized by the VP capacity, is given as:

Let instantaneous Vp utilization VP(t) is defined as

VP(t) = [summation over VCi] Ri/C

Where VCi is active at time t,

and Ri is the peak rate of Vci

C is the bandwidth capacity of VP

The no. of active VC's change with time which changes the VP utilization. The cells arriving over a VP during a cell transmission slot are counted and this count is then converted into instantaneous VP utilization. Instantaneous VP utilization is calculated for every cell slot and are tracked for a monitoring period (Tm). Then the maximum instantaneous VP utilization observed during the monitoring period is used as admission of new connection. The maximum instantaneous VP utilization [VP.sub.max](t) is defined as

[VP.sub.max](t) = max VP(t')

Where t' [member of] (t - Tm, t)

Then the remaining bandwidth is defined as 1 - [VP.sub.max](t)

Now if the requested bandwidth by the new connection is R than the request will be accepted if

R/C < 1 - [VP.sub.max](t)

Otherwise rejected

If the new connection is accepted the changed instantaneous VP utilization is increased by an amount equal to the accepted peak rate R of the new VC that is

[] (t') = VP(t') + R/C

This way this mechanism tries to bridge the gap between static allocation and dynamic allocation by using the network resources efficiently without the computational complexity of dynamic allocation

The dynamic self-sizing network [13] goes with the current demand rather than using inflexible static allocation. Dynamic VP bandwidth allocation uses the VP flexibility characteristic of ATM and it itself continuously adjusts the resource allocation to each connection of each class which is connected. The self-sizing network uses measurement driven techniques, and use simple algorithm depending on regular traffic measurement and monitoring. Thus the flow of a dynamic VP bandwidth control system is given as:

The dynamic VP bandwidth allocation is done using measurement driven traffic technologies as the traffic load is heavy or light and QoS is poor or good. A simple algorithm is used to increase or decrease the VP bandwidth according to the result of measurement driven traffic technologies. Performance estimation based Bayesian approach gives another very good algorithm for VP dimensioning. The new VP bandwidth is accepted when there is enough space for transmission path required by the new connection accommodating the VP.


No VP bandwidth allocation algorithm is good enough for all classes of traffic requiring different QoS. E.g. over dimensioning may good for one VP, where as high utilization of resources is required by another. A short allocation cycle may be allowed for one VP, while a longer allocation cycle is preferable for another. Dynamic VP bandwidth allocation gives an edge over static allocation especially when the reference model of the incoming data is not properly defined, because of its bursty nature. Many different algorithms are needed because the best algorithm to use depends on the class of traffic accommodated in the VP and on which part of the network it is applied. An ideal dynamic VP bandwidth allocation can be a combination of various simple algorithms.


[1.] Dominique Gaiti and Pujolle.1996. "Performance Management Issues in ATM Networks : Traffic and Congestion Control", IEEE/ACM Trans. On Networking, Vol. 4 (No. 2) April 1996.

[2.] Elwalid, A.I. and Mitra, D. 1993."Effective Bandwidth of General Markovian Traffic Sources and Admission Control of High Speed Networks". IEEE/ACM Trans. On Networking, Vol.1 (No. 3) June, 1993: 329-341.

[3.] Elsayed, K. 2005. "A Framework for End-to-End Deterministic-Delay Service Provisioning in Multiservice Packet Networks", IEEE Trans. on Multimedia, 2005.

[4.] Epic Livermore, Richard P. Skillen, Maged and Marek Wernik. 1998." Architecture and control of an Adaptive High-Capacity Flat Network", IEEE Communications Magazine, May 1998

[5.] Handel, R., Hubes, M.N. and Schrocles, S. 1999. "ATM Networks: Concepts Protocols Applications". Pearson Education Asia, New Delhi, 1999.

[6.] Honggyi Zhang, Oliver W. Yang and Mouftah, H..2000. "A hop by hop Flow Controller for a Virtual Path", Computer Networks and ISDN Systems, 32 (2000):99-119.

[7.] Jose M. Barolo, Jorge Gracia Vidal and Olga Casals.2000. "Worst--Case Traffic in a Tree Networks of ATM Multiplexers", IEEE/ACM Trans. On Networking, Vol. 8 (No. 4) Aug. 2000.

[8.] Klaus Peter May, Pierre Semal, Yonggang Du, and Christoph Hernann. 1995. "A Fast Restoration System for ATM--Ring Based LANs", IEEE Communications Magazine, Sept. 1995.

[9.] Kawamura, R. and Ikou, To Kizawa. 1995. "Self-Healing Virtual Path Architecture in ATM Networks", IEEE Communications Magazine, Sep. 1995.

[10.] Jain, R.1996. "Congestion Control and Traffic Management in ATM Networks : Recent Advance and a Survey", Computes Networks and ISDN Systems (1996): 1723-1738

[11.] Jedijanto, J.E. and Gun, L.1993. "Effectiveness of Dynamic Bandwidth Management Mechanisms in ATM Networks", INFOCOM, 1993: 401-410.

[12.] Mayer, W.D. and Grover, Y. Zheng.1998. "VP-Based ATM Network Design with controlled over-subscription of Restoration Capacity". Proc. Ist Int'l Workshop on Design of Reliable Communication Networks, Brugge, Belgium, May 1998.

[13.] Saito, H. 1996. "Innovations o Circuit/Path Operations in ATM Networks--Self Sizing Networks", NTT Rev. Vol. 8 (No. 1) 1996: 56-65

[14.] Shiomoto, Shinichiro Chaki and Naoaki Yamanaka 1998." A Simple Bandwidth Management Strategy Based on Measurements of Instantaneous Virtual Path Utilization in ATM Networks", IEEE/ACM Trans. On Networking, Vol. 6 (No.5) Oct. 1998.

[15.] Song Chang, San qi Li and Joydeep Ghosh. 1995. "Predictive Dynamic Bandwidth Allocation for Efficient Transport of Real--Time VBR Video over ATM", IEEE Journal on Selected Areas in Communications, Vol. 15 (No. 1) Jan. 1995.

[16.] Steve Rosenberg, Mustapha Aissaoui, Keith Galway, and Natalie Girous, "Functionality at the Edge Designing Scalable Multiservice ATM Networks", IEEE Communications Magazine, May 1998.

[17.] Zui Rosberg.1996. "Cell Multiplexing in ATM Network", IEEE/ACM Trans. On Networking, Vol. 4 (No. 1) Feb. 1996.

Vivek Tyagi and Vinay K. Tyagi * N.A.S. (PG) College, Meerut (U P), India

* Dept. of Statistics, M.M. (PG)College, Modinagar-201204 (U P), India
COPYRIGHT 2008 A.K. Sharma, Ed & Pub
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2008 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:asynchronous transfer mode
Author:Tyagi, Vivek; Tyagi, Vinay K.
Publication:Bulletin of Pure & Applied Sciences-Mathematics
Article Type:Report
Geographic Code:9INDI
Date:Jan 1, 2008
Previous Article:Some results on skolem mean graphs.
Next Article:On estimating population mean using post-stratification and auxiliary information.

Related Articles
New ATM products help LAN migration.
Switched networks extend life of today's internetworks.
Merging voice and data.

Terms of use | Privacy policy | Copyright © 2022 Farlex, Inc. | Feedback | For webmasters |