The Dimensioning and Provisioning of Customer Access Links in a multiple network and multiple service environment.

M. Tierney*, A. Newcombe+, T. Curran+, D. O'Sullivan*
 

Abstract

At present the provisioning process of access services is a very significant proportion of the operating cost for a service provider (or telecom operator). Even when the cable plant is installed, it is still necessary to designate connection points, route cables, install customer premises interfaces, select a port access to the local switching equipment and instantiate the customer with the appropriate class of service requested. This situation will remain for some years to come and even the introduction of fibre will not greatly effect the relative cost of this operation. For any particular request, there are a range of possibilities available to a telecom operator for provisioning. The solution required is the one which satisfies both the telecom operator and customer goals with respect to the business objectives and quality of service, respectively.

This paper describes software that has been developed in the DESSERT (R2021) project, which handles the dimensioning and provisioning of access links given a multiple network scenario. The mixture of network technologies handled, reflecting the current trend of access possibilities, includes: X.25 CCITT Recommendation, Integrated Services Digital Networks (ISDN) and Metropolitan Area Networks (MAN). All these networks can afford a high level of automation in their access link provisioning with resultant productivity and economic benefits.

1 Introduction

With the predicted increase in telecommunication service demands [1], there is an envisaged need for highly skilled and knowledgeable personnel. Such personnel will need to respond in a timely manner to satisfy customers' orders, and provide quick solutions to the lengthy and complicated task of service provisioning. Powerful decision support systems are prime candidates to help personnel cope with these escalating telecommunication demands. Such decision support systems are software modules that help automate the decisions faced in service provisioning [2].

State of the art problem solving techniques based on Knowledge Based Systems (KBS) and the mathematical methods of Operations Research (OR) are currently being employed to develop decision support systems for service provisioning. Three Decision Support Systems (DSSs) have been developed within DESSERT1 which aid to address different stages of service provisioning2. Firstly, the Customer Requirements Capture (CRC) DSS which helps to establish a technical specification of the customers service requirements. Secondly, the Generation and Selection of Alternative Configurations (GSAC) DSS which helps to produce a network configuration (resource assignment) that supports the given technical specifications. Thirdly, the Resource Scheduling (RS) DSS which helps schedule and allocate the resources necessary for these services to be initiated [4, 5, 6].
 

2 The GSAC DSS Scenario

For the purposes of developing the DSS along practical guidelines, a scenario has been adopted to model the system operating within expected future network domains [5]. Figure 2.1 shows the general domain given within the scenario which can be broken down in to further sub-domains; for example the Access Network, Transit Network, etc. This decomposition also allows for a more manageable approach to the problem of resource assignment.

Figure 2.1: The GSAC Scenario domain and sub-domains. Note that the Access Networks can also be subdivided in to sub-domains. The Service Access Controller (SAC) which provides the required interfaces to the customer premises equipment, the Access Link (AL) which provides the connectivity between the SAC and ATI, and the Access Network/Transit Network Interface (ATI) which provides a gateway to transit connection related functions (e.g. switching, routing, etc.)

2.1 Modelling the problem

As can be seen in Figure 2.1, the GSAC scenario is a move towards a network (technology) independent view of resource assignment. Further abstraction, however, is required in order to develop a complete network independent view of the problem. Figure 2.2 shows the layers of information modelled within GSAC work [5].
Figure 2.2 The mapping from technical requirements to physical entity types.

The major elements of this abstraction are:

Figure 2.3: Functional breakdown for an X.25 configuration.

The model shown in Figure 2.2 provided the basis for developing the GSAC DSS software. This software is comprised of several software modules called tools ,for example, Access Network Generation (ANG) tools, Transit Network Configuring tools, etc. [5] Such tools, for instance the MAN-ANG tool and the X.25-ANG tool address the problem of dimensioning and provisioning access links over different network technologies.

3 Access Network Generation (ANG) Tools

The MAN, X.25 and ISDN Access Network Generation (ANG) [7] tools help to produce a set of ranked access network configurations that meet the customer's requirements. Each 'configuration' represents a possible assignment of network resources. This task which the tool aims to support, is traditionally undertaken as a pen and paper exercise and relies heavily upon the knowledge, expertise of an individual to determine an optimal solution. The first step in determining an optimal solution is the task of access link dimensioning.

3.1 Access Link Dimensioning

Dimensioning is the task of determining the bandwidth necessary to implement a set of service requests. The task is further constrained in meeting a set of usage (profiles) parameters and quality of service parameters. In the X.25, MAN and ISDN ANG tools dimensioning is achieved by a combination of KBS and mathematically modelling techniques.

In the X.25 ANG tool the customer's usage parameters are examined and the peak usage and mean holding time (duration) are determined. The tool then groups the services that are similar in holding time and probability of loss and combines each group to establish the 'call intensity'. This is used, with the probability of loss, to solve the Erlang B formula [1, 9] which gives the number of circuits that the service provider must allocate to meet that probabiltiy of loss i.e., the number of concurrent X.25 'calls' that the service provider must be prepared to cater for.

Different service types have different packet statistics, (e.g. a terminal application might have a very bursty packet arrival rate and a file transfer application might have a very constant packet arrival rate) hence it is necessary to establish the upper bound on packet delay (i.e., the maximum delay that a packet should experience). To find this upper bound the packet traffic is visualised through a queuing model. The queuing model is solved to find the bitrate on the access link that makes the packet delay (in the queue) less than the specified upper bound on packet delay.

Another constraint on bandwidth is whether a service originate or terminate at the customer site. Furthermore, service characteristics such as, bi-directional symmetric, asymmetric or unidirectional can influence the bit rate requirements.

When the bandwidth of the access link is determined it is presented to the user who can decide to increase or decrease the value. In future versions of the X.25 ANG tool, it is envisaged, that a user may explore the quality of service vs. bandwidth trade-off, namely, by changing the bandwidth and determining the effect on the quality of service.

Since some services may not be suitable for implementation on a particular network technology (say X.25), a single service request may be implemented over a combination of different network technologies. Dimensioning over different networks could involve the expertise of several experts, however the GSAC DSS only requires the input of a single expert.

3.2 Access Network Configuration

Having dimensioned the access link, access network configuration is the next task which needs to be addressed. Configuring a customer's access network is essentially determining the arrangement of resources (if any) to be allocated to satisfy the customer's requirements. The solution should be favourable to the Service Provider yet still meet the customers needs. The type and number of possible configurations will depend on the constraints placed upon the access network (see Figure 2.1). Such constraints are for example, the bandwidth required on the access link, conformance to a certain interface types (or equipment), and that customer access links on one site can influence, the solutions offered on other sites. As part of the configuring process the overall access network needs to be considered.

Some miscellaneous considerations may arise in configuration, for example, some solutions that meet customer requirements may seem cost effective in the short term, however a configuration that provides spare capacity to other potential customers may prove far more cost effective in the long term.

The complexity of the configuration process becomes quite apparent when new services and technologies are introduced into the telecomm domain. An experts task can be further complicated when s/he has to make decisions (probably based on rules of thumb), that may have to conform to company policy. Here, DSS tools (e.g. ANG tools) can make the experts job more streamline.

Within the configuration process, many sub-tasks are supported by the ANG tools. Figure 3.1 provides a breakdown of the configuration process. The user can be involved in a variety of ways, for instance, s/he has the freedom to activate any selection of sub-tasks and their order of application. Furthermore, the user can perform a manual configuration (i.e. where s/he inputs the configuration information) of SAC, AL and ATI.. Here, the feasible (functional entity) options are presented to the user, the infeasible options are discounted, and the user is guided towards the best solution set.. The user is not allowed to build an overall access configuration, but only individual SAC, AL and ATI configurations, which the system will check for consistency.

Each of the configuration solutions will meet the requirements of being able to support (a) the transport capacity and communication characteristics required and (b) the quality factors specified by the customer's service request.

Figure 3.1: A breakdown of the configuration process within the ANG tools. Note that the customer's existing configuration is checked for adequate free capacity to implement the new services. If the free capacity exists then configuration may not be necessary.
 The ANG tools produce a set of configurations (called a solution set) that each satisfy the customer's requirements, Hence, it is therefore necessary, to establish the best (optimal) solution from this set. This task is performed by a configuration evaluator

3.3 Configuration Evaluator

Based on fuzzy logic and Multi-Criteria Decision Making (MCDM) techniques [10], the "configuration evaluator", is a software tool that establishes the optimal configuration solution from the configuration solution set. The mechanism behind the technique examines the performance of every possible solution over the set of desired criteria. (This optimal solution is understood to be the best solution from a service provider's point of view). In fact, the tool provides a ranking of the solution set, based on the set of desired criteria. One obvious criterion for example, is cost but in a realistic scenario several factors may influence the overall optimal solution, for instance, installation time ,the degree of reliability, the ability of the solution to evolve, etc. [1]. The tool also allows the service provider to "weigh" the importance of each criterion. This facilitates the service provider in finding solutions that are for example, most importantly, cheap and least importantly, secure.

Weighted averages is an approach which aids decision making in the context of multiple criteria evaluation. The technique can be extended, using fuzzy set theory, to incorporate the modelling of imprecise natural language expressions (or linguistic terms, for example,. high, medium, low, etc.) [1, 10].

Linguistic terms are used in the evaluation stage to firstly, quantify criteria that can't be easily quantified with a numerical value (e.g., security), and secondly, to help to provide a more 'user friendly' presentation of numerically quantifiable criteria.

Incidentally, configuration evaluation can be performed either during the generation of configurations or during the selection of the most appropriate configuration.

4 Conclusions

With ISDN, MAN, etc. the service options available to the customer are expected to expand greatly. It will not be possible for such services to be manually provided in a cost effective manner. Indeed such provisioning will need to be automated. This paper has presented a decision directed approach to the problem and which allows user the selection and control of the appropriate provisioning tasks.

The work undertaken in the development of the ANG tools for GSAC demonstrates shows how DSS techniques may be effectively employed to achieve service provisioning tasks. The approach taken allows a much more efficient use of resources by the service provider and it allows dimensioning and provisioning tasks over a multiple network and service environment. The service provider decision system is realised as a combination of fuzzy logic decision making with user selected goals and graphical support display presentations.

Acknowledgements

We would like to acknowledge the support of the various people associated with the DESSERT project, all of whom have contributed to the project's results. Special thanks are due to Alan Clarke and Terry Turner of Broadcom. DESSERT is a RACE II project, partly funded by the Commission of the European Communities, DG XIII, Telecommunications, Information Industries and Innovation. The DESSERT consortium includes Broadcom Éireann Research Ltd., BT plc, Framentec, SEMA Group, Inform, Trinity College Dublin, PTT Netherlands and Queen Mary and Westfield College.

References

[1] Minoli, Daniel; Broadband Network Analysis and Design, Artech House, 1993.

[2] Davison Robert, et al; A Decision Support System for the Allocation of Resources during Service Provisioning, RACE IS&N Conference, November 1993.

[3] RACE Common Functional Specifications, Issue C, Publication of RIC, Brussels, December 1992.

[4] Phase I Demonstrator Customer Requirements Capture Application, DESSERT Deliverable, R2021/BRI/NMD/DS/R/019/b1, October 1993.

[5] Phase I Demonstrator Generation & Selection of Alternative Configurations DSS, DESSERT Deliverable, R2021/BRI/NMD/DS/R/019/b1, October 1993.

[6] Phase I Demonstrator Resource Scheduling DSS, DESSERT Deliverable, R2021/FTC/NMD/DS/R/020/b1, October 1993.

[7] Phase I Toolkit Prototype and Report, DESSERT Deliverable, R2021/QM/ELE/DS/P/0014/b1, August 1993.

[8] Kleinrock, Leonard; Queuing Systems Volume I: Theory, Wiley & Sons, 1976.

[9] Girard, Andréw; Routing and dimensioning in circuit-switched based networks, Addesson-Wesley, 1990.

[10] Brown, Tim; A New Fuzzy Weighted Average Algorithm, Proc. QUARDET 1993.


* Broadcom Éireann Research Ltd., Kestrel House, Clanwilliam Place, Dublin 2, Ireland.
+ School of Electronic Engineering, Dublin City University, Glasnevin, Dublin 9, Ireland.
1 DESSERT is a RACE II project (R2021), partly funded by the Commission of the European Communities, DG XIII, Telecommunications, Information Industries and Innovation.
2 DESSERT has aligned itself with the stages of provisioning that has been agreed by the RACE community in the Common Functional Specification (CFS) H404 [3].