Next Generation Network (NGN) – Availability and Resilience Research.

SCHEDULE 1

Project Description: Next Generation Network (NGN) Availability and Resilience Research Project

(This document is derived from the material in the proposal “Next Generation Network (NGN) Availability & Resilience Research – A Proposal to Alcatel & Telecom, Revised November 14 2005” but has been amended to include project team updates and decisions from the project workshop held 28/Nov/2005. It is not itself the Proposal document.)

1Research Background

Telecom and Alcatel have expressed interest in commencing a research programme that will examine characteristics of Next Generation Network (NGN) services. On February 7, 2005 Telecom, Alcatel and MediaLab met to discuss potential research objectives.

The outcome of this meeting included agreement there are some important aspects of NGNs that are not sufficiently understood, particularly those relating to resilience and service availability. Subsequently, Alcatel drafted a preliminary research scope which was circulated to NZ researchers who were invited to attend a Workshop in April. This is contained in Appendix I.

The Workshop provided an excellent opportunity for Alcatel and Telecom to engage with the New Zealand ICT research community, which included representatives from the Universities of Auckland, Waikato, Massey, Canterbury, Victoria and Otago. Research proposals were provided by all the university participants. After an evaluation phase, a combination of Waikato, Massey and Canterbury were selected for the initial stage of the programme. Auckland, Otago and Victoria were asked to participate at a later stage of the programme.

In August, Telecom, Alcatel, MediaLab and the Universities of Waikato, Massey and Canterbury met and reviewed the research requirements. Subsequently, the university researchers collaborated to produce the following proposal.

2Research Introduction

The ultimate of goal of this project is to develop tools that allow informed planning and capacity decisions about the deployment of network components. At the heart of the project is the prediction of service availability despite component failures. The research aspect of this project comes from two changes from traditional telecommunication networks. First, it is common for the services to have a greater range of performance possibilities. For example, in a mixed traffic environment it is unavoidable that data and video traffic will have some impact on the quality of VoIP traffic over a shared medium.

It is also possible that the failure of control systems will impact the possibility or performance of establishing and/or terminating new VoIP calls without impacting existing calls. The long range dependence of data traffic makes existing telecommunications models less valid in a mixed traffic scenario.

Secondly, and perhaps more importantly, there are more complex interactions between components of the network that need to be modelled for accurate prediction of availability. For example, an increase in latency in the connection to a SIP server might be multiplied many times by the transport protocols used to connect to a SIP server causing unacceptable call establishment performance. Such interactions include those between protocols, such as the recovery mechanisms that are included at multiple levels of the network hierarchy.

3Approach

In the initial stages of this project, two approaches to this problem will be pursued.

  • The first is to extend analytical tools to include more facilities for modelling these interactions.
  • The second is to extend and deploy research work being undertaken within the group in code-based simulation to include network components being proposed for the NGN.

A third steam, network measurement and analysis, will provide raw data and performance insights to the behaviour of the live network as it is deployed.

The deliverables of the project will be jointly owned by Telecom and Alcatel. Subject to protection of any party’s confidential information contained in the deliverables, Telecom and Alcatel will have unfettered rights to use and exploit the intellectual property comprised in the deliverables. The Universities will have the right to use such intellectual property for internal, non commercial purposes. Any commercial use of the intellectual property by the Universities will be subject to the prior written approval of Telecom and Alcatel.

4Initial Proposal

The primary goal of the first three months of the project is to undertake preparatory work that will enable and support the development of a code based simulation of parts of the NGN network and analytical analysis of the network. Simulations of this type can be used to estimate metrics that can, in turn, provide a basis to answer questions about service availability of services that have complex interactions between service components, between protocol layers and with physical transport mechanisms.

4.1Three Month Outputs

4.1.1Output 1 – Watching Brief

This output involves development of a watching brief, or literature review of the relevant research and industry experience. A presentation on this material will be prepared for Alcatel /Telecom staff. The study will cover:

  • Massey and Canterbury will coordinate to both undertake a literature review and critique (watching brief) regarding network & resiliency, topologies, QoS strategies, service resiliency related issues and Multilayer restoration and protection.
  • Waikato will investigate and critique the relevant literature on simulator technologies, especially those with real code facilities.
  • Victoria will also contribute as is found appropriate during the project (sharing of material across the project team members is important to achieve the best results.)
  • Alcatel / Telecom will support the development of the Watching Brief with regular reviews of the material collected and critiqued.

The Watching Brief documentation will be stored in the project Data Repository, which will be developed by Waikato.

4.1.2Output 2 – Simulator Review

This output involves a review of thevarious simulators that are available for research and the selection of the most appropriate tool(s) for ongoing research. For each simulator, the project will need to determine their various strengths and weaknesses, especially:

  • Which of the NGN protocol have models available
  • Is there a path for the development of application andtraffic models
  • Can they support the proposed NGN QoS architecture
  • Will they scale to the required dimensions
  • Simulation of a simple scenario using those simulators thatare potentially useful.

The project team will develop a report on the results of the simulator review, with particular focus areas for the universities:

  • Massey will report on the service modelling process implications of using simulation techniques.
  • Waikato will report on real code simulation, specifically the embodied code for implementation of protocols. This will include the I/O formats of the simulators. The estimated elapsed time frame for completion of this work is 8-10 weeks.
  • Canterbury will report on simulators and tools for evaluating the status of performance in NGNs. This will be specifically related to the modelling abilities and credibility of any models and simulation results i.e. the scalability, errors, costs, extensibility and I/O formats. The estimated elapsed time frame for completion of this work is 4-6 weeks.

Alcatel / Telecom will support this work by confirming protocols to be deployed.

  • QoS architecture
  • Scale information (topology, traffic loads)
  • Discussion of appropriate simple simulation
  • Brief review of simulation results
  • Output 3 – Presentation

­This output was proposed previously, but has been removed from the agreed research project proposal.

4.1.4Output 4 – Measurement Plan

This output involves establishment of the measurement requirements and an initialmeasurement plan. The plan will probably include some active and passive measurement components and a proposal for initial trial deployment.Actual deployment of measurement devices and review of results will be part of a follow on stage of the project.

Waikato are responsible for the production of this output, which will produce:

  • data for this project particularly, and
  • measurements for the long term management of the Telecom network.

The actual output of this piece of work will primarily be:

  • Measurement definitions,
  • Identified uses for different types of measures,
  • Actual network measurements,
  • Identification of points in the network where measurements should be made,
  • An outline of costs to provide the required measures.

Canterbury will contribute to the Measurement Plan by investigating the definitions of what needs to be measured, particularly as it relates to their preferred simulation tool(s).

Massey will contribute to the Measurement Plan by investigating the distinctions for measurements, between network management and dimensioning requirements, relating to the physical availability and service characteristics.

Victoria will contribute to the Measurement Plan by investigatingthe relationship with the PSTN resiliency project that was recently undertaken by Victoria and Auckland, to see whether there are some useful synergies.

This Output 4 – Measurement Plan is needed for completion of the work for Output 6 – Service Availability Model Definition.

Alcatel / Telecom will support the work for this Output with:

  • Discussion of measurement needs and plan,
  • Support for deployment of measurement devices
  • Output 5 – Data Repository

Output 5 involves the establishment of a common data repository, including a neutral data storage format. The first step is to establish a list of the data required and generated by different parts of the project. Areas of overlap and interaction will be identified. An XML based schema for storage of the data will be developed, most probably heavily based on existing schema, such as the Global Grid Forum Network Measurement Working Group (GGF/NMWG) schema.

Waikato will develop and host a data repository. Items to be stored in the Repository are:

  • the project documentation e.g. administration, references, watching briefs and reports (a Wiki was suggested), and
  • any data e.g. topology information, measurements and parameter data.

The data must have a common format to avoid problems in working with the data in multiple applications.

The Data Repository is expected and required to be up and running by December 7th.

Canterbury, Victoria and Massey will all contribute to the development of the project Data Repository by identifyingtheir I/O data preferences required in the repository, definingformats/tools/schemas for the data, templates for various project documents, and recommending suitable approaches to developing the shared repository. (Victoria has a masters student who is keen to contribute to this output and has good XML skills.)

4.1.6Output 6 – Service Availability Model Definition

Output 6 involves the development of a proposal for a working definition of a NGN "Service Availability" model. The output is a specification for a model to determine availability of services on the NGN, and as far as possible, development of this model. The model will consist of metric definitions that make it possible to answer questions about the availability of the NGNservice.

The ultimate result of this project will be data that conforms to this model. Forexample, for voice it might provide raw performance characteristics of voice connections (latency, jitter, loss,rejected calls, dropped calls and termination failures), or it might combine these raw performance characteristics into composite scores such as a synthetic MOS score or a GoS score. This is an active research topic.

While a proposal will be developed in the initial three month project, it is likely that further development or refinement will be required.

Massey have lead responsibility for the Service Availability Model Definition output. The actual output will be a report on the service availability definitions and methodology, with associated models.

Massey will be investigating mathematical models which can be evolved to meet any service needs, capturing the concept of availability in the broadest sense of both the service and physical aspects. The initial focus will be on voice in relation to other NGN services.

Canterbury will contribute a PhD student to help with this effort. This output will be highly collaborative, between Canterbury and Massey.

A draft output from Output 6 will be required for progression of Outputs 7 & 8, however it is accepted that the refinement of this output will be ongoing.

Service availability relates to user perceptions. Waikatooffer their human usability laboratory for potential studies as part of this output – this is expected to be useful as part of a follow on phase in the project.

Alcatel / Telecom support required:

  • Discussion of needs of model
  • Iterative review of possibilities
  • Output 7 – Network Component Selection

Output 7 involves the selection of a network portion for an initial detailed simulation study. This selection process will be driven by the intersection of the capabilities ofthe simulation technology and what can be reasonably developed as part of the project.

Waikato have overall responsibility for developing this Network Component Selection output.

The actual output will be a document describing the network architecture to be simulated, including some measurement data in the repository related to the chosen architecture segment(s).

Massey, Canterbury and Victoria will contribute to the Network Component Selection output as found appropriate during the project.

This network component selection output will be dependent for completion on Outputs 2, 4, 5 and 6 and might relate the demonstration from Output 2.

Alcatel / Telecom support:

  • Topology information, including existing/possible active component locations
  • Iterative discussion of possible network subset and failure scenarios for study
  • Output 8 – Prototype Development

Development and preparation of a prototype analytically based network resiliency analysis simulation. The function will include a simple test for network (physical and IP layer) resiliency and use the proposed NGN network portion (or part thereof) as a test case.

Massey will build prototype tool/s based on the results of Output 6 - Service Availability Model Definition. It is expected to be a Proof of Concept only. If practicable it will include a simulation and associated results. The analysis will commence with physical availability and evolve into service availability.

Canterbury will aid in the development of the tools, and will helpwith the provision of programming support if necessary.

The actual output will be Proof of Concept tools based on Output 6. The work will be starting immediately and will evolve throughout the project.

Alcatel / Telecom support:

  • Topology Information as in Output 7
  • Iterative discussion of GUI components.

4.2Use of OPNET Tool

An important aspect of the work will be designing and coding a purpose-built simulator for commercial use according to Telecom’s needs and interests. We will employ OPNET just for proof of concept purposes. As the University of Canterbury OPNET licence terms only allow its use for research. A commercial-use licence could be an order of magnitude more expensive.

5Schedule

The project kick-off meeting was held on 16 Nov 2005. The following table illustrates the estimated completion of the three month outputs described in section 4.1Three Month Outputs. These will be reviewed during the course of the project. The elapsed time for Phase 1 will include an allowance of 4 weeks for the Christmas holiday break, approximately indicated below. Based on a start date of 16 November 2005, the expected delivery date for Output 8 is 27 March 2006.

Figure 51: Research Activities

Activities / Week 1-2 / Week 3-4 / Christmas Break / Week 5-6 / Week 7-8 / Week 9-10 / Week 11-12 / Week 13-14
Output 1
Output 2
Output 3
Output 4
Output 5
Output 6
Output 7
Output 8
Consider Ph-2

5.1Phase 2

While this proposal outlines a three month project, we expect that once completed, a second phase of the research will commence, possibly partially supported by the Foundation for Research Science and Technology (FRST) which has already expressed interest in supporting the research through their Technology NZ schemes.

5.2Project Team

The project team will consist of the following leaders of the various Universities and MediaLab. Post graduate student support will be provided by all organisations.

People / University / Role / Experience
Dr Murray Pearson / Waikato / Technical leadership / Computer science, network management and measurements
Dr Tony McGregor / Waikato / Technical leadership / Computer science, network management and measurements
Dr Richard Nelson / Waikato / Technical leadership / Computer science, network management and measurements
Dr Richard Harris / Massey / Technical leadership / Computer science, network engineering and modelling
Dr Harsha Sirisena / Canterbury / Technical leadership / Computer science, network engineering and modelling
Dr Krys Pawlikowski / Canterbury / Technical leadership / Computer science, teletraffic analysis and modelling
Dr Peter Komisarczuk / Victoria (VUW) / Post Graduate support / Peter will provide VUW post graduate support to the project
Dr Murray Milner / MediaLab / Technical Advisor / Technical and industry management ICT experience
Steve Hoare / MediaLab / Project Management / Project Management

5.3Resource Effort

The spread of human resource effort (FTE basis) for the universities, estimated at the project commencement, is summarised in the table below. These are estimates only: they have been used to develop the ‘price cap’ for each university for this project; invoicing will be on a ‘time & materials’ basis, based on timesheets for actual work done. (The effort initially planned for Output 3 is expected to be distributed across the other Outputs.)

The total days effort are for the range of lecturers and postgraduate students who will be involved in the research. Therefore a blended daily rate has been assumed for pricing.

The Universities will each provide a weekly progress report and timesheet of work effort to MediaLab.

The project disbursement costs, including travel, accommodation, research tools and printing, are not exactly known. However, these costs have been estimated as below (excl. GST):

Other Project Expenses– to be charged at cost
Travel and Accommodation / $12,000
Research Tools / $2,000
Reporting / $1,000

6Next Steps

Telecom and Alcatel accepted the proposal at mid-November 2005, and project planning and work is underway.

Appendix I: Previously Suggested Background & Scope