Questions for Measurement Mtg Consideration.doc
Waikato Questions for Consideration at the Proposed
Measurement Workshop
1. What resiliency & availability measurements are taken by Telecom planning groups.
2. Are there any specific long term goals for NGN measurement.
3. Approximately, what are the constraints regarding measurement; e.g., funding, deployment, etc.
4. What suggestions are there for possible network component/s for detailed simulation and analysis. Any suggestions for important components?
5. What opinions/thoughts are there regarding the possibility of defining a “quantified measurement” (a ‘metric’) of Service Resiliency as a numerical value between 0 and 1, where 0 is ‘no service’ and 1 is ‘perfect/full service’? It does not matter whether there is a continuous range between 0 and 1 or discrete steps in this range. We require the metric to be objectively measurable rather than being a subjective assessment of service quality in order to be useful in this project. (A linear scale between 0 and 1 is not expected, and that each service would likely have a different mathematical characteristics.)
Massey Questions for Consideration at the Proposed
Measurement Workshop
Q# Question / Comment
1 From the proposed NGN topology, are there some estimates of link capacities and
costs available? (We need to know the capacities and costs of each link as in COST
239 networks.) What are the likely service or physical link protection mechanisms that
may be employed?
2 What is considered to be an acceptable level of end-to-end VoIP service downtime?
An end-to-end availability model for VoIP may be derived from classical PSTN end-toend
availability model given in the following reference:
Telecommunication-Network Dependability: A Baseline on Local-Exchange
Availability, C. M. Hamilton, et al. IEEE Reliability and Maintainability Symposium,
1991.
3 Are there some typical values for Defects per Million in VoIP systems that are used in
practical systems by Telecom/Alcatel?
4 What transport level protocols are being used or proposed for use in VoIP by Telecom
NZ or Alcatel?
5 For VoIP service fulfilment, the performance thresholds must be specified to define
what constitutes a defective call. How can such thresholds be defined for multimedia
services on the Telecom NZ NGN? What will be the basis for deciding such
thresholds?
6 What are any other parameters which contribute to call impairment that are used by
Telecom NZ? Some possibilities are: Random packet loss, consecutive packet loss,
excessive delay and delay variations.
How can we set thresholds for these parameters?
7 Can you comment on the value of the following methods for providing QoS on VoIP
services:
[1] Constraint based routing and traffic engineering (function provided by MPLSTE,
OSPF-TE and ISIS-TE)
[2] Bandwidth management and admission control (RSVP-TE and DS-TE)
[3] QoS for established calls (function provided in IP-DiffServ)
[4] Adaptive coding (require enhancements to SIP)
8 How does Telecom NZ propose to automate fault isolation and root cause analysis
when there are defects in VoIP quality? Current methods are manual. It is relevant to
this project that there should be a potential mechanism to generate fully automatic
report generation.
9 Is the E-model used or proposed for use by Telecom – or is this dependent on the
outcomes of the present study?
10 Mean Opinion Scores seem to be country/culture dependent. Are there any studies on
this for NZ? Would such a study be useful in aligning parameters for E-models or
related techniques in the current project?
11 What are characteristics of codecs to be used in VoIP on Telecom NZ NGN?
12 Is Telecom is using MPLS or GMPLS in its current networks? Is it anticipated that this
technology will be used in the NGN? (Noted that this is discussed in the QoS
architecture documents, of course)
13 Is it envisioned that the proposed NGN will be linked with Internet-2? (This would have
an impact on the measurement techniques that are to be developed in the project.)
Page 1 of 2