| Thread Links | Date Links | ||||
|---|---|---|---|---|---|
| Thread Prev | Thread Next | Thread Index | Date Prev | Date Next | Date Index |
|
Dear Colleagues,
I have found your July 05 contribution very helpful and was motivated to review it again in the past few days, as well as to study the referenced ITU-T G.xxx Recommendations. I have noted that more recent versions of G.108, G.113 and G.175 have since been published and you may want to update your contribution, as follows:
G.108: 03/2004 G.113: 02/2001 G.175: 05/2000
The current version of G.175, in particular, has some significant changes which the "E-Model" section in your contribution no longer agrees with. I would urge you to review G.175 (05/2000) carefully and update your contribution C802.20-05/36.
Reflecting on the comments I made in the July 05 meeting, I am still searching for the right extent to which the E-Model approach can be applied to the 802.20 evaluation methodology. We need to keep in mind that the scope of the 802.20 standard will be limited to Layer 1 and Layer-2 of the air interface and hence it will have no control over the major portions of the end-to-end E-Model values. In addition, we must also consider the fact that none of the equipment-related parameters will be known at the 802.20 EC stage and, thus, I would suggest that we focus on a small (but critical) subset of the prescribed ITU-T E-Model parameters, namely, the air interface's contribution to:
It is clear that, in the 802.20 evaluation criteria, we would need to analyze and quantify the added transmission-impairments caused by a given proposed air interface when it operates in the defined mobility environments. Along these lines, the actual impact on the voice quality is determined by the aggregate end-to-end transmission impairments, including the effects of equipment design (such as the speech/audio codec) and the acoustic environments, which are clearly out of scope for the 802.20 evaluation. Thus, we cannot include these non-air interface elements in our analysis and should consider narrowing the usage-scope of the ITU-T E-Model and apply it with all but the above three parameter-types set to some uniform, predefined typical values which would make sense for mobile wireless networks.
On the question of audio/speech codec: ------------------------------------------------------------- The application of the E-Model involves determining the equipment impairment factor, Ie which also depends on the choice of speech codec. Appendix I of ITU-T Recommendation G.113 (02/2001) provides provisional planning values for codec-dependent Ie factor. We could choose from Table I.2 - "..Ie under conditions of random packet loss, codecs G.729-A+VAD, G.723.1-A+VAD and GSM EFR". As to the impact of the delay, we should consult ITU-T G.108 Amendment 2 (03/2004) and agree on similar E-Model Id parameter values. I am not sure how to handle the echo issue; we could agree on some typical values that would reflect the extra delay (when compared to wireline systems) introduced by mobile wireless.
Thus, if we eventually decide to adopt the E-Model for the 802.20 EC, under the aforementioned constrains, technology proposals will not have to specify their actual codec and the EC should assume one uniform codec chosen from ITU-T G.113 tables and the calculated R rating should be based on the Ie values which correspond the % Packet Loss values derived from the individual proposal simulation results.
Your thoughts and comments on these proposals will be greatly appreciated.
Best regards, Dan
|