The document describes research using genetic programming to develop models for real-time, non-intrusive evaluation of VoIP quality. Researchers collected data on VoIP traffic parameters and quality ratings to train and test models. The best models estimate quality as a function of bitrate, loss rate, frame duration, and other variables. The models perform comparably to PESQ, the standard quality estimation method, and better than prior approaches, allowing real-time quality monitoring without intrusive measurement.
Top Rated Pune Call Girls Budhwar Peth โ 6297143586 โ Call Me For Genuine Se...
ย
Realtime, Non-Intrusive Evaluation of VoIP Using Genetic Programming
1. REAL-TIME, NON-INTRUSIVE EVALUATION OF
VOIP
USING GENETIC PROGRAMMING
A. Raja1 A. Azad2 C. Flanagan1 C. Ryan2
1Wireless Access Research Centre
Department of Electronic and Computer Engineering
2Bio-Computing and Developmental Systems
Department of Computer Science and Information Sysmtems
University of Limerick, Limerick, Ireland
5. RESEARCH GOAL
Derivation of a VoIP listening Quality estimation model as a
function of transport layer metrics.
Genetic Programming based Symbolic Regression is used
Using the ITU-T Recommendation P.862, PESQ , as the
reference system
PESQ is the de-facto standard for speech quality
estimation.
6. ON DATA COLLECTION
The VoIP traf๏ฌc parameters include loss rate, burstiness
measure, frame duration, packetization interval and
bit-rate.
A total of 3360 distorted speech ๏ฌles were created for each
combination of network traf๏ฌc parameters.
1177 35% were used for training
503 15% were used for testing
1680 50% were used for speaker independent validation
10. SIGNIFICANCE OF MODELS
1 The model is a good approximation to PESQ.
2 Suitable for real-time and non-intrusive estimation of
speech quality whereas PESQ is NOT.
3 Simple models; depend on 3 and 1 variable respectively.
4 Performs signi๏ฌcantly better than ITU-T P.563
11. THE HUMAN COMPETITIVE ELEMENT
1 The results are better than past approaches that include:
Various ๏ฌavors of Arti๏ฌcial Neural Networks based models.
Non-linear least squares estimation.
Multi-state Markov Chains.
Look-up tables
2 All the above approaches (excluding ANNs) used a
restricted set of network traf๏ฌc parameters.
3 By far it is the ๏ฌrst ever application of GP on the problem of
speech quality estimation, and has produced better results.
12. THE HUMAN COMPETITIVE ELEMENT
1 The results are counter-intuitive.
2 The results solve a formidable problem and outperform
previously existing solutions.