(1) A Methodology for Deriving VoIP Equipment Impairment Factors for a mixed NB/WB Context (2) (i) Muhammad Adil Raja, F1-OP-08 Foundation Building University of of Limerick Ireland adil.raja@ul.ie +353-61-202715 (ii) Raja Muhammad Atif Azad CSIS University of Limerick Ireland atif.azad@ul.ie +353-61-202763 (iii) Colin Flanagan F2-0-04 Foundation Building University of Limerick Ireland colin.flanagan@ul.ie +353-61-202622 (iv) Conor Ryan CSIS University of Limerick Ireland conor.ryan@ul.ie +353-61-202755 (3) Muhammad Adil Raja (4) This paper proposes a novel approach to quantifying the quality degradation of Voice over IP (VoIP) telephony in the presence of codec and network-related impairments. This approach differs from the baisc ITU-T E-Model for VoIP quality estimation [1] in that it addresses mixed narrowband/wideband scenarios. It makes novel use of instrumental models and sym- bolic regression via Genetic Programming (GP) to enable the evolution of degradation models from a modest set of initial parameters. Here, a two-step approach has been used. First, values of impairment factors are derived using WB-PESQ as a reference model. Secondly, a GP based symbolic regression approach has been utilized to automatically evolve the functional form of equipment impairment factors from a set of variables. Very few a priori assumptions are made about the model structure. The effectiveness of the approach is demonstrated by a number of generated models which compare favorably with WB-PESQ and outperform the traditional E-Model in terms of prediction accuracy when compared using WB-PESQ. A significant advantage of the approach is that new models are easily generated to account for continuing evolution of the VoIP standards. (5) B, D, E, F, G (6) (B): To facilitate speech quality estimation for VoIP networks in the wake of transition from Narrow Band (NB) to Wide Band (WB), the existing estimation model for NB speech (ITU-T Recommendation G.107, termed E-Model) was extended by Moller et. al. which appeared in IEEE Transactions on Audio, Speech and Language Processing in November 2006. GP evolved models demonstrate a superior performance across a wide range of modern telephony codecs when compared with that of the original E-Model. (D): This paper has been unconditionally accepted for publication in a domain specific journal corroborating our claims that the evolved models: - improve upon the International Telecommunication Union-Telecom (ITU-T) current standard (ITU-T Recommendation G.107); - demonstrate acceptably superior performance than the last reported best result; - demonstrate a successful (and parsimonious) use of a richer feature space than the previous results; - are suitable for real time evaluation of speech; - generalise against a variety of telephony codecs and are not tailored specifically to any one of them as previously reported in the relevant literature. (E): Lingfen and Ifeachor (2002), Mohamed et. al. (2004) and Hoene et. al. (2005) correlated the speech quality estimation to codec bit rate whereas Cole and Rosenbulth (2001) and Sun and Ifeachor (2006) gave codec specific formulations considering other factors. The latest of the attempts that evaluates speech quality in the mixed NB and WB traffic that amended the ITU-T standard was proposed by Moller et. al. (2006). However, the contributory impairment factors were much less than considered by Genetic Programming while evolving the models. GP managed to produce the results that outperform the ITU-T recommendation (E-Model) as amended by Moller et. al. (F): GP evolved models are better than E-Model (G.107). (G): Speech quality estimation is an area of on going research as it is crucial to the delivery of optimal quality of service (QoS). (7) Adil Raja, R. Muhammad Atif Azad, Colin Flanagan and Conor Ryan, "A Methodology for Deriving VoIP Equipment Impairment Factors for a mixed NB/WB Context",To Be Published unconditionally in its present form in IEEE Transactions on Multimedia, Michael Christel ed. IEEE Press. (8) To be divided equally among the authors. (9) This paper undertakes a real world problem and upstages the industrial standard set by the relevant body i.e. ITU-T. The results are scrupulously tested on a variety of network operating conditions including a diverse set of modern telephony codecs. The results are also compared with the current state of the art in this domain. The results are impressive because they give an average performance gain of upto 36.7% on the training data and delightfully generalise on the test data (prediction gain of up to 35.6%). The utility of the results is summarised as follows: - performance superior to industrial standard and the latest reported research at the time of submission; - suitable of real time evaluation of speech as against expensive and intrusive approaches unfit for such purposes; - demonstration of a successful use of Genetic Programming in a novel domain encouraging its use in future.