Issues around Quality of Service (QoS) and security for Voice over IP (VoIP) have been extensively investigated separately, due to the great attention this technology currently attracts. The specific problem this work addresses centres upon the selection of optimal parameters for QoS and security for VoIP streams integrating both network impairments and user perception metrics into a novel empirically-driven approach. Specifically, the simulation model seeks the optimal parameters in terms of variable VoIP payloads, iterative block ciphers, codecs and authentication mechanisms to be used, so that optimum tradeoff between a set of conflicting factors is achieved. The model employs the widely used Transmission Rating Factor, R, as the methodology to predict and measure the perceived QoS based on current transmission and network impairments. The R factor is then used to map perceived QoS to the corresponding Mean Opinion Score value, which gives the average estimation of perceived voice quality (Quality of Experience). Furthermore, a genetic algorithm (GA) has been developed that uses the output from the simulation model as an input into an offline optimisation routine that simultaneously maximises the VoIP call volumes and the Level of Encryption (LoE) per call basis, without degrading the perceived quality of service under a specific threshold as dictated by the R factor. The solutions reflect the optimum combination of parameters for each codec used and due to the small size of the search space the actual speed of GA has been validated against an exhaustive search algorithm. The results extracted from this study demonstrate that under strict and pre-defined parameters the default payload size supported by the codecs is not the optimal selection in terms of call volume maximisation and perceived QoS when encryption is applied.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:570918 |
| Date | January 2010 |
| Creators | Epiphaniou, Gregory |
| Publisher | University of Bedfordshire |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Source | http://hdl.handle.net/10547/142250 |
Page generated in 0.0018 seconds