Internet access has evolved in the last decade with the availability of smart handheld devices and high bandwidth offered by mobile networks at reduced costs. Multimedia traffic primarily video became the main share of global Internet traffic due to the popularity of multimedia supported social applications, online gaming, and IPTV. Thus, the success of any network based business model is dependent on service quality and Quality of Experience (QoE) is widely accepted as a means to describe user satisfaction with a service. QoE can be influenced by numerous factors ranging from content liking to system and context-related artifacts.

In this thesis, a holistic approach is adopted to evaluate the QoE of an end-user regarding video quality based on system, contextual and human factors. At the system level, an impact of the Maximum Transmission Unit (MTU), and the comparison of widely used video codecs in error-prone networks is investigated by using diverse streaming protocols on an emulated network. It is found that the myth of using small MTU for better performance in high-loss scenarios comes at a cost of increased latency at the intermediate nodes and doesn’t provide any gain in video quality. The performance of the Google royalty-free VP8 codec was also shown to be on par with the widely adopted and proprietary H.264, especially in scenarios involving high jitter. At a human level, user delight towards the shown stimulus, his or her mood along with the frequency of watching online content (contextual), and their impact on quality ratings are investigated. It is found that human-related personality traits tend to influence subjective ratings and highlight the requirement of more cohesive QoE models to estimate user perception.