Medical Quality of Service and medical Quality of Experience for ultrasound video streaming over small cell networks

Mobile healthcare (M-health) is an evolving paradigm that brings together the evolution of emerging wireless communications and network technologies with the concept of "connected healthcare". The introduction of fourth generation (4G) technologies, along with the improved communication through the latest mobile devices and smart phones has enabled successful deployment of m-health applications worldwide. However, there are two major technical challenges that have not been adequately investigated. First, the challenge of achieving robust and beyond clinically acceptable diagnosis for high bandwith demanding m-health applications, such as medical ultrasound video streaming. Second, the varied Quality of Service (QoS) and Quality of Experience (QoE) requirements that are needed in order to ensure uninteruppted diagnosis for m-health users (e.g. patients, specialists and healthcare providers). This thesis presents the medical Quality of Service (m-QoS) evaluation of medical ultrasound video streaming m-health application over 4G and beyond small cell networks. The performance analysis carried through simulation for m-QoS shows that the proposed small cell network outperforms the traditional macrocell network in terms of system capacity and network performance for m-health users. Furthermore, this thesis proposes the medical Quality of Experience (m-QoE) prediction model utilising small cell technology for evaluating the video quality of medical ultrasound videos at the hospital's end. The proposed model is based on Fuzzy Inference System (FIS) that explores the relationship between the m-QoS parameters and the m-QoE. The proposed model learns the behaviour of m-QoS to predict m-QoE. Another aspect of this thesis is to include content awareness and context awareness, along with m-QoS in the m-QoE prediction framework. The prediction model is based on Multi-Layer Pereptron (MLP) Neural Network, which incorporates network conditions (obtained over small cell networks), ultrasound video content type, and display device characteristics to predict m-QoE. In addition, a part of this study explores using the m-QoE prediction model to control and optimise the m-QoE levels through device-aware adaptive mechanism, which ensures that the predicted m-QoE levels are maintained above acceptable diagnostic quality. The prediction accuracy of both models (i.e. FIS and MLP) is validated through subjective tests obtained from the medical experts. The validation results infer that the proposed m-QoE models exceed in imitating human judgements on medical ultrasound video quality