Computational Inference and Control of Quality in Multimedia Services (eBook)

Supervisor's Foreword 7
Preface 9
Contents 11
1 Introduction 14
1.1 QoE Management Framework 14
1.2 QoE Definition 16
1.3 Factors that Affect QoE 17
1.4 Resources Versus Quality 19
1.5 Handling the Complexity of QoE Modelling 22
1.6 Learning Versus Deterministic Design 22
1.7 Main Contributions 23
References 24
2 Objective QoE Models 28
2.1 Objective Aspects of QoE 28
2.2 State of the Art 29
2.2.1 PSNR 29
2.2.2 SSIM 30
2.2.3 VQM 31
2.2.4 MOVIE 33
2.2.5 Reduced and No Reference Methods 34
2.3 Models for JPEG2000 and MPEG4/AVC 35
2.3.1 Characterizing the Video Content 38
2.4 Experimental Analysis of Objective QoE 39
2.4.1 PSNR Results 41
2.4.2 SSIM Results 41
2.4.3 VQM Results 42
2.4.4 MOVIE Results 43
2.4.5 SI, TI, C and M Results 43
2.5 Discussion and Conclusions 43
References 45
3 Subjective QoE Models 50
3.1 Subjective QoE Approaches 50
3.2 State of the Art 51
3.2.1 Rating the Quality 51
3.2.2 Limits and Noticeable Differences 53
3.3 Maximum Likelihood Difference Scaling 55
3.3.1 The Video Subjective Study 59
3.3.2 MLDS Subjective Results 60
3.4 Adaptive MLDS 62
3.4.1 Adaptive MLDS Method 63
3.4.2 Learning Convergence 65
3.4.3 Experimental Setup and Results 66
3.5 Conclusions 72
References 72
4 QoE Management Framework 77
4.1 Existing Approaches 77
4.2 QoE Management for Video Streaming 79
4.2.1 Architecture of a Video Streaming Systems 79
4.2.2 A Hybrid QoE Management Framework 81
4.3 QoE Management for a Mobile IPTV Service 82
4.3.1 Objective Measurements 83
4.3.2 Subjective Measurements 83
4.4 Computational Inference of QoE Models 84
4.4.1 Supervised Learning Background 85
4.4.2 QoE Models for Mobile IPTV 88
4.5 On-line Inference of QoE Models 95
4.5.1 Online Supervised Learning Background 95
4.5.2 Subjective QoE Models with Continuous Learning 98
4.6 QoE Remedies 101
4.6.1 Estimating the QoE Remedies 101
4.6.2 QoE Remedies for Mobile IPTV 105
4.7 Conclusions 106
References 106
5 QoE Active Control 111
5.1 QoE Active Control Framework 111
5.2 HTTP Adaptive Streaming Client 112
5.2.1 DASH Standard 113
5.2.2 Existing Heuristic Strategies 114
5.3 The Intelligent Streaming Agent 118
5.3.1 Reinforcement Learning Background 119
5.3.2 Intelligent Streaming Agent Architecture 124
5.3.3 Reward/Penalty Function 125
5.3.4 Estimating Network Throughput Trends 128
5.3.5 Simulating the Background Traffic 130
5.3.6 Agent Performance 131
5.4 Conclusions 132
References 133
6 Conclusions 137
6.1 The Principal Message 137
6.2 Latest Developments 140
6.3 Future Directions 142
References 143