An Unequal Error Protection Scheme for Uncompressed Wireless Video Transmission Using Iterative Joint Source-Channel Decoder

由於57-66GHz高頻微波家用網路已發展致足以提供數Gbps大的無線網路傳輸頻寬，因此無壓縮高解析視訊的無線傳輸系統開始受到矚目。然而，相對於傳統的低頻（2.4GHz或5GHz）無線訊號而言，微波訊號將會受到更大的干擾。換句話說，此類無線視訊系統將會在相當於非常低的訊噪比下的通道進行傳輸，而如何在如此高干擾的無線通道下有效率的利用發送端的能量以及維持接收端收到的視訊品質成為一個相當具有挑戰性的問題。在本論文中，我們提出了一個在發送端做傳送能量分配管理的非平等資料保護架構。相對於傳統的平等式資料保護，我提出的架構能夠更有效率的分配發送端有限的資源給不同位元層，使得相對重要的資訊在傳輸期間可以受到更好的保護，最後因此提高接收端收到的視訊品質。我們所提出的非平等資料保護架構是針對利用基於三維馬可夫場模型的代迭式共同視訊-通道解碼系統的接收端所設計，其中三維馬可夫場是一個基於位元層的馬可夫場模型，其能夠成功的描述視訊中空間與時間的冗餘性。為了在發送端準確的評估接收端收到的視訊品質，所提出的非平等資料保護架構也運用了相似的馬可夫場模型來預測接收端的資料冗餘性。在能夠精準估測出接收端解碼視訊品質後，我們進一步利用拉格朗日乘數法來解決在固定的平均傳輸能量條件下，如何分配能量給各位元層以最佳化估測解碼品質的問題。除此之外，我們也同時探討如何最小化傳輸端能量以達到給定的解碼品質。其中影片品質是以視訊的平均峰值訊噪比當作評比。我們的模擬結果顯示，比起傳統平等式資料保護，所提出的非平等資料保護架構能夠更有效的管理傳輸端資源並且提升無線無壓縮視訊傳輸系統的效能及解碼品質，此外，由於影片資料分級受到保護，所提出的系統更能穩定視訊解碼品質。Uncompressed high-definition video transmission over wireless personal area networks (WPANs) has been drawing interest due to development of the 57-66GHz millimeter-wave (mmWave) WPANs that can provide multi-Gbps transmission data rate. However, mmWave signals have stronger attenuation than the conventional low-frequency signals (2.4 or 5GHz), and therefore supporting such uncompressed HD video streaming and maintaining the received video quality under harsh wireless link become challenging problems. In this thesis, we propose an unequal error protection (UEP) scheme which manages the transmission power allocation for each bit-plane level in encoder. Efficiently transmission energy management can offer better received video quality over conventional equal error protection (EEP) scheme under the same average transmission power. Besides, this encoder is designed for an iterative joint source-channel decoder (ISCD) based on a 3D-MRF soft-in soft-out (SISO) source decoder. The 3D-MRF based source decoder utilizes a bit-plane level MRF model which successfully reveals the spatial and temporal redundancy of uncompressed video sequences. To estimate the received video quality before transmission, our proposed UEP scheme also exploits such MRF model to exploit the spatial and temporal redundancy of uncompressed video and to optimize the resource usage for better error protection through fragile wireless channel. To accomplish this system, we formulate and solve the optimization problems that correspond to maximizing the expected video quality in terms of average peak signal-to-noise ratio (PSNR) under the given average transmission power constraint. Besides, since the received video quality can be perfectly predicted in encoder, we also formulate and solve the optimization problems for the minimum energy required for the certain video quality. Computer simulations show that the proposed UEP scheme can further enhance the video quality and mitigate the unstable quality varying in comparison to the conventional EEP scheme.口試委員會審定書 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i 致謝 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii 中文摘要 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .iv Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vi List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xii 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.1 Wireless Uncompressed Video Transmission . . . . . . . . . . . . . . 1 1.2 Related Works Improving Wireless Uncompressed Video Transmission . . 2 1.3 Main Work in This Thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 2 MRF-Based ISCD for Uncompressed Video . . . . . . . . . . . . . . . . . . 5 2.1 3D MRF Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 2.1.1 Overview of Low-Level MRF Model . . . . . . . . . . . . . . . . . . . . 6 2.1.2 Neighborhood System and Conditional Distribution of 3D MRF Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 2.2 Overview of 3D-MRF based SISO decoding . . . . . . . . . . . . . . . . 11 2.2.1 Parameter Estimation in Decoder . . . . . . . . . . . . . . . . . . . . . . . 11 2.2.2 SISO source Decoder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 2.3 3D-MRF based ISCD structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 3 Optimization Problem of UEP Scheme for Uncompressed Video Transmission . . . . . . 19 3.1 UEP Scheme at Bit-Plane Level of Uncompressed Video . . . . . . . . . 20 3.1.1 Estimation of Decoded Video Distortion . . . . . . . . . . . . . . . . . . . . 23 3.2 Solving Constrained Optimization Problem with Lagrange Multiplier . . . . . . . . 26 3.2.1 Minimizing the Video Distortion with Given Transmission Power Constraint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 3.2.2 Minimizing the Required Power with Certain Video Distortion Constraint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 3.2.3 Numerical Result . . . . . . . . . . . . . . . . . . . . 29 4 Estimation of Decoded Video Distortion with MRF-based ISCD . . . . . . 35 4.1 Estimation of Distortion for ISCD Decoder with MRF Source Decoder . . . . . . . . 35 4.2 Numerical Result . . . . . . . . . . . . . . . . . . . . . 40 4.2.1 Distortion for ISCD Decoder with Spatial MRF Decoder . . . . . 41 4.2.2 Distortion for ISCD Decoder with Temporal MRF Decoder . . . . 43 4.2.3 Distortion for ISCD Decoder with 3D-MRF Decoder . . . . . . . 45 5 Optimization of UEP Scheme for MRF-based ISCD Decoder . . . . . . 47 5.1 UEP Scheme for ISCD Using MRF Source Decoder . . . . . . . . . . . . 47 5.2 Numerical Result . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 5.2.1 UEP Scheme for ISCD with Spatial MRF Decoder . . . . . . . . 52 5.2.2 UEP Scheme for ISCD with Temporal MRF Decoder . . . . . . . 57 5.2.3 UEP scheme for 3D MRF based ISCD . . . . . . . . . . . . . . . 61 5.3 Complexity Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 6 Conclusions and Future Works . . . . . . . . . . . . . . . . . . 66 6.1 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 6.2 Future Works . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 A Trellis based code decoding algorithm . . . . . . . . . . . . . . . . . . 69 A.1 Convolutional code encoder . . . . . . . . . . . . . . . . . . . . . . . . 69 A.2 BCJR decoder with MAP/log-MAP decoding rules . . . . . . . . . . . . 72 Bibliography . . . . . . . . . . . . . . . . . . 7