Backward-Shifted Strategies Based on SVC for HTTP Adaptive Video Streaming

Zakaria Ye,Rachid El-Azouzi,Tania Jimenez,Eitan Altman,Stefan Valentin

DOI: https://doi.org/10.48550/arXiv.1605.03815

2016-05-12

Abstract:Although HTTP-based video streaming can easily penetrate firewalls and profit from Web caches, the underlying TCP may introduce large delays in case of a sudden capacity loss. To avoid an interruption of the video stream in such cases we propose the Backward-Shifted Coding (BSC). Based on Scalable Video Coding (SVC), BSC adds a time-shifted layer of redundancy to the video stream such that future frames are downloaded at any instant. This pre-fetched content maintains a fluent video stream even under highly variant network conditions and leads to high Quality of Experience (QoE). We characterize this QoE gain by analyzing initial buffering time, re-buffering time and content resolution using the Ballot theorem. The probability generating functions of the playback interruption and of the initial buffering latency are provided in closed form. We further compute the quasi-stationary distribution of the video quality, in order to compute the average quality, as well as temporal variability in video quality. Employing these analytic results to optimize QoE shows interesting trade-offs and video streaming at outstanding fluency.

Multimedia

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to avoid video playback interruption in HTTP adaptive video streaming when there is a sudden drop in network capacity, while maintaining a high - quality user experience. Specifically, the paper proposes a Backward - Shifted Coding (BSC) strategy based on Scalable Video Coding (SVC). By pre - fetching the content of future frames in the case of highly variable network conditions, it maintains a smooth video stream and improves the Quality of Experience (QoE).

### Main contributions of the paper
1. **Proposing a new coding scheme**: The BSC scheme adds time - shifted redundant layers to the video stream, allowing future frames to be downloaded at any time. When there is a sudden capacity drop, these pre - fetched frames can be played, thus maintaining a smooth video stream with sufficient quality.
2. **Providing an explicit expression for QoE metrics**: Using the Ballot theorem, the paper derives explicit expressions for QoE metrics such as initial buffering delay, re - buffering time, and content resolution.
3. **Proposing an optimization scheme**: This scheme takes into account not only the waiting time but also the average video quality to optimize QoE.
4. **Outperforming existing adaptive bit - rate algorithms**: Through theoretical analysis and numerical experiments, the paper proves that the BSC scheme is superior to the existing DASH adaptive bit - rate algorithms in reducing the probability of video playback interruption.

### Paper structure
- **System design**: Describes how the BSC scheme is combined with video codecs following the H.264/SVC standard and how it is integrated into an adaptive streaming system based on the MPEG - DASH standard.
- **Mathematical performance evaluation**: Uses the Ballot theorem to mathematically model the performance of the BSC scheme and analyzes QoE factors such as initial buffering delay, re - buffering time, and content resolution.
- **Optimization problem**: Proposes an optimization problem aimed at minimizing the re - buffering time while maximizing the average video quality.
- **Verification of theoretical results**: Verifies the correctness of the theoretical analysis through simulation and shows the performance under different packet arrival processes.
- **Conclusion**: Summarizes the main findings of the paper and future research directions.

### Formula presentation
- **Ballot theorem**:
\[
\text{Ballot theorem}: \frac{N_A - N_B}{N_A + N_B}
\]
where \( N_A \) and \( N_B \) are the number of votes obtained by candidates A and B respectively.

- **Starvation probability**:
\[
P_<^s(N, \phi, x)=\sum_{k = x+\phi - 1}^{N - 1}\frac{x+\phi - 1}{2k - x-\phi + 1}\binom{2k - x-\phi + 1}{k - x-\phi + 1}p^{k - x-\phi + 1}q^k
\]
\[
P_>^s(N, \phi, x)=P_{s1}+(1 - P_{s1})\cdot P_{s2}
\]
where
\[
P_{s1}=\sum_{k = x}^{\phi - 2}\frac{x}{2k - x}\binom{2k - x}{k - x}p^{k - x}q^k
\]
\[
P_{s2}=\sum_{k = 2\phi - 2}^{N - 1}\frac{x+\phi - 1}{2k - x-\phi + 1}\binom{2k - x-\phi + 1}{k - x-\phi + 1}p^{k - x-\phi + 1}q^k
\]

- **Initial buffering delay**:
\[
E[\tau_x]=\frac{x + \phi}{\mu - \lambda}
\]

- **Average bit rate**:
\[
b_{\text{avg}}=\frac{T_Lb_L + T_Hb_H}{T_L + T_H}
\]

These formulas and analysis methods together form...