Optimal bit allocation for fine-grained scalable video sequences in distributed streaming environments

Abstract

We present optimal schemes for allocating bits of fine-grained scalable video sequences among multiple senders streaming to a single receiver. The senders are heterogeneous in their outgoing bandwidth and they hold different portions of the video stream. We first formulate and optimally solve the problem for individual frames, then we generalize to the multiple frame case. Specifically, we formulate the allocation problem as an optimization problem with an objective function to minimize the distortion.By using the piece-wise linear rate-distortion model, we transform the general (nonlinear) optimization problem into an integer linear programming (ILP) problem. We further design a simple rounding scheme that transforms the ILP problem into a linear programming (LP) one, which could be solved efficiently using common optimization techniques such as the Simplex method. We prove that our rounding scheme always produces a feasible solution, and the solution is within a negligible margin from the optimal one. We also propose a new algorithm (FGSAssign) for the single-frame allocation problem that runs in O(n log n) steps, where n is the number of senders. We prove that FGSAssign is optimal. Furthermore, we propose a heuristic algorithm (mFGSAssign) that produces near-optimal solutions for the multipleframe case, and runs an order of magnitude faster than the optimal one. Our extensive experimental study shows the effectiveness of our allocation algorithms in improving the video playout quality.

Rate-distortion optimized streaming of fine-grained scalable video sequences

“…However, because mOPT does not consider quality fluctuations, it may actually yield wild fluctuations as m increases, which are not desirable in streaming systems. Similar results obtained for other sequences and scenarios are shown in Hsu and Hefeeda [2006b]. …”

“…We plot the minimum, average, and maximum per-frame running times for executing mFGSAssign and mOPT for different block sizes in Figure 15 for Foreman sequence in two scenarios. Similar results for other sequences and scenarios are given in Hsu and Hefeeda [2006b]. The figure illustrates the efficiency of the mFGSAssign algorithm versus the mOPT algorithm, which uses the highly-optimized Matlab implementation for solving LP problems.…”

“…We compute the average error per frame, and we repeat that for all frames in the considered sequences. The results, plots are given in Hsu and Hefeeda [2006b], confirm the accuracy of the linear R-D model: the average error is less than 2% in almost all cases.…”

“…Notice that the nonscalable algorithm could not leverage collaboration from other senders, because they carry different versions of the stream. The results for other scenarios with different sequences are given in Hsu and Hefeeda [2006b]. 6.2.2 Optimality of FGSAssign.…”

“…Truncation is a simple operation which can be done in real time. Due to the space limitation, we refer readers to Hsu and Hefeeda [2006b] for a brief discussion, and to Li [2001] and Radha et al [2001] for a detailed treatment on FGS coding.…”