Efficient allocation of seed servers in peer-to-peer streaming systems with scalable videos
read more
Citations
Live peer-to-peer streaming with scalable video coding and networking coding
Analysis of peer-assisted video-on-demand systems with scalable video streams
Capacity Management of Seed Servers in Peer-to-Peer Streaming Systems With Scalable Video Streams
Scalable video streaming over P2P networks: A matter of harmony?
Deftpack: A Robust Piece-Picking Algorithm for Scalable Video Coding in P2P Systems
References
Overview of the Scalable Video Coding Extension of the H.264/AVC Standard
CoolStreaming/DONet: a data-driven overlay network for peer-to-peer live media streaming
A Measurement Study of a Large-Scale P2P IPTV System
A survey on peer-to-peer video streaming systems
Related Papers (5)
The streaming capacity of sparsely connected P2P systems with distributed control
Frequently Asked Questions (14)
Q2. Why do they need to be able to fetch video data from other peers?
Due to their adaptability to bandwidth variations, naively fetching video data from other peers may result in frequent variations in the number of video layers.
Q3. What is the way to allocate seed servers?
Allocating seeding resources optimally will lead to better utilization of seed servers, and higher video quality for users, especially during periods with excessive loads which are typically the most difficult to handle in real systems.
Q4. What is the utility of the algorithm in Figure 2?
Since its solution is a subset of the solution obtained by their algorithm in Figure 2, the utility obtained by case (ii), denoted by z0, is a lower bound on their obtained utility z, i.e., z0 ≤ z.
Q5. What is the problem of requesting from a set of heterogeneous senders?
An algorithm is presented in [15] to be run on each peer independently that decides how to request video layers from a given set of heterogeneous senders, assuming layers have equal bitrate and provide equal video quality.
Q6. What is the upperbound on the total utility that can be gained?
C/Cmin is an upperbound on the number of sub-requests that can be served, and The author= CCminB 0 max is an upperbound on the total utility that can be gained.
Q7. What is the reason why the algorithm did not increase the satisfaction of peers?
Figure 4 also shows that for a very large seeding capacity such as 200 Mbps, which is nearly enough for fully satisfying all peers even with the FCFS method, the BitTorrent-like method still could not increase the satisfaction as expected.
Q8. How much of the quality range of the video is attributed to cooperating peers?
With their algorithm compared to FCFS and BTlike algorithms, cooperating peers receive up to 7 dB higher quality, which is even more than half of the entire quality range of the video.
Q9. Why did the experimental ratio for this case be higher than 96%?
This is due to dynamics of the network that were not involved in the approximation analyses, i.e., the experimental ratio would have been always higher that 96% if all peers stayed in the network as expected, they let the tracker (and the tracker was able to) decide and update their partnerships at every 10-second step, and all peers did obey their assumptions about sharing their upload bandwidths among layers, which the authors intentionally made them disobey by deviating by up to 50% from Eq. (11).
Q10. What is the algorithm for sorting sub-requests?
The algorithm consists of sorting sub-requests, which runs in O(K0 logK0) where K0 isSRA GREEDYGreedyAllocation (K, C, n[], b[][], c[][])
Q11. How many peer may leave the network?
Each peer may leave at any time according to an exponential probability distribution, by which 25% of peers leave the network before they finish watching the video and doing the expected seeding.
Q12. What is the result of the proposed seed server allocation algorithms?
The results show that the proposed seed server allocation algorithms result in peers receive more video layers, and thus an enhanced video quality (over 2 dB).
Q13. What is the approximation factor for the seed server allocation problem?
The authors present another approximation algorithm for the seed server allocation problem, whose running time is independent of the seeding capacity and only depends on the number of requests in the queue.
Q14. What is the value of y[k, i] for nonzero k?
Each y[k, i] for nonzero k and i values holds the number of sub-requests served from the request reqk in the solution to subproblem (k, i). y[0, i] and y[k, 0] are set to zero.