The small-world phenomenon: an algorithmic perspective
STOC '00 Proceedings of the thirty-second annual ACM symposium on Theory of computing
Improved Bounds for Mixing Rates of Marked Chains and Multicommodity Flow
LATIN '92 Proceedings of the 1st Latin American Symposium on Theoretical Informatics
Pastry: Scalable, Decentralized Object Location, and Routing for Large-Scale Peer-to-Peer Systems
Middleware '01 Proceedings of the IFIP/ACM International Conference on Distributed Systems Platforms Heidelberg
Gossip-Based Computation of Aggregate Information
FOCS '03 Proceedings of the 44th Annual IEEE Symposium on Foundations of Computer Science
Proceedings of the twenty-third annual ACM symposium on Principles of distributed computing
Fastest Mixing Markov Chain on a Graph
SIAM Review
Distributed Uniform Sampling in Unstructured Peer-to-Peer Networks
HICSS '06 Proceedings of the 39th Annual Hawaii International Conference on System Sciences - Volume 09
On unbiased sampling for unstructured peer-to-peer networks
Proceedings of the 6th ACM SIGCOMM conference on Internet measurement
Understanding churn in peer-to-peer networks
Proceedings of the 6th ACM SIGCOMM conference on Internet measurement
Uniform Data Sampling from a Peer-to-Peer Network
ICDCS '07 Proceedings of the 27th International Conference on Distributed Computing Systems
ACM Transactions on Computer Systems (TOCS)
The Convergence-Guaranteed Random Walk and Its Applications in Peer-to-Peer Networks
IEEE Transactions on Computers
T-Man: gossip-based overlay topology management
ESOA'05 Proceedings of the Third international conference on Engineering Self-Organising Systems
Hi-index | 0.00 |
Selecting a random peer with uniform probability across a peer-to-peer (P2P) network is a fundamental function for unstructured search, data replication, and monitoring algorithms. Such uniform sampling is supported by several techniques. However, current techniques suffer from sample bias and limited applicability. In this paper, we present a sampling algorithm that achieves a desired uniformity while making essentially no assumptions about the underlying P2P network. This algorithm, called doubly stochastic converge (DSC), iteratively adjusts the probabilities of crossing each link in the network during a random walk, such that the resulting transition matrix is doubly stochastic. DSC is fully decentralized and is designed to work on both directed and undirected topologies, making it suitable for virtually any P2P network. Our simulations show that DSC converges quickly on a wide variety of topologies, and that the random walks needed for sampling are short for most topologies. In simulation studies with FreePastry, we show that DSC is resilient to high levels of churn, while incurring a minimal sample bias.