1 / 8

Rumors and Routes

Rumors and Routes. Rajmohan Rajaraman Northeastern University, Boston May 2012. Outline. Rumor spreading Bounds on cover time Small-world networks Low-degree low-diameter models Decentralized search or network navigability. Rumor Spreading. Information dissemination in a network

ting
Télécharger la présentation

Rumors and Routes

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Rumors and Routes Rajmohan Rajaraman Northeastern University, Boston May 2012 Rumors and Routes

  2. Outline • Rumor spreading • Bounds on cover time • Small-world networks • Low-degree low-diameter models • Decentralized search or network navigability Rumors and Routes

  3. Rumor Spreading • Information dissemination in a network • Push paradigm: • Every node that has the rumor forwards it to a random neighbor • Theorem: Push completes in O(n log(n)) steps whp • Pull paradigm: • Every node that does not have the rumor currently queries a random neighbor • Push-pull paradigm Rumors and Routes

  4. Push-Pull Rumor Spreading • Conductance of a graph • Theorem [Giakkoupis 11]: The push-pull process completes in O(log(n)/Φ) steps with high probability • Weaker bounds given earlier by [Chierichetti-Lattanzi-Panconesi 09, 10] Rumors and Routes

  5. Small-World Networks • Six degrees of separation: • Diameter of “real” even very large networks appears to be small • How do we model them? • G(n,p) graphs, with p = Ω(log(n)/n): • Diameter O(log(n)) • Average degree Ω(log(n)) • Random 3-regular graphs are expanders • Small-world graphs: • Model with constant degree, locality, and low diameter Rumors and Routes

  6. Models with Low Diameter • [Watts-Strogatz 00] • Consider a grid over n nodes • Underlying low-degree graph capturing locality • For each node, add a long-range edge • To a neighbor chosen uniformly at random • Theorem [Flaxman-Frieze 04]: If we add a random “1-out” edge for each node in any connected graph G, we obtain an expander whp. Rumors and Routes

  7. Decentralized Search • Myopic shortest paths: • If long-range takes you closer to destination, use long-range path • Otherwise follow along shortest path in short-range subgraph • Exercise: Though the Watts-Strogatz model yields low-diameter, myopic navigation takes long paths • Alternative model [Kleinberg 00]: Suppose long range edge from u to v with probability proportional to 1/d(u,v)α, where d(u,v) is short-range distance • Theorem: If α = 2, then myopic navigation completes in O(log2n) steps Rumors and Routes

  8. Take Away Messages • Combination of local strategies can yield exponential improvement • A single random edge per node can yield an expander • Small diameter may not imply navigability • Appropriate long-range contact distributions lead to good navigability • Techniques: • Random graph models • Martingales and large deviations • Emphasis on local algorithms for search Rumors and Routes

More Related