Opportunistic Spatial Gossip over Mobile Social Networks

1. Opportunistic Spatial Gossip over Mobile Social Networks Augustin Chaintreau Pierre Fraigniaud Emmanuelle Lebhar Thomson CNRS CNRS Paris Universite Paris Diderot Universite Paris Diderot Paper Discussed by: Ranjeet Sangle

2. Before we begin .. • Less than 6 Pages long • Out of which, Heavy Math on about 3 Pages. • Can’t really talk too much about the Mathematical Details. • Remaining Pages =3. • Not a single Figure !!

3. An Outline • Review of Basic Terms • The Paper analyzed - Section-by-section • Applications • Questions/Comments

4. Review of Basic Terms Random Walk a mathematical formalization of a trajectory that consists of taking successive random steps [1] Mates each static node maintains some privileged relationships with one or few mobile nodes that it has seen in the past. Mobile Social Networking A type of social networking where one or more individuals of similar interests or commonalities, conversing and connecting with one another using the mobile phone [2] Random Walk example Source: http://en.wikipedia.org/wiki/Random_walk

5. The Abstract explained .. • The Paper investigates how the principles underlying online social network services could be used to take advantage of node mobility in an opportunistic manner. • The model includes static nodes, and mobile nodes which follow random walks. • A simple connection scheme enables to execute sophisticated tasks (e.g., routing) and mechanisms (e.g. spatial gossip) • Future online social networks can exploit mobility as long as they forget connections appropriately

6. Introduction … • Online Social Networks (OSN) can take advantage of nodes mobility by appropriate connections maintenance. • Mates are forgotten and replaced by other nodes met in function of time and opportunistic encounters. • Two types of nodes • Static – positioned at vertices of lattice • Moving – perform random walks in lattice

7. Introduction (Cont’d) … • A mobile node can be the mate of at most one static node, and a static node can have up to m mates, where m ≥ 1 is a parameter. • A static node can communicate with its mates at all times. • The proposed scheme simply requires the static nodes to reinitialize their mates according to a simple random process consisting in • (1) deciding when a mate should be forgotten, and • (2) how a forgotten mate should be replaced by a new one.

8. Opportunistic Connection Scheme • ‘m’ Mates are acquired opportunistically by static nodes when they happen to meet mobile nodes. • If a static node x that has less than m mates happens to meet a free mobile node y, then y becomes the mate of x. • A forgetting process between the static nodes and their mates enables a renewal of the set of mate. • Forgetting process depends on age of the mate connection.

9. Opportunistic Connection Scheme (Cont’d) • Assume that the mobile node y became the mate of the static node x at time t0. • Every Δt period of time, the static node x determines whether y should remain its mate. • More precisely, at time t0 + a · Δt, with a ≥ 1, node x forgets y independently from the past with probability φ(a).

10. Opportunistic Connection Scheme • Setting the Forgetting function

11. Opportunistic Connection Scheme (Cont’d) • Setting the Forgetting Function (Cont’d) …

12. Opportunistic Routing • Opportunistic routing extends the idea of geographic routing, by using some node that is awake and available for routing at the time the packet needs to be transmitted[5]. • Greedy routing finds source-destination paths whose length grows polylogarithmically as a function of the distance between the source and the destination in the lattice

13. Opportunistic Spatial Gossip • Spatial Gossip :– Gossip protocols in which the node selection mechanism uses the spatial distribution of the nodes. • Gossip protocol is based upon two mechanisms: a node selection mechanism and a message selection mechanism • every node u repeats infinitely the following: • Node selection: u selects q nodes v1, . . . , vq; • Message selection: u sends Mi to vi, i = 1, . . . , q; • Where Mi denotes a message prepared by u for node vi, and q ≥ 1 is a fixed parameter..

14. Opportunistic Spatial Gossip(Cont’d) • Two node selection mechanisms: • 1. Geographic Gossip • Based upon the principle of flooding. • Uses links of the lattice and connections with the mates.

15. Opportunistic Spatial Gossip(Cont’d) • 2. Social Gossip • Uses connections with the mates.

16. Discussion • What if all nodes are mobile? • problem of using opportunistic shortcut even more complex. • Assuming that the movements of the nodes are mutually independent, the shortcut distribution would simply result from the forgetting mechanism applied to the combination of two independent random walks. • What if nodes follow other mobility processes?

17. Applications • Gossip-based communication protocols are efficient methods for designing robust and scalable communication schemes in large distributed systems [3]. • Resource location • Construction of approximate spanning tree • In a variety of contexts, the use of randomization to propagate information has been found to provide better reliability and scalability than more regimented deterministic approaches. [4]

18. References: • [1] http://en.wikipedia.org/wiki/Random_walk • [2] http://en.wikipedia.org/wiki/Mobile_social_network • [3] A. Demers, D. Greene, C. Hauser, W. Irish, J. Larson, S. Shenker, H. Sturgis, D. Swinehart, and D. Terry. Epidemic algorithms for replicated database maintenance. Proc. of ACM PODC, 1987. • [4] D. Kempe, J. Kleinberg, and A. Demers. Spatial gossip and resource location protocols. J. ACM 51(6):943–967, 2004.

19. Questions / Comments??