1 / 13

Massively Distributed Database Systems In-Network Query Processing (Ad-Hoc Sensor Network)

Massively Distributed Database Systems In-Network Query Processing (Ad-Hoc Sensor Network). Spring 2014 Ki- Joune Li http://isel.cs.pusan.ac.kr/~lik Pusan National University. Basic Concepts – in-network query processing. each node has - a local & tiny DB and - sensors.

Télécharger la présentation

Massively Distributed Database Systems In-Network Query Processing (Ad-Hoc Sensor Network)

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. Massively Distributed Database SystemsIn-Network Query Processing(Ad-Hoc Sensor Network) Spring 2014 Ki-Joune Li http://isel.cs.pusan.ac.kr/~lik Pusan National University

  2. Basic Concepts – in-network query processing each node has - a local & tiny DB and - sensors a query "find the nodes where temperature is higher than 35oC" How to process it?

  3. Why in-network query processing ? • scalable • No need to store the entire DB • Interact with neighbor nodes • A node failure is not critical • Issue • Query processing time  determined by # of hops • Energy consumption • Battery is normally limited • Energy consumption for communication is relatively high • SQL-like query

  4. Energy Consumption in S. Banerjee, A. Misra, http://pages.cs.wisc.edu/~suman/pubs/winet03.pdf

  5. Energy Consumption Ptx Ptx Prx . . . r N r What does it imply?

  6. Multi-hop instead of infrastructure network • No global network topology like TCP/IP • Network topology with its neighbors  local stateless routing algorithm

  7. Unit-Disk Graph • UDG: Graph G(N,E) where N is the set of nodes (sensors) andE is set of edges whose length is less than 1 (unit) • Types if UDG • RNG • Gabriel Graph • Delaunay Graph • Each node in V maintains the node IDs connected via edges in E

  8. Gabriel Graph • Graph GG(V,E) • V is a set of nodes n (id, p) where p is a point in Euclidean space • E is a set of edge (a, b) that there is no other node within the closed disk of (a, b)

  9. RNG – Relatively Neighborhood Graph • Graph RNG(V,E) • V is a set of nodes n (id, p) where p is a point in Euclidean space • E is a set of edge (a, b) that two points aand bby an edge whenever there is no third point cthat is closer to both aand bthan they are to each other (there is no other point within the intersection of the circles centered at a and b with radius the distance d(a, b))

  10. Delaunay Triangulation Graph • Graph DTG(V,E) • V is a set of nodes n (id, p) where p is a point in Euclidean space • E is a set of edge e (a, b) where e is a side of triangle constructed byDelaunay Triangulation. • Delaunay Triangulation: for a set P of points in a plane is a triangulation DT(P) such that no point in P is inside the circumcircle of any triangle in DT(P)

  11. Routing - GPSR • in Brad Karp and H.T. Kung in MobiCom 2000, pp.243-254 • GPSR (Greedy Perimeter Stateless Routing) • A node x • broadcasts a query message with destination point D • the closest node y receives and forwards the message.

  12. Example RNG with 200 nodes (subset of GG) full UDG with 200 nodes GG with 200 nodes (subset of full UDG) GG with 200 nodes over 2Km X 2Km where radio range is 250 m

  13. Routing - GPSR circle(xD) • Problem • both of node x are fartherfrom the destination D • Right-Hand Rule: Perimeter • Combination of UNG and Perimeter routing

More Related