A Framework of Distributed Indexing and Data

Dissemination in Large Scale Wireless Sensor

Networks

[Author Yingshu Li][Presenter Harsh Achrekar]

Proposal

• Integrated distributed Connected dominating set Based Indexing (CBI) data dissemination scheme – Handling of large amount of sensing data–Min. usage [limited n/w computational resources]– providing timely responses to queries– scalability , load balance and adaptivity during

dynamic changes.

Connected Dominating Set [CDS] Concept

• For a graph G(V,E), a Dominating Set S of G is a subset of V where each node in V \ S is adjacent to at least one node in S.

• Connected subgraph of G • Nodes in C are dominators,

the others are dominatees.• Candidate for virtual

backbones in WSNs.• Routing tasks - dominators

• A k-hop dominating set D in G is a set of nodes such that every node in G is at most k hops away from at least one of the nodes of D.

• Fig. (b) shows 2-hop dominating set.

k-Hop Dominating Set [k-hop DS] Concept

Broadcast v/s data dissemination framework

• Without CDS – A node broadcasts the packet whenever it first

receives one• With CDS – All dominatees only need to send packets to their

closest dominator(s) – dominators forward the packets towards their

destinations.

ArchitectureCONNECTED DOMINATING BASED INDEX (CBI)

ArchitectureCONNECTED DOMINATING BASED INDEX (CBI) • Storage nodes – Sensing data is collected and stored close to sensing nodes – to decrease the communication cost – form a k-hop dominating set of the whole network.

• Index nodes – Information of high level semantically rich data are pushed

and maintained – formed by a connected m-hop dominating set.

• Queries are routed to the appropriate index nodes instead of flooding into the whole network.

• Storage and index nodes usually do not sense raw data

A Layered Approach• Bottom layer - sensing nodes - monitor the targets to

generate raw sensing data.

• Middle layer - storage nodes - store the high level semantically rich data.– max. dist(sensing nodes ,storage nodes )<= k hops. – raw data need not travel across the entire n/w.

• Top Layer – index nodes - store the index information for those high level semantically rich data– use connected m-hop dominating set as index node set to

dominate the storage nodes . – According to property of connected m- hop dominating set,

max. dist (storage nodes ,index nodes) <= m hops – index node set is strictly connected.

Storage Nodes Determination

• Query Injection time from sink to index node– Query_inject_time(sink ,index )<=(k+m) hops (constant).

• construct k-hop DS of n/w - obtain storage node set• Randomly select a node(usually a center of n/w) as a

root. Initialize BFS search on root.

• Every/All node exchanges info with its k-hop neighbors about the level[?], degree[?] and ID[?]. {? – explain}

• To obtain Storage node of small sizesubgraph composed by all storage nodes have small

diameter any leaf node should not be a storage node

• if (L1 < L2) or (L1 = L2 && D1 > D2) or (L1 = L2 && D1 = D2 && ID1 < ID2)then(L1,D1, ID1) >(L2,D2,ID2)

• At first, Leaf node u =smallest (Level,Degree,ID) among its neighbors sends a DOMINATING message to its exact k-hop away Parent node v =largest (Level,Degree,ID) to request v become a dominator (storage node).

• When v receives this DOMINATING message, it becomes Black and broadcasts a BLACK message to all of its k-hop neighbors

• On receiving a BLACK message from its parent, u becomes Gray and broadcasts a GRAY message to all of its k-hop neighbors.

• Next node with smallest(Level,Degree,ID) among its k-hop neighbors that have not decided its status start this procedure.

• Terminate - root ={ Gray,Black}.

• Node 26 ,a leaf nodes with variable (4,1,26) sends a DOMINATING message to Node 7 with largest (Level,Deg, ID) within Node 26’s 2-hop neighbors. Node 7 becomes Black and broadcasts a BLACK message.

Index Nodes Determination• Construct connected m-hop dominating set I to

dominate all storage nodes only and use all the nodes in I as index nodes.

• Form new graph G’ of all the storage nodes and their parents in G.

• Construct a connected m-hop dominating set. - all the storage nodes should be dominatees

• Add connectors to connect the red nodes which are index nodes.

Index Construction

• Each index node stores one copy index of its dominatees (storage nodes).

• Flood query to all index nodes for query result• flood overhead is much lower – size (index node set)< size( nodes in whole n/w)

• When a query is injected into the sink (Node 29) it is forwarded to the Storage Node 4 which stores the data for Node 29.

• Then, the query is flooded to all the index nodes. After that, the query result is returned to the sink after getting the result.

Results