An Integrated Approach to Sensor Role Selection by

An Integrated Approach to Sensor Role Selection by Mark Perillo and Wendi Heinzelman

Outline l Motivation l Background l Modeling l Solution (DAPR) l Analysis l Comparison l Conclusion

1. MOTIVATION (WSN Energy Efficiency) l Limited energy supply VS long network lifetimes l Hardware, Operating System, Low-level protocol design l Balance/reduce energy consumption l Reduce redundancy but ensure Qo. S requirement l dynamic sensor selection, in-network aggregation, distributed source coding

BACKGROUND l Abundant data l l Multi-hopping l l Filter sensors Design routing Diverse importance l Assign duties

WHAT HAVE BEEN DONE ?

Relevant Work -- Sensor Selection(Ｉ) Principle : desired coverage • PEAS activeness probing/querying; • Gur game paradigm state switching according to base station • Sensing coverage protocol sleep/wake time scheduling according to neighbors’ with differentiated surveillance neighbor redundancy, coverage redundancy • CCP coverage and connectivity

Relevant Work -- Sensor Selection(ＩＩ) Principle: Considering routing less sensors + some routings + short path = desired coverage

Relevant Work – Routing Protocols(Ｉ) Principle: Shortest path • Table-driven routing protocol • destination-sequenced distance vector routing • clusterhead gateway switch routing • the wireless routing protocol • Source-initiated on-demand routing • ad hoc on-demand distance vector routing • dynamic source routing • temporally ordered routing algorithm • associativity-based routing • signal stability routing

Relevant Work – Routing Protocols(ＩＩ) Principle: considering energy efficiency • Power-aware MAC layer routing • route through nodes with sufficient remaining power • route through lightly-loaded nodes • Maximizing the network lifetime • minimize the energy consumed every packet

PROPOSAL Sensor selection + Energy conserved routing PRINCIPLE To use the sensors not as important as data generators more liberally as routers

MODELING – Assumptions • Power consumption largely from traffic transmitted and received • Portion or entirety of an area A needs to be monitored by any one or multiple sensors • There may be one or several data sink locations

MODELING -- Varianbles

MODELING – Formalization(Ｉ) Coverage: Number of nodes: Data flow: Nt = Ns + Nsink

MODELING – Formalization(ＩＩ) Energy consumption: Scheduling: Lifetime:

MODELING – Coverage-Aware Routing Cost Common cost (energy-aware cost): Total energy in a subset area x:

MODELING – Worst Coverage-Based Cost Finds out the least-covered subregion

E(Xa) = 2; E(Xb) = 3; E(Xc) = 2; E(Xd) = 1; Cwc(S 1) = ½; Cwc(S 2) = ½; Cwc(S 3) = 1;

MODELING – Comprehensive Coverage. Based Cost Weighted sum (in terms of area of subregion) of 1/E(x) It provides a more balanced view of a nodes importance to the sensing task.

E(Xa) = 2; E(Xb) = 3; E(Xc) = 2; E(Xd) = 1; Ccc(S 1) = area(A)/2 + area(B)/2 Ccc(S 2) = area(A)/2 + area(B)/3+are(C)/2 Ccc(S 3) = area(B)/3 + area(C)/2+are(D)/1

MODELING – Combining Cost Functions Most effective in extending lifetime with 100 percent coverage: Effective in providing long network lifetimes with graceful degradation:

SOLUTION – DAPR (Distributed Activation with Predetermined Routers) The decision made in sensor selection and route discovery are influenced each other. Procedure: Route Discovery Phase Sensor Selection Phase Sensor query

DAPR – Route Discovery Phase(Ｉ) Assumption: • Nodes have location information of neighbors with redundant coverage regions; • Low power wakeup system Cost of a link = routing cost nodei x energy for transmission + routing cost nodej x energy for reception Cost of a route = sum of links in the route

DAPR – Route Discovery Phase(ＩＩ) Data Sink Node Initiate query flood query receive query Calculate link cost Update query packet forward query (delay scheme): proportional to Clink(Si) Next Node

DAPR – Sensor Selection Phase* Initial: be inactive to sense and generate data • Assign activation delay (proportional to the route cost) • Check received activation beacon check if the neighborhood is already covered • Send activation beacon send activation beacon to neighbors if possible • (Send deactivation beacon) send deactivation beacon if in high redundancy and in the highest cost route * DAPR: A Protocol for Wireless Sensor Networks Utilizing an Application-based Routing Cost

DAPR (Cont’d) • Awareness of neighbors • Location • Redundant coverage • Given highest priority • Nodes along highest route • Reserving opt-out/deactivation beacon • Sending beacon • In single hop (Dtransmission_range >> Dsensing_range) • Forwarding (no guarantee)

SIMULATION&ANALYSIS – Simulation result

SIMULATION&ANALYSIS – Experiment Result deployment Uniform Clustered Video C(Si) 1 Cea(Si) Cwc(Si) Ccc(Si) 100% coverage 362 1094 1178 904 98% coverage 521 1198 1184 1200 100% coverage 62 247 365 376 98% coverage 81 260 377 388 100% coverage 381 855 1063 717 98% coverage 585 1097 1108 921

SIMULATION&ANALYSIS – Experiment Result for Sensor Selection Configuring activation/backoff delay in Worst Coverage-Coverage Routing Cost Selection Criteria Random Individual Cost Cumulative Routing Cost Uniform 1036 1088 1178 Clustered 364 365 Video 818 800 1063

SIMULATION&ANALYSIS – Experiment Result for Combing routing cost Worst coverage-based + energy-aware routing cost β 0 Cwc 0. 05 0. 25 0. 5 1 Cea Uniform 1178 1192 1224 1093 1094 Clustered 365 360 245 247 Video 1062 1082 1106 853 855 U+ routers 1368 1572 1635 1549 1402 C+ routers 476 567 576 525 403 V+ routers 1214 1299 1306 1083

COMPARISON – to Centralized Approach l Assuming subject to these conditions: data flow, energy consumption constraints and scheduling constraints, we try to maximize the operation time of the system

COMPARISON – to Centralized Approach Uniform scenario: worst coverage + energyaware cost with DAPR gains 14% over the nonintegrated approach. 56% closer to centralized solution.

COMPARISON – to Centralized Approach (Cond’t) In clustered scenario: Worst coverage routing cost with DAPR improves lifetime by 56%. 77% closer to centralized solution, Due to use of the coverage-aware routing cost.

COMPARISON – to Centralized Approach (Cond’t) In video scenario: DAPR with the combined worst coverage and energy-aware cost makes lifetime gain 50%, Closing gap by 76%, Because the selection of sensors based on the cumulative route cost.

CONCLUSION Contribution Incorporation of coverage information into the routing protocol and the priority for sensor selection Worst coverage-based cost – maintaining 100% coverage for the maximum lifetime Comprehensive coveraged-based cost – giving a more balanced interpretation of a node’s value to the sensing task

Thank You