Landmark-Based Information Storage and Retrieval in Sensor Networks Qing Fang Department of Electrical Engineering, Stanford University Jie Gao Department of Computer Science, Stony Brook University Leonidas J. Guibas Department of Computer Science, Stanford University INFOCOM 2006
Outline Introduction Related Work Landmark-Based Data Centric Simulation Conclusion
Background sink
Background sink
Background sink
Geographical Hash Table (GHT) lion
Disadvantage of GHT No distance-sensitive producer lion consumer
Disadvantage of GHT No distance-sensitive Communication bottleneck
Disadvantage of GHT No distance-sensitive Communication bottleneck Bad for queries the cross-type data
GHT with Structured Replication quad-tree d = = 4 d = = 16
GHT with Structured Replication d = 1
Related Work Title – GLIDER: Gradient Landmark-Based Distributed Routing for Sensor Networks Author – Qing Fang, Jie Gao, Leonidas J. Guibas, Vin de Silva From – INFOCOM 2005
GLIDER a b c t s d
Overview producer h consumer T1T1 T2T2 T3T3 T4T4 replication path retrieval path
Replication Path producer h LaLa LcLc LdLd a LeLe LfLf b LgLg c consumer No distance-sensitive LbLb
Distance-sensitive
Optimal Principle a → b → c → d → e → f → g c → d → e → f → g
Optimal Principle producer h LaLa LcLc LeLe LfLf LgLg consumer LbLb LdLd
Finger Tree producer h LaLa LcLc LeLe LfLf LgLg consumer LbLb LdLd
Finger Tree LaLa LcLc LeLe LfLf LgLg LbLb LdLd s
Replication Path producer h LaLa LbLb LcLc LeLe LfLf LgLg a consumer LdLd
Replication Path producer 1 h LaLa LbLb LcLc LeLe LfLf LgLg a producer 2 LdLd
Simulation Network Enviroments – 316m × 316m sensing field – 2000 nodes – 11m communication range – 6.2 degrees – 23 landmarks – Simulated with C++
Simulation Data structure and storage requirement – The landmarks – Neighborhood distances to its reference landmarks – The hash function – A bit to record it’s on boundary or not – The IDs of its neighboring sensors
Simulation Compare producer cost with GHT
Simulation Compare consumer cost with GHT
Simulation Retrieval Path Length GHT100% Landmark-based70.2%
Simulation Load distribution of landmark-based Load distribution of GHT
Conclusion A location-free, landmark-based information brokerage scheme for sensor networks – Distance-sensitive – Load-balanced
Thank You !