Author：Jianliang Xu, Xueyan Tang, and Wang-Chien Lee A New Storage Scheme for Approximate Location Queries in Object Tracking Sensor Networks Author：Jianliang Xu, Xueyan Tang, and Wang-Chien Lee Reference： Jianliang Xu; Xueyan Tang; Wang-Chien Lee,” A New Storage Scheme for Approximate Location Queries in Object Tracking Sensor Networks”, in Parallel and Distributed Systems, IEEE Transactions on Volume 19,  Issue 2,  Feb. 2008 Page(s):262 - 275 報告者：林育弘 2018/5/29

Outline Introduction System model Energy-conserving Approximate Scheme Performance Analysis Performance Evaluation Conclusions 2018/5/29

Introduction(1/3) Object tracking sensors networks applications Military control Wildlife Monitoring A large range of sensor network Research 如何追縱object 如何降低update traffic 2018/5/29

Introduction(2/3) 目標追蹤的感測網路經過多個節點的傳輸工作 Many applications 易使資料出現誤差 Many applications Can tolerate a certain degree of error To make good use of fault-tolerant space Enhance energy efficiency Effectively improve the performance 2018/5/29

Introduction(3/3) Energy-conserving Approximate SchemE (EASE) Scheme 尋找出最佳半徑 並結合傳統的Centralized Storage(CS)與Local Storage(LS) 效能 可降低網路通信量96% 可延長sensor network 的生命期 2018/5/29

System model(1/4) 透過定位讓每個sensor可以知道自己的所在位置(by GPS) 讓一群sensor nodes組織為一個群體 每個群體有一個sensor node 為 cluster heads 功能性比較 : cluster heads > sensor nodes 比其他的node有更大的儲存空間 可以與其他群的cluster heads做溝通 2018/5/29

System model(2/4) Approximate Location Queries 任何一個 sensor node可傳送Query Each inquiry will have the goal to record Data structure <object_id , p> id : 被標記起來目標的編號 p : 查詢此物體可以接受的容錯範圍 2018/5/29

System model(3/4) Centralized Storage(CS) Use on 優點 缺點 Not very strict requirements must be very precise queries 優點 Query costs are low 缺點 update costs are too high CS node Update answer query D Q 註： CS node : Centralized Storage node D : Detecting node Q :Querying node 2018/5/29

System model(4/4) Local Storage (LS) Use on 優點 缺點 Strict requirements must be very precise queries 優點 Do not update the cost 缺點 query costs are too high LS node Q 註： Ls node : Local Storage node Q : Querying node answer 2018/5/29

EASE -Architecture CS node LS node r Approximate area 2018/5/29

EASE CS node Answer 1 update LS node Querying node Answer 2 Approximate area Answer 1 : P >= r Answer 2 : P < r 2018/5/29

EASE - Location Updating(1) Local update 當物體移動之後(仍在原先的範圍之內) detecting node則會改為離物體最近的node detecting node在Update Report給LS node C L Update 註： C : Centralized Storage node L : Local Storage node D : Detecting node 2018/5/29 D

EASE - Location Updating(2) Remove update 當物體移動之後(離開了原先的範圍) detecting node則會改為離物體最近的node detecting node會轉變為此目標物的LS node LS node再Update 給CS node Update C L 註： C : Centralized Storage node L : Local Storage node D : Detecting node L 2018/5/29

Performance Analysis(1/2) 2018/5/29

Performance Analysis(2/2) C(r)：計算出最佳的半徑 query rate * CS cost query rate * LS cost Remote update rate * Remote update cost Sampling rate * Local update cost 2018/5/29

Performance Evaluation(1/4) 2018/5/29

Performance Evaluation(2/4) 2018/5/29

Performance Evaluation(3/4) 2018/5/29

Performance Evaluation(4/4) 2018/5/29

Conclusions 使用EASE做準確度的查詢處理作研究 EASE in Object Tracking Sensor Networks 有效地降低網路通信量 節省感測器節點的電力 改進感測網路的生命期 2018/5/29