Collaborative Target Detection in Wireless Sensor Networks with Reactive Mobility

Size: px

Start display at page:

Download "Collaborative Target Detection in Wireless Sensor Networks with Reactive Mobility"

May Summers
5 years ago
Views:

1 1 / 24 Collaborative Target Detection in Wireless Sensor Networks with Reactive Mobility Rui Tan 1 Guoliang Xing 1 Jianping Wang 1 Hing Cheung So 2 1 Department of Computer Science City University of Hong Kong 2 Department of Electronic Engineering City University of Hong Kong

2 Outline 2 / Motivation 2. Preliminaries 3. Problem Formulation 4. Near-optimal Solution 5. Performance Evaluation

3 3 / 24 Challenges for Mission-critical Sensing Applications Stringent QoS requirements Target detection/tracking, security surveillance High detection probability Low false alarm rate Bounded detection delay Unpredictable network dynamics Coverage holes caused by death of nodes Changing physical environments Different spatial distribution of events

4 4 / 24 Exploit Mobility in Target Detection Sense better signal by moving sensors closer to targets Adapt to the changes of network condition and physical environments Example: fire detection

Mobile Sensor Platforms 5 / 24 Robomote @ USC Koala

manufacturing cost A small number of mobile sensors

5 Mobile Sensor Platforms 5 / 24 USC NASA GRC irobot.com Challenges Low movement speed (0.1 2m/s) Increase detection latency High manufacturing cost A small number of mobile sensors available High energy consumption Locomotion consumes much higher power than wireless communication

6 Overview of Our Approach 6 / 24 Data-fusion based target detection Explore the collaboration between mobile and static sensors Near-optimal sensor movement scheduling algorithm Reduce moving distance of sensors Satisfy QoS requirements: Low false alarm rate High detection probability Bounded detection delay Performance evaluation using real data traces

7 Outline 7 / Motivation 2. Preliminaries 3. Problem Formulation 4. Near-optimal Solution 5. Performance Evaluation

8 Signl Energy Model and Noise Model 8 / 24 Signal Energy ( 10 2 ) /distance 2 ( 10 2 ) Occurrence Noise energy ( 10 4 ) Plotted using real data traces from DARPA SensIT experiments e(x) = initial target energy x 2 noise N(µ,σ 2 ) Measurement = e(x) + noise

9 Single-sensor Detection Model 9 / 24 Local decision of sensor i { 1 if ei λ = 0 if e i < λ noise measurement The false alarm rate of sensor i ( ) λ µ PF i = Q σ The detection probability ( ) λ µ PD i = Q e(xi ) σ CCDF: Q(x) = 1 R x φ(t)dt µ λ µ + e(x i ) closer to the target, higher P D e i

10 Decision Fusion Model 10 / 24 System detection decision Majority Rule: { 1 if more than n/2 sensors decide 1 0 otherwise The system false alarm rate P F = Q The system detection probability P D = Q n 2 n i=1 Pi F n i=1 Pi F + n i=1 (Pi F )2 n 2 n i=1 Pi D n i=1 Pi D + n i=1 (Pi D )2

11 Outline 11 / Motivation 2. Preliminaries 3. Problem Formulation 4. Near-optimal Solution 5. Performance Evaluation

12 Target Detection with Mobile Sensors 12 / 24 Long distance movement can quickly deplete the battery of a mobile node disrupt the network topology Problem formulation: minimize the moving distance of sensors subject to P F α, e.g., 5% P D β, e.g., 95% Average detection delay D, e.g., 15s

13 A Two-phase Detection Approach 13 / 24 target e 1 e 2 e 1 +e 2 >λ 2 terminate! e 1 <λ 2 1st phase: each sensor makes local decision by e 0 λ 1 If the system decision is 1, the 2nd phase is initiated 2nd phase: mobile sensors move and periodically sense A sensor terminates the detection and decides 1 if Make final detection decision e 1 + e e j λ 2

14 Advantages of Reactive Mobility 14 / 24 Sensors move reacting to positive decision in the 1st phase Avoid unnecessary movement by consensus check in the 1st phase Reduce the probability of movement when the target is absent Terminate moving once enough signal energy is obtained If a loud target appears, mobile sensors can terminate movement quickly

15 Problem Formulation 15 / 24 Objective: Find the two detection thresholds λ 1, λ 2 and a movement schedule to minimize the expected moving distance: P a P D1 L 1 + (1 P a ) P F 1 L 0 correct detection false alarm Constraints: P a : the probability that a target appears L 0 (L 1 ): the expected moving distance when the target is absent (present) P F 1 P F 2 α P D1 P D2 β

16 Outline 16 / Motivation 2. Preliminaries 3. Problem Formulation 4. Near-optimal Solution 5. Performance Evaluation

17 The Structure of Optimal Solution 17 / 24 Theorem 1: Total moving distance decreases with the system detection probability in the 2nd phase, i.e., P D2 Linear approximation using the 1st order Taylor expansion n Q 1 2 (P D2 ) = n i=1 Pi D2 n i=1 Pi D2 n i=1 (Pi D2 )2 2 n P i n D2 + constant i=1 P D2 increases with n i=1 Pi D2 with high probability Simplified problem formulation Maximize n i=1 Pi D2 subject to the constraints: P F 1 P F 2 α P D1 P D2 β

18 The Structure of Optimal Solution (Cont.) 18 / 24 Combination of sensor movement is exponential Finding maximized n i=1 Pi D2 is exponential Theorem 2: In the optimal solution, each mobile sensor move in parallel and consecutively Implication n i=1 Pi D2 can be maximized by Dynamic Programming

19 Dynamic Programming: An Example 19 / 24 Two sensors: A and B Budget: two sensor moves Suppose: PD A(0) = 0.40, PA D (1) = 0.50, PA D (2) = 0.60 PD B(0) = 0.46, PB D (1) = 0.60, PB D (2) = 0.67 PD A(2) + A PB D (0) = 1.06 B PD A(0) + A PB D (2) = 1.07 B PD A(1) + A PB D (1) = 1.10 B This procedure can be implemented via Dynamic Programming

20 Outline 20 / Motivation 2. Preliminaries 3. Problem Formulation 4. Near-optimal Solution 5. Performance Evaluation

21 Simulation Settings 21 / 24 Data: public dataset of DARPA SensIT experiment Targets: Amphibious Assault Vehicles (AAVs) Sensors are randomly deployed in a 50m 50m field 75m 50m 2 50m fixed sensor surveillance spot mobile sensor

22 Impact of The Number of Mobile Sensors 22 / 24 Total 12 sensors Detection probability (%) % to 35% 60 performance 50 improvement 40 by 6 mobile sensors 30 static 1/2 mobile all mobile False alarm rate (%)

23 Conclusions 23 / 24 Propose a two-phase detection approach Reactive mobility Collaboration between static and mobile sensors Develop a near-optimal movement scheduling algorithm Provide insights into detection system design Efficient movement schedule of a small number of mobile sensors significantly boost the detection performance

24 Thanks! 24 / 24

Data Fusion Improves the Coverage of Wireless Sensor Networks

Data Fusion Improves the Coverage of Wireless Sensor Networks Guoliang Xing ; Rui Tan ; Benyuan Liu 3 ; Jianping Wang ; Xiaohua Jia ; Chih-Wei Yi 4 Department of Computer Science & Engineering, Michigan