1 VISA VIRTUAL SCANNING ALGORITHM FOR DYNAMIC PROTECTION

1 VISA: VIRTUAL SCANNING ALGORITHM FOR DYNAMIC PROTECTION OF ROAD NETWORKS Jaehoon Jeong (Paul), Yu Gu, Tian He and David Du Computer Science and Engineering University of Minnesota {jjeong, yugu, tianhe, du}@cs. umn. edu April 23 rd, 2009 IEEE Infocom’ 09, Rio de Janeiro, Brazil

Problem Definition 2 Detect Wireless Sensor Deployment Entrance point Protection point Take action

Virtual Scan vs. Legacy Schemes Entrance Protection Point l Point (a) Always-Awake 3 Sensor network is always awake. Sleeping Time Detect! (b) Duty Cycling Sensor network sleeps during the silent time (i. e. , sleeping time). 0 3 1 2 Detect! (c) Virtual Scan Each sensor sleeps during the scan time and the silent time. 0 3 1 2 Detect!

VISA Scheduling Procedure 4 Wireless Sensor Network Sensor Deployment Road Network Graph Working Schedule Sleeping Schedule

Road Network Graph 5 The vertices in G are intersections, protection points,

6 Working Schedule Establishment Working Schedule during the Scan Period When each sensor wakes up and how long it works for sensing. Working Schedule = (work start time, work end time) Working Schedule Setup Each time. sensor’s work start time is set to the scan arrival

7 Sleeping Schedule Establishment Sleeping Schedule How long all the sensors can sleep with detection guarantee. How often to generate the Virtual Scan on Road Network? Sleeping Time 0 3 1 2 Protectio n Points Entrance Points One global sleeping time. Objective is computed using the Maximize the sum of Shortest Path. scan time and silent time.

8 Sleeping Schedule Establishment When the scan arrives at e 1, the path of (e 1, p 2) is vulnerable to the intrusion. This formulation guarantees the target detection. the shortest path Maximum sum of scan time and silent time for detection

Handling Sensing Holes 9 Two Cases for Sensing Holes Deployment Time Sensor Energy Depletion Our Approach Labeling into E or P Linear Network

Hole Labeling in Linear Network 10 How to label two sensing holes? Case 1: Case 3: The Shortest Path for Sleeping Time The Sleeping Time longer than Case 1 and Case 2: The Sleeping Time is the same as with Case 1 Case 4: The Maximum Shortest Path

11 Minimum Spanning Tree (MST) based Clustering for Hole Labeling The idea is to maximize the inter-distance between Entrance Cluster (E) and Protection Cluster (P). This maximum interdistance determines the sleeping Maximum Interdistance

Performance Evaluation 12 Performance Metrics Network Lifetime Average Detection Time System Behavior Sleeping Time over System Lifetime Investigated Parameters Working Time Sensor Density Silent Time Road Network for Simulation

13 Comparison of Sleeping Time of Three Approaches over System Lifetime Comparison • Virtual Scan: 28. 2 hours • Duty Cycling: 1. 4 hours

Impact of Working Time 14 (a) Network Lifetime for Working Time (b) Average Detection Time for Working Time 1. Network Lifetime Comparison for 1 -second Working Time Ø Virtual Scan’s lifetime is 18 times longer than Duty Cycling’s. 2. Average Detection Time Comparison for 1 -second Working Time

Conclusion 15 In the surveillance for road networks, the scan-based sensing algorithm (VISA) opens a promising direction. VISA takes advantages of road network’s characteristics for a maximum sleeping time. Vehicles move along the roadways whose maps are available. Contributions Energy-efficient Sensing Schedule for Road Networks VISA uses the optimal sleeping time for an arbitrary road network. Sensing Hole Handling VISA guarantees target detection while sensor network deteriorates. Basis for Quality of Surveillance VISA can control the trade-off between the network lifetime and average detection time.