1 ROME: Road Monitoring and Alert System through Geo-Cache Bin Zan, Tingting Sun, Marco Gruteser, Yanyong Zhang WINLAB, Rutgers University
Road Monitoring in Mobile Sensor Networks New road monitoring services are enabled by cars with sensing, communication, computing and storage capabilities. System Assumption Location-based interests Data items are accessed through queries after they are generated. 2 camera GPS rotation sensor accelerometer vibration sensor cell phone DSRC Communication Sensing Positioning Skid Resistance Surface Distress Road Roughness
Data Collection Challenges in Mobile Networks Centralized Bandwidth (Cellular) Cost (DSRC) Storage 3 Query-response Difficult to locate data source afterwards Local aggregation Data are stored around where they are generated by passing cars Short-ranged radios have larger bandwidth than cellular links
Location Aggregation Three local aggregations. Aggregation content format 4 LocationTimeEvent IdDetector (−92315:74000)2:00PM1VD 3245 (−92316:73989)2:30PM2SX 1443 (−92316:73989)2:45PM3MK 3098 Duration 2 hours
Geo-Cache Concept Anchor Location: Locations of interest, where sensed results are aggregated. Geo-Cache Aggregated information around the Anchor Location. Geo-Cache Collection Protocol Protocols to hand over information through inter-vehicle communication as nodes move away. 5
Geo-Cache Collection Protocol Broadcast Geo-Cache (Baseline) Periodically broadcasting Geo-Cache to those closer to the anchor location, until time-to-live timer expires. High overhead Relay Geo-Cache Acknowledgment Overhearing Delayed Geo-Cache 6
Relay Geo-Cache Relay Protocol Acknowledgement 7 A B C ACK
Relay Geo-Cache 8 A B C Relay Protocol Acknowledgement
Relay Geo-Cache 9 A B C ACK Handoff Relay Protocol Acknowledgement
Relay Geo-Cache 10 A B C Relay Protocol Overhearing
Relay Geo-Cache 11 A B C Relay Protocol Overhearing
Relay Geo-Cache 12 A B C Handoff Relay Protocol Overhearing
Delayed Geo-Cache (Boomerang) Relay protocol incurs high overhead when detection probability is low Delayed protocol: cars keep their Geo-Cache locally, and later hand off to an opposite collector 13 AB Detect C
Delayed Geo-Cache (Boomerang) 14 ABC D Relay protocol incurs high overhead when detection probability is low Delayed protocol: cars keep their Geo-Cache locally, and later hand off to an opposite collector
Delayed Geo-Cache (Boomerang) 15 AB D C Relay protocol incurs high overhead when detection probability is low Delayed protocol: cars keep their Geo-Cache locally, and later hand off to an opposite collector
Delayed Geo-Cache (Boomerang) 16 B D C Geo-Cache back to Anchor Location A Relay protocol incurs high overhead when detection probability is low Delayed protocol: cars keep their Geo-Cache locally, and later hand off to an opposite collector
When Nobody is Around … What if the handoff process cannot find a candidate (the current carrier is the only node in the area)? Infrastructure: cellular, WiFi, road-side DSRC units, etc. 17
Performance Evaluation Performance metrics Communication Overhead Geo-Cache anchoring delay (the time between when the geo-cache is carried away from the anchor and when it returns) NJTP Trace Represents relatively sparse off-peak traffic from 6 to 8 AM, south NJ. Synthetic workload 70km two lane road. Cars’ arrival rate as Poisson distribution. Average speed: 30 m/s. Radio range: 100m for dense traffic and 250m for sparse traffic NS2 b as PHY and MAC Two-ray-ground propagation model 18
Performance Evaluation Broadcast Geo-Cache vs. Relay Geo-Cache. 19
Performance Evaluation Delayed (Boomerang) vs. Relay Geo-Cache. 20
Conclusion New mobile sensing applications are being enabled Local aggregation through the Geo-Cache structure provides a cost-effective method for mobile sensing Delayed Geo-Cache protocol can significantly reduce the communication overhead while anchoring the Geo-Cache especially with lower detection probabilities. 21
Questions & Answers 22