Energy-Efficient Computing for Wildlife Tracking

1. Energy-Efficient Computing for Wildlife Tracking Design Tradeoffs and Early Experiences with ZebraNet Created by Chris Roedel Presented by Israel Martinez

2. Overview • Background Information • Other Sensor Networks • ZebraNet Design Goals • Life as a Zebra • Collar Design • Protocol Design • Experimental Results • Discussion Questions Some Graphics Courtesy of Princeton University ZebraNet Project

3. Background Information • Biology researchers want to track and monitor wild animals • Long term • Over long distances • Need detailed information about what the animals do to see how environment affects them • Current tracking techniques are far too primitive and are not very useful to researchers Some Graphics Courtesy of Princeton University ZebraNet Project

4. Why Other Networks Won’t Suffice • No infrastructure to support Wired Networks => Wireless • Flying over an area and looking for VHF ‘ping’ signals • Can miss interesting events • Limited to daylight hours • Hard to obtain data from reclusive species • GPS + Satellite Uploads • Most sophisticated system can only store 3000 position entries & no biometric data • Satellite uploads are slow, power hungry and $$$$ • Peer to Peer offers opportunity to improve Some Graphics Courtesy of Princeton University ZebraNet Project

5. Contributions • First system to employ both mobile nodes AND mobile base stations • Specialized communication models that funnel data towards a base station and optimized for a high degree of latency tolerance • Examine energy tradeoffs using real system energy measurements from prototype in operation Some Graphics Courtesy of Princeton University ZebraNet Project

6. GPS Position taken every 3 minutes Detailed activity logs for 3 minutes / hour 1 year ‘untouched’ operation Operation over 100s or 1,000s of sq km Latency is not critical, but high probability for delivering all data eventually Zebra collar weight limit of 3-5 lbs. No fixed base stations, antennas, or cell service Design Goals for ZebraNet Some Graphics Courtesy of Princeton University ZebraNet Project

7. Day to Day as a Zebra • Social Structure • One type of Zebra moves in ‘Harems’ • Generally, only one male in the ‘harem’ => reducing the number of collars need to track a large number of zebras • Groups of ‘Harems’ form Herds • These dynamics challenge ecologists, but will help ZebraNet transfer information between ‘harems’ • Movement Patterns • Distance Moved • Net distance moved in a 3 minute period • One of three groups: Grazing, Graze-Walking, Fast Moving • Turning Angle • How far does the animal turn during each of the 3 phases • Water Sources and Drinking • Need to find water sources at least once per day • Sleep • Must rely on keeping watch and fleeing from predators Some Graphics Courtesy of Princeton University ZebraNet Project

8. Zebra Movement Speeds • From Field Data • Grazing: • 0.017m/s • Graze-walking: • 0.072 m/s • Fast: • 0.155 m/s • Turns ~ < 60° Some Graphics Courtesy of Princeton University ZebraNet Project

9. Non-Intrusive Necklace Design • GPS, Flash RAM & CPU • Short Range Radio • Long Range Radio w/ Packet Modem • Does not show: • Packaging • Batteries • Solar Array • Power Management Circuits Some Graphics Courtesy of Princeton University ZebraNet Project

10. Power and Weight Information Total Weight Goal 3-5 lbs. Energy Goal: 5 days if no recharge Some Graphics Courtesy of Princeton University ZebraNet Project

11. Harem A Harem B A,B B,A Basic Protocol in Action Harem A and Harem B come within short range radio range. They transfer their own information with each other Some Graphics Courtesy of Princeton University ZebraNet Project

12. Harem C Harem A Harem B C A,B B,A Basic Protocol in Action Harem A and Harem B move away from each other, but Harem B moves within range of Harem C, transferring both B’s and A’s information to C. Harem C transfers its information to B. Some Graphics Courtesy of Princeton University ZebraNet Project

13. Harem C Harem B C,B,A B,C,A Basic Protocol in Action Now Harem C is within Long Range Radio range of the mobile base station and can transfer its information along with B’s and A’s. The base station has the information from all the animals even though it only came within range of Harem C. Some Graphics Courtesy of Princeton University ZebraNet Project

14. Protocol Design • Two peer-to-peer protocols evaluated here • Flooding: Send to everyone found in peer discovery. • History-Based: After peer discovery, choose at most one peer to send to per discovery period: the one with best past history of delivering data to base. • Compared to “direct”: no peer-to-peer, just to base • Success rate metric: Of all data produced in a month, what fraction was delivered to the base station? Some Graphics Courtesy of Princeton University ZebraNet Project

15. ZNetSim It’s a zebra mobility model and simulation environment for ZebraNet It was armed with facts and field observation about zebra behavior and reasonable assumptions of the terrain and operating characteristics of the Mpala Research Center in Kenya It returns two metrics: • Success rate – the percentage of data that gets back to the base station • Energy consumption Some Graphics Courtesy of Princeton University ZebraNet Project

16. Experimental Results • Used their own ZNetSim simulator to vary parameters and determine best solution • These graphs show the difference between direct connections and peer-to-peer multihop Some Graphics Courtesy of Princeton University ZebraNet Project

17. Experimental Results (cont) Direct Connection proves to be the least reliable type of connection Some Graphics Courtesy of Princeton University ZebraNet Project

18. Experimental Results (cont) Some Graphics Courtesy of Princeton University ZebraNet Project

19. Experimental Results (cont) • Radio range key to data • homing success: ~3-4km for 50 collars in 20kmx20km area • Success rate: • Ideal: flooding best • Constrained bandwidth: history best • Energy trends make selective protocols best • Mobility model key to protocol evaluations • Fast random moves hurt history • Chicken and Egg: • mobility model is the biology research goal Some Graphics Courtesy of Princeton University ZebraNet Project

20. Future Plans • Hardware: • Nov ‘02: waterproof prototype for local animal tests • Spring ’03: build final rugged version • Software: • Impala middleware: • Allow for wireless software updates & adaptivity • Beyond wildlife tracking: • Resource recovery, traffic management, security, surveillance… Some Graphics Courtesy of Princeton University ZebraNet Project

21. Conclusions • ZebraNet as Engineering Research: • Early detailed look at mobile sensor net with mobile base stations • Demonstrates promise of large-extent, long-life sensor networks with GPS • Detailed look at power/energy concerns • ZebraNet as Biology Research: • Enabling technology for long-range migration research • Good view of key inter-species interactions Some Graphics Courtesy of Princeton University ZebraNet Project

22. Discussion Questions? • Does this model what they want? • What other animals could be fitted with these sensors? Some Graphics Courtesy of Princeton University ZebraNet Project