Bluetooth networking to get data to a central data sink By: Bradley Clayton Supervised by: Alfredo Terzoli and Peter Wentworth http://www.cs.ru.ac.za/research/students/g01c2974/ g01c2974@campus.ru.ac.za Recap What do we want to do? + = Collect interactions between animals Using short range Bluetooth radios Using Ad-hoc networks to collect data from animals Focus of this project: Focus on the data collection at strategic points (watering holes) Advantages of Bluetooth: Low power consumption Data capabilities Small in size 1
Current Tracking Technologies VHF, GPS, GMS or GPS and GSM Large and heavy Consume a large amount of power Transmits a simple signal over a long distance Human probe effect Is hindered by trees above an animal Transmits over a long range Relies on nearby cell phone towers Collars are big Pin-points an exact location but does not tell us who else was there. Current Tracking Technologies Little Bluetooth Robots Bluetooth enabled Robots Let loose in a 1000x1000cm room Position information was exchanged via scatternets Collisions depended on the performance of the scatternet 2
Piconets Piconets 3
Scatternets Shared Termonology: The Degree of a device is the number of piconets to which the device belongs. Piconets and Scatternets 4
Migration and Merging Shared Migration and Merging Shared 5
Migration and Merging Migration and Merging 6
Identified Risk and Outcomes RISKS: Will we be able to get the windows Bluetooth SDK to work with our r devices? Will we be able to get into Bluetooth and have full control over it? The Widcomm SDK looks like the only option with the devices we have. h Will it work? Will we have to control Piconets and Scatternets ourselves? Will the current state of Bluetooth technology support what we want w to do? Outcomes: No, And the Windows SDK only allows basic Bluetooth profiles and sockets. Yes! Yes! The Widcomm SDK allowed us to make use of all the Bluetooth profiles as well as access to different levels of the Bluetooth stack. No. Bluetooth devices collect them self s s into Piconets and Scatternets on their own. Is this a good or a bad thing? Stay tuned! More happened before we found the answer to this question. The AWIS Simulator Developed while we did not have an SDK Simulates our system Enables us to scale the system Tested our communication algorithms and gave us an idea on how we should develop the real system when we finally found a suitable SDK 7
The AWIS Simulator Simulated Variables: Battery Life Memory Sizes Rate of generated redundant data Random movement of devices Encounter collection protocols The AWIS Tag The client side of the network Is tied to an animal Collects and swaps encounters (Adam s area of research) Deposits encounters at upload stations via piconets and scatternets 8
The AWIS Tag Device inquiry Service Discovery Notes Upload and AWISTag services Check if the Upload service was found AND if we have any closed encounters Send the encounters to the upload station An encounter can be in one of four states, Open, Closed, Uploaded and Deleted. The Upload Station Advertise the upload service and wait Allow a connecting client (AWISTag( AWISTag) Accept closed encounters from the client Send these closed encounters, via TCP/IP, to a central point. 9
The Final System Will the current state of Bluetooth technology support what we want to do? YES! We are successfully realizing encounters through the use of Bluetooth inquiries. And we are successfully depositing closed encounters. 10
BUT Not all the time It takes 10 seconds to do a proper Bluetooth device inquiry It takes 2 or more seconds to do a service discovery on each inquired device It takes 10 to 40 seconds to upload closed encounters to an access point. Sometimes devices, although physically present, are not found during the inquiry Sometimes services, although present, are not discovered. In conclusion Our system works! Bluetooth does not broadcast data like a LAN. Animals need to stay put for a while. There is a risk that encounter data can be lost. Lost encounter data may not be noted However, redundancy does reduce encounter loss. 11
In conclusion We now have a good understanding of Bluetooth. Bluetooth is getting better all the time. Bluetooth standards need to be followed a little better. More data management can be written into the AWIS system to counter these small issues. Makes nice future improvement work. Any Questions? 12