Questions/Discussions
=====================


Q: What was the resolution of the images?
A: 200x200 pixels, black and white.

Q: Was our system enough for the biologists?
A: Of course, the biologist always want more; 
   higher resolution, color images, and more number of images per day.
   Although we can never fully satisfy their desire (unless wireless
   technology matches the performance of wired devices at same or lower 
   prices), we believe that we have made a progress towards the future.
   Compare 3 images-per-hour to 1 labor-intensive-manual-check-per-day!!

Q: What is the actual power usage profile of the system?
A: We don't have the exact nor run-time data on power usage,
   but our preliminary in-lab test showed current draw of
   30mA during idle period, and 60~70mA when active.
   (active meaning cyclops taking image with IrLed on?)

Q: Wouldn't it be better to use WiFi(2.4GHz)?
A: We selected 433MHz (CC1000) because it had better propagation compared to 
   2.4GHz (CC2420) in our deployment topology where there were dense foliage 
   and no line-of-sight. Of course, this limited our radio bandwidth.
   If there are better radios that have higher data rate and works well in
   our environment, we are willing to use them.
   We did not use WiFi because they require much more energy. We wanted our
   nodes to run indefinitely with solar panel.

Q: Multihop communication?
A: Our software supported multihop routing, and our deployment 'allowed'
   multihop routing. In fact, we have observed transient multihop routes.
   Our deployment results do not show long-term multihops because we used
   the same bird nest locations from the previous deployment 2 years ago 
   without changing their locations. Back then, the birdnests were 
   placed so that they can communicate in 1-hop. Our current software 
   supports multihop routing and you can flexibly move the nodes.

Q: What problems can we identify from end, remotely?
A: The problems that we can see remotely are;
   - server down
   - stargate down
   - mote reachable, but does not return image
   - mote un-reachable
   Server down is due to power outage at James Reserve. (powered by generator)
   Stargate down is suspected to be power problem. The problem could have been
   alleviated if we had robust recovery-after-reboot mechanism.
   A mote that is reachable (pingable) but does not return image had two main
   reasons. Majority of the cases were low battery: a mote works correctly
   when supplied voltage is above ~2.7. If the battery level is around 2.4~2.6,
   they are pingable (can perform simple operations) but cannot invoke camera
   module or access the flash memory. Another reason is loose connection
   between a mote and the cyclops camera.
   Un-reachable mote (that does not route through crashed stargate) is usually
   due to depleted battery. When we deploy our nodes, we carefully select the 
   locaion of the solar panel so that they get sufficient sun light. However,
   as time goes by, the tree grow, foliage increases, and sun moves. Also, the
   weather can be cloudy for several days. Hence, solar pannels may not get
   sufficient sun light and the batteries may get depleted. Our reasons for 
   unreachable mote include broken antenna connector, corroded power connector,
   loose cables, mal-functioning solar panel, etc.

Q: Did we use any duty-cycling or low-power-listening kind of technique?
A: No, we did not. We were sending images 'as-fast-as-possible'. There were not
   much 'idle' time for duty cycling. And we believe that this will not change
   in the near future since no matter how many images we transfer, biologists
   will always want more and more images.

Q: What is the main lesseon learnt?
A: 'Energy' is indeed important. As we have desribed above, most of the 
   problems that we've encountered are power or energy related. Using solar 
   panels does not guarantee anything.


Q: How long did it take to identify the problems?
   Why did it took 2 weeks to get back?
A: We can identify above 'remotely-identifiable' problems in near real-time.
   It took us 2 weeks to fix the problems only because James Reserve was at 
   distant location and several people needed to coordinate to schedule a trip 
   to the deployment site.

Q: How much was tailored for birds?
A: Our packaging was tailored to fit on the bird nest. However, other parts of
   the system were not. The hardware configuration (mote+cyclops+sensors) can
   be (and have been) used at the pitfall-traps to monitor lizards. All 
   software running on the motes and the stargates are generic; they can be 
   re-used in other applications. Only the end-software running at the 
   server needs to be modified to fit the application requirements (such as
   image resolution, how often to take images, how to store that, etc.)

Q: Routing and transport protocol?
A: We used Tenet, which uses MultihopRSSI (for mica2) for mote routing.
   We used RCRT protocol in Tenet for rate-controlled reliable data transfer.

Q: Will this work at James Reserve be extended? 
A: We are not sure since many member and the funding situation has changed.