Router2 – which links our network over the mountains to the peat research zone had fallen over in high winds. Olivia and Emma successfully repaired it on a cold day in November:
Winter was setting in and soon travel up to the sites will be very difficult. A replacement box/node/battery linked up the peat zone into the network once this was installed. Data which had been stored on peat nodes started to flow back to the server every hour (as expected) via four network hops!
In September Arthur and Kirk went to reprogram the nodes with bug fixes.
We had reduced the RAM usage of the code and removed some bugs. It also allowed us to recover readings from “lost” nodes and recharge the batteries.
Last year a laser scan of part of the mountain side, to establish a base line. So this year we have performed a second laser scan to compare to the first one. This data will be analysed later to discover how the rock face has moved in the last year.
In order to cover the study area over the ridge – we placed this sensor node in a strategic place so it passes traffic across. By this time the packets have come 3km up to the mountain then 1km across to this node – then another 600m down to the peat area.
All this with 868MHz radios running IPv6 (6LowPAN) and CoAP to gather the data!
Arthur – our superb intern team member this summer has worked very hard to get the CoAP layer working for the system and has now seen it in action on the mountain.
We’ve just installed a new IP camera at the estate. This will give us regular images so it’s possible to see what the weather is like, to help plan field trip days, and give information about the snow fall and melt throughout the year.
In 2014 we implemented an HTTP/web interface to the sensor nodes to test how much easier they would be to manage. This was successful but the large number of packets exchanged made it slow (not surprising).
In 2015 we are testing CoAP instead – as it is essentially a compressed binary form of HTTP – and only uses UDP packets – so there are far fewer radio exchanges. Initial tests in the lab with our CC1120 868MHz driver show it is very efficient – and we are now porting large parts of our code to use it. We will still be using Protocol Buffers – as they are also small binary compressed objects for data – so we should have the best of both worlds!
What this shows is a “ping” from a computer in Southampton going all the way to Glen feshie (last part via satellite link), through our “border router”, across to a node via the low power radio. It took around a second for the reply to get back to Southampton. (The IPv6 addresses have been blurred to prevent attacks).
For techies: the interval is set to 10s to prevent overloading the low power network and the packet size is set to 25 so it is small (6LowPAN prefers small packets)