I recently had the opportunity to attend NERCs (Natural Environment Research Council) third, and final, “Networks of Sensors” Annual Technology Showcase Event. It was particularly interesting as my area of research falls under sensor networks, in particular real-word deployments. It was a good opportunity to see what others had done with a fair bit of funding. There were six talks/projects in total ranging from air pollution monitoring at Heathrow to glacial dynamics monitoring in Greenland.
My favorite projects were a deployment of 20 high precision GPS nodes on Helheim glacier in East Greenland and aircraft deployable GPS stakes developed by BAS. Over the course of two months, the nodes in Greenland relayed position data every 7 seconds to help understand how glaciers (in particular the end terminating in a fjord) move. Glaciers are an incredibly harsh enviroment for sensor networks. Crevasses, combined with the difficulty in mounting equipment on ice, makes it incredibly hard to set up any equipment, let alone networks. The approach used, was to use multiple “base stations” on the rocky terrain either side of the glacier to cover the whole 5km wide glacier. Each node would attempt to communicate with two base stations in an attempt to ensure that one base station would be able to communicate successfully with the node. Early paper.
The BAS Aircraft deployable GPS stakes were simply very clever and nice bits of engineering. They resembled 3m tall sleek aluminium missiles. Designed to be dropped from an aircraft, they first descend by parachute and then free fall. To stop them from making too big a hole in the snow/ice, the stakes are fitted with folding “snow brakes”. Some 50 nodes had been deployed and were still operational almost a year later. The deciding factor in their survivability was the rate at which snow accumulated and, therefore, how quickly the nodes were buried. ADIOS.
Something I found a little surprising, although in hindsight less so, is that most of the successful, large-scale deployments weren’t really “networks”. Most used GSM, Iridium or WiFi on each node to sidestep the problem of creating/managing/maintaining your own infrastructure. While it was, just about justifiable, in the applications presented there are clearly a huge number of other applications where such an easy route isn’t an option. Whilst Iridium coverage is almost global, the cost of an Iridium SBD modem is around £300 and data costs are prohibitively expensive unless you are lucky enough to get a discount. Equipping 20 nodes with such a modem isn’t ridiculous, equipping 1000-2000? The initial cost alone would be around £500,000 for the modems alone.
GSM is a lot more sensible, as the costs per device is a lot lower, the same applies to the data costs. The problem with GSM is that it relies on the environment you are sensing being covered by a mobile network. For deployments in cities this is a perfectly reasonable approach. It would even be a sensible idea in some rural areas, in countries with good GSM infrastructure. Anywhere else; off -shore, in forests, up mountains, in deserts, on glaciers or ice sheets it’s just not going to work. WiFi also suffers exactly the same problems, more severely than GSM in fact. If your nodes are near a city, or built up area, fine, otherwise woops – no WiFi.
I’m actually not having a go at the achievements made by the teams, in fact most of the environments they worked in suit such technologies. The fact that they have actually deployed, for months or years, large numbers of devices in the real world is absolutely brilliant and I applaud them. I should admit to being a little jealous I can’t sidestep all these “hard bits” in my own research. However, what really took me back was that this sort of research should be about pushing the envelope. Sensing the natural environment is going to take you away from cities and developed areas. Creating reliable, flexible and maintainable networks for sensors can’t be avoided in this field and I’d hoped to see some impressive work in that direction.
All in all, it was an interesting and worthwhile trip. It’s always good to have a chance to talk to other researchers and see what they have done.