Self-organising sensor network

A University College London spinout has launched a versatile wireless sensor network inspired by the behaviour of creatures that act together, such as flocks of birds, termite mounds or beehives.

Senceive’s Flatmesh product allows hundreds of sensors to form a robust, non-hierarchical radio network ‘mesh’, where each unit communicates with its neighbours rather than via central routers. In the same way as a ‘biological network’ of a mass of animals acts together, simple rules applied to each component part allow behaviour more intelligent than the sum of its parts.

Flatmesh components can be fitted with a variety of sensors which allow them to be used for applications as diverse as monitoring railway embankments for slippage or preventing precious artworks being exposed to excess levels of light.

The original concept sprang from a UCL project investigating novel ways of building short-range radio networks for applications where a great deal of sensors are needed in a densely-packed area. The Self-Organising Collegiate Sensor Networks (SECOAS) project deployed a dense network of sensors mounted on buoys in the North Sea to monitor factors such as the turbidity of the water that could indicate a sandbank shifting near a wind farm. Each buoy carried a short-range radio – just enough to get to the next two or three buoys.

Dr Matthew Britton, chief executive officer and co-founder of Senceive, said: ‘The idea is that instead of building a big radio that blasts information back to a central point, you have just one of them that does that and the others can be miniaturised so they only need to talk to each other.

‘As our network modes are miniature, the challenge was to give these very limited components simple rules to allow a network of them to behave cleverly together. The individual sensors collaborate, and if they observe something interesting, only then do they communicate that.’

Steven Schooling, director of physical science and engineering at UCL Business, added: ‘The rationale for the research was the deficiency observed in current wireless systems such as Zigbee.’

Zigbee can be complicated for customers to set up, and the remote battery-operated sensors need to be within range of a constantly powered coordinating device. Senceive developed its own network architecture where each node talks to its neighbours and one unit has a gateway capability that sends the information to a PC.

Britton said: ‘Every device is the same and there’s no tree structure like Zigbee, just a blob of devices that send the information. Like a biological network, it’s a robust system with little structure. But if something goes wrong, as long as each device is not too far away from others, the information will find its way round, even if half the network is destroyed.’

Senceive overcame early battery-life problems by building tight synchronisation schedules into the wireless network. Each sensor wakes up for a few milliseconds every 10 seconds or so, collaborates with its neighbours then goes back to sleep if nothing untoward is reported. They also report back on battery status and give plenty of advance notice when they need to be replaced.

Senceive has carried out a number of monitoring projects for Network Rail to detect such events as bridge strikes, changes in track geometry, or embankment landslides.

Britton said: ‘For the prototype embankment-monitoring project, we installed 30 nodes fitted with sensitive tilt-monitoring devices fitted on poles stuck deep in the ground. Installation needed to be quick as it costs a lot of money to get track possession. We deployed it in one day either side of a track.

‘We simulated a landslide by tilting nodes over manually. If two neighbouring devices measured a tilt, we called that a landslip. If it was just one, it could be that someone just knocked it over. If one disappeared from the network, neighbouring devices reported that as a catastrophic failure as the device could have been buried.’

The water industry has expressed interest in using Flatmesh to monitor pressure at pumping stations, across bound aries managed by different authorities and coming into people’s homes. Britton said: ‘If just one node sees a drop in pressure, there may be a leak, but if several see it, there may not be a problem, just standard flow deviance.’

There are also potential military and security applications, such as a perimeter network that can detect intrusion through breaking a light beam, or triggering a tripwire or proximity sensor.

The Technology Strategy Board has sponsored an 18-month, £500,000 Flatmesh project with a number of partners. These include the British Museum, which is concerned about volatile chemicals from some objects damaging others, and Historic Royal Palaces, which wants to monitor damp over hundreds of locations in the Tower of London. The National Trust wants to keep track of the ‘light budget’ in rooms and around objects – there’s only a certain amount of light exposure allowed over time, after which they need to be stored in the dark.

The next step for Senceive is to release a GPRS version of Flatmesh that will work over the mobile phone network. In the longer term, the company aims to develop Flatmesh into its proposed Intellimesh product, where more intelligence is embedded into the network, interpreting the results as it operates so that less human interaction is required.

Berenice Baker