Self Organising Collegiate Sensor Networks (SECOAS)

Project Summary

SECOAS was a BT-led, 3-year DTI collaborative project that finished at the end of March 2006. The main aims of this project were, firstly, to develop a wireless sensor network for oceanography that would give unprecedented spatial resolution of a complex coastal environment, and, secondly, to develop and test lightweight techniques for pervasive ICT (information communication technology) that would have a range of properties, including resilience, adaptation, self-configuration and responsiveness to user requirements.

Intelisys, now trading under the name Salamander, built the sensor module that was suspended from a pyramidal cage, beneath a customised buoy. On the 60 cm diameter mooring buoy was mounted a box housing data loggers, PIC microprocessors, and a low-power 173.25 MHz radio, to which a 50cm antenna was attached. Some preliminary radio trials were carried out with partners Essex University, Plextek, and Kent University.

Whilst the sensor devices were developed to be deployed off the coast, they could also be deployed as fluvial sensors, as they have a small footprint, that could be further scaled down when operating in a more protected environment.

A number of unique capabilities was demonstrated by putting lightweight software on to the devices to optimise sampling rates, forwarding of data and in-situ data compression. The AI (artificial intelligence) running on the low-cost, low-power microprocessors, enabled devices to be deployed without a priori knowledge of the environment to be monitored. That is, the devices exhibited adaptation to the environment in which they were deployed, in order to give the best characterisation of the environment of interest, within user-specified parameters of required lifetime and measurement priorities.

The devices were able to

share operating parameters with their neighbours, in order to refine their management of sampling rate, forwarding rate and data-compression, within available resources of battery power and bandwidth.

respond to differential measurement priorities, resulting in a measurement quality of service guarantee.

respond to further user preferences, dynamically, such as the volume of data required, or the length of the monitoring experiment, by sending short messages into the deployed sensor network.

demonstrate real-time data-processing, at the node level, such that a subset of data points could be forwarded that was still able to represent the key features of the shape of the curve of data, and thereby conserve battery power.

As part of two wireless sensor network trials at sea, BT gathered detailed radio and networking data. This resulted in an unexpected observation: that there was an excellent correlation between wave activity on the sea and the quality of radio communication between the wireless sensor node buoys. The sheer simplicity of the devices, whose data-gathering and communication of that data are inextricable, means that that they are very robust, because they do not have any sub-surface or other mechanical sensing elements, and the electronics consists of one simple radio microprocessor.

Current oceanographic devices give detailed information about a very limited number of points, perhaps 1 device for every 100km or more of coastline. Our technology would be cheap enough to enable sea-state monitoring devices to be distributed all around coastal waters, at short intervals, of the order of 1 km. Furthermore, the sea-state information gathered would refer to the sea between buoys, as well as at the buoys themselves. So, the consequence of this technology is the ability to create an extremely rich, real-time observation tool. This real-time data could be used to enhance current state-of-the-art modelling of oceanographic processes, and to deliver rapid warnings of sea conditions to shipping and coastal residents, in the event of high seas and flood conditions.