Algorithms improve water pollution measurement
Chemical Field Effect Transistor (CHEMFET) sensors offer a number of benefits over more traditional chemical sensors. They are compact, easily and cheaply manufactured and have low power and supplementary electronics requirements. The SEWING project consortium took advantage of these features when designing a new water pollution monitoring prototype based on CHEMFET sensors. One of the disadvantages of CHEMFET sensors that the consortium had to overcome is the issue of signal processing. The sensors are highly sensitive to interference from other ionic species and care must be taken to remove this noise from the signal. Furthermore, the signal is non-linear and strongly affected by the ambient temperature. The SEWING consortium brought together signal-processing experts from a number of different countries to address the problem. A multi-step measurement cycle was implemented, incorporating various techniques to increase the signal-to-noise (S:N) ratio. This included injecting Ionic Strength Adjuster and potassium, the primary interfering species, into the sample at different stages. An optimisation algorithm was developed to accurately measure ion concentrations from a number of CHEMFET sensors. Redundant sensors can be employed to further reduce measurement uncertainty. Another approach applied an inverse algorithm to a time series of measurements. A third and final scheme to increase the S:N ratio was drawn from parallel research in multiple interference cancellation. The signal processing software was successfully integrated with the prototype's operating software. In this way, accurate pollutant ion concentrations are generated in quasi real-time enabling the prototype with early warning functionality. Offline signal processing is also possible. Information about publications regarding these algorithms can be found online at: http://www.sewing.mixdes.org(opens in new window)
