The Virtual Sensor Concept - Separating Sensor Software from the Hardware

Abstract

Aquatic environmental sensors are expensive which limits the ability to undertake widescale remote monitoring. In most instances, too much logic and functionality are contained within the sensor device itself. This approach is expensive and computationally restrictive. This paper presents a new architectural paradigm for remotely deployed sensors whereby the calibration logic (and other functionality) is separated from the physical sensor hardware. A sensor device is only responsible for taking raw unprocessed sensor readings, which are transmitted back to a central server. All processing occurs on the server where computational capability and sophistication are unbounded. This approach allows for significant flexibility in terms of dynamic calibration adjustments to be applied to sensor data (e.g., in the case of sensor fouling or device decay) and for statistical quality assurance/data production algorithms to be applied. We present an example of how this paradigm can be adopted in this context to a turbidity sensor.