Wireless sensor networks can provide farmers with real-time measurements of parameters such as soil moisture, crop health, and animal movement. This data will help inform practices for sustainable agriculture.
Therefore, helping farmers to more accurately and effectively control activities such as irrigation, planting, stock movement, and pesticide application. Eventually networks of nodes with both sensors and actuators will eventually not only monitor the agricultural environment but also control it intelligently.
The DSS Group has worked extensively on digital agriculture, from tracking livestock position, real-time classification of their activities through wearable sensors, in-rumen wireless gas sensing. The virtual fencing technology that we developed and patented is being commercialised.
A significant research challenge of this work is processing the large streams of sensor data captured by the embedded devices. For instance, our project on understanding the behavioural aspects of the animals involves placing accelerometers, magnetometers, and microphones on collars or ear tags. The sampling rate of these sensors easily runs into the tens of Hz, and is much higher for audio (in the order of Khz). Sending all this data in real-time through the wireless interface is impossible due to bandwidth and energy limitations. To address this issue, we develop real-time classification algorithms that run in-situ on the embedded nodes and are computationally efficient to avoid overloading these battery or solar power devices. The figure below shows an activity-labeled trace from one of our experiments.