Data Reduction in multi-hop collection of agriculture data
Résumé
Recent droughts have once again proven the threat of climate change on our food production systems. Building efficient and frugal soil monitoring networks may be part of the solution. Providing real-time data on temperature, humidity and soil moisture can help decision makers to optimize irrigation, fertilization and pest control schedules. Additionally, such networks can be used to raise alert in case of a disease outbreak allowing for a swift response. One of the great challenges to make these tools widely available lays in building frugal and efficient data collection. In part, these networks should have low energy requirements to allow for deployment in remote areas and reduce maintenance costs. Networks with large amounts of transmissions often involve redundancies in the transmitted data, which results in high energy consumption for low utility. For example, the measurement of temperature at a point A and another measurement of temperature at a second point close to A. Using a well-chosen prediction algorithm, estimations of a node's measurements could be an acceptable alternative to transmitting the real values. This work extends our previous work designed for leveraging correlation between two nodes to explore this correlation between multiple nodes over a multi-hop network. Results show that we can reach up to 250 times less data while not losing the information which validates our approach as we can see in the evaluation section.
Domaines
Informatique [cs]Origine | Fichiers produits par l'(les) auteur(s) |
---|