For six weeks a network of 39 MyriaNed nodes equipped with three different relative humidity and temperature sensors will have gathered data in a greenhouse when the network is removed on Tuesday, November 2. One of the use cases being tackled in the STORM project concerns climate control in greenhouses, and after the experiment in a senior apartments building, this experiment is the first attempt at the real deal. Via Priva, contact was established with Lans, a large greenhouse company growing different kinds of tomatoes. They were kind enough to host our experiment in one of their greenhouses, and give valuable feedback on the application of wireless sensor networks in greenhouses.

The experiment serves three goals. First, to test the quality of current digital humidity sensors. The traditional analog sensor is currently still used in greenhouses, and for good reason, as it is very precise, also under extreme circumstances. While the quality of digital sensors, especially the kind that is cheap enough to be used in bulk in a wireless sensor network, has increased a lot, it remains to be seen if they can rival analog sensors when humidity levels go over 90%. For this reason, the nodes in the greenhouse are equipped with three different humidity sensors, from Sensirion, General Electric and Smartec.
The second reason is to evaluate the performance of the network in a hostile environment like a greenhouse. The MyriaNed platform was designed to be highly self-organizing, so that it can in principle be deployed in any environment and any configuration, and still function. However, these self-organizing capabilities do depend on the nodes being able to communicate with each other frequently enough. As we use a 2.4Ghz radio, the high humidity typically found in greenhouses thus make the greenhouse a challenging environment for the MyriaNed platform.
Third, climate control in greenhouses is anticipated to be a commercial application for the MyriaNed platform. By doing experiments in a greenhouse while the platform is still in development, we will be able to take into account findings from these experiments and specific wishes from the greenhouse farmer in the further development of the platform.

As the experiment is still ongoing, not too many results are available yet. However, a sneak preview of the performance of the network can be found below. In the picture, the amount of data points received from each node is plotted, according to the location of the node in the greenhouse. The sink node is located off the (1, 1) corner, and along the axis labelled "Pad" there is a path in the greenhouse. As one can see, not all nodes provide the same amount of data; in fact the differences can be quite drastic. However, there is no very strong correlation between nearness of the node to the sink and the amount of data that was received from the node, as one might expect. During the experiment, the nodes keep track of the amount of radio messages they receive from neighbouring nodes, and save this data in their data flash. Thus, once the nodes have been collected, this information will hopefully give more insight into the connectivity of the network.