At the beginning of May, we spent some days "in the field" doing some preliminary test of data collection protocols. The Istituto Agrario at S.Michele all'Adige in Italy (luckily 30 minutes far away from our lab) made available for our experiments a part of their vineyards. In particular, we have access to an area of 200 x 60 meters located on a slope (a map is shown in the picture), over which we are free to install our boxes containing either TMote or 3Mate! nodes measuring temperature, humidity and light.
We had already deployed a small network in January. This first installation already taught us how hard a deployment in this environment can be. Before leaving the lab directed to the vineyard, we had to paint part of the box cover to protect the batteries from direct sunlight and we made holes to enable air recycling. However, the hard part came when we reached the place; in fact, the very same week when we had planned the installation, a lot of snow fell down, making the placement of boxes on the poles more expensive especially in terms of physical energies... something we, of course, never experienced working in Torre Aquila.
In this first tests we were interested in analyzing connectivity between nodes exploring different distances. Therefore, we installed 20 nodes (a node in the field can be seen in the picture) along two parallel lines, as shown in the map, for the maximum length of the vineyard. To these nodes we added some more needed to connect the network running the experiments with the building where we had access to the internet; this enabled the control of the system remotely from the lab. In this experiments we essentially confirmed previous results already published in the literature.
This month, we went to the vineyard with the intent to double the number of nodes installed and run quick experiments on some routing protocol. As soon as we arrived, we realized that Mother Nature had worked "tirelessly" since the winter deployment. Grapevines had grown in the meantime and the entire area, previously brownish had become incredibly green thanks mainly to the emerged foliage. This time, either a steady drizzle, in the first day we visited the vineyard, or a strong sun, in the following week, made our life in the field hard. Nevertheless we managed to install 45 nodes with which we started running some experiments...
...which returned poor results. The cause: a drastic change in the connectivity among the nodes due to the presence of dense foliage. Essentially the connectivity range did not fulfill the expected 20 meters coverage, requiring the routing protocol to spend a lot of effort to identify an actually good parent in the collection tree and then successfully deliver data to the base station. While discussing this behavior, someone rightly made me notice that in all the pictures of WSN deployments in vineyards, an external antenna attached to the sensor node is always clearly visible; we, instead, were still relaying on the ones integrated in the board of the nodes.
Despite the failure, we are not giving up and we are already discussing about going back to the vineyard... as soon as possible :-)
Last February 10th, Luca and Gian Pietro (in the pictures, taken from here) presented a demonstration of the adaptive lighting we expect to achieve in the TRITon project, thanks to the usage of a Wireless Sensor Networks. At ETTX 2009 (The First European TinyOS Technology Exchange), we introduced the idea of exploiting the fine-grained sensing of light in the tunnel, made possible by the WSN technology, to perform online adaptation of the light intensity generated by the lamps.
The demonstration consisted of few TMote-like devices measuring light and reporting data to the sink. The gateway, a gumstix to which the sink has been attached, makes data available to an algorithm controlling the light intensity of a small lamp affecting the measures at some reference node. This control loop adjusts the output of the lamp so that the sensor readings stabilize at some value, in the case of the demo, defined a priori. The poster presented at ETTX can be downloaded here.
Our group already convinced the local authorities in Trento of the suitability of WSN for the challenges posed by the TRITon project. We made the same demonstration with 20 nodes (far more than what we used at ETTX), in a non-operational tunnel, as shown in the picture on the left. Each of the nodes were reporting light samples at a frequency of one every 30 seconds. In the tests we ran in the tunnel, we always experienced a delivery rate higher than 99% at the sink, demonstrating the effectiveness of the routing and reliability mechanisms. Even simulating the situation of nodes failures due to damage or battery exhaustion, the routing protocol nicely adapted to get rid of the no-longer available nodes in the process of data forwarding.
Next demo? At IPSN in San Francisco; we are showing the system we deployed in Torre Aquila. See you there!
Debugging and testing code for WSNs is known to be an hard and everlasting task. Usually the activity ends up in spreading motes everywhere around the office where they attract the attention of the cleaning ladies, who think them as nice blinking useless DIRTY toys that need to be cleaned.
Excited by the last year full of new deployments, we decided to spend the first day of the year creating our own private WSN. After some minutes spent thinking about a possible placement for the nodes, someone turned on the lights... yes... there it is the best place!
In an hour, with several rolls of tape, USB cables and some equilibrium we installed 14 TMote Sky along the lamps placed on the ceiling. The sink is a 3Mate!, equal to the nodes we placed both in Torre Aquila and in the tunnels of the TRITon project, made by 3TEC, a manufacturer in Trento which provides us with customized TMote-like nodes.
Now we have our own stable deployment available to debug and test the new features we are planning to add to our TeenyLIME middleware. Do you want to know more? Stay tuned!