Sensor Webs Monitoring Microclimates
One of the key problems with setting up sensor webs in the wilderness, is figuring out a way to power them. Solar cells is one solution, but solar cells don't work in every environment. They are useless underground for example. Above-ground in areas where sunlight is blocked, they are not completely useless, but don't provide enough power for continuous operation through both day and night with what little power is available.
One such location is under the canopy of a forest or jungle. In such an environment, the solar panels are competing for sunlight with nature's solar generators – leaves. The latter are much more efficient than the former, and have the advantage of being able to grow to cover all possible light sources. So much so, that very little light reaches the forest floor ,where sensors are placed.
With solar out of the question, and the small size of wireless sensors, batteries are really the only currently practical solution. Stored power that slowly drains over time, and eventually needs to be replaced. Flycatcher sugar cells might be a third option, but current varieties of insect-powered robot are simply too big and bulky to be considered for field use just yet. So the goal is to make wireless sensor battery systems last as long as possible on a full charge, so that someone can be sent out to replace them as infrequently as possible.
The German Fraunhofer Institute's department for Microelectronic Circuits and Systems (IMS) have been working on this and a few other problems for such wireless sensor webs. Their researchers have reworked the way sensors gather data so that the battery life is maximised.
“We adapted the software design accordingly and now have operating times of 12 months,“ says Hans-Christian Müller, group manager at the IMS. By pruning the sensor design right down to bare bones basics in terms of operating software, and ensuring the majority of the sensors spend most of their time in sleep mode, battery life is preserved.
Only a handful of the motion sensors are fully powered at any time, monitoring the environment for any signs of change. If something is spotted, the rest of the network is notified and powered up to check on things. After a time, all but a handful are powered down again, and they share the tasks as a collective.
Other types of sensor power up at regular intervals, again spending most of their time in sleep mode. Parameters that change slowly such as soil moisture need not be measured as often as air temperature, so the sensors don't power up as often. Since data transmission requires the most energy, the measurement values are calculated as early as possible, inside the sensor itself. The extra power drain of occasional use on-board processing circuitry is more than offset by the reduced transmission length; only sending processed data rather than raw results.
In addition, the IMS have been integrating breakthroughs by other teams into their sensors to give a more cohesive web. One of the big issues in previous webs was the need to wire the sensors together so they could communicate. However, with each sensor assigned an RFID and a broadcast transceiver, they can talk to one another. Place a new one in the ground, and it broadcasts its location (and ID code, containing data about its type) out to any other sensors in the area, who activate and respond, establishing relative positions with each other. At the same time as broadcasts back to base are made, they regularly stay in contact with one another, now that positions are known. If one doesn't respond, the others can alert base of the sensor's failure to respond.
In practical terms, this means that placing new sensors into an existing web is as simple as taking them out of the truck, poking them into the ground, or tying them to a tree. They will coordinate and mesh with other sensors in the area themselves, with no further effort required from the person placing them. Depending on which values they are to measure – for example, soil moisture content, air temperature or the moisture in the leaves – different sensor nodes are inserted into the soil or affixed to branches. If required, the measuring positions can be changed without much effort – remove, transport, place – the unit will broadcast it's ID again, and all other units in the area automatically update their own data as to where that unit is.
Along with the sensors in any given forest area are one or more mobile cellular modems. These act as another sensor as far as the web is concerned, but these are the ones all radio data is given to. These then dial out on the cellphone network, and send the actual data back to base. Again, this approach saves power on the sensors themselves, as none of them have to transmit very far.
In practical terms this now means the batteries have to be changed once every twelve months. The web can go all year without maintenance otherwise. If a unit fails, the others will let base know. If a unit develops a fault but can still transmit, it will let base know on its next broadcast.
The web is currently undergoing testing in the Northwest German Forestry Testing Facility in Göttingen, Germany, as part of the EMMON project. If all goes as expected, they will be deployed in forests across Europe, and most of the technologies will find their way into other sensor web projects.