Smart sensors embedded in the wireless network could help reduce pollution, says Deborah Estrin, and even improve our response to natural disasters
Sensor networks are an exciting class of computing systems that combine distributed sensing, computation and wireless communication. This technology is touted as being as disruptive and enabling as the Internet, with broad applications such as monitoring public exposure to contaminants, managing land use and supporting safer structures. Why is this field generating so much attention, and how soon might we be able to use these applications? What will it mean to have self-configuring, autonomous distributed systems playing such a critical role in our civil infrastructures?
The Internet transformed the way in which individuals and organizations interact with each other and the virtual world. The emerging embedded network will transform the way in which we understand and manage our physical world. Sensor networks combine the wireless technologies that have revolutionized communications with sensor technologies that have revolutionized medical and industrial technology. Microsensors, on-board processing and wireless interfaces can now be integrated at a very small scale, and at relatively low power, to enable up-close monitoring of a wide range of physical phenomena, thereby enabling spatially and temporally dense environmental monitoring.
Like the Internet, these large-scale, distributed systems, composed of smart embedded sensors, will eventually cover the planet, monitoring and collecting information on such diverse phenomena as endangered species, soil and air contaminants, patients, and man-made environments. Across this wide range of applications, sensor networks promise to reveal previously unobservable phenomena and might eventually help us understand and manage an increasingly interconnected and fragile physical world.
Many of the early applications for this technology are scientific. Ecological phenomena such as biocomplexity, the carbon cycle, climate change and harmful algae cannot currently be observed at adequate spatial and temporal resolution. Public health officials are also taking an increasing interest in this technology, for improved modelling and forecasting of public health risks, as well as for real-time monitoring of existing exposure.
As an example, one of our projects at the Center for Embedded Network Sensing (CENS) involves monitoring the use of nitrate-laden treated water for secondary agriculture. Current received wisdom posits that the impact of crop growth and the microbes in the soil will together so diminish nitrate concentrations that by the time they reach the ground water, the nitrates will be at safe levels. However, the public health concern is that the nitrate level might, rather, contaminate the ground water. This is a very difficult process to verify because the soil is so heterogeneous that sampling at any one position is not necessarily representative. By deploying a dense three-dimensional grid, we will be able to make in situ observations of such important processes.
This technology presents many challenges and has captured the attention and imagination of researchers around the world. Sensor network systems will comprise many distributed elements, often deployed in physically difficult environments. Consequently they will need to be adaptive and self-configuring in order to operate autonomously for long periods of time in dynamic, heterogeneous environments. Perhaps the greatest challenge is that deployed systems will need to exploit increasing amounts of intelligence inside the network in order to scale up. In particular, as the number of sensing points increases it will be infeasible to run these systems by bringing back every sensory input to a centralized location for human inspection or computer processing. Instead, deployed systems will need to process the sensor data close to where the data are observed, and thereby filter out the interesting events from the uninteresting ones.
Sensitive and moving
Embedded sensor networks will increasingly exploit motion in addition to sensing. By introducing even small amounts of mobility into these observing systems, their coverage and effectiveness can increase tremendously. This holds true for both imagers and communication antennae (consider how your field of view is increased with small movements of your eyes or neck, or how your cellphone reception is improved with very small movements of orientation or position). We have several projects under way that exploit robotic technology to support high-density sensing and sampling in both air and water. For example, to determine the preconditions for the development of harmful algae blooms in our marine biology application, robotic nodes move around on the water surface and autonomously collect samples of microorganisms at both the water surface and below which can then be correlated with sensed environmental micro conditions.
At CENS we are also creating ad hoc networks of wireless seismometers that hop data out in the presence of potentially massive disruptions, and at challenging data rates. We are designing the technology so that scientists can quickly deploy large wireless grids for data collection in response to important seismic or volcanic events. Ultimately, we hope to support systems that are programmed with event detection inside the network so that these systems can provide real-time monitoring and alerts, in addition to their essential data gathering function.
Real world applications
As these dense monitoring systems are deployed they will present tremendous opportunities for integration with remote sensing capabilities. Establishing “ground truth” for interpretation of remote sensing images is an obvious application. However, the possibilities also include real-time adjustment of in situ assets, position, focus, attention and so on, in response to more global phenomena observed via remote sensing.
Although significant information technology research and development is needed to realize the potential of this technology (in particular in areas such as robustness, scaling, data integrity, data fusion, as well as additional sensor development), we are now at the point where we can begin applying early versions of these dense sensing and mobile technologies to engineering and commercial applications. The technology developed to characterize contaminant transport in soils and microorganisms in aquatic environments has direct applicability to precision agriculture and water quality monitoring worldwide. In addition, the structural and seismic monitoring technologies, now being used to model structural response, will be applicable to monitoring structural safety.
Who will monitor the monitors?
As embedded sensor networks move out of the laboratory we must take on several pressing social issues. Concerns are being raised about pervasive monitoring of individuals as they work and play, and questions are being asked about monitoring and validating the monitoring systems themselves. However, if we are relying on a sensor network to warn us of public exposure to harmful elements, we must have the means by which to verify system integrity.
Sensor network technology is on the brink of transforming important branches of science, such as ecology and environmental science. To this end, exciting “big science” projects are in the works, such as the National Ecological Monitoring Network (https://neoninc.org). However, as we have discussed, the impact of this technology will be felt far beyond scientific inquiry. Ultimately, high-resolution observing systems could dramatically reduce public exposure to polluted air and water, provide access to safer food and shelter, and allow first responders to magnify their effectiveness in reacting to natural and human disasters.
Deborah Estrin is a professor of computer science at the University of California, Los Angeles, and director of the National Science Foundation’s Science and Technology Center for Embedded Networked Sensing (CENS). Her research addresses protocols for autonomous, distributed and physically coupled wireless systems, with particular application to environmental monitoring.