A sense of things to come – smart sensors and the environment

Key text

How do you manage a unique natural resource like the Great Barrier Reef when it's threatened on so many fronts? Coral bleaching caused by rising water temperatures due to climate change, is possibly the biggest and most immediate threat.

Coral bleaching – one of many threats to the reef
(image: Great Barrier Reef Marine Park Authority)

But then there's also coral disease; sediment, fertiliser and pesticide pollution from mainland run-off; and the growing spectre of acidification. The reef ecosystem is already susceptible to the effects of climate change; pollution only makes the problem worse. Each of these threats comes from a different source, operates at a different scale and interacts in a variety of ways.

Appropriately managing the many challenges confronting the reef requires reliable and timely information on its changing environment. Where are sea temperatures rising? How fast are changes happening? Where are excess nutrients and sediment coming from? Is it a gradual process or do they come in pulses following heavy rain? Are there any changes in the reef environment prior to an outbreak of coral disease?

But perhaps the bigger question is how do you collect the information to answer these questions? And don't forget, the Great Barrier Reef is a very big place covering 300,000 square kilometres and made up of a network of around 3000 reefs. Environmental monitoring may be needed continuously at multiple locations and over long periods. Deploying an army of scientists is out of the question in terms of budget and resources. Remote sensing by satellite can be useful but often doesn't provide the required level of detail. For example, satellite images can only reveal water surface temperatures at resolutions of a kilometre or more. This does not provide sufficient detail to investigate the cause of damaging events such as coral bleaching. Measurements at smaller scales and at various depths are needed to better understand the nature of bleaching events.

Detail is one thing, being able to see changes in real-time throughout the reef would also help researchers and environmental managers to preserve this environmentally and economically important asset. And the only way this can be realistically achieved is by deploying networks of sensors over the reef to monitor its many processes.

And that's exactly what is happening. Sensors of all shapes and sizes are being deployed across the Great Barrier Reef to monitor different processes. The monitoring is being likened to covering the Great Barrier Reef with a digital skin, enabling scientists and managers to pick up real-time information about the reef.

This may be the way of the future for all forms of environmental management, be it catchment management, bushfire risk management, agriculture or monitoring invading pests (Box 1: Environmental applications of smart sensor technology). When the task involves understanding environmental changes over large or multiple spatial scales, sometimes in areas where humans are few and far between, then smart sensor networks may be the answer.

Sensors getting 'smarter'

Humans have been using environmental sensors for centuries. For example, consider the rain gauge. All over the world we have been putting out a variety of rain-collecting vessels that can accurately tell us how much rain has fallen at any particular location. Rainfall is one of the most important environmental variables around but, as sensors go, the traditional rain gauge is not very 'smart'. It collects information but it can't do anything with that data, and it needs regular checking. What's more, data collected at one location may not reflect what's happening elsewhere.

A traditional rain gauge is a good reliable sensor but it's passive. It collects, you record. In recent decades automated weather stations have been developed that have the capacity to collect meteorological data automatically – 24 hours a day, seven days a week – and then transmit that information to a central location. While these stations are much 'smarter' than the old passive rain gauge, weather stations are expensive and you can only set up so many.

But all that is changing as an exciting mix of nano-engineered materials, miniaturised computers and rapid wireless communications is giving rise to a new generation of environmental sensors. These new sensors are smaller, cheaper and more versatile than anything we've seen before and their deployment in intelligent, interacting networks (Box 2: Intelligent sensor networks) is opening up a new age of environmental monitoring.

These new sensors will be able to process information they collect and adapt to changing conditions. They'll be able to measure an amazingly wide range of environmental variables, and they'll be able to share this information with a network of sensors. And because they'll be smaller and cheaper, we'll be able to build more of them and create networks with a greater coverage than has ever been possible before.

The nuts and bolts of a smart sensor

Smart sensors come in all shapes and sizes, detect many different things and process the data they collect in different ways.

The name 'smart' or 'intelligent' sensor relates to a sensor's ability to detect information of interest (eg, a sudden change in water temperature), process the data it collects, communicate with other sensors and have some capacity to adapt to changes in the environment.

To be 'smart' there's a basic set of components that a sensor can use. The first component is the unit that senses or measures some environmental variable. This variable might be sound, light, temperature, pressure, magnetic field or some chemical compound, or a combination of variables (eg, temperature and pressure).

Next, our sensor needs some capacity to process the raw data being collected, and this is usually done through its own tiny computer processor. Combined with this processing capacity is also the ability to store this information in its computer memory and only transmit information of interest.

The sensor needs to be able to communicate with other sensors so the data it collects and the information it processes can be passed through a network of sensors to a central computer. For maximum convenience this communication may be wireless; if so, the sensor also needs some form of radio transmitter.

Sensing, processing and communicating – these are the three basic requirements that make a sensor smart. On top of this, however, it'll also need a power supply to drive these processes.

The challenge now is to refine the developing smart sensor technology.

Making more of smart sensors

Like many areas of emerging technology, the science of smart sensors and smart sensor networks is strongly multidisciplinary in nature. It may involve electrical, mechanical and materials engineers as well as biologists and chemists. In addition, putting together smart sensor networks needs good information technology skills and interpreting collected information involves statistics and network modelling expertise. To help bring these people together, the Australian Research Council (ARC) has established the ARC Research Network on Intelligent Sensors, Sensor Networks and Information Processing (or ISSNIP).

Related site: ARC Research Network on Intelligent Sensors, Sensor Networks and Information Processing (ISSNIP) Provides details of smart sensor research being conducted through ISSNIP. (Australian Research Council, Australia)

ISSNIP is developing smart sensor solutions for the environment, defence, security, transportation and health monitoring. Some of the many fascinating applications of the research conducted through ISSNIP are the development of navigation systems for aircraft based on the visual systems of bees, sensors to detect drugs, helping unmanned aerial vehicles find targets and facial recognition for security systems.

Research on ways to improve smart sensors is advancing on many fronts. This includes development of more sensitive detectors, smarter communication between sensors and more efficient ways of processing information by sensors and sensor networks.

Internationally, the push is on to develop smaller and cheaper sensors. How small can a sensor become? Well, believe it or not, researchers are working on smart sensors, referred to as 'smart dust', that are the size of grains of sand.

Monitoring the Great Barrier Reef

Research trials like those being conducted on the Great Barrier Reef are an important way to refine smart sensor technology for a range of applications. Once established, smart sensor networks that provide access to real-time information about the reef will be valuable tools for its management.

Related site: Microsoft® SensorMap Provides interactive maps with up-to-date information from sensor sites around the World, including Australia. (Microsoft® Research, USA)

The Great Barrier Reef Ocean Observing System will see the installation of sensor networks across seven Great Barrier Reef sites. Wireless sensor networks, for example, are being tested at Davies Reef. They involve the placement of a number of environmental sensors that measure temperature, salinity, light and oxygen. The information from some of these sensors is presented in an easy to use web-based format via SensorMap.

There's also a wireless sensor network being installed in Nelly Bay at Magnetic Island. This network consists of temperature sensors vertically positioned below the ocean surface 2 metres apart. The network will beam back temperature information to a base station real-time to help understand upwellings and their impact on the productivity of the Great Barrier Reef. Another sensor is also being deployed there to measure wave frequency.

Heron Island sensor network, one of seven proposed networks on the Great Barrier Reef
(image: ISSNIP)

Scientists and environmental managers need detailed, real-time information about the reef environment if they are going to fully understand the processes occurring there. Eventually smart sensor networks across the reef will be providing some of the answers so that it can be better protected against pollution and climate change.

Related Nova topics:

Wireless but not clueless

Coral bleaching – will global warming kill the reefs?

Box 1: Environmental applications of smart sensors

One of the big environmental challenges of our time is managing freshwater in an age of climate change. Sensor networks can be used to help manage this precious resource by continuously monitoring water quality and water flow into catchment areas that are often remote and broad-ranging. The resulting information can also be combined with other sensor networks monitoring soil condition to assist with the fair and efficient distribution of water for irrigation, as well as for management of the surrounding environment.

For example a sensor network is proposed for a North Queensland catchment that empties into a reef system. This network will provide information about the effects of nearby land use on the immediate catchment environment and in the adjacent reef system.

Similar technology could also provide early warning of flash flooding. Indeed, networks of smart sensors have already been deployed in a river for just this purpose in the UK by researchers at Lancaster University. Some of the sensors measure pressure from below the waterline in order to determine depth. Others monitor the speed of the river flow. Each sensor node is smaller than a human fist and powered by batteries and solar panels. Each is also accompanied by a computer unit about the size of a packet of chewing gum, which contains a processor about as powerful as those found in modern mobile phones.

When do plants need a drink?

And it’s not just catchment managers who are interested in monitoring water. If farmers could take the guesswork out of when their crops and pasture plants need water they could make their operation a lot more efficient. Scientists at CSIRO have been investigating the installation of wireless sensor networks over farms to measure soil moisture to determine the most effective irrigation for a particular field.

Researchers at the University of Colorado have taken water monitoring on farms one step further. They have developed clip-on sensors the size of a fly's wing that are attached to plant leaves during the growing season. The sensor monitors moisture content and chemical signatures that can indicate when the plant is undergoing water stress. The chemical signs, such as an increase in salt and sugar content in the cells, occur much earlier than physical signs, such as drooping leaves.

Because of its tiny size, the sensor can only transmit a signal about half a metre away to the next sensor. The signal passes along the network to a base station and is then transmitted to the farmer’s computer, which analyses the data. The result is either emailed to the farmer (for example, ‘turn irrigation on now’) or may even switch on the water for the farmer. The system is expected to save up to 40 per cent of water use.

Preparing for bushfires

Every Australian city is vulnerable to bushfires, and as our urban zones increasingly extend into bushland areas it’s a problem that’s only growing. Low cost sensor networks set up around our cities could monitor local moisture levels, humidity, wind-speed and direction. Integrating this information with satellite imagery and long term weather forecasting would enable better understanding of fire risks and bushfires as they develop. And such information could prove invaluable in planning a coordinated disaster response or providing early warning to high-risk areas.

Listening for invading toads

Cane toads have been spreading across northern Australia for decades but it’s always been difficult to monitor the invasion front because it’s so large and is often in unpopulated areas. Smart sensors are now being trialled in an attempt to better monitor when the noxious toad first enters a region. Researchers at the University of NSW together with National ICT Australia have developed wireless sensor nodes consisting of an acoustic sensor, a micro-processor and a wireless communication unit. Each sensor picks up toad or frog calls and relays them to a microserver that also forms part of the network. The microserver receives the data, determines whether a cane toad is present and then relays this information back to a central control room. The system is being progressively trained to distinguish different frogs and so far can recognise the calls of up to nine frog species from northern Australia. By making the data collection automatic with wireless transmission of the results, pest management has become more feasible across remote and widespread areas.

Related sites

• Australian Broadcasting Corporation
  • Plants tell sensor when they need a drink (News in Science, 23 July 2007)
  • Tracking cane toads from the sofa (News in Science, 12 July 2005)
  • Future farms (ABC Rural, 1 February 2007)

• Water catchment flow monitoring (ISSNIP, Australia)

Box 2: Intelligent sensor networks

Each sensor node has the ability to collect a variety of environmental data, process that information and communicate it to other sensor nodes in the network. That makes the nodes pretty clever, but when they're working together the network they form has its own special abilities. The data being collected from all the sensors can be sent back to 'headquarters' from a single node, rather than using multiple streams of data. This reduces band width requirements as well as power consumption for the communication of the network's information.

In addition, the fact that the network is passing on information from a number of sensors improves the accuracy of the information gathered compared to that obtained from a single sensor.

Sensor networks can also spot and deal with anomalous information from individual sensors – that is, information that isn't consistent with information coming from other nodes. It could be that the sensor delivering the unusual data has malfunctioned, as is often the case when a sensor is operating in a hostile environment such as the ocean. These so-called ad hoc networks can isolate faulty sensors and continue to provide quality information from the remaining sensors while, at the same time, signalling that the malfunctioning sensor needs attention.

However, it might be that the anomalous readings are the first sign of some type of environmental event taking place. For example, the signal might be the first appearance of colder or warmer water over a coral reef. If neighbouring sensors verify this 'event' then the network may be programmed to increase the monitoring capacity of the affected sensors (for example, double the number of temperature recordings being made). The network is sensitive to changes in environmental conditions and responds to ensure important information is captured – it has the capacity to self organise in response to a changing environment. Australian research is refining the technology for detecting unusual readings in sensor networks through the Great Barrier Reef Project.

The intelligent network can tell you what's happening at a variety of scales simultaneously. It can report back on changes happening around individual nodes but it can also interpret data from a cluster of nodes to demonstrate if some impact is happening at a larger scale.

So an intelligent network of sensors offers many advantages including variable network scales, reliability, efficient use of resources and lower power consumption.

Related sites

• Ad hoc networking in the forest (National Science Foundation, USA)

• Ad hoc Networks (HowStuffWorks, USA)

• March of the motes (New Scientist, 23 August 2003)

Activities

• Higher Education Authority (Ireland)
  • Sensors, motes and environmental monitoring – contains information, practical activities and a quiz on sensor technology

• National Aeronautics and Space Administration (USA)
  • Rain sensor – students build a rain sensor and discuss basic sensor technology

• National Oceanic and Atmospheric Administration (USA)
  • Keeping watch on coral reefs – students learn about threats to coral reefs. They then use satellite data to assess the ‘alert level’ of reefs and discuss the value of environmental monitoring programs.

Activity 1. Using sensor data from the Great Barrier Reef

The graphs on the left (red line) show the number of days that the reef water has been at or above the temperature shown on the x-axis. When this crosses the threshold number of days (black line) the reef is at risk of a coral bleaching event.

1. According to the data, which reef site is currently at the highest risk of a coral bleaching event? Explain your answer.

2. Using the information on the web page explain why the threshold number of days is different for each reef.

3. Explain the trend in the threshold curves with increasing temperature.

At the top of the page, change the time period to 2001-2002 and click ‘generate charts’

4. Compare the overall risk of a coral bleaching event at Davies Reef in 2001-2002 to that in 2007-8.

The graph on the right for Davies Reef shows the average water temperature over the selected time period compared to the long-term average temperature. Coral bleaching events often occur when temperatures are 1ºC or more above the long-term average.

5. Is it likely that Davies Reef suffered a coral bleaching event in 2001-2002? Explain your answer using the data available.

Using SensorMap hover over the sensor symbol at Townsville, Queensland to open up data for that site.

6. List the different conditions that sensors measure at Davies Reef.

7. Find out what PAR stands for and explain the benefit of having both PAR data and temperature data when assessing productivity on the Great Barrier Reef.

8. Using no more than 400 words, evaluate the use of intelligent sensor networks for environmental monitoring on the Great Barrier Reef compared to data loggers.

Activity 2. Helping sensor technology with biomimetics: Research assignment

Evaluate the role of biomimetics in the field of sensor technology using Australian examples of biomimetics research.

Assessment rubric:

Category	excellent	highly competent	competent	needs improvement
Structure	Key points are clearly presented and synthesised from the various sources. Paragraphs are well structured and progress logically.	Most key points are clearly presented. Paragraphs are mostly well structured and progress logically.	Presentation of key points may at times be unclear. Paragraphs are mostly well structured.	Presentation of key points from the various sources is often unclear. Paragraphs are not well structured.
Information	Information is highly relevant to the topic. The use of biomimetics has been clearly explained and thoroughly evaluated using 2-3 Australian examples.	Information is relevant. The use of biomimetics has been clearly explained and evaluated using 1-2 Australian examples.	Information is mostly relevant. The use of biomimetics has been explained and evaluated using 1-2 Australian examples.	Information from the various sources is at times irrelevant.  The use of biomimetics has not been explained and evaluated effectively using Australian examples.
Research	8 or more sources of information have been cited. Information is from books, reliable internet sources and scientific articles	6-7 sources of information have been cited. Information is from books, reliable internet sources and scientific articles	4-5 sources of information have been cited. Information is from books, reliable internet sources and/or scientific articles	0-4 sources of information have been cited. Information is not from a range of sources eg books, reliable internet sources and scientific articles
Bibliography	Bibliography handed in and set out using the correct format.	Bibliography handed in with few errors.	Bibliography handed in with several errors.	Bibliography contains many errors or no bibliography included.

Useful sites to get you started:

• ISSNIP, Australia
  • Biomimetics
  • Research projects

• Australian Broadcasting Corporation
  • Srini’s Bees (Catalyst, 15 February 2007)
  • New robot brain takes to the skies (Catalyst, 18 December 2003)

• Aggressive bees may track future of flying robots (University of Queensland)

• Fly on the wall inspires vision technology (Cosmos, 29 August 2006)

Further reading

Useful sites

Clearly reviews sensor technology and its applications (includes animations).
http://www.nsf.gov/news/special_reports/sensor/overview.jsp

ARC Research Network on Intelligent Sensors, Sensor Networks and Information Processing (Australia)

• Anomaly detection in wireless sensor networks
  Covers research into intelligent sensors and sensor networks on the Great Barrier Reef.
  http://www.issnip.unimelb.edu.au/research_program/sensor_networks/core_technologies/anomaly_detection/anomaly_detection_in_wsn

• SensorMap for the Great Barrier Reef
  Describes the SensorMap project which links sensor data from the Great Barrier Reef to users through an internet site.
  http://www.issnip.unimelb.edu.au/research_program/sensor_networks/environmental_monitoring/sensor_map/sensormap_for_the_gbr

• Projects
  Lists a range of Australian research projects involving intelligent sensors, sensor networks and information processing.
  http://www.issnip.unimelb.edu.au/research_program

Australian Broadcasting Corporation

• Srini's Bees (Catalyst, 15 February 2007)
  Describes the use of the bee visual system in sensors for aircraft navigation.
  http://www.abc.net.au/catalyst/stories/s1848841.htm

• Smart dust (The Science Show, 5 April 2003)
  Explains the development and potential applications of smart dust sensors.
  http://www.abc.net.au/rn/science/ss/stories/s800732.htm

Facility for Automated Intelligent Monitoring of Marine Systems (Integrated Marine Observing System, Australia)

Explains the use of sensor networks as a component of the Great Barrier Reef Ocean Observing System.
http://www.imos.org.au/faimms.html

How motes work (HowStuffWorks, USA)

Explains wireless sensing networks using smart dust or motes.
http://computer.howstuffworks.com/mote1.htm

Glossary

coral bleaching. Loss of colour of corals due to an environmental stress such as increased water temperature, pollution or sedimentation. Environmental stress can cause corals to expel microscopic algae from their tissues. These symbiotic algae provide up to 90 per cent of the coral’s energy needs. Loss of these algae results in the bleached appearance of corals as they provide most of the coral’s colour. Bleached corals often starve then die if the stress persists.

coral disease. Coral diseases can have a range of causes including bacteria, fungi and algae. Coral disease has had a significant effect on reefs in the Caribbean and the incidence in some areas of the Great Barrier Reef has increased in recent years. The coral disease White Syndrome has been linked to increasing sea temperatures on the Great Barrier Reef.

intelligent sensor. A sensor that has an inbuilt ability to sense information (eg, light, temperature, salinity), process the information and send selected information to an external receiver (including to other sensors). Also called a smart sensor, mote or smart dust. Intelligent sensors are able to detect particular conditions such as abnormal temperature and react according to programmed instructions (eg, by increasing the sampling rate).

microserver. A self-contained computer system that can be integrated into a remote network with limited support. A micro-server can receive, process and forward information (eg, from sensor nodes).

nano-engineered. Constructed at the atomic or molecular level, generally at 100 nanometres or smaller. One nanometre is equal to one-billionth of a metre.

node. In computing a node is a network junction or connection point. In intelligent sensor networks each sensor can therefore be considered as a node in the network.

smart sensor. A sensor that has an inbuilt ability to sense information (eg, light, temperature, salinity), process the information and send selected information to an external receiver (including to other sensors). Also called an intelligent sensor, mote or smart dust. Smart sensors are able to detect particular conditions such as abnormal temperature and react according to programmed instructions (eg, by increasing the sampling rate).