Sensor Networks
Wireless sensor networks are a new technology for collecting data about the natural or built environment. Our research is creating technologies that will radically improve the cost and quality of data gathering to enhance the understanding of our natural environments and provide the ability to manage and exploit Australia's resources.
Research
 |
 |
 |
 |
 |
One of the key challenges for deploying sensor networks that can operate for years without maintenance is that the nodes are able to operate using energy harvested from the environment – typically using solar panels. This energy-neutral operation is complicated by the fact that the amount of solar energy available can be severely restricted during periods of significant cloud cover or as foliage grows around a sensor node. By looking at the typical trends of solar energy patterns in covered and open areas, we are developing new techniques for predicting future energy levels and adapting the behaviour of nodes (e.g. back off duty cycle) in order to ensure there is sufficient energy resources for nodes to maintain basic functionality at all times. Read more... |
 |
In-Network Processing of Complex Data The ability to handle complex data such as that from audio or video sensors will mean that sensor networks can be used for increasingly sophisticated environmental monitoring applications. Our research is addressing the challenges of processing complex data within a wireless sensor network where there are severe resource constraints in terms of processing power, bandwidth, and available energy resources. We are developing methods for efficient processing of audio/video information and distributed event detection via feature, output or decision fusion techniques. We are also developing and testing techniques for in-network storage of complex data types, and in-network interpolation strategies for dealing with missing data. Read more... |
 |
Robot – Sensor Network Interaction We are exploring ways of using robots to augment the capability of fixed sensor networks. Static nodes have a limited sensing horizon and typically a low-cost sensor, low-computational power and low energy budget. By contrast a mobile agent (robot) has more energy and computation available, and can carry larger and higher quality sensors. Further, since it is capable of returning to base the robot can recharge, download data over a high bandwidth link and recalibrate its sensors. Robots can also physically interact with the environment and with the static sensor network for tasks such as deployment, redeployment and repair of nodes. Meanwhile, the sensor network can act as a communications fabric for mobile agents, and can provide information beyond the range of a robot's on-board sensors. Read more... |
 |
Our research into sensor networks is informed by the challenges we discover in deploying practical sensor networks. Critical to this is our ability to produce robust, reliable hardware, as network nodes (including the electronics, sensors, and housing) must be sufficiently robust to survive for many years in all weather conditions. Our Engineering Support team provides us with essential expertise in circuit board design, sensor integration, and the construction of novel housings and mounting systems. Read more... |
CSIRO's Smart Sensor Network Technology
Our node hardware platform is designed to be extremely versatile from top to bottom – enabling a vast range of possible applications while maintaining standardized communication capabilities for interoperability. The latest revision of this hardware supports IEEE 802.15.4 on both 2.4GHz and 900MHz which allows it to avoid interference on a whole frequency band. The platform allows data-rates from 20 kbps (with range larger than 12km) to 3 Mbps (short ranges), and nodes can carry many kinds of sensors, including miniature cameras and microphones.
Early versions of our sensor nodes operated with the Fleck Operating System (FOS), a cooperative thread-based operating system which was developed in-house at CSIRO. We have now moved to the widely supported TinyOS operating system, enabling us to share and collaborate easier with the community. We aim to further leverage the potential of combining TinyOS with state of the art industrial research by porting many of our FOS tools to TinyOS. In parallel, we are developing online anomaly detection tools that can alert network operators and managers of specific issues in their deployment for better monitoring and maintenance of long-term deployments. These efforts are complemented by our ongoing development of over-the-air network reprogramming to enable easier information retrieval and code updates for our deployments.
