Wireless sensor networks (WSN) consist of small devices (e.g. sensor nodes) with low cost (typically $1 - $20), low power (resulting in a long battery and shelf life), low data rate (up to approx. 250 KB/s or 2% the speed of Wi-Fi) and low processing capabilities (based on embedded 8/16-bit systems). WSNs are typically ad-hoc (there is no fixed infrastructure) and isolated (there is no reference to the Internet or any other network).
This technology is increasingly used in diverse areas like patient monitoring (i.e. wireless biosignal monitoring), residential / commercial control (i.e. ambient temperature/humidity measurement), industrial control (i.e. remote machine monitoring), environmental monitoring (i.e. pH of a waterway) and smart grid systems (i.e. advanced meters).
The WSN market is gaining momentum with 100s of millions of sensor nodes to be deployed world-wide over the next 10 years. One of the most important drivers for this development relates to increasing energy costs / environmental concerns and subsequent massive worldwide investment in smart grid technology in its broadest sense. Sensor networks have a key role to play in this initiative, from improved building and environmental monitoring to smart metering. Data centres (DC) consume 3% of global energy and this is predicted to increase significantly with evolution of cloud computing. A gain in energy efficiency in DC energy consumption will translate into significant savings. Other drivers include the medical domain where there is growing need for better ambulatory patient monitoring at home or in step-down facilities and more general company savings through better asset management.
Security of these networks, however, is critical. End users do not want their health data being exposed, while manufacturers must comply with legislation regarding privacy such as HIPAA (Health Insurance Portability and Accountability Act). Similarly, in the energy domain broad security concerns within the evolving smart grid represent a key challenge to its success and sensor security is one aspect of this. OSNA will look into these issues and provide an authentication solution for such networks.
Jonathan Hanley, Fanie Meiring, Hugh Melvin, Michael Schukat
Enterprise Ireland Commercialisation+ Funding
09/11 – 08/13
This research deals with the problem of multi robot exploration and object recognition. We propose a system to accurately map an environment and to model and classify objects within the environment. Our research is intended to be of use in the development of service robots to carry out tasks in domestic or industrial environments.
To model an environment accurately, we aim to utilise the ability of robots to accurately recognize and localise each other. A consistent map is thus achieved by maintaining visual contact between robots, with different robots assigned roles of explorer and observer. While maintaining contact between some team members, the team as a whole spreads out, exploring the environment as efficiently as possible.
To accurately model an objects appearance requires that sensory information be combined from different orientations. Our proposed coordination technique thus instructs team members to navigate around objects and gather as much information about them as possible. This information can be used to calculate the shape of the object and to model its appearance. For service robots to manipulate or move objects, the objects 3D shape must be known. To this end, our proposed coordination approach will facilitate the implementation of shape from silhouette. Objects shapes are determined by obtaining the silhouette or outline contour from different orientations, and fusing these to form a bounding hull. To classify a known object, our coordination approach will gather as much information about the object as is necessary to determine the objects identity with sufficient certainty. To reduce uncertainty, the views of an object that best discriminate between the objects possible matching identities are observed.
Declan O’Beirne, Michael Schukat
IRCSET Ph.D. Scholarship
Completion in 2012
Approximately one-third of all independently living elderly people are not compliant with their drug therapy. This lack of medication management results either in undermedication or overmedication, causing unnecessary and often serious health risks.
So far care schemes and retrospective assessment procedures have been the only applicable solution to tackle this problem, but with the increasing aging of the population and cost constraints in the health sector other more (cost-) effective methods must be introduced.
This project deals with a computer-assisted screening and management service that allows health care providers and drug-treated individuals to tackle this issue.
Michael Schukat, Tim Meagh, et al.