Coordinating military attacks, managing aerial drones, and monitoring terrorists online could all become easier with new software under trial by BAE Systems; a team of researchers from BAE and the universities of Southampton, Oxford, Bristol and Imperial College London spent five years developing a series of algorithms that allow different computer systems to co-ordinate their actions without a central authority
Coordinating military attacks, managing aerial drones, and monitoring terrorists online could all become easier with new software under trial by BAE Systems.
The family of computer technologies was originally created as part of the U.K. £5.5 million ALADDIN project, which was designed to improve the response of emergency services during a disaster.
Stephen Harris writes that a team of researchers from BAE and the universities of Southampton, Oxford, Bristol and Imperial College London spent five years developing a series of algorithms that allow different computer systems to co-ordinate their actions without a central authority.
The project, the title of which stands for Autonomous Learning Agents for Decentralized Data and Information Networks, topped the aerospace and defense category at last year’s The Engineer Technology & Innovation Awards.
BAE Systems is now simulating versions of the ALADDIN algorithms applied to ten operational situations that use computers, including working out the logistics of a supply line and detecting terrorist threats by monitoring internet social networks.
“What we’re trying to do is put intelligence into our systems,” said BAE Systems’ managing director of strategic capability solutions, Simon Jewell. “We are now in the position to pull through the learning and new technology [of ALADDIN], and integrate the capability into current and future projects.”
Central to the project is the idea of “agents,” pieces of software such as the command module of a ship’s computer, the control system of an unmanned aerial vehicle (UAV), or an analysis program on the lookout for unusual online activity.
The algorithms allow the agents involved in an operation to exchange information with one another and build a bigger picture of the situation than they would have on their own.
Then the agents effectively negotiate with each other to decide the best course of action for the whole team, rather than just themselves. This approach could be particularly useful during a cyber attack when communication from a central authority is knocked out.
Most of the applications involve integrating ALADDIN into existing software, but in some cases the project has led to the creation of new programs, such as environmental modeling.
BAE Systems hopes to deploy the first ALADDIN-supported systems within the next couple of years.
“ALADDIN provides a toolbox of techniques, both for the design of single agents and for how those agents will be designed to work together to drive the overall system performance,” said Dr. Robert Johnston, head of technology for BAE Systems Integrated System Technology.
The ten applications are at different stages of testing but the most advanced is known as Force Threat Evaluation and Weapon Assignment (F-TEWA)
Harris notes that the ALADDIN algorithms have been inserted into BAE Systems’ combat management computer system to help co-ordinate military equipment and resources on a battlefield as the situation changes.
Another program plans and monitors the supply chain for forces on the ground, predicting what resources will be needed and reducing the distribution of supplies that are not used.
ALADDIN is also being used to analyze online social networks to monitor terrorist groups and other threats. By studying communications, the software can help identify sub-groups within an organization, as well as anomalous behavior.
The military is increasingly looking for decentralized approaches to its operations, despite its command structure, said Dr Simon Case, BAE Systems’ capability technology leader.
However, many of the applications being studied still involve human input, with computers making suggestions rather than operational decisions. The next stage of the program, known as Project ORCHID, will look at improving the interaction between humans and agents.
The RoboCupRescue disaster simulation platform version 1.0 was developed as part of the ALADDIN project.Learn more about RobocupRescue v1.0 »
Gaussian Processes have been developed to predict sensor readings on the Bramblet sensor network on the south coast of England.Learn more about LiveSensors! »