When: December 16, 2011 at 10:30
Where: VUB campus Etterbeek, Building F, 10th floor, room 10F720
Duration: 45 min. presentation + 15 min. questions
Title: Machine Learning Techniques for Communication in Wireless Sensor Networks
Presenter: Dr. Anna Förster
Abstract: Communication protocols remain one of the main challenges in developing and deploying wireless sensor networks (WSN). Despite the overwhelming number of existing protocols, schemes and algorithms, there is still no consensus on a standardized solution. Adapting the existing protocols and algorithms to the various environmental changes and requirements remain a hard and manual task. Machine learning techniques offer a simple and elegant solution to this. Several algorithms and approaches from this family are able to easily recognize changing or changed environmental conditions and to adapt the behavior of communication protocols to them. In this talk I will quickly present some basic ML techniques and then discuss their usage in various existing WSN communication protocols, as well as sketch some future ideas and applications.
Short bio: Dr. Anna Förster has received her MSc. degree in Computer Science and Space Engineering from the Freie Universität Berlin and Technische Universität Berlin, Germany, with major in Artificial Intelligence Techniques for Mobile Robotics. She received her PhD in Informatics from the University of Lugano, Switzerland, where she focused on applications of Machine Learning to optimize routing and clustering techniques in WSNs. She is the author and co-author of several publications on ML applications in WSNs, e.g. the most complete survey of the usage of artificial intelligence for sensor networks (Kulkarni, R.V., Förster, A. and Venayagamoorthy, G. K.: Survey on Applications of Computational Intelligence for Wireless Sensor Networks, IEEE Communications Surveys and Tutorials, 2011). Currently she is a researcher in the Networking Laboratory at the University of Applied Sciences in Southern Switzerland. One of the major goals of her work in the area is to adapt existing ML techniques to meet the severe memory and processing restrictions of wireless sensor networks, as well as to enable user-friendly programming, testing and deployment of WSNs.
