Motivation:
A current deficiency of sensor networks today is the requirement to have an in-depth understanding of programming embedded systems in order to utilize them effectively. This makes it difficult for professions outside of computer science to be able to use a sensor network for their purposes. The result is a group of computer scientists working in tandem with the other group, to empower the other profession to achieve its objective.
A specific field that can contribute greatly to sensor networks is that of statisticians. Sensors networks are unreliable, and using probabilistic and statistical methods are very natural ways to handle these problems. Many research projects are beginning to use statistical methods to reason about sensor networks, for instance BBQ.
Goal:
The purpose of this project is to come up with an API that allows statisticians to more easily adapt their algorithms to sensor networks. The statisticians should not be worrying about things like resource allocation or use of the MAC layer. The systems people should not need to worry about the theoretical algorithms. However, there is a tension between giving the statistician sufficient flexibility to describe an algorithm, and yet not make them deal with typical systems issues. We are looking for a common thread of statistics problems that apply to sensor networks.
Ideally, the statisticians would then be able to develop their algorithms with much less needed understanding of what is going on “under the hood”. Even if this is not the case, it should make the responsibilities of systems programmers interacting with statistical algorithms clearer, and thus make the mutual creation of these systems more efficient, with less required knowledge-transfer between the two groups.
Members:
Jeremy Schiff, Dominic Antonelli, David Chu, and Georgios Alexandros Dimakis,
Related Material:
Class Project - Sensor Networks (CS294) & Statistical Learning Theory: Graphical Models (CS 281A) - StatSense Scheduling and Robustness Poster [258KB .ppt]
Class Project - Sensor Networks (CS294) & Statistical Learning Theory: Graphical Models (CS 281A) - StatSense Scheduling and Robustness Paper [831KB .pdf]
Class Project - Advanced Topics in Computer Systems (CS 262B) & Statistical Learning Theory (CS 281B) - StatSense Localization Poster [227KB .ppt]
Class Project - Advanced Topics in Computer Systems (CS 262B) & Statistical Learning Theory (CS 281B) - StatSense Localization Paper [872KB .pdf]