Visible to the public Community Sense and Respond: Theory and Applications


This talk describes preliminary results on implementing Community CPS: CPS in which members of different communities - ordinary residents, power and water utilities, and government agencies - participate by hosting sensors, actuators, and computation engines. CCPS deal with extremes of heterogeneity, uncertainty, distribution, poor deployment, unreliable operation, and high load variability. This talk describes a CCPS system that senses and responds to earthquakes. The talk describes a system deployed in Southern California, measurements, and lessons learned about the theory and implementation of CCPS. The talk describes CPS at the intersection of CS, earth sciences and civil engineering. Preliminary work applying the CCPS framework to sensing and responding to air pollution is discussed.

The talk describes a CCPS architecture in which inexpensive sensors are connected to the Cloud which detects critical events, determines appropriate responses, and then sends control signals to responders. In the current implementation, the sensors are cheap accelerometers connected to the Google App Engine and the responders are phones. Algorithms for detecting earthquakes, and lessons learned with using distributed computing in the Cloud for fusing data from large numbers of sensors is described. Results in detecting very small earthquakes (magnitude 2+) including earthquakes that are barely perceptible are described. The extension of these algorithms to other CCPS applications is discussed.

Data streams from sensors and other sources provided by members of the community are highly heterogeneous, very noisy, and their behavior varies significantly over time. The talk describes algorithms to learn changing behavior of data sources, and to adapt data fusion methods to give appropriate weights to data sources with varying quality. Preliminary results on the vibration modes of a 9-storey building on campus, using the CCPS system, are presented. This analysis was conducted by civil engineers on the project. Initial deployments of the system to other campuses (USC and UC Irvine in Southern California, and IIT-Gandhinagar in India) are described.

The talk describes preliminary work on generalizing results to other CCPS applications using inexpensive personal hazard stations. A prototype version of the personal station has been developed, and the system uses the same Cloud framework as the Community Seismic Network. Lessons learned in developing a general framework for CCPS are described.

Award ID: 0932392

Creative Commons 2.5

Other available formats:

Community Sense and Respond: Theory and Applications