Presentation Title
An Analysis of Real Time Streaming of Sensor Fusion in Smart Buildings
Degree Name
Master of Computer Science and Systems (MCSS)
Department
Institute of Technology
Location
UW Tacoma - William W. Philip Hall, Milgard Assembly Room
Event Website
http://guides.lib.uw.edu/tactalks
Start Date
18-5-2017 6:05 PM
End Date
18-5-2017 6:10 PM
Abstract
As buildings and devices becoming increasingly “smarter,” the need for more data requires more sensors in order to provide more precise and insightful interpretations of daily applications. As a result, servers are required to become stronger and more advanced in order to compute and process the abundance of data arriving at high frequencies and to meet real-time requirements and expectations. The naive approach would be to buy new, or to upgrade, existing servers. This can potentially become costly and inconvenient for end users. In addition, placement of new equipment is constrained by building blueprints that must meet safety compliance, and in order to be efficient, the equipment must be placed in the vicinity of the sensors. However, the need for additional or upgraded servers can be prevented if you can effectively direct the incoming data and reroute it into proper queues before computation.
My research compares the computation of incoming data across multiple software topology designs. The system will work under various workload intensities depending on the way data is routed. However, the trade-off of changing the way the data is routed could potentially mean a loss of accuracy. While keeping accuracy of computations a high priority, one component of my research will be implementing a system that can compute occupancy predictions on environmental sensor data that is compared across multiple software topology designs, to determine under which conditions we can reduce server workload intensity while still preserving accuracy.
COinS
An Analysis of Real Time Streaming of Sensor Fusion in Smart Buildings
UW Tacoma - William W. Philip Hall, Milgard Assembly Room
As buildings and devices becoming increasingly “smarter,” the need for more data requires more sensors in order to provide more precise and insightful interpretations of daily applications. As a result, servers are required to become stronger and more advanced in order to compute and process the abundance of data arriving at high frequencies and to meet real-time requirements and expectations. The naive approach would be to buy new, or to upgrade, existing servers. This can potentially become costly and inconvenient for end users. In addition, placement of new equipment is constrained by building blueprints that must meet safety compliance, and in order to be efficient, the equipment must be placed in the vicinity of the sensors. However, the need for additional or upgraded servers can be prevented if you can effectively direct the incoming data and reroute it into proper queues before computation.
My research compares the computation of incoming data across multiple software topology designs. The system will work under various workload intensities depending on the way data is routed. However, the trade-off of changing the way the data is routed could potentially mean a loss of accuracy. While keeping accuracy of computations a high priority, one component of my research will be implementing a system that can compute occupancy predictions on environmental sensor data that is compared across multiple software topology designs, to determine under which conditions we can reduce server workload intensity while still preserving accuracy.
https://digitalcommons.tacoma.uw.edu/tactalks/2017/spring/4