No abstract

No abstract

Commercial buildings consume nearly 19% of delivered energy in the U.S, nearly half (42%) of which is consumed in buildings with digital control systems [23] comprised of wired sensor networks. These sensors have scant metadata, and are represented by "tags" which are obscure, buildingspecific and not machine parseable. We develop a human-inthe-loop synthesis technique which uses syntactic and datadriven steps to parse these sensor tags into a common namespace, which can enable portable building applications. We show that our technique allows an expert to fully parse a large fraction (70%) of the tags with 24, 15 and 43 examples for three large commercial buildings comprising 1586, 2522 and 1865 sensors respectively, and deploy three portable applications on two buildings with less than 30 examples.

Most large, commercial buildings contain thousands of sensors that are manually deployed and managed. These sensors are used by software and firmware processes to analyze and control building operations. Many such processes rely on sensor placement information in order to perform correctly. However, as buildings evolve and building subsystems grow and change, managing placement information becomes burdensome and error-prone. An automatic verification process is needed. We investigate empirical methods to automate spatial verification. We find that a spatial clustering algorithm is able to classify relative sensor locations -for 15 sensors, spread across five rooms in a buildingwith 93.3% accuracy, 13% better than a k-means clusteringbased baseline method. Analysis on the raw time series data has a classification accuracy of only 53%. By decomposing the signal into intrinsic modes and performing correlation analysis, an observable, statistical boundary emerges that corresponds to a physical one. These results may suggest that automatic verification of placement information is possible.

No abstract