Search

Smart homes form an integral part of smart grid infrastructure. Recently, large numbers of sensors have been added within smart homes to enhance home appliance automation and monitoring. This addition raises questions about where the data generated within the home should be processed. The data can be processed either by one central processor or through multiple distributed processors closer to the sensors. This paper proposes a smart home distributed architecture involving home sensors talking directly to a smart gateway installed within the home. The gateway then decides which data should be forwarded to the central processor for further analysis. A test bed is designed to highlight the advantages of this approach. An open data set is used to feed sensor data into the test setup. It is shown that the local processing of data can improve efficiency by effectively utilizing available network bandwidth. Furthermore, local processing is favorable for time-critical smart home applications, since local processing has a faster data communication round trip time as compared with that of central processing. Moreover, we argue that certain calculations, like energy usage prediction for home appliances, can effectively be done locally while the central processor can be used for coordination between different local processors.