A low-load stream processing scheme for IoT environments
Tomoki Yoshihisa, Takahiro Hara · 2016
Recently, IoT (Internet of Things) environments have begun to attract a great deal of attention. In IoT environments, various “things”, such as sensors, are connected to the Internet and act as distributed data sources to continuously generate big data (stream data). One of the main research issues for such stream data is fast execution of continuous queries. Most conventional schemes assume that processing servers receive data streams from remote data sources. Therefore, when large numbers of data sources are distributed in IoT environments, large amounts of communications traffic from those data sources are produced. This is the main cause of long stream processing times. To tackle this key problem, we take this feature of IoT environments into account and try to reduce the stream processing time by reducing the stream processing load. Our proposed scheme extracts conditional expressions, and determines the order of evaluation of these expressions, in order to reduce processing and communication loads. Our evaluation shows that the proposed scheme can reduce the maximum number of communication hops and the average amount of communication traffic in IoT environments.