IoT, or Internet of Things, is booming in recent years, where smart devices communicate with each other through M2M (Machine to Machine) protocols. These protocols typically adopt a hierarchical data format to specify the meta-data, such as devices, sensors, categories, and other properties, together with the sensory data; consequently, queries on the sensory data are in the form of path expressions, possibly with wild cards. Unfortunately, current database systems cannot support these path queries on stored IoT data efficiently.

In this thesis, we propose a Composite Bitmap (CBM) method to index the stored IoT data. Specifically, we allocate a fixed number of bits to each level of the IoT data using the domain knowledge about the level, and combine the bits for all indexed...[ Read more ]

IoT, or Internet of Things, is booming in recent years, where smart devices communicate with each other through M2M (Machine to Machine) protocols. These protocols typically adopt a hierarchical data format to specify the meta-data, such as devices, sensors, categories, and other properties, together with the sensory data; consequently, queries on the sensory data are in the form of path expressions, possibly with wild cards. Unfortunately, current database systems cannot support these path queries on stored IoT data efficiently.

In this thesis, we propose a Composite Bitmap (CBM) method to index the stored IoT data. Specifically, we allocate a fixed number of bits to each level of the IoT data using the domain knowledge about the level, and combine the bits for all indexed levels into a 64-bit integer. With the stored data indexed in CBM, processing an incoming query becomes efficient bit-wise operations on the index. We have implemented our method and compared it with a NoSQL database based implementation under various workloads. Our results show that CBM outperforms the database based implementation by one to two orders of magnitude.