Pattern discovery from dynamic data streams using frequent pattern mining with multi-support thresholds

Discover the frequent patterns from streams of data is challenging due to the characteristics of such streams: a continuous unbounded high speed data of evolving nature which must be processed on the fly with bounded computing memory and limited storage capacities. On other hand, the growth of the Internet of Things (IoT) has helped to sense everything around us at home, work and in the streets. The IoT is a network of physical devices, vehicles and other items which are embedded with sensors, software and network connectivity that enables these objects to collect and exchange data. Data from a large numbers of sensors, deployed in infrastructure (such as roads and buildings) or to report on environmental conditions, can give decision makers a heightened awareness of real-time events. In this work, we develop methods to exploit real-time streaming data that flows from IoT devices. In particular, we aim to discover the interesting frequent patterns from these heterogeneous streams. Therefore we will classify the items in the joint stream over titled time windows to reduce the number of emerging patterns, then we will mine the frequent pattern over the classified items using multi-support threshold. We conduct the experiment on a car parking lots environment with three simulated streams from sensors, smart payment machines and a mobile application.