A new Learning Automata based approach for online tracking of event patterns

Detecting spatiotemporal pattern from noisy sequences of events plays a very important role in presence sharing, Internet of Things (IoT) and many other fields. As pointed out in existing literature, the core activities of these applications involve event notifications. However, excessive number of event notifications will lead to user׳s intolerability. Existing literature proposed a Spatiotemporal Pattern Learning Automata (STPLA) to solve this problem effectively in both stationary and non-stationary environments. However, one limitation of the STPLA is that it cannot be both memory balanced and bias toward any of the two actions, i.e., “suppress” or “notify”. To solve this problem, this paper proposed a new Learning Automata based approach, named as Spatiotemporal Tunable Fixed Structured Learning Automata (STP-TFSLA), for online tracking of event pattern. Furthermore, we also show that the STP-TFSLA is with small memory footprint and is able to cope with non-stationary environment.