How Wireless Sensor Networks Can Benefit from Brain Emotional Learning Based Intelligent Controller (BELBIC)

Wireless sensor networks (WSNs) are composed of small sensing and actuating devices that collaboratively monitor a phenomena, process and reason about sensor measurements, and provide adequate feedback or take actions. One of WSNs tasks is event detection, in which occurrence of events of interest is detected in situ whenever and wherever they occur. Some examples of these events include environmental (e.g. fire), personal (e.g. activities), and data-related (e.g. outlier) events. Simply speaking, event detection is a classification process, in which membership of data measurements to each event class is determined. Neural network is one of the classifiers that have often been used for detecting events with known patterns. One of the techniques to maximise the neural network performance during classification process is enabling a learning process. Through this learning process, neural network can learn from errors generated in each round of classification to gradually improve its performance. In this paper we investigate applicability of Brain Emotional Based Intelligent Controller (BELBIC) to improve neural network performance. Empirical results show that incorporating the BELBIC with neural networks improves the accuracy of event detection in many circumstances.