A survey of machine learning in Wireless Sensor netoworks From networking and application perspectives

Abstract

Wireless Sensor Networks (WSNs) are used to collect data from and make inferences about the environments or objects that they are sensing. These sensors are usually characterized by limited communication capabilities due to energy and bandwidth constraints. As a result, WSNs have inspired resurgence in research on machine learning methodologies with the objective of overcoming the physical constraints of sensors. In this paper, machine learning methods that have been applied in WSNs to solve some networking and application problems are surveyed. Fundamental limits of learning algorithms will be addressed and future machine learning research direction are highlighted.