A Fog Computing Approach for Localization in WSN

The Fog Computing paradigm proposes an extension of the cloud-based computing to the network edges in the Internet of Things. It facilitates localized analysis closer to the data sources for improved responsiveness of the system as well as cloud-based learning for historical analysis. In this paper, we present our fogenabled Wireless Sensor Network (WSN) system for activity monitoring and localization in the context of Ambient Assisted Living. Our WSN architecture consists of two types of devices - a wearable sensor device and a cloud gateway node. We discuss our Edge Mining approach for real-time activity classification on the sensor device as well as the Genetic Algorithm used for cloud-based analysis. The design of our analytical framework together with the communication model addresses the challenge of sensor-cloud integration. We evaluate the performance of our system for outdoor localization of the elderly. The analysis is based on acceleration data collected using our wearable device across different activity sequences obtained from the Kasteren dataset.