Smart things: Conditional random field based solution for context awareness at the IoT edge

摘要

The realization of Smart Cities requires a huge number of connected devices, sensors and actuators as instances of Internet of Things (IoT) acting autonomously to collect data and provide different services to users. However, the environment in which these devices or sensors are deployed can largely impact the collected data and the provided services quality. We believe it is important that such devices are made conscious of the context of their surroundings, particularly readings from nearby sensors while collecting data. To do this, in this paper we propose the concept of “Context Awareness at the IoT Edge” by allowing an object to compute the spatio-temporal influence of the environment observed by nearby IoT devices, on its provided service. Specifically, we use Conditional Random Field (CRF) as a prediction model to analytically derive such an influence relation for a device. Our CRF-based model considers both the spatial and temporal impact of nearby sensors on a collected data by a specific device. We perform simulations and experimentation to evaluate the proposed CRF-based model. Results show that the proposed model not only estimates the context of a sensor with high accuracy (up to 98.5%) but also shows the effect of spatio-temporal variations in sensor readings.