2014/06/15 by Jaziar Radianti, Julie Dugdale, Radianti, Jaziar +5
Computer Science · Social Sciences · Engineering · #Context-Aware Activity Recognition Systems #Human Mobility and Location-Based Analysis #Evacuation and Crowd Dynamics
paper · pdf · doi:10.48550/arxiv.1406.3848
The increasingly sophisticated sensors supported by modern smartphones open up novel research opportunities, such as mobile phone sensing. One of the most challenging of these research areas is context-aware and activity recognition. The SmartRescue project takes advantage of smartphone sensing, processing and communication capabilities to monitor hazards and track people in a disaster. The goal is to help crisis managers and members of the public in early hazard detection, prediction, and in devising risk-minimizing evacuation plans when disaster strikes. In this paper we suggest a novel smartphone-based communication framework. It uses specific machine learning techniques that intelligently process sensor readings into useful information for the crisis responders. Core to the framework is a content-based publish-subscribe mechanism that allows flexible sharing of sensor data and computation results. We also evaluate a preliminary implementation of the platform, involving a smartphone app that reads and shares mobile phone sensor data for activity recognition.