Distributed Machine Learning with Self-Organizing Mobile Agents for Earthquake Monitoring

Ubiquitous computing and The Internet-of-Things (IoT) raises rapidly in today's life and is becoming part of self-organizing systems (SoS). A unified and scalable information processing and communication methodology using mobile agents is presented to merge the IoT with Mobile and Cloud environments seamless. A portable and scalable Agent Processing Platform (APP) is an enabling technology that is central for the deployment of Multi-agent Systems (MAS) in strong heterogeneous networks including the Internet. A large-scale distributed heterogeneous seismic sensor and geodetic network used for earthquake analysis is one example, which can be extended by ubiquitous sensing devices like smart phones. To simplify the development and deployment of MAS in the Internet domain agents are directly implemented in JavaScript (JS). The proposed JS Agent Machine (JAM) is an enabling technology. It is capable to execute AgentJS agents in a sandbox environment with full run-time protection, low-resource requirements, and Machine Learning as a service. A simulation of a seismic network and real earthquake data demonstrates the deployment of the JAM platform. Different (mobile) agents perform sensor sensing, aggregation, local learning and prediction, global voting, and the application.