TY - JOUR TI - Uncertainty-aware autonomic framework for resource management in mobile computing grids DO - https://doi.org/doi:10.7282/T3833QBX PY - 2014 AB - Mobile platforms are becoming the predominant medium of access to Internet services due to the tremendous increase in their computation and communication capabilities. Also, as more and more of these mobile devices are coupled with in-built as well as external sensors capable of monitoring ambient conditions, biomedical and kinematic information, and location, they can provide spatially distributed measurements regarding the environment in their proximity. Cumulus, a mobile computing grid, which harnesses the heterogeneous sensing and computing capabilities of mobile devices in the field as well as that of servers in remote datacenters is envisioned. Cumulus can be exploited to enable innovative mobile applications (defined by workflows) that rely on real-time in-situ processing of sensor data. However, enabling applications that require real-time in-the- field data collection and processing using mobile platforms is still challenging due to the following concerns: the inherent uncertainty associated with the quality and quantity of data from mobile sensors as well as with the availability of mobile computing resources in the field, security, and privacy. The goal of this research is to design and develop a uni ed uncertainty-aware (robust), secure, and privacy-preserving framework for data and computing resource management in the Cumulus in order to enable execution of mobile application workflows in real time and in situ and, hence, to generate actionable knowledge from raw data within realistic time bounds. In order to achieve the stated goal, an autonomic (self-organizing, self-optimizing, and self-healing) middleware that aids in the organization of the sensing, computing, and communication capabilities of static and mobile devices in order to form Cumuli is proposed. As the relevance of the output of workflows rely heavily on the quality and quantity of raw data coming from the underlying sensing infrastructure as well as on the computing resources available to execute them in real time, a uni ed data and computing resource management mechanism for data- as well as task-parallel applications is proposed. Finally, Maestro, a robust, secure, and privacy-preserving framework for concurrent mobile application management in Cumulus is proposed. The proposed framework is evaluated using experiments on a prototype testbed as well as through simulations on a purpose-built Java-based simulator. KW - Electrical and Computer Engineering KW - Mobile computing KW - Mobile apps KW - Computing platforms LA - eng ER -