High-performance computing in mobile services

With the ever increasing popularity of smart phones, mobile services have been evolving rapidly to allow users to enjoy localized and personalized experiences. Users can discover local information and keep connected with family and friends on the go, and ultimately to experience the convergence of cyber space and physical world where digital technologies are interwoven into the day-to-day life. A pivotal component of such a cyber-physical convergence is the contextual intelligence. The extraction and dissemination of contextual information around users is the key for the cyber capabilities to be applied to physical activities and for the cyber world to better reflect the physical reality. In this talk, we shall address some issues arising from context-based mobile services. In particular, we discuss how mobility impacts contextual relevancy and personalization in mobile services. The relevancy and timeliness of contextual information not only are essential for these services to deliver great user experiences, but also put significant computation pressure on service infrastructure that processes continuous data streams in real time and disseminate relevant data to a large amount of mobile users. This talk will explore the challenges and opportunities for high-performance computing in mobile services. Based on key findings from large-scale mobile measurement data, the talk will analyze the tradeoff of different computing architectures, present case studies of scalable system design and implementation for personalized mobile services, and conclude with open challenges for the broad research community in performance measurement and modeling.