Our work on shadowing smart phone for disaster management, entitled “Shadow Phone: Context Aware Device Replication for Disaster Management”, was accepted to ACM RACS 2013. Below is the abstract of the paper.
Mobile devices are easy to carry but have limited resources. Due to its mobility and proliferation, they have been used in many disaster management systems to deliver disaster alerts and to assist disaster rescue. In this work, we design and implement a context-aware synchronization framework to assist people to better utilize mobile devices for disaster management. Today, disaster warning systems provide general information of disaster events and broadcast to all people. This simple approach adopts one-way communication model in which information is always transferred in a single direction from server-side to client-side. However, for better support of disaster warning and rescue, the system is supposed to leverage informative content by producing user-context dependent information. That means the two-way communication model should be deployed, server-side learns user-context and returns user-dependent feedbacks to client-side. The feedbacks are not only warning info, but also the instructions to guide users how to behave in the emergency cases. To achieve this idea, it requires lots of computing power to customize information to meet demands of large-numbers of users. Therefore, cloud computing is the key technology that contributes to the idea implementation. In this work, the proposed framework provides two important features: (1) context-aware transmission policy which controls how to transmit data efficiently according to data priority, battery level and network data rate; (2) an architecture that supports the model of mobile interactive applications in which clients and servers can interact with each other in an autonomic manner. Compared to Dropbox-like transmission scheme, the context-aware transmission protocol can prolong the battery lifetime for up to seven times and increase the amount of transmission for up to seven times. Moreover, this scheme is not limited to single type of disaster scenario.