Optimizing upload time of data from mobile devices

The technological advances in mobile devices have resulted in increasing numbers of applications on them, and users use these applications to create lots of data. Due to the truly ubiquitous nature of these devices, the data is created potentially at all times of the day and in different wireless environments, e. g., at home, on the subway, at a coffeeshop. These data must eventually be uploaded to the server cloud from the mobile device by connecting to a wireless network. In different environments, the available wireless networks change, and the cost of connecting to them in order to upload these data also changes. For different applications, the preference for the uploading time for their data may be very different. We study the problem of optimizing the time for uploading based on the preferenc measure as well as the cost of the wireless network available at the time of uploading. Furthermore, because the wireless environment is dynamic and nondeterministic in the future, we use machine learning techniques for modeling user mobility patterns and predicting the wireless network in the near future, to assist in finding the optimal uploading time.