An adaptive policy-driven middleware for mobile computing

The recent trend in mobile computing has led to complex environments, and requirements which are changed dynamically. Adaptation is a solution for complex, and flexible mobile computing systems, it can not only dynamically adapt to a variety of changes inside of the physical layer, but also adapt to changes outside of their applications, and requirements. In this paper, we present an adaptive middleware system, which can dynamically relocate software components between computers to provide adaptation. On top of it, we also design a policy interpreter to execute user-defined adaptation to control the location of program at Runtime. This paper describes the two ideas with several scenarios, such as location measurement, and preference recommendation components in a sensor network. In the last, we demonstrate several sets of round-trip time through the proposed middleware system, and policies on Google Cloud Platform, which are verified in three instance groups at different zones.