The past two decades of explosive growth in wireless networking, mobile computing and web technologies has profoundly influenced society at large. Almost anyone with access to a mobile device has access to services on the Internet and has reaped the benefits of instant accessibility to Internet-enabled technologies such as social networks, media streaming applications, location-based services, instant messaging, etc.;In this thesis we aim to synergistically exploit mobile and cloud computing to enable services that can enrich the experience and capabilities of mobile users in a pervasive environment. While mobile computing empowers users with anywhere, anytime access to the Internet, cloud computing harnesses the vast storage, computing, and software infrastructure resources of large organizations into a single virtualized infrastructure within reach of the general population. We argue that a tiered approach that synergistically exploits local and public clouds to achieve application QoS and scalability is a well suited architecture for the mobile cloud computing paradigm.;In this thesis, we studied the problem of optimal and fair service allocation for a variety of mobile applications (single or group/collaborative mobile applications) in a mobile cloud computing paradigm. Specifically, we concentrate on three main issues: (i). Modeling of the MCC systems and formulation of the MCC service allocation problem, (ii) Service and resource provisioning algorithms, (iii) System performance testing.;The first section of this dissertation develops a novel framework to model mobile applications as a location-time workflow (LTW) of tasks; here user mobility pattern are translated and mapped to mobile service usage patterns. We show that an optimal mapping of LTWs to 2-tiered cloud resources considering multiple QoS goals such application delay, device power consumption and user cost/price is an NP-hard problem for both single and group-based applications. Next, we designed a range of heuristics and approximations, in particular based on techniques such as greedy, simulated annealing and genetic algorithms to solve the formulated optimization problems. We considered the optimality of the heuristic approaches (as compared with an optimal solution) using running time and scalability as performance metrics. We also developed a MapReduce-based algorithmic model using Pig Latin to address scalable resource provisioning when the search space for optimization is large. We developed a prototype middleware platform, MAPCloud to orchestrate the components of a 2-tiered mobile cloud computing system. MAPCloud was evaluated by implementing a range of mobile applications that span compute, storage and bandwidth intensive applications. A detailed simulation study using measurements and trace data obtained from application profiling was used to further assess system performance at scale.
展开▼
