A mobile middleware for quality of service adaptations

Abstract

Computation and networking resources in mobile operating environments are much scarcer and more dynamic than in desktop operating environments. Mobile applications can use adaptive computing to optimize the quality-of-service (QoS) delivery based on dynamic contextual situations. Fuzzy control models have been successfully applied to various distributed network QoS management systems. However, existing models are either application-specific or limited to abstract modeling and simple conceptual scenarios that do not take into account the overall scalability of these models. Specifically, the large number of QoS parameters in a mobile operating environment causes the rule-explosion problem, in which an exponential increase in the number of rules correspondingly increases the demand for processing power to infer the rules. Hierarchical fuzzy systems were introduced to reduce the number of rules using hierarchical fuzzy control, in which correlated linguistic variables are hierarchically inferred and grouped into abstract linguistic variables. In this thesis, we propose a mobile QoS management framework that uses a hierarchical fuzzy control model to support a highly extensible and structured adaptation paradigm. The proposed framework integrates several levels of QoS abstractions derived from user-perceived requirements. It also maps these abstractions to appropriate QoS resources that drive the development of mobile services that mitigate the effects of varying mobile environments. This framework provides an optimal overall service by synergistically balancing the QoS requirements of users and applications with the dynamic allocation of resources and chaining of services. Our proposal is novel in that it looks at QoS from a holistic, systematic, and pragmatic perspective. This thesis demonstrates the flexibility and efficiency of our QoS management framework in adapting to different users, applications, and platforms operating in wireless environments that are characterized by dynamic and constrained resources.