Model Reference Adaptive Control in Fuzzy-Based Context-Aware Middleware

Owing to the dynamic characteristics of mobile environments, a mobile application needs to adapt to changing contexts to improve performance and resource utilization. We have developed an adaptive middleware infrastructure that simultaneously satisfies the individual needs of applications while maintaining overall system performance. A Model Reference Adaptive Control mechanism has been implemented in the middleware using control theory and fuzzy-based techniques. With reference to the model reference adaptive control theory, we also present a Self-Adaptive Fuzzy-based Service Adaptation Model (SA-FSAM) by taking historical adaptation information into account, and utilizing a closed-loop control mechanism to fine-tune adaptation decisions. The SAFSAM and a conventional threshold-based linear-control model have been evaluated using a campus assistant mobile application. With the introduction of self-adaptive elements in the control model, the SA-FSAM shows significant improvements in service adaptation decisions.