Towards an ideal service QoS in fuzzy logic-based adaptation planning middleware

• We propose an approach based on fuzzy logic for adaptation planning middleware.
• Addressed problem: the explosion in the number of rules that increases execution time.
• We separately calculate each QoS value of an ideal variant for a given context state.
• We define a fuzzy similarity method to select the most similar variant to the ideal.
• Experiments show our approach both reduces the number of rules and execution time.

Mobile applications require an adaptation phase to adapt to the user's and application context. Utility functions or rules are most often used to make the adaptation planning or decision, i.e. select the most adapted variant for each required service. Fuzzy controllers are used when it is difficult or even impossible to construct precise mathematical models. In the case of mobile applications, the large number of Quality of Service (QoS) and context parameters causes an exponential increase in the number of rules (aka. rule explosion problem), that increases the processing time of the adaptation planning. To reduce the processing time and simplify the fuzzy control system, we propose the concept of ideal QoS. Fuzzy values of ideal QoS parameters are calculated using several fuzzy control systems to fit the context state and user preferences. A fuzzy logic similarity metric based on fuzzy sets and fuzzy operators is proposed to select the service variant having the nearest QoS values to the ideal. Experiments show that our approach can significantly improve both the number of rules and the processing time when selecting the variant that well adapts to environment changes.