Policy Driven Adaptation of Context-aware Services with preferences supporting Mohcine Madkour Ecole Mohammadia des Ingenieurs University Mohamed V Rabat, Morocco mouhcine.madkour@gmail.com Abdelilah Maach Ecole Mohammadia des Ingenieurs University Mohamed V Rabat, Morocco, maach@emi.ac.ma Driss El Ghanami Ecole Mohammadia des Ingenieurs University Mohamed V Rabat, Morocco, elghanami@emi.ac.ma Abstract—Ubiquitous computing has revolutionized the way we use computing. The widespread of wearable devices and the seamless connectivity between them have transcended the traditional computing era to the pervasive computing and enable new opportunities for a user to perform his/her operation all the time and everywhere. These new environments are characterized by large quantity of heterogeneous services and context-aware applications that dynamically join and leave the network. Such environment will constitute a large-scale internetworking infrastructure and likely to provide a new level of openness and dynamics. For achieving its goal, ubiquitous computing needs the context data brought by the seamlessly connected devices either mobile handset or embedded in the surrounding physical environment and imperceptible to a user. Indeed, context and context-awareness provide computing spaces with the capability to usefully adapt the services they provide. We present our mechanism of adaptation of services by a policy-driven, context- aware manner. The adaptation is based on enumerating the different policies of a service object. As the execution environment, user context and preferences change, the service object will be adapted to use different behaviors, driven by user preferences. The adaptation process is able to capture and to process context-dependent preferences and qualitative uncertainty labels which are used to determine which set of preferences should be considered in a given context. To demonstrate this mechanism we will provide an example scenario and evaluate our solution. Keywords—service adaptation; user preferences; context- awareness; pervasive computing; fitness function I. INTRODUCTION Policies can be used for representation of all types of adaptation and monitoring activities in pervasive computing. The term 'policy' is used in different ways in the literature. A general definition is that a policy is a declarative, high-level description of goals to be achieved and actions to be taken in different contexts and situations. Context-awareness mechanisms and, in particular, user- related information awareness, are one of those required extensions for the Ubiquitous Computing to fulfill present and future services demands. Probably, an important part of the context information for a service is related to the user preferences. We can define user context preferences as the subset of the context information influencing a service that model user-related aspects. Preferences models have an utmost importance for service adaptation, either reported in their profiles or inferred from their actual behavior, preferences are the underlying criteria and the most intuitive to make the service adaptable and useful. On the other hand QoS properties in service description can be specified as QoS parameters and QoS policies. QoS parameters are QoS attributes that can be expressed in quantifiable measurements or metrics. A service usually possesses a set of QoS parameters, though many of them are of dynamic nature, i.e., related to the service execution environment, a service can still advertise its assumed QoS. QoS policies are rules related to QoS parameters. A service can provide different QoS classes of service depending on their providing guarantees a set of QoS parameters. For example in systems which depend on service classes users subscribed, a Gold Class user may have access to Gold Class Service, which guarantees a set of QoS parameters, such as bandwidth and response time, much better than a user in Silver Class. QoS policy of services compose both functional and non-functional properties. Functional properties can be measured in terms of throughput, latency, response time, whereas non-functional properties address various issues such as integrity, reliability, availability and security of web services [1]. For ensuring an acceptable QoS and allowing for adaptation to user preferences, we define the policy-based adaptation of context-aware service as the automatic selection of the best policy for delivering the service. The adaptation process use fuzzy sets represented with linguistic variables and membership degrees to define the user’s preferences over a set of context attributes. Afterword the matching between preferences and policies intrinsic‟ attributes should be done. The remainder of this paper is outlined as follow: In the second section we present our context-aware service adaptation scheme introduced by a motivating use case scenario, the third section introduces tools, definitions and formulas for our fuzzy- based context aware service adaptation, thereafter section four presents the distance-based fuzzy adaptation process algorithm. Section five explains use case example application and makes some discussions and the penultimate section presents some background and related works. Last section is the conclusion. 978-1-4799-0792-2/13/$31.00 ©2013 IEEE