A Context-Aware Service Selection Mechanism Based on Hidden Markov Model

It is an important yet challenging problem that how to provide personalized services in pervasive environments. This paper explores the relationship between context and service requirements of customers, and proposes a service selection mechanism based on a modified Hidden Markov Model. Based on the contextual information, we build a novel Hidden Markov Model, and adjust its parameters dynamically by continuously changing context and invoked services. For a user, the service calling probabilities at current state can be calculated based on this model, which realizes personalized service selection. Meanwhile the user´s next state can also be predicted, which realizes selecting in advance by roulette wheel algorithm. The prototype was tested in a campus environment, which verified the validity of the proposed method.