Abstract: Context awareness is a term that refers to knowledge of where, when and what the user does. The study implements all the above forms of context awareness, using only RFIDs in contrast to various expensive technologies like video/audio sensing, pressure pads etc. The paper implements the next room prediction of users based on Fuzzy Timed K-Markov model. Various values of K have been used and the optimal value is suggested. The study also models the uncertainty of user behavior recognition using Bayesian Belief Networks (BBNs) and predicts the user`s action using negative reinforcement learning technique. Unlike in previous models where the implementations invariably had the users wear the tags, or the Bracelet readers, but, we use a combination of both as it seemed to help us derive hierarchical contextual information, that is, the context changes that accompany people interacting with each other. The model is also unobtrusive during the learning period. The study in all, aims to realize a context aware Ambient Intelligent system in Smart Home though RFIDs.
A.S. Meenakshi Sundaram , M. Mahesh Babu , C. Manikandan , V. Rhymend Uthariaraj and R. Shriram , 2007. Improving Prediction Accuracy in Context Aware Smart Homes Using K-Markov Model . International Journal of Soft Computing, 2: 273-278.