Even more, reacting too late to context changes may worsen the user experience by showing information out of context selleckchem or performing unexpected actions for the user. Anticipation and improvement of user experience are some of the key features needed by Ambient Intelligence technologies in order to bring this intelligence to the environments and make them sensitive to users [1�C3]. To achieve these features we could add information about the most probable next context, so that those applications would have more time to make the necessary adjustments and be ready when the users’ context actually changes. This work is focused on a particular aspect of the context, the users’ location, thus aiming to offer services based on users’ future destinations.
More precisely, we are going to study some tools for estimating those future locations: the so-called location prediction algorithms.Location predictions may be an interesting improvement for ubiquitous computing applications, such as Location Based Services (LBSs). The prediction of users’ next location would allow to provide services related not only to their current location, but also to their future destinations. This way each user could be aware of information related a certain place (restaurant, museum) and decide whether to stop by that place or not right before getting there. The mobile phone itself may also be aware of users’ future location, thus being able interact with that location (e.g., an office or home) so it is prepared somehow when the user gets there (computer, lights or heat turned on).
Ambient Intelligence Drug_discovery applications can also exploit the movement patterns learned and predicted, for example for anomaly detection in elderly people care systems to determine if they get lost [4].Some authors http://www.selleckchem.com/products/mek162.html propose to calculate the predictions in the network [5,6]. However, in this case we are interested in learning and predicting using the mobile terminal itself because of several reasons, namely: (i) the advantages drawn from the fact that each user (terminal) learns and predicts her location, thus making the process distributed (with respect to the option of the network doing all the work); (ii) the improvement in privacy, since there is no need for sending location data through the network (the device obtains that information and process it); and (iii) the possibility of choosing the preferred technology for location tracking among the many ones integrated in mobile devices (GPS, WiFi or GSM/UMTS).However, making any kind of data processing in mobile devices is tied to a concern on the limited memory and processing speed. Therefore, the selection of the prediction algorithm to use needs to take into account such restriction.