| This thesis studies the monitoring of continuous multi-dimensional queries. A continuous query runs over long periods of time and requests reporting of its result as the data dynamically change. Initially, we focus on nearest neighbor (NN) monitoring, a problem that typically arises in location-based services. In this context, the query points and the data objects move arbitrarily, and the task of a central server is to report the k closest objects (i.e., the k NNs) to each query point. First, we propose a method that aims at minimizing the communication cost spent in message exchanges between the data objects and the server. Our second contribution is an algorithm that minimizes the processing cost at the server, assuming that the distance between objects and queries is defined according to the Euclidean metric. Next, we present efficient NN monitoring techniques targeted to applications where the objects and the queries move in a road network, and their distance is defined as the length of the shortest path connecting them in the network. Finally, we consider a class of non-spatial queries and show that geometric reasoning can be used for their efficient monitoring. In particular, we study continuous top-k processing over sliding windows. In this setting, a stream of multi-dimensional tuples arrives at a central server. Among these tuples, only the most recent ones are considered valid. Given a function that assigns a score (i.e., a real number) to each tuple, the task of the server is to continuously report the k valid tuples with the highest scores. |
