| Content-based publish-subscribe (CBPS) framework acts as the basic element in context-aware applications due to its potential to provide an efficient event distribution, better separation of concerns and extensibility. This thesis attempts to evaluate a proximity relationship algorithm presented in the system Fulcrum [BG05] in an actual publish-subscribe implementation called Ubibot [VEG09]. The Fulcrum makes CBPS subscriptions first-class by allowing the subscriptions to have computations associated with them and making those computations to be full-flown, including the ability to advertise their own events and subscribing to multiple events from multiple sources. The Ubibot is modified to support content-based routing and then implement the idea presented in Fulcrum. The solution is evaluated by using a buddy proximity scenario covering a wide range of behavioral possibilities, looking for anomalies. An overall performance of lg2(distance) is achieved from the algorithm and lg2(Original Events) of reduced event traffic. |
