A Performance Evaluation of Spatial Indices for Geospatial Publish/Subscribe Systems

In a publish/subscribe system, publishers publish information content which they want to distribute among subscribers, who, on the other hand, need to subscribe to publications of interest by activating appropriate subscriptions. Traditional content-based publish/subscribe systems define a publication as a set of properties (i.e., name-value pairs), while a subscription is defined as a set of various constraints on publication properties. Therefore, they basically define a publication and subscription as a point and hyper-rectangle in a multidimensional attribute space, respectively. In this paper we focus on a geospatial publish/subscribe system in which each publication and subscription content is extended with a geospatial object (e.g., a point or polygon). Such an object is used to express the location of a published event or subscription interest. To efficiently match incoming publications in real-time with a set of stored subscriptions, it is necessary to index subscriptions in an appropriate data structure. Spatial indices are typically used in literature for indexing of geospatial objects. In this paper we experimentally evaluate the performance of several spatial indices when used for indexing of geospatial subscriptions in order to efficiently match incoming publications with stored subscriptions.