Mobile Spatial Subscriptions for Location-Aware Services

Abstract

Spatial subscriptions have been used to specify locations of interest in Distributed Event-based Systems (DEBSs). However, current DEBSs representations to support spatial subscriptions are not expressive enough to describe some forms of subscriptions in mobile settings. For instance, users are not allowed to specify a spatial subscription that refers to other more well-known locations, in case they are not familiar with the names of their current locations. In addition, the middleware in existing DEBSs does not support changes at runtime, and modification to these middleware systems to support spatial subscriptions are highly coupled with specific DEBS infrastructures. In this thesis, I argue that by enhancing the expressiveness of spatial subscriptions, a new model of mobile spatial subscriptions for location-aware services can be defined and a reusable plug-in implementation approach that supports existing DEBSs can be developed. This thesis first summarizes the essential abstractions to specify mobile spatial subscriptions, and analyze the expressiveness of existing DEBSs to support these abstractions. Second, it proposes a three-level mobile spatial subscription model, which supports the essential abstractions used to specify spatial subscriptions. The first level of the model handles subscriptions consisting of geometric coordinates; the second level supports subscriptions with location labels; the third level interprets subscriptions which specify locations by stating their dynamic properties. Next, a plug-in implementation approach is introduced, and hence, the three-level model can be integrated with different DEBSs with minimal modification to the middleware. The subscription model is implemented as a subscriber/publisher component, instead of directly modifying the existing DEBS. Finally, I develop a prototype system, Dynamic Mobile Subscription System (DMSS), and illustrate the usefulness and applicability of the three-level model and the plug-in implementation approach.