Pervasive open spaces: an intelligent and scalable pervasive environment for providing contextual resource sharing
Scalability imposes itself as a great setback for pervasive computing research. We present a novel approach called Open Spaces that provides users characterized with various mobility patterns with both scalable and intelligent resource allocations based on user context.Resource sharing typically includes memory, processing, and secondary storage. To provide such resource sharing in terms of context, we discuss in this work the idea of physical or logical structures called domes that form the Open Spaces environment and encapsulate both user resources and context information. We present resource sharing as an application to illustrate how computing resources can be allocated inside domes, and how user mobility patterns affect the re-allocation of resources inside the domes themselves. We present a way by which computing resources can be dynamically allocated and shared between users within the environment in a transparent and efficient manner. We use secondary storage such as main memory as our primary resource sharing criteria due to its speed advantage. Its primary usage is for holding application data loaded into memory per user device. This in turn would allow providing a shared memory model that can also be reused among the sharing users in the system. We then discuss how we can predict future resource acquisitions by learning the user navigational patterns inside Open Spaces. Our results show that learning user resource sharing patterns within Open Spaces creates a better prediction model than conventional resource sharing systems.