Design and optimization of spatial organizations for context exchange and surveillance

Abstract

Modern environments have huge quantities of contextual information that is mainly relevant/useful when a user/mobile is situated at particular locations. This paper investigates organizations for exchanging localized context and surveilling the corresponding areas that minimize bandwidth/energy requirements. We propose a simple model where context is associated with spatial regions, cells, and is transferred to/from mobiles as their trajectories traverse the cells. Key elements of the model are the granularity (e.g., area) of the cells and the amount of context associated with each cell. The model uses random tessellations to generate spatial partitions of the space into cells of varying granularity, and an associated context content function to capture possible ways in which contextual data associated with a cell might scale. We discuss a simple taxonomy for such scalings giving a crude idea of how representative applications might behave. Based on this model we analyze optimal flat aggregative and hierarchical organizations for mediating context exchange revealing some fundamental principles relating the optimal granularity of cells to different types of applications' spatial content scaling.