Synthesis and Analysis of Spatio-Temporal Spectrum Demand Patterns: A First Principles Approach

Modeling and analysis of Primary User (PU) spectrum requirement is key to effective Dynamic Spectrum Access (DSA). Allocation of long term licenses, as well as opportunistic spectrum usage by secondary users (SU) cannot be done without accurate modeling of PU behavior. This is especially important in the case of cellular network traffic, which exhibits a significant spatio-temporal variation. Recently, there has been a lot of interest in modeling PU behavior in cellular networks by means of detailed analysis of proprietary data from wireless providers (e.g., Willcomm et al., IEEE DySpan 2008). While such analysis gives useful insights, major shortcomings of such an approach include (i) unavailability of data for open scientific study, and (ii) hard to predict future trends, and changes resulting from behavioral modifications. In this paper, we develop a methodology to generate synthetic network traffic data to model primary usage, by combining a number of different data sets for mobility, device ownership and call generation in a large synthetic urban population. Unlike simple random graph techniques, these methods use real world data sources and combine them with behavioral and social theories to synthesize spatial and dynamic relational networks. We use our tool to model the network traffic in the region of Portland, Oregon, calibrated by using published aggregate measurements of Wilcomm et al. As an illustration of our approach, we study the variation in demand as a result of changes in calling patterns based on user activities, and the impact of increased user demand on hotspots and their cascades within the region.