Content diffusion in wireless MANETs

An intriguing approach to increasing throughput, owering latency, extending network coverage and reducing load on infrastructure is wireless content sharing via Mobile Ad hoc NETworks (MANETs). The potential efficacy of multi-hop MANET content diffusion is heavily influenced by small-scale patterns of mobility and content demand; a topic about which relatively little is known. We infer device encounters from a large multi-site wireless access point association trace to analyze the impact of i) time of day ii) day of week iii) site and iv) number of content sources on universal diffusion potential. In addition, we draw upon an empirical trace of application usage from a popular mobile campus maps application to model realistic content demand. Using trace-driven simulations we find that universal diffusion potential varies widely over the analyzed parameters, favoring larger more active sites and weekdays over weekends. More content sources proves beneficial, though mostly over the short-term. Our analysis of content demand for campus maps suggests that up to a third of application requests could be served from the MANET over a 12-hour period and again that weekdays are more amenable to content diffusion than weekends.