Google services are not necessarily the answer for group communication among mobile users

Recent trends are that resource rich laptops are ubiquitous and are becoming the primary computing platform for many users. USA today [2] described the emergence of about 30 million (American) nomadic users. Gartner Dataquest predicted a yearly growth of 10% of these users. Newer laptops are also boasting resources as high as 2.6 GHz dual core processors, 320 GB of storage, 4 GB of memory and 300 Mbps 802.11n wireless LANs. Consider an university setting; student groups work on academic and social projects while groups of students and faculty collaborate on course and research projects. These collaborations are enhanced by using rich media such as audio and video. For example, faculty members distribute audio and video of the lecture material [1] to students while student groups work on a group report. Traditionally wireless laptops relied on services provided by the wired infrastructure. Either the university provided the shared storage space or the users used services such as Google Docs 1 to share contents. However, even though raw storage is inexpensive, it is expensive for the university to provide large amounts of managed storage. Even though services such as Google Docs are free, the users might express privacy concerns. Besides, using Internet storage could require the university to upgrade its Internet capacity, a potentially expensive proposition. Laptops rarely shared and utilized the resources available among other laptops even though the potential for large amounts of unmanaged storage is high. In our campus, we observed about 12,500 active mobile devices. Even if each laptop volunteered 10 GB of shared storage, the aggregate storage can be quite large (over 100 TB). Hence, we investigated the behavior of collaborations systems built among wireless users in a campus environment. We based our analysis on the empirical availability patterns of a large number of wireless users in two universities. We focus on two different kinds of group communications. In direct delivery systems, shared contents were directly served by the providers (or their replicas); clients are not required to maintain a local copy. Applications such as Apple iTunes follow this sharing model. Instructor can distribute their lectures using this model. Our analysis [4] showed that laptop users require a large number of replicas to service contents to members of the class or to Internet users. Depending on the time of day,