Uncovering vicinity properties in disruption-tolerant networks

International audienceMost existing proposals in the area of disruption-tolerant networking rely on the binaryassertion that when two nodes are not in contact, they are necessarily in intercontact. Sucha monolithic notion is, in our opinion, too limitative. In this paper, we advocate the use of theneighborhood of a node beyond one hop to help design more efficient communication solutions.We provide a formal definition of j-vicinity and associated measures, namely j-contactand j-intercontact. These measures allow better understanding the proximity betweennodes as they are not restrained solely to the direct contact situation.Weobserve unexpectedbehaviors in j-contact distributions and point out their dependency on node density. Wealso observe that a significant share of pairs of nodes spend a non-negligible amount of timein each other’s vicinity without coming into direct contact.We showthen that using a small j(between 2 and 4) is enough to capture a significant amount of communication possibilitiesthat are neglected by existing approaches. Finally, we provide a rule of thumb to derive thepopulation in the j-vicinity by observing only the direct contacts of a node