ProxiMol: Proximity and Mobility Estimation for Efficient Forwarding in Opportunistic Networks

A series of opportunistic contacts in space and time among devices carried by mobile users, can be utilized to forward messages from one user to another in the absence of an end-to end connected path. Existing routing metrics exhibit efficient performance in either homogeneous (users have similar mobility characteristics) or specific heterogeneous (users exhibit varied mobility characteristics) scenarios. However, in practice, behavior of users changes at different locations and times, making it hard to generalize any one routing algorithm. The adaptive forwarding scheme, ProxiMol, proposed in this paper leverages two simple facts some users have better likelihood of message delivery due to higher mobility, while others do due to their location's proximity to destination. Key contributions of ProxiMol include: i) a model to infer user's location over time from its diffusion (a measure of mobility); ii) an analytical result to estimate distance between users; and iii) an empirical method to estimate diffusion of a user. These are used to compute the likelihood of delivery taking into account both the mobility of a user and her proximity to destination. In addition to this robust forwarding scheme, a novel concept of disconnected distance that captures partial paths in networks with moderate levels of connectivity is introduced. ProxiMol improves delivery ratios (10-20%) and reduces delays by up to 50%, when compared against previously proposed algorithms, in user environments that range from relatively homogeneous to highly heterogeneous settings.