Delay tolerant networks (DTNs) are a special type of wireless mobile networks, which may lack continuous network connectivity. Routing in DTNs is very challenging, as it must handle network partitions, long delays, and dynamic topology in such networks. Recently, the consideration of social characteristics of mobile nodes provides a new angle of view in the design of DTNs routing protocols. In many DTNs, a multitude of mobile devices are used and carried by people (e.g. pocket switched networks and vehicular networks), whose behaviors are better described by social models. This opens the new possibilities of social-based routing, in which the knowledge of social characteristics is used for making better forwarding decision. However, the social relations do not necessarily reflect the true device communication opportunities in a dynamic DTN. On the other hand, the increasing availability of location technologies (GPS, GSM networks, etc.) enables mobile devices to obtain their locations easily. Consider that an individual's location history in the real world implies his/her social interests and behaviors to some extent, in this dissertation, we study new social based DTN routing protocols, which utilize location and/or social features to achieve efficient and stable routing for delay tolerant networks. We first incorporate the location features into the social-based DTN routing methods to improve their performance by treating location similarity among nodes as possible social relationship. Then, we dis- cuss the possibility and methods to further improve routing performance by adding limited amount of throw-boxes into the networks to aid the DTN relay. Several throw-boxes based routing protocols and location selection methods for throw-boxes are proposed. All pro- posed routing methods are evaluated via extensive simulations with real life trace data (such as MIT reality, Nokia MDC, and Orange D4D).
