I. Introduction
In communication environments with intermittent connectivity between source and destination, the domain of Delay/Disruption Tolerant Networking (DTN) [1] has gained significant attention. In such networks, nodes exchange information in an opportunistic manner as no permanent end-to-end connectivity between the source and destination is assumed. Data packets may be relayed over single or multiple hops as and when a pair of nodes come in each other's radio vicinity. The challenge posed by user non-cooperation has been a subject of immense interest [2]–[6] as DTN communication paradigm inherently assumes user cooperation in spreading a message. In this work, we consider a content dissemination network with multiple distributed sources that generate mutually exclusive localized contents. Such scenarios arise in a gamut of DTN applications [11]–[14] like dissemination of digital contents (news, multimedia, guidance instruction, safety information at the time of disaster, promotions offers or advertisements etc.) from distributed sources (which are capable of generating the contents and/or collecting and storing contents from users) to a population of mobile users. Such public content dissemination system is facilitated by mobility and cooperative-caching of contents by users. Aiming to study the scenario that emerges when rational users are forced to be altruistic in message forwarding, this work starts with the following crucial assumptions: (a) a scheme (like Direct Reciprocity) exists that forces a user to share her cached contents with any encountered node or source to be able to collect any content from the other party, (b) users have heterogeneous interest in contents, thus a user has motivation to delete contents that she received from another party but is of no interest to her (in other words, receive contents selectively) so as to make space for her preferred contents in buffer. In this premise, rational nodes try to make optimized use of their resources to judiciously pull out their content of interest from the network. In stark contrast to prior work, we introduce mobility as a user strategy alongside selective caching to characterize the set of actions and model this strategic interaction among users as a game. An objective of this work is to find the social optimal, Nash Equilibrium conditions in this scenario and to study the limit beyond which non-cooperation is no longer beneficial to an individual.