UC Santa Cruz

Tactical and emergency-response networks require efficient communication without a managed infrastructure in order to meet the requirements of mission critical applications. In these networks, mobility, disruption, limited network resources, and limited host resources are the norm instead of the exception. Despite these constraints, applications must quickly and reliably share data collected from their environment to allow users to coordinate and make critical decisions. Our previous work demonstrates that applying information-centric paradigms to the tactical edge can provide performance benefits over traditional address centric approaches. We expand on this work and investigate how social relationships can be inferred and exploited to improve network performance in volatile networks.

As a result of our investigation, we propose SOCRATIC (SOCial RATe control for Information Centric networks), a novel approach to dissemination that unifies replication and network coding, which takes advantage of social content and context heuristics to improve network performance. SOCRATIC replicates network encoded blocks according to a popularity index metric that captures social relationships, and is shared during neighbor discovery. The number of encoded blocks that is relayed to a node depends on its interest in the data object and its popularity index, i.e., how often and for how long it meets other nodes. We observe that nodes with similar interests tend to be co-located and we exploit this information through use of a generalization of a data object-to-interest matching function that quantifies this similarity. Encoded blocks are subsequently replicated towards the subscriber if a stable path exists. We evaluate an implementation of SOCRATIC through a detailed network emulation of a tactical scenario and demonstrate that it can achieve better performance than the existing socially agnostic approaches.