Beyond Traditional DTN Routing: Social Networks for Opportunistic Communication

D Elay-tolerant networks (DTNs) [1] are partitioned wire- less ad hoc networks with intermittent connectivity. Additional terminology in this family of dynamic networks includes disruption-tolerant networks, intermittently connected networks, and opportunistic networks. DTNs are never fully connected at any point in time, but points of disconnection may be predictable as in vehicular networks following transportation schedules or networks with satellites traversing orbits [2]. In an intermittently connected network (ICMAN) or an opportunistic network, nodes rarely have information on the changing network topology [3] [4]. Nodes may not know the availability of future encounters, but the network may benefit from learning such patterns over time. Thus, subsets of nodes in transmission range leverage cooperation during pairwise contacts to forward data towards a destination [4].

The designers of these dynamic networks often rely on the mobility of nodes to route messages and bridge partitions. Intermediate relays may be required to store messages and deliver them to destinations as they are encountered, i.e. enter into radio range. Mobility-assisted routing in DTNs is enabled by this “store-carry-forward” paradigm. A variant of this approach is the store-carry-replicate strategy, which replicates the routed packets, thus increasing the number of copies in the network. As investigated in [2], traditional ad hoc routing protocols must be adapted within a DTN architecture. Classical proactive or reactive routing approaches proposed for regular ad hoc networks do not work for these challenged DTNs, due to the fact that an end-to-end path may not be available at the time of transmission. However, over time, as different links come up and down thanks to mobility (or other environmental characteristics), the dynamic evolution of connectivity graphs 1 The first author is now with LGS Bell Labs Innovations, Florham Park, NJ, USA. Email: maschurg@lgsinnovations.com over a longer time interval may lead to an asynchronous endto-end path.

Existing DTN routing protocols evolved from enabling the transfer of any amount of data to carefully selecting intermediate nodes to efficiently carry information. Forwarding schemes were adapted over time to address different performance measures: delivery ratio, message latency, and overhead. The design of DTN routing algorithms may be application-specific, but generally all schemes should balance the overhead from redundant copies with successful delivery and minimal delay. In this work we emphasize the need for multi-objective optimization to better understand performance tradeoffs in opportunistic networks.

Improved performance amounts to identifying suitable carriers for a specific destination. Nodes may be drawn to particular geographic regions or influenced by the behavior of other nodes. With an underlying assumption that the mobility process is ergodic and stationary, algorithms have been designed to predict the future from past behavior. This assumption may not always be valid and slower changing attributes, like social connections, may be leveraged to enable efficient message delivery. Social relationships are expected to vary slower than the transmission links between mobile nodes [5]. In fact, the application of social network theory to model delaytolerant networks has led to the design of a new class of routing solutions. Forwarding algorithms like SimBetTS [6] and BUBBLE [5] consider a node’s role in the social structure of the network to make routing decisions.

Social-based protocols may quantify the social network structure, identify socially-similar nodes, and/or utilize context information [4] like shared interests or community affiliations. Social-based routing is a particularly relevant solution for opportunistic networks with a social component like pocketswitched [5] and mobile peer-to-peer networks [7].

In this work, we present the evolution of DTN routing protocols and highlight the application of social network theory to communication systems. Previous tutorials and surveys focused on formally defining a DTN architecture and discussing routing solutions. Unique to our review is the classification of routing protocols based on the network graphs we define: the dynamic wireless graph composed of every available link in time; the contact graph calculated from the aggregation of past wireless links; and finally the social graph formed by interpersonal relationships. The intent of this article is not to provide a comprehensive review of all DTN protocols. Instead, we chose a cross section of protocols that chronicles the development of sophisticated routing solutions for intermittently connected ad hoc networks. We begin with a description of the wireless graph and associated protocols; then transition to the contact graph. The social graph is then introduced and defined from two perspectives. The article concludes with a discussion of open research issues and c

This content is AI-processed based on open access ArXiv data.