Network Management of Disruption-Tolerant Networks: A Systems Engineering Approach

Disruption Tolerant Networks (DTNs) maintain communications despite the temporary loss of one or more nodes in the path, including cut vertices and path endpoints. These networks utilize store-and-forward, persistent storage, and intelligent timer techniques to overcome planned and unplanned transmission obstacles. DTNs are favored in deep space missions where constraints on power, infrastructure, and physics preclude high-speed, robust communication. Network management (NM) describes the configuration, state, fault, and performance management of a network node. It is likely that NM applications operating over DTNs cannot make the bandwidth and longevity assumptions codified in non-DTN protocol NM applications. This paper describes a systems engineering approach to analyzing the DTN NM problem. We present four major developmental areas (system architecture, data definition/usage, application architecture, and tool implementation) critical to the maturation of this ability. Within each developmental area we present implementation recommendations and current threads of research. Ultimately, we conclude that closed-loop network management of DTNs is not viable and architectures that promote status telemetry over control loops, autonomous decision making over static configuration, and locality over global administration must be engineered for these networks. We further conclude that a per-network system analysis is necessary to balance the desire for an implementation-specific optimal solution versus a generalized best-practice solution.