NASA developed the Delay or Disruption Tolerant Networking (DTN) approach to enable data forwarding in networks with intermittent connectivity, extreme distances, and data loss. Although the network topology changes are mostly predictable, dynamic network features such as network congestion and data loss are challenging to account for when computing paths. This can lead to decreased performance and efficiency. This paper discusses the implementation of a reinforcement learning solution called the Cognitive Space Gateway (CSG) approach for the High Rate Delay Tolerant Networking (HDTN) project along with method improvements. The method is based on spiking neural networks (SNN) that dynamically optimize the assigned goal using estimations and observations of the network congestion and loss. The paper also defines a class of service for bundles, leading to the customization of the routing goal as a multi-objective expression. An experimental evaluation of the method shows cases where the CSG outperforms the standard Contact Graph Routing (CGR) making it suitable for different space applications, where data reliability and speed are crucial factors. By using spiking neural networks and reinforcement learning, the Cognitive Space Gateway approach offers a promising solution to the challenges of dynamic network features and the optimization of routing goals in Delay or Disruption Tolerant Networking.
Implementing a Cognitive Routing Method for High-Rate Delay Tolerant Networking
2023-06-20
816439 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Delay/Disruption Tolerant Networking
NTRS | 2021
|Delay/Disruption Tolerant Networking
NTIS | 2016
|Delay/Disruption Tolerant Networking
NTRS | 2016
|CFDP in delay-tolerant networking
NTRS | 2002
|