Marino Lent
University of Houston
Forthcoming space networks are expected to be multi-tenant with multiple service providers carrying bundle flows with differentiated classes of service. These requirements, coupled with a higher uncertainty of the resulting system state, e.g., worse contact communication conditions than initially expected or planned contacts that fail to realize, demand a revised approach to space delay-tolerant networking (DTN). The project objectives aim to enhance the routing capabilities of DTN gateways with situated artificial intelligence techniques, bringing (A) user-defined multivariable utility functions that become self-optimized via learning; (B) scalability through a continual adaptation to unplanned events and network topology changes; and (C) compatibility with the Bundle Protocol and non-cognitive, standard DTN routing regions. These capabilities are integrated with an open architecture and experimentally demonstrated. Outcomes of this work are expected to help increase space networking autonomy with levels of communication performance and efficiency well above what can be achieved today.