Cognitive Networking for NASA
- Created on Tuesday, 01 June 2010
NASA’s Space Communications and Navigation Program (SCaN) is exploring potential benefits of adding “intelligence” (reasoning and learning) to SCaN’s Integrated Network Architecture in order to increase network efficiency, provide tailored user services, reduce network operating costs through automation, and increase security and resiliency in an emerging environment shaped by:
- Small mobile mission clusters,
- Traditional large spacecraft and launch vehicles,
- Complexities of commercial and international partnering on missions and networks,
- Increasing threats to US space assets, and
- NASA’s need for 40% reduction in network operating costs.
Cognitive networks monitor changes in network conditions and the environment (by tasking nodes and sensors). They are guided by network end-to-end goals and policies, and have cognitive engines to reason about options to improve network end-to-end performance and service to users. Centralized or distributed cognitive engines can command reconfiguration of network operating characteristics (at any layer in the C3I stack, in any reconfigurable element authorized) or support human operators with statistical data and recommendations. Cognitive elements remember and learn from actions, and diagnose or correct anomalies.
What is NASA Doing?
NASA is surveying state-of-the-art in cognitive networking, which is generally for future Internet, terrestrial mobile wireless, or tactical battlefield applications, and only beginning to wrestle with adaptation to NASA space requirements. NASA plans technology studies in FY11. An optimal entry point for new technology, or “hooks and scars,” would be the 2018 Integrated Architecture, when NASA will converge Deep Space Network, Space Network (TDRSS), Near Earth Network, and future Exploration Destination networks into an Integrated Architecture, to provide unified support across all solar system and near-Earth domains. This transition will bring commonality and build on investments to space-qualify software defined radios, disruption tolerant networks, and optical as well as high-rate RF communications.
Technologies and Architecture Insights of Interest
The Network Goals Layer contains the network end-to-end goals to be optimized, operating policies, and network representation. Architectural decisions include human interface and degree of control, cost-benefit of optimization goals, security, and trust issues.
The Cognitive Layer contains software agents with reasoning and learning algorithms, and knowledge databases. Architecture decisions include degrees of centralization, adaptive or intelligent, efficient algorithms, and where to do the cognitive processing.
The Network Elements Layer is the sensing and reconfiguration layer, i.e. nodes, sensors, and software-reconfigurable network elements. Architecture decisions include cost-effective bit-wise degrees of reporting, and extent of crosslayer reconfiguration of the C3I stack.
What Applications Can NASA Envision?
The potential benefits of cognitive networking include enhanced services to users such as dynamic bandwidth allocation, differentiated QoS, improved mobility, and event-driven service; autonomous remote operations and scheduling; advanced routing and flexible topologies for exploration destinations; and radio optimization of spectrum, modulation, coding, power, and interference.
Improved network performance will enable users to optimize network performance proactively with network awareness; optimize quality of service and connectivity; decide on use of partner networks for connectivity and cost; and enable hooks and scars to keep pace with future Internet innovations.
Reduced network operations costs and future capital expenses will reduce human oversight needed and support human troubleshooting with detailed network status, statistics, and recommendations. It also will reduce transmission to Earth of remote mission data, which could be managed locally; manage complexity of international and commercial network partners; and increase network efficiency to avoid capital expense of expansion.
Increased security and reliability would enable proactive network monitoring and statistical trends; fault detection and repair; and visibility into behaviors, quicker detection of intrusion, and pinpointing. It also would enable access controls and authorization/trust mechanisms, and flexibility in partner interoperability for cross-support.