Not listed AS DEMANDS ON SPACE NETWORK SYSTEMS CONTINUE TO INCREASE NEW APPROACHES ARE NEEDED TO FULFILL THE END-TO-END RELIABILITY AND PERFORMANCE EXPECTATIONS. LONG SEPARATION DISTANCES COMPLEX ORBITAL MECHANICS AND NODE TRAJECTORIES PRODUCE DRASTIC COMMUNICATION CONDITIONS WITH EXTREME DELAYS RANDOM DISRUPTIONS AND SEVERE DISTURBANCES THAT ARE DIFFICULT TO AVOID. THE AIM OF PROJECT IS TO DESIGN AND CREATE COGNITIVE GATEWAYS (CG) THAT WILL AUTONOMOUSLY DECIDE HOW TO ROUTE DATA OBJECTS OVER DIFFERENT SPACE COMMUNICATION TECHNOLOGIES. EACH CG LEARNS THE BEST NEXT-HOP DECISION FOR DATA OBJECTS USING OBSERVATIONS OF THE OPERATIONAL ENVIRONMENT CROSS-LAYER METRICS PAST PERFORMANCE LINK SCHEDULES PRE-ASSIGNED RULES AND USER REQUIREMENTS OF END-TOEND QUALITY OF SERVICE FOR DIFFERENT COMMUNICATION TYPES. THE PROJECT DEVELOPS FAST LEARNING METHODS BASED ON RECURRENT SPIKING NEURAL NETWORKS (SNN) OF RANDOMLY CONNECTED NEURONS (RESERVOIR) AND REINFORCEMENT LEARNING THAT ADJUST AT RUNTIME THE SYNAPTIC STRENGTH OF A READOUT LAYER. SNNS PROVIDE THE LEARNING AND DECISION-MAKING FUNCTIONS FOR CGS. JOINTLY DISTRIBUTED CGS CREATE A COGNITIVE ROUTING LAYER THAT CAN INTEGRATE DIFFERENT COMMUNICATION TECHNOLOGIES FOR SPACE MISSIONS INCLUDING CONVENTIONAL RADIO COGNITIVE RADIO DELAY-TOLERANT NETWORKS AND FREE OPTICAL COMMUNICATIONS. THE PROJECT ALSO DEVELOPS AND EVALUATES A COMPOSABLE SYSTEM PROTOTYPE. THE DESIGN IS FLEXIBLE TO ALLOW SYSTEM MODIFICATIONS AND FUTURE EXTENSIONS. FOR EXAMPLE A SOFTWARE SNN COMPONENT COULD BE SWAPPED WITH HARDWARE IMPLEMENTATION ON A NEUROMORPHIC PROCESSOR FOR GREATER SPEED AND ENERGY EFFICIENCY. THE VALIDATION IS CARRIED OUT ON A LABORATORY TESTBED THAT EMULATES THE ESSENTIAL FEATURES OF SPACE COMMUNICATIONS AND CONDITIONS. THE EVALUATION IS COMPLEMENTED BY A THOROUGH HIGH-PERFORMANCE COMPUTING SIMULATION TO OBTAIN ACCURATE EVIDENCE OF THE EFFICACY OF THE PROPOSED METHODS UNDER A LARGER NUMBER OF DIFFERENT CONDITIONS AND PARAMETERS. CREATING AUTONOMY IN THE NETWORK THROUGH COGNITION CAN POTENTIALLY BRING MANY BENEFITS TO FUTURE SPACE MISSIONS. IT CAN DRASTICALLY REDUCE ERRORS AND THE NEED OF HUMAN ASSISTANCE IN THE RECURRENT TASK OF DECIDING OPTIMAL NETWORK ROUTES AND PARAMETERS AFTER A CHANGE IN THE GEOGRAPHICAL OR OPERATIONAL STATE OF THE NETWORK. ALSO BECAUSE A COGNITIVE NETWORK CAN RECONFIGURE ITSELF WHENEVER NEEDED TO ACHIEVE ITS ASSIGNED GOALS THIS LEADS TO IMPROVED RELIABILITY AND SYSTEM PERFORMANCE AS WELL AS ASSET EFFICIENCY. $0 1/5/21 P00004 AS DEMANDS ON SPACE NETWORK SYSTEMS CONTINUE TO INCREASE, NEW APPROACHES ARE NEEDED TO FULFILL THE END-TO-END RELIABILITY AND PERFORMANCE EXPECTATIONS. LONG SEPARATION DISTANCES, COMPLEX ORBITAL MECHANICS, AND NODE TRAJECTORIES PRODUCE DRASTIC COMMUNICATION CONDITIONS WITH EXTREME DELAYS, RANDOM DISRUPTIONS, AND SEVERE DISTURBANCES THAT ARE DIFFICULT TO AVOID. THE AIM OF PROJECT IS TO DESIGN AND CREATE COGNITIVE GATEWAYS (CG) THAT WILL AUTONOMOUSLY DECIDE HOW TO ROUTE DATA OBJECTS OVER DIFFERENT SPACE COMMUNICATION TECHNOLOGIES. EACH CG LEARNS THE BEST NEXT-HOP DECISION FOR DATA OBJECTS USING OBSERVATIONS OF THE OPERATIONAL ENVIRONMENT, CROSS-LAYER METRICS, PAST PERFORMANCE, LINK SCHEDULES, PRE-ASSIGNED RULES, AND USER REQUIREMENTS OF END-TOEND QUALITY OF SERVICE FOR DIFFERENT COMMUNICATION TYPES. THE PROJECT DEVELOPS FAST LEARNING METHODS BASED ON RECURRENT SPIKING NEURAL NETWORKS (SNN) OF RANDOMLY CONNECTED NEURONS (RESERVOIR) AND REINFORCEMENT LEARNING THAT ADJUST, AT RUNTIME, THE SYNAPTIC STRENGTH OF A READOUT LAYER. SNNS PROVIDE THE LEARNING AND DECISION-MAKING FUNCTIONS FOR CGS. JOINTLY, DISTRIBUTED CGS CREATE A COGNITIVE ROUTING LAYER THAT CAN INTEGRATE DIFFERENT COMMUNICATION TECHNOLOGIES FOR SPACE MISSIONS, INCLUDING CONVENTIONAL RADIO, COGNITIVE RADIO, DELAY-TOLERANT NETWORKS, AND FREE OPTICAL COMMUNICATIONS. THE PROJECT ALSO DEVELOPS AND EVALUATES A COMPOSABLE SYSTEM PROTOTYPE. THE DESIGN IS FLEXIBLE TO ALLOW SYSTEM MODIFICATIONS AND FUTURE EXTENSIONS. FOR EXAMPLE, A SOFTWARE SNN COMPONENT COULD BE SWAPPED WITH HARDWARE IMPLEMENTATION ON A NEUROMORPHIC PROCESSOR FOR GREATER SPEED AND ENERGY EFFICIENCY. THE VALIDATION IS CARRIED OUT ON A LABORATORY TESTBED THAT EMULATES THE ESSENTIAL FEATURES OF SPACE COMMUNICATIONS AND CONDITIONS. THE EVALUATION IS COMPLEMENTED BY A THOROUGH HIGH-PERFORMANCE COMPUTING SIMULATION TO OBTAIN ACCURATE EVIDENCE OF THE EFFICACY OF THE PROPOSED METHODS UNDER A LARGER NUMBER OF DIFFERENT CONDITIONS AND PARAMETERS. CREATING AUTONOMY IN THE NETWORK THROUGH COGNITION CAN POTENTIALLY BRING MANY BENEFITS TO FUTURE SPACE MISSIONS. IT CAN DRASTICALLY REDUCE ERRORS AND THE NEED OF HUMAN ASSISTANCE IN THE RECURRENT TASK OF DECIDING OPTIMAL NETWORK ROUTES AND PARAMETERS AFTER A CHANGE IN THE GEOGRAPHICAL OR OPERATIONAL STATE OF THE NETWORK. ALSO, BECAUSE A COGNITIVE NETWORK CAN RECONFIGURE ITSELF WHENEVER NEEDED TO ACHIEVE ITS ASSIGNED GOALS, THIS LEADS TO IMPROVED RELIABILITY AND SYSTEM PERFORMANCE, AS WELL AS ASSET EFFICIENCY. Other Administrative Action $0 1/5/21 P00003 AS DEMANDS ON SPACE NETWORK SYSTEMS CONTINUE TO INCREASE, NEW APPROACHES ARE NEEDED TO FULFILL THE END-TO-END RELIABILITY AND PERFORMANCE EXPECTATIONS. LONG SEPARATION DISTANCES, COMPLEX ORBITAL MECHANICS, AND NODE TRAJECTORIES PRODUCE DRASTIC COMMUNICATION CONDITIONS WITH EXTREME DELAYS, RANDOM DISRUPTIONS, AND SEVERE DISTURBANCES THAT ARE DIFFICULT TO AVOID. THE AIM OF PROJECT IS TO DESIGN AND CREATE COGNITIVE GATEWAYS (CG) THAT WILL AUTONOMOUSLY DECIDE HOW TO ROUTE DATA OBJECTS OVER DIFFERENT SPACE COMMUNICATION TECHNOLOGIES. EACH CG LEARNS THE BEST NEXT-HOP DECISION FOR DATA OBJECTS USING OBSERVATIONS OF THE OPERATIONAL ENVIRONMENT, CROSS-LAYER METRICS, PAST PERFORMANCE, LINK SCHEDULES, PRE-ASSIGNED RULES, AND USER REQUIREMENTS OF END-TOEND QUALITY OF SERVICE FOR DIFFERENT COMMUNICATION TYPES. THE PROJECT DEVELOPS FAST LEARNING METHODS BASED ON RECURRENT SPIKING NEURAL NETWORKS (SNN) OF RANDOMLY CONNECTED NEURONS (RESERVOIR) AND REINFORCEMENT LEARNING THAT ADJUST, AT RUNTIME, THE SYNAPTIC STRENGTH OF A READOUT LAYER. SNNS PROVIDE THE LEARNING AND DECISION-MAKING FUNCTIONS FOR CGS. JOINTLY, DISTRIBUTED CGS CREATE A COGNITIVE ROUTING LAYER THAT CAN INTEGRATE DIFFERENT COMMUNICATION TECHNOLOGIES FOR SPACE MISSIONS, INCLUDING CONVENTIONAL RADIO, COGNITIVE RADIO, DELAY-TOLERANT NETWORKS, AND FREE OPTICAL COMMUNICATIONS. THE PROJECT ALSO DEVELOPS AND EVALUATES A COMPOSABLE SYSTEM PROTOTYPE. THE DESIGN IS FLEXIBLE TO ALLOW SYSTEM MODIFICATIONS AND FUTURE EXTENSIONS. FOR EXAMPLE, A SOFTWARE SNN COMPONENT COULD BE SWAPPED WITH HARDWARE IMPLEMENTATION ON A NEUROMORPHIC PROCESSOR FOR GREATER SPEED AND ENERGY EFFICIENCY. THE VALIDATION IS CARRIED OUT ON A LABORATORY TESTBED THAT EMULATES THE ESSENTIAL FEATURES OF SPACE COMMUNICATIONS AND CONDITIONS. THE EVALUATION IS COMPLEMENTED BY A THOROUGH HIGH-PERFORMANCE COMPUTING SIMULATION TO OBTAIN ACCURATE EVIDENCE OF THE EFFICACY OF THE PROPOSED METHODS UNDER A LARGER NUMBER OF DIFFERENT CONDITIONS AND PARAMETERS. CREATING AUTONOMY IN THE NETWORK THROUGH COGNITION CAN POTENTIALLY BRING MANY BENEFITS TO FUTURE SPACE MISSIONS. IT CAN DRASTICALLY REDUCE ERRORS AND THE NEED OF HUMAN ASSISTANCE IN THE RECURRENT TASK OF DECIDING OPTIMAL NETWORK ROUTES AND PARAMETERS AFTER A CHANGE IN THE GEOGRAPHICAL OR OPERATIONAL STATE OF THE NETWORK. ALSO, BECAUSE A COGNITIVE NETWORK CAN RECONFIGURE ITSELF WHENEVER NEEDED TO ACHIEVE ITS ASSIGNED GOALS, THIS LEADS TO IMPROVED RELIABILITY AND SYSTEM PERFORMANCE, AS WELL AS ASSET EFFICIENCY. Other Administrative Action $0 4/14/20 Not listed AS DEMANDS ON SPACE NETWORK SYSTEMS CONTINUE TO INCREASE NEW APPROACHES ARE NEEDED TO FULFILL THE END-TO-END RELIABILITY AND PERFORMANCE EXPECTATIONS. LONG SEPARATION DISTANCES COMPLEX ORBITAL MECHANICS AND NODE TRAJECTORIES PRODUCE DRASTIC COMMUNICATION CONDITIONS WITH EXTREME DELAYS RANDOM DISRUPTIONS AND SEVERE DISTURBANCES THAT ARE DIFFICULT TO AVOID. THE AIM OF PROJECT IS TO DESIGN AND CREATE COGNITIVE GATEWAYS (CG) THAT WILL AUTONOMOUSLY DECIDE HOW TO ROUTE DATA OBJECTS OVER DIFFERENT SPACE COMMUNICATION TECHNOLOGIES. EACH CG LEARNS THE BEST NEXT-HOP DECISION FOR DATA OBJECTS USING OBSERVATIONS OF THE OPERATIONAL ENVIRONMENT CROSS-LAYER METRICS PAST PERFORMANCE LINK SCHEDULES PRE-ASSIGNED RULES AND USER REQUIREMENTS OF END-TOEND QUALITY OF SERVICE FOR DIFFERENT COMMUNICATION TYPES. THE PROJECT DEVELOPS FAST LEARNING METHODS BASED ON RECURRENT SPIKING NEURAL NETWORKS (SNN) OF RANDOMLY CONNECTED NEURONS (RESERVOIR) AND REINFORCEMENT LEARNING THAT ADJUST AT RUNTIME THE SYNAPTIC STRENGTH OF A READOUT LAYER. SNNS PROVIDE THE LEARNING AND DECISION-MAKING FUNCTIONS FOR CGS. JOINTLY DISTRIBUTED CGS CREATE A COGNITIVE ROUTING LAYER THAT CAN INTEGRATE DIFFERENT COMMUNICATION TECHNOLOGIES FOR SPACE MISSIONS INCLUDING CONVENTIONAL RADIO COGNITIVE RADIO DELAY-TOLERANT NETWORKS AND FREE OPTICAL COMMUNICATIONS. THE PROJECT ALSO DEVELOPS AND EVALUATES A COMPOSABLE SYSTEM PROTOTYPE. THE DESIGN IS FLEXIBLE TO ALLOW SYSTEM MODIFICATIONS AND FUTURE EXTENSIONS. FOR EXAMPLE A SOFTWARE SNN COMPONENT COULD BE SWAPPED WITH HARDWARE IMPLEMENTATION ON A NEUROMORPHIC PROCESSOR FOR GREATER SPEED AND ENERGY EFFICIENCY. THE VALIDATION IS CARRIED OUT ON A LABORATORY TESTBED THAT EMULATES THE ESSENTIAL FEATURES OF SPACE COMMUNICATIONS AND CONDITIONS. THE EVALUATION IS COMPLEMENTED BY A THOROUGH HIGH-PERFORMANCE COMPUTING SIMULATION TO OBTAIN ACCURATE EVIDENCE OF THE EFFICACY OF THE PROPOSED METHODS UNDER A LARGER NUMBER OF DIFFERENT CONDITIONS AND PARAMETERS. CREATING AUTONOMY IN THE NETWORK THROUGH COGNITION CAN POTENTIALLY BRING MANY BENEFITS TO FUTURE SPACE MISSIONS. IT CAN DRASTICALLY REDUCE ERRORS AND THE NEED OF HUMAN ASSISTANCE IN THE RECURRENT TASK OF DECIDING OPTIMAL NETWORK ROUTES AND PARAMETERS AFTER A CHANGE IN THE GEOGRAPHICAL OR OPERATIONAL STATE OF THE NETWORK. ALSO BECAUSE A COGNITIVE NETWORK CAN RECONFIGURE ITSELF WHENEVER NEEDED TO ACHIEVE ITS ASSIGNED GOALS THIS LEADS TO IMPROVED RELIABILITY AND SYSTEM PERFORMANCE AS WELL AS ASSET EFFICIENCY. $0 4/14/20 Not listed AS DEMANDS ON SPACE NETWORK SYSTEMS CONTINUE TO INCREASE NEW APPROACHES ARE NEEDED TO FULFILL THE END-TO-END RELIABILITY AND PERFORMANCE EXPECTATIONS. LONG SEPARATION DISTANCES COMPLEX ORBITAL MECHANICS AND NODE TRAJECTORIES PRODUCE DRASTIC COMMUNICATION CONDITIONS WITH EXTREME DELAYS RANDOM DISRUPTIONS AND SEVERE DISTURBANCES THAT ARE DIFFICULT TO AVOID. THE AIM OF PROJECT IS TO DESIGN AND CREATE COGNITIVE GATEWAYS (CG) THAT WILL AUTONOMOUSLY DECIDE HOW TO ROUTE DATA OBJECTS OVER DIFFERENT SPACE COMMUNICATION TECHNOLOGIES. EACH CG LEARNS THE BEST NEXT-HOP DECISION FOR DATA OBJECTS USING OBSERVATIONS OF THE OPERATIONAL ENVIRONMENT CROSS-LAYER METRICS PAST PERFORMANCE LINK SCHEDULES PRE-ASSIGNED RULES AND USER REQUIREMENTS OF END-TOEND QUALITY OF SERVICE FOR DIFFERENT COMMUNICATION TYPES. THE PROJECT DEVELOPS FAST LEARNING METHODS BASED ON RECURRENT SPIKING NEURAL NETWORKS (SNN) OF RANDOMLY CONNECTED NEURONS (RESERVOIR) AND REINFORCEMENT LEARNING THAT ADJUST AT RUNTIME THE SYNAPTIC STRENGTH OF A READOUT LAYER. SNNS PROVIDE THE LEARNING AND DECISION-MAKING FUNCTIONS FOR CGS. JOINTLY DISTRIBUTED CGS CREATE A COGNITIVE ROUTING LAYER THAT CAN INTEGRATE DIFFERENT COMMUNICATION TECHNOLOGIES FOR SPACE MISSIONS INCLUDING CONVENTIONAL RADIO COGNITIVE RADIO DELAY-TOLERANT NETWORKS AND FREE OPTICAL COMMUNICATIONS. THE PROJECT ALSO DEVELOPS AND EVALUATES A COMPOSABLE SYSTEM PROTOTYPE. THE DESIGN IS FLEXIBLE TO ALLOW SYSTEM MODIFICATIONS AND FUTURE EXTENSIONS. FOR EXAMPLE A SOFTWARE SNN COMPONENT COULD BE SWAPPED WITH HARDWARE IMPLEMENTATION ON A NEUROMORPHIC PROCESSOR FOR GREATER SPEED AND ENERGY EFFICIENCY. THE VALIDATION IS CARRIED OUT ON A LABORATORY TESTBED THAT EMULATES THE ESSENTIAL FEATURES OF SPACE COMMUNICATIONS AND CONDITIONS. THE EVALUATION IS COMPLEMENTED BY A THOROUGH HIGH-PERFORMANCE COMPUTING SIMULATION TO OBTAIN ACCURATE EVIDENCE OF THE EFFICACY OF THE PROPOSED METHODS UNDER A LARGER NUMBER OF DIFFERENT CONDITIONS AND PARAMETERS. CREATING AUTONOMY IN THE NETWORK THROUGH COGNITION CAN POTENTIALLY BRING MANY BENEFITS TO FUTURE SPACE MISSIONS. IT CAN DRASTICALLY REDUCE ERRORS AND THE NEED OF HUMAN ASSISTANCE IN THE RECURRENT TASK OF DECIDING OPTIMAL NETWORK ROUTES AND PARAMETERS AFTER A CHANGE IN THE GEOGRAPHICAL OR OPERATIONAL STATE OF THE NETWORK. ALSO BECAUSE A COGNITIVE NETWORK CAN RECONFIGURE ITSELF WHENEVER NEEDED TO ACHIEVE ITS ASSIGNED GOALS THIS LEADS TO IMPROVED RELIABILITY AND SYSTEM PERFORMANCE AS WELL AS ASSET EFFICIENCY. $145.4k 5/11/19