Dissimilarity Measures for Clustering Space Mission Architectures - PowerPoint PPT Presentation

Dissimilarity Measures for Clustering Space Mission Architectures Cody Kinneer Institute for Software Research, Carnegie Mellon University Sebastian J. I. Herzig Jet Propulsion Laboratory, California Institute of Technology 18 October 2018 – ACM/IEEE MODELS Conference, Copenhagen, Denmark The cost information contained in this document is of a budgetary and planning nature and is intended for informational purposes only. It does not constitute a commitment on the part of JPL and/or Caltech. All content is public domain information and / or has previously been cleared for unlimited release.

Robotic Space Exploration Voyager 1 & 2 (1977) j p l . n a s a . g o v 2

The JPL Product Lifecycle Source: Nichols & Lin, 2014 j p l . n a s a . g o v j p l . n a s a . g o v 3

The JPL Product Lifecycle Source: Nichols & Lin, 2014 j p l . n a s a . g o v j p l . n a s a . g o v 4

Networked Constellations of Spacecraft • Small spacecraft enable innovative low-cost multi-asset missions • Goal of initiative is to develop new technologies that support novel mission concept proposals j p l . n a s a . g o v 5

Motivating Case Study Spacecraft-Based Radio Interferometry Radio interferometers : • Radio telescopes consisting of multiple antennas • Achieve the same angular resolution as that of a single telescope with the same aperture  Typically ground-based Source: http://www.passmyexams.co.uk/GCSE/physics/images/radio- telescopes-outdoors.jpg Want to do this in space: • Frequencies < 30Mhz blocked by ionosphere • Cluster of spacecraft (3 – 50) functioning as telescopes in LLO  CubeSats or SmallSats are promising enablers for this j p l . n a s a . g o v 6

Which Architecture is Optimal? Opt. 1 3U 3U 3U 3U To 3U 3U Groun d Challenge: transmit very large data volume from LLO to Earth j p l . n a s a . g o v 7

Which Architecture is Optimal? Opt. 1 Opt. 2 3U 3U 3U 3U 3U 3U 3U 3U SmallSat SmallSat (~100kg) (~100kg) To 3U 3U 3U Groun 3U 3U 3U d To Ground Challenge: transmit very large data volume from LLO to Earth j p l . n a s a . g o v 8

Which Architecture is Optimal? Opt. 1 Opt. 2 3U 3U 3U 3U 3U 3U 3U 3U SmallSat SmallSat (~100kg) (~100kg) To 3U 3U 3U Groun 3U 3U 3U d To Ground Challenge: transmit very large data Opt. 3 volume from LLO to Earth 3U 3U 6U 6U 3U 6U 3U 6U To Ground j p l . n a s a . g o v 9

Which Architecture is Optimal? Opt. 1 Opt. 2 3U 3U 3U 3U 3U 3U 3U 3U SmallSat SmallSat (~100kg) (~100kg) To 3U 3U 3U Groun 3U 3U 3U d To Ground Challenge: transmit very large data Opt. 3 volume from LLO to Earth 3U 3U 6U 6U • How many spacecraft? • Are all equipped with interferometry payload? Are some just relays? 3U 6U 3U 6U • Who communicates with Earth? • What frequency bands? Multi-hop? To • … Ground • Optimal w.r.t. cost? Science value? j p l . n a s a . g o v 10

Mission Architecture Trade Space Exploration Mechanized Exploration Which interferometry Model 1 Model 1 Model 4 Model 4 Abstraction of Abstraction of missions are Model 2 Model 2 optimal with Domain Model 3 Domain Model 3 Model n Model n respect to cost & scientific benefit? “A constellation mission consists of at least 2 spacecraft and at most 100” Solution Problem “A spacecraft can, but does not have Generation Description to contain the interferometry payload” Models in domain Which models in the domain are we “Constellation mission A with 3 “Operation of the interferometry looking for? spacecraft, one of which has a payload operation requires power” payload and solar cells” j p l . n a s a . g o v 11

Mission Architecture Trade Space Exploration Mechanized Exploration Which interferometry Abstraction of Abstraction of missions are optimal with Domain Domain respect to cost & scientific benefit? “A constellation mission consists of at least 2 spacecraft and at most 100” Solution Problem Search “A spacecraft can, but does not have Description to contain the interferometry payload” Models in domain Which models in the domain are we “Constellation mission A with 3 “Operation of the interferometry looking for? spacecraft, one of which has a payload operation requires power” payload and solar cells” In practice, too many possible In practice, too many possible solutions to generate & compare all solutions to generate & compare all  View as a search problem  View as a search problem j p l . n a s a . g o v 12

Application to Case Study Domain model in Ecore + OCL (Excerpt) 20 concepts, 9 associations, 15 attributes / parameters 20 concepts, 9 associations, 15 attributes / parameters > 48 10 possible models > 48 10 possible models j p l . n a s a . g o v 13

Application to Case Study Domain model in Ecore + OCL (Excerpt) 20 concepts, 9 associations, 15 attributes / parameters 20 concepts, 9 associations, 15 attributes / parameters > 48 10 possible models > 48 10 possible models j p l . n a s a . g o v 14

Application to Case Study Domain model in Ecore + OCL (Excerpt) 20 concepts, 9 associations, 15 attributes / parameters 20 concepts, 9 associations, 15 attributes / parameters > 48 10 possible models > 48 10 possible models j p l . n a s a . g o v 15

Problem: Too Many Architectures! j p l . n a s a . g o v 16

Idea: Clustering Similar Architectures j p l . n a s a . g o v 17

Overview of Approach j p l . n a s a . g o v 18

Overview of Approach PAM: Partitioning Around Medoids j p l . n a s a . g o v 19

PAM j p l . n a s a . g o v

PAM j p l . n a s a . g o v

PAM j p l . n a s a . g o v

PAM j p l . n a s a . g o v

PAM j p l . n a s a . g o v

Distance Measure? j p l . n a s a . g o v 25

Distance Measure? j p l . n a s a . g o v 26

Distance Measure? How to determine distance is non-trivial How to determine distance is non-trivial  We investigate three approaches  We investigate three approaches j p l . n a s a . g o v 27

Feature Selection Feature Vector CubeSat3U CubeSat3U Number of 4 Assets MX MX Cost 4.97 SmallSat CubeSat3U Coverage 0.28 MX HK Mission 22.97 Duration Deep Space Network ... ... j p l . n a s a . g o v 28

EMF Compare j p l . n a s a . g o v 29

Graph-edit Distance CubeSat3U MX MX CubeSat3U SmallSat MX HK Deep Space Network j p l . n a s a . g o v 30

Feature Selection j p l . n a s a . g o v 31 31

Validation • Manual clustering task • Given pairs, assign a distance score • Caveats 31 pairs, two groups of 2-3  j p l . n a s a . g o v 32

Results Group Group Features Features Features Graph- EMF 1 2 (All) (Assets) (Objectives) edit Compare Distance Group 1 1 Group 2 0.501 1 Features 0.364 0.386 1 (All) Features 0.263 0.560 0.436 1 (Assets) Features 0.304 0.223 0.869 0.341 1 (Objectives) Graph-edit 0.276 0.217 0.464 0.289 0.429 1 Distance EMF 0.029 0.123 0.536 0.147 0.424 0.789 1 Compare j p l . n a s a . g o v 33

Insights from human designers Keyword Group 1 Group 2 • Presence or absence of SmallSat relay 2 5 • Number of incoming / outgoing connections (relay) bands 2 3 • Number of bands of communication layers / 2 6 • Difference influenced by: levels  Background SmallSats 2 2  Goals threads 0 2 j p l . n a s a . g o v 34

Conclusions • Clustering has the potential to enable more through analysis of the architectural trade space • Dissimilarity measures for space mission architectures are non- trivial, and have trade-offs in granularity, extensibility, and types of considered information • Discussed insights from human clustering task, importance of a range of options • Clustering is a promising approach for design space exploration Cody Kinneer ckinneer@cs.cmu.edu j p l . n a s a . g o v 35

jpl.nasa.gov Government sponsorship acknowledged. All technical data was obtained from publicly available sources and / or is fictitious.

Backup Slides ACM/IEEE MODELS 2018 Presentation on “Dissimilarity Measures for Clustering Space Mission Architectures”

EMF Compare j p l . n a s a . g o v 38

Graph-edit Distance j p l . n a s a . g o v 39

Example Mission Architecture • Number of spacecraft CubeSat3U CubeSat3U • Type of spacecraft • Directed communication links • Communication equipment MX MX Gain   Band SmallSat CubeSat3U • Ground station • Payload MX HK Deep Space Network j p l . n a s a . g o v 40

Implementation Open Source Technologies Used in Implementation • Representation of Domain  Ecore / Eclipse EMF + OCL • Exploration Rules  Henshin • Analyses / Fitness Functions  Java • Optimization Using Genetic Algorithms  MOMoT, MOEA j p l . n a s a . g o v 41