Program Analysis

• Running it twice may make sense, but time constraints on EDL would be very bad
  • Checkpoint and restart-type program will probably be the best
• Prior work on analyzing SDC behavior from bit flips assume HPC scenarios and cannot generalize
  • Static + dynamic analysis + generalization might be a lot stronger
  • So far, no prior work has done analysis on how runtime correlates with radiation risk, since the chances of an SEU are so low on Earth that it’s far more relevant to think about computation size
  • But this is significant for a space setting