2017-11-01 Meeting notes

Status on follow-ups for last weeks action items (5 min)

• Not complete (week 40) Mariusz Eivind Grøtte Get preliminary results from EOIR tool in STK (or MODTRAN) on "what can we see from space"

• Not complete (week 40) Unknown User (julianve) First revision of PCB

• Not complete (week 40) Unknown User (rogerbi) A cost benefit analysis of power usage vs compression for comms.
• Not complete (week 40) Unknown User (rogerbi) Present architecture for ground station, how will it work. Decide when upgrade happens, does that involve getting the total package such as h/w and s/w? And need a data system for storing and archiving. 
• Not complete (week 40) Unknown User (sivertba) Find a student to do thermal analysis and coordinate with Mariusz

• Not complete (week 40) Unknown User (joaoff) Report of what has been done with HSI imaging
• Not complete (week 40) Give feedback on Gantt chart (ALL) - Still want a list of activities from each - respond to emails

Data from HSI

Struggling to find the real SNR. Joao have calculated some SNR values based on the parameters based on the camera specification. It seems to be a high SNR for space application, and we could expect more noise from real usage of the camera. Should try to find the total SNR. in the visual spectra we have theoretical good values for SNR. A lower frame rate gives a better SNR, and an realistic value can be 15, which is below what has been used in current calculations.

Additional notes by Roger:

IdleTechs & Software

• Software, data processing pipeline
• Good simulations in Matlab, implement it in FPGA
• Precision loss
  • Cut to a certain precision 
  • 16 bit or less (is it viable to cut the precision - what is lost)
  • Computations are using 64 bit in Matlab, 16 bit values from the image
  • Custom floating point (24 bit)

Communications

• Got initial funding for upgrading ground station
• Coarse calculations on data budget, reasonable calculations fit with S-band link
• Tyvak says that we can have comms. link horizon-to-horizon 
  • Need precise pointing of the satellite
• 100-200 kb/s is good
• New PhD students, one will join on Monday - she will work on the communications (PhD in SDR) - contribute more on general communications
• A lot of work to be done on data budget vs. duty cycle vs. comms. link

New PhD. student mainly supposed to focus on SDR, but could perhaps help us on this project.

Mission Design

• Some requirements iterations to be done still based on NASA AMES meeting
• Wrote a HSI memo to Steve Dunagan (NASA Ames - optics expert) on optics and SNR
  • Still need to know actual performance of camera (raw images)

• Will lose spatial resolution due to slewing manoeuvre (not a function of height - but slant range)

• Have 30 x 30 km area to map (potentially larger in x-direction, e.g. 50x30 km)
  • dx=250 m, movement of satellite makes dx=500 m (30 fps)
  • dy=52 m (# pixels = 580)
  • assumes f0=50 mm, f/4
• Did sensitivity analysis on exposure time vs. aperture size. viewing angles on SNR
• Viewing angle considerably degrades performance and signature we may see
  • Now spatial resolution is function of slant range (at 80 deg the range is 3000 m, dx=1900 m, dy = 300 m).
  • 1 overhead pass is where we should put our expectations. (should not be bold)
  • Go easy on mission requirements, one nadir-looking satellite might be fine - everything else should be extra
• test actual performance of camera when not looking directly at region of interest.
  • looking form an angle and such
• Look into energy budget
• Look into how GSD<100 m should be achieved, currently it is at 248 m with current slew rate required and optical characteristics
• 15 fps gives higher SNR but lower spatial resolution - might have to go up to 30 fps - depends what is "good enough"

Actions ahead

• Determine the precision on numbers for implementation in FPGA; custom-made precision (24/32 bit); Match the Matlab precision (64 bit) with the image (16 bit)
• Actual performance of real HSI camera (take raw images)
  • Determine TOTAL SNR - only have to find a way to model noise, i.e. the variance of signal-dependent noise (variance of photons)
    • Definition: total signal received on the sensor - what can the camera see. Indicator of water-leaving photons - but not the answer.
    • Look at a reference
    • Subtract image from a dark frame (close the camera lid) - then you have all signal-independent noise (dark noise, read-out noise etc.)
    • Grating/optical efficiencies should be based on the manufacturer's numbers
  • test performance of camera when not looking directly at region of interest - Looking at a reference target from different angle and such
  • Test SNR for different exposure times (even may subtract two images at different exposures to see results)
• Look into how GSD<100 m should be achieved, currently it is at 248 m with current slew rate required and optical characteristics
• Look into energy budget
• Iterate on HSI memo to Steve