Missing Packets in MASCS VIRS Data

An investigation by John Hayes reveals:
... there is [likely] an off-by-one bug in the MASCS software. The MASCS 
software packs as many spectra as it can into the VIRS VIS science packet. 
To save space, after the first spectrum, subsequent spectra are differenced 
from their predecessors. The MASCS software uses a common software library 
to bit-pack the differences. The MASCS software predicts how many bits the 
packed result will need, rounded up to a multiple of 8 bits to get a whole 
number of bytes. This length is tested against the remaining space in the 
packet. If there is room, the spectra difference is computed, the result is 
bit-packed, and then copied into the packet; otherwise, a new packet is 
started. However, the bit-packing library pads the result out to a multiple 
of 16 bits, i.e. an even number of bytes. Therefore, the MASCS prediction of 
the length could be one byte short. In the worst case, MASCS could generate 
a packet one byte larger than the maximum allowed.

The MASCS VIRS VIS and VIRS NIR science packets use the same format and the 
same spectra packing algorithm. However, we have only observed dropped VIS 
packets so far. It is possible we may expect occasional missed packets in 
NIR as well as VIS.

A one word patch is sufficient to fix the problem. The code that tests the 
length against the remaining space in the packet includes a constant value 
4096, the maximum packet payload size. Reducing this value by a few bytes 
would solve the problem. If the theory is incorrect, the patch would be 
harmless; it would just reduce the maximum packet size slightly. The patch 
could be uplinked in real-time. Or, it could be made "permanent" by adding 
it to the start-up macro. Note: a one word patch would be needed for the VIS 
code and for the NIR code.

A patch has been developed and is awaiting brassboard testing pending 
available resource time. The missing packets are rare, and considered a 
small data processing problem. The patch will be implemented if the effort 
and risk of doing so are both low enough.

The list of EDRs (and thus also CDRs) missing packets for delivery 6 is 
below. The companion excel spreadsheet 
(virs_vis_edr_missing_packet_analysis.xls) shows a detailed breakdown of the 
edrs and exactly which packets are missing.

VIRSVE_ORB_11116_174442
VIRSVE_ORB_11119_020502
VIRSVE_ORB_11120_061451
VIRSVE_ORB_11122_183617
VIRSVE_ORB_11123_055943
VIRSVE_ORB_11124_185335
VIRSVE_ORB_11125_065755
VIRSVE_ORB_11128_085914
VIRSVE_ORB_11129_200841
VIRSVE_ORB_11130_201914
VIRSVE_ORB_11133_094858