80 likes | 258 Views
DSN Raw Data Incompatible Binary Formats. PDS4 Operational Readiness Review and Acceptance Review 30 November 2011. DSN Radio Science Data Flow. 1980 ANTENNA. 2012 ANTENNA. ATDF (TRK-2-25). TNF (TRK-2-34). RSR (0159-SCIENCE). ODR (RSC-11-3). ARCHIVED. TDM 0212-TRACKING. ODF
E N D
DSN Raw DataIncompatible Binary Formats PDS4 Operational Readiness Review and Acceptance Review 30 November 2011
DSN Radio Science Data Flow 1980 ANTENNA 2012 ANTENNA ATDF (TRK-2-25) TNF (TRK-2-34) RSR (0159-SCIENCE) ODR (RSC-11-3) ARCHIVED TDM 0212-TRACKING ODF (TRK-2-18) ODF (TRK-2-18) ANALYSIS SPK FILES MASS/DENSITY GRAVITY MODELS GRAVITATIONAL WAVES ATMOSPHERIC DRAG ATMOSPHERIC TEMP/PRESSURE ELECTRON DENSITY PROFILES RING OPACITY/PARTICLE DENSITY SURFACE ROUGHNESS/DENSITY BINARY ARCHIVED ASCII
Case 1:ODF Data Record Format(36 8-bit bytes = 288 bits) Time Tag fits in 32-bit word Time Frac + D/L Delay fit in 32-bit word but don’t align with 8-bit bytes 1, 2, 3, 6, and 7-bit fields RF H/P wraps over 32-bit boundary and does not align
Case 2: RSR Data Sample Packing Real and Imaginary parts of complex samples are segregated Time order of samples within 16-bit words is reversed One of these structures is repeated thousands of times per record
Option 1 • Allow arbitrary bit field binary formats (allow ‘descriptive’) • Data provider decides bit field lengths • Data provider decides start bits • Data provider describes interpretations • Pros and Cons: • Preserves instrument output • Violates prescriptive vs descriptive philosophy • Applicability to other data providers (MIPL, MSSS, …)? • Makes PDS4 tool development harder/impossible • No changes needed by DSN or RS community • Discourages ‘new investigator’ browsing
Option 2 • Require conversion to compliant formats (current default) • Ad hoc PDS conversions (dangerous?) • Work with DSN on conversion tools (unlikely) • ASCII, FITS, HDF5, or TBD which could be judged compliant • Pros and Cons • DSNandRS community unlikely to use archive • Improved browsing capabilities • Major conversion effort • Software • 5 binary products (ATDF, ODF, ODR, TNF, and RSR) • TNF hasover a dozenrecord types (SIS includes 77 pages of descriptive tables) • ODF is in its 6th generation (6 versions of SIS); each ODF version has multiple record types • Who will validate, how, and when? High probability of conversion error • ‘Inverter’ software, needed for at least some validation, could be public? • Operations • Virtually every radio science raw data archive is built around these products • DSN has been generating some products for more than 30 years • Deliveries of TNF and RSR will continue indefinitely (ODF limited to ‘old’ missions)
Option 3 • DSN binaries accepted as encoded byte streams • PDS points to DSN documentation • Pros and Cons • Preserves instrument output • Applicability to other data providers (MIPL, MSSS, …)? • No PDS4 tool development required (?) • No changes needed by DSN or RS community • Discourages ‘new investigator’ browsing • Raw data will be less accessible than under PDS3
Recommendation • Accept arbitrary bit field formats for instrument output only (the most raw data) • Notes: • This does not allow ODF, the most popular radio tracking data type • But the ODF is being phased out. • Convert non-compliant binaries to compliant formats as interest demands and resources allow • Requires PDS to develop, maintain, and make available conversion software (but only for the PDS4 ‘platform’ by EN programmers) • Expect ‘browsing’ users to work with the converted products • Pros and Cons • Preserves instrument output • Makes ‘usable’ products available, but only as interest dictates • PDS can choose format of converted data (such as ASCII, FITS, or HDF5) • Meshes with current DSN and RS community needs • Doubles(?) volume of converted data sets