been substantially improved in software or hardware technology in more than 20 years1,2 even though, internally on the astrodynamics support workstation off-line platform, considerable improvement of the special perturbation orbit determination and orbit propagation has taken place over the past 10 years. (These improvements have been reflected only in updates to the astrodynamics standardized codes distributed to specific users who could take advantage of them.)
Unlike in conventional, commercial software companies, in the Air Force system the concept of “pushing” or forcibly distributing improved software to user hardware is not practicable. Users of AFSPC software have different needs and do not have standard hardware, and, equally important, most users bury the standardized software in an overarching system that serves their particular uses. For example, a radar system3 has a large body of command, control, and processing software within which AFSPC software is only a small percentage (<10 percent) of the code. AFSPC also has to be cautious about International Traffic in Arms Regulations (ITAR) restrictions, further complicating the model of automatically distributing and upgrading software for external users.
Finding: The Air Force recognizes the need for maintaining compatibility with existing systems in any new system development, and is committed to supporting its legacy customers.
Examples of problems caused by a rigid, older architecture abound, and a few are described below:
• New software technologies. The standardized software was originally written in FORTRAN 77 as used by the SPADOC system and has since been somewhat improved to Fortran 95 and C with only SGP4 available in Java. However, modern computer science practices are moving away from monolithic codes and toward architectures and languages that support advanced software technologies such as object-oriented design, modularity, parallel processing, etc.
• New hardware technologies. SGP4 and the SP standardized algorithms were written for single-thread, single core computers. Adaptation to modern, multi-thread and multi-core architectures to support parallelism will require explicit investment in research and development and in implementation.
• New database technologies. Use of the standardized software requires databases of orbital elements and tracking data. While a database is used on the SPADOC system, flat files were the only technology supported in the astrodynamics support workstation environment and the distributed versions of the standard algorithms, and so modern relational generalized databases cannot be used without modifying the software.
• New algorithms. Some of the standardized software has been frozen for more than 20 years. Other parts have slowly evolved during the past 10 years. However, many of the modern improvements in dynamical systems, in computational algorithms, and uncertainty assessment have not yet been included.
• New types of data. A principal purpose of space situational awareness is to monitor, and warn of, close approaches to operational satellites as mandated by the 2010 National Space Policy and recently accepted by U.S. Strategic Command on behalf of AFSPC/JSpOC. The accuracy and the immediacy of such warnings would be significantly enhanced if the JSpOC could automatically accept and process data from satellite owners/operators—military, civilian, and commercial. However, such an automated capability does not exist in the JSpOC, largely because of lack of funding to automate this capability, information assurance issues (i.e., computer security), and concern over the potential to compromise internal systems and processes by corrupted data.
The population of cataloged objects continues to rise, and with that increase comes a corresponding increase in computational demand. Catalog growth is anticipated from improvements in observational capabilities as well as from higher frequencies of launches and collisions. The forecasted growth in computational demands can be
1 F.R. Hoots, P.W. Schumacher, and R.A. Glover, History of analytical orbit modeling in the U.S. space surveillance system, Journal of Guidance, Control and Dynamics 27(2):176, 2004.
2 This does not imply that the products have not improved: they have because of more accurate and more prolific sensor data as well as more frequent data processing. However, the theory in the algorithms remains as it was 20 years ago—and does not reflect advances in astrodynamics and in computation systems.
3 From verbal communications with the software and space operations personnel at the Space Surveillance radars at MIT Lincoln Laboratory.