Software engineering: a Defense Department bugaboo

When it comes to developing the complex and costly software to control the snazziest new weapons and communications systems, the Department of Defense clearly needs some help.

The Government Accountability Offiice said today that about half of the DoD’s weapons and communications programs it examined had at least have had at least a 25% growth in their expected lines of code since system software development started.  That fact translates into millions of dollars in cost overruns and project delays, the GAO said. 

The GAO highlighted one program in particular, the Army’s Future Combat Systems (FCS) communications program where the number of lines of software code has nearly tripled since development of the system began. Software has in part caused the program’s costs to hit over $128 million in 2006 from $88 million in 2003 a 46% increase, the GAO said.

The FCS program is an integrated family of advanced, networked combat systems; unmanned ground and air vehicles; and unattended sensors and munitions being built by primary contractor Boeing. FCS features 14 major systems and other programs along with an overarching network for information superiority and survivability, the GAO said.

Since the GAO examined FCS last year, the Army deleted four systems and made several other adjustments to the FCS development program based largely on budgetary constraints. The Army also reduced the annual FCS production rate and stretched out the production phase by about 5 years, also due to budget limitations. As a result, total cost estimates for the program were slightly reduced, the GAO noted.

The Army’s FCS development cost estimate depends on a number of assumptions. Historically, programs using such assumptions tend to underestimate costs. Program officials stated they will not spend more in development than the current value of the FCS development.

Software issues were discussed as part of a wide-ranging, often scathing, 205-page report assessing the status of 72 key Defense Dept. weapons and communication systems.

The DoD’s investment in weapon systems is humongous.  The GAO said weapon acquisition programs are now at their highest level in two decades.  The DoD expects to invest about $900 billion over the next 5 years on development and procurement with more than $335 billion, or 37%, going specifically for new major weapon systems. Every dollar spent inefficiently in developing and procuring weapon systems is less money available for many other budget priorities—such as Social Security, Medicare, and Medicaid.

But efficient software development is critical as modern weapon systems are increasingly more dependent on software than anytime before, and the development of complex software represents a potential leap forward in operational capability for any number of DOD defense acquisitions, the GAO stated.

Technological advancements have even made it possible for software to perform functions once handled by hardware. As this demand for software grows, the use of disciplined, structured development processes that measure, manage, and control software requirements is essential to delivering software-intensive systems on time and within budget, the GAO said.

The GAO said that one key metric used by leading software developers is to measure changes to the amount of software code developed for the program. Size metrics, such as lines of code, are used to compare the amount of software code produced with the amount originally estimated. Changes to the size needed can indicate potential cost and schedule problems.

The GAO said it has found cases where programs continue to have difficulties in managing software development for weapon systems. “We found that many programs are very susceptible to discovering costly problems late in development, when the more complex software and advanced capabilities are tested. Of the 33 programs that provided us data about the overlap between system development and production, almost three-quarters still had or planned to have system demonstration activities left to complete after production had begun. For nine programs, the amount of system development work remaining was estimated to be over 4 years. This practice of beginning production before successfully demonstrating that the weapon system will work as intended increases the potential for discovering costly design changes that ripple through production into products already fielded, and usually require substantial modification costs at a later time,” the GAO said.  

Making design changes to achieve reliability requirements after production begins is inefficient and costly. For example, during flight tests in 2007, the Air Force’s Joint Air-to-Surface Standoff Missile failed four times in four tries, resulting in an overall missile reliability rate of less than 60% despite being more than 5 years past the production decision. The failures halted procurement of new missiles by the Air Force until the problems could be resolved.  

The GAO also noted the Marines’ Expeditionary Fighting Vehicle program experienced software growth during system development, and it identified software test failures as a factor affecting the system’s reliability.  The Marine Corps’ EFV is designed to transport troops from ships to shore at higher speeds and from longer distances than current  Assault Amphibious Vehicle.

Software issues were only part of the GAO’s report which took a look at the way such high-profile weapons systems, such as the F-22 and a host of high-technology communications systems, were being bought, developed and implemented.  The overall development of most systems needs to be greatly improved the GAO said. 

The GAO did say that the DoD has recently taken actions that could help move the department toward more sound, what it calls knowledge-based acquisition processes. For example, a new concept decision review initiative, guidance for determining acquisition approaches based on capability need dates, and the establishment of review boards to monitor weapon system configuration changes could enable department officials to make more informed decisions in the early stages of a program and better match program requirements and resources, a key first step, the GAO said.   

Layer 8 in a box

Check out these other hot stories:

Researchers unravel mystery of lightning diversity 

DARPA wants a quantum entanglement revolutionUS lacking secure national space strategy, GAO says 

FTC takes aim at  prepaid calling card giant 

Pentagon awards $15.7M to advance university science and engineering research 

International cyber-cop unit girds for uphill battles 

Join the Network World communities on Facebook and LinkedIn to comment on topics that are top of mind.

Copyright © 2008 IDG Communications, Inc.