Congress orders F-35 Software Plan


Congress ordered the Pentagon to establish an independent team consisting of subject matter experts to review the development of software for the Joint Strike Fighter program.

The House Armed Services Tactical Air and Land Forces Subcommittee asked the Pentagon to submit a report by March 3, 2014 as part of the committee’s markup of the 2014 defense budget. The F-35 software program has served as one of the largest challenges for program engineers to keep on schedule.

“The committee continues to support the F-35 development and procurement program, and believes a software development review by the Department will ensure that the F-35 program remains on schedule to provide a fifth generation capability in support of our national security strategy,” the Congressional language states.

The JSF program developmental strategy is, in part, grounded upon a series of incremental software “drops” — each one adding new capability to the platform. In total, there are more than 10 billion individual lines of code for the system, broken down into increments and “blocks,” F-35 program office officials explained.

“Software development remains a focus area of the joint program office. We have a solid baseline and we need to be able to execute on that,” said Joe DellaVedova, F-35 program office spokesman.

Software drop 2B is undergoing flight testing at Edwards Air Force Base, Calif., and Naval Air Station Patuxent River, Md; software Block 2B builds upon the enhanced simulated weapons, data link capabilities and early fused sensor integration of the Block 2A software drop, DellaVedova added.

“With Block 2B you can provide basic close air support and fire an AMRAAM {Advanced Medium Range Air to Air Missile}, JDAM [Joint Direct Attack Munition] or GBU 12 [laser-guided aerial bomb]. This allows the plane to become a very capable weapons system,” he said.

Overall, DellaVedova said the F-35 program office has been making substantial progress. Software drop 3I, which is a technical refresh of Block 2B, is slated to by ready by 2016.

“This is complicated and labor intensive work but this has leadership focus from industry and government to deliver on the promise of the F-35. With its stealth and its enhanced situational awareness, the F-35 will provide a backbone for our forces for generations to come. Our progress continues at a slow and steady pace and we are focused on completing things within the schedule and budget we’ve been given.”

About the Author

Kris Osborn
Kris Osborn is the managing editor of Scout Warrior.
  • tee

    This should be fun, just think Microsoft Vista and how well that worked out for everyone. As a coder for over 20+ years, they are trying to make new code work with very old hardware ( weapons ) in some cases that were designed in the 80’s & 90’s. It may work with the new latest state of the art weapons built with more modern code & circuit boards, but the older 8/16 bit code & boards will be fun.

    • Menzie

      I understand the point you are trying to make but maybe you used the wrong example of Vista. I have had it since day one of release and have it on all my laptops and my desktop. I have had nothing but good come from it. So maybe try another example.

      • oblatt1

        LOL you obviously have very low expectations. Vista is widely acknowledged to be a dog, even by Microsoft

    • greg

      I don’t think they are having weapons integration issues at all as of right now. The main issue I see is with the sensor fusion which is all new hardware.

      The more graphics and video the more buggy, which is evident by every advancement is OS released to date. For example Mountain Lion is much more buggy then Snow Leopard…Because they are throwing things in.

      I don’ t think the old weapons would be a problem first off…One of the weapons they mentioned will probably be one of the oldest used the GBU-12,how old is that again? Secondly that is the purpose of plugins.

      Honestly it seems like the F-35 is farther along then most of the naysayers would like to admit at this point.

    • Brownie

      Yup. We need to expose the most secret of technologies to the Red Chinese once again. Let’s put some more fingers into the espionage pie. Smart.

    • blight_

      If it makes you feel better, the F-35 is mostly new code hardware. What I think is funny is that it’s talking to hardware much older than it, and perhaps originally designed to talk to software written in Ada.

      Of course, all the older missiles have been shot off, so most of the “old” weapons are on newer blocks that have steadily improved in electronics.

    • Restore Palestine

      Vista was a big flop. Slow, buggy, incompatible with many peripherals … It’s an eye-candy and not much else, just like the F-22 and F-35.

      Old hardware tend to be slow when running latest programs mainly because of processor speeds and smaller RAM. As long as the native compiler is current with recent enhancements and new functionality, problems in the program are often logical errors of the programmer, not technical problems of the hardware.

  • CharleyA

    Yada yada yada

  • oblatt1

    You have to remember that Lockheed deliberately chose one of the most bug prone languages to develop the 10 billion lines of code in. They then hired as many cheap inexperienced developers as they could find. Even a conservative estimate will be that there are 50 million bugs in the code.

    While this means the mean time between failure of major systems in the F-35 is 15 minutes it also ensures decades of revenue for Lockheed. Its a tradeoff – the future of Lockheed for the future of American air-power.

    We can be confident that the F-35 will never work as advertised, because it simply was never designed to.

  • brownie

    Folks can rant and rave, just like they did against the M1, Bradley, and F-22. This acquisition is a done deal. Move on.

  • Taylor

    Oblatt1 must be Tokyo Rose reincarnated.

  • XYZ

    On the bright side: Picture = saved. F-35s flying in formation is pretty sexy lookin’, and that’s coming from a guy who loves to make jokes at Lockheed’s expense.

  • Daniel Tebar

    Does anyone have info on the methodologies used for the development of these advanced avionics systems? Is UML in the mix? Is it OOA/D?

    • SMSgt Mac

      Heh. You ask an intelligent question beyond the ken of the Anti-JSF crowd. If any of these doomsayers were able to answer they wouldn’t be doomsayers. UML? Yes. Rational and Rhapsody are the biggies. In the early days, requirements definition was in Rational Rose and Rhapsody was for code building, because that was where the strengths of each were. Over the years, with version updates and software business acquisition events moving all of it under IBM (I think), it may all be Rhapsody now. (But don’t tell the locals, they still envision legions of gnomes hand-coding. Grab a beer and watch the Engineering Illiterati do their tribal dance.)

      • oblatt1

        On the other hand some of us talked to Cody and Ivar back in the day and don’t think much of where Rational went. A failed technology for a failed design sounds about right for the F-35

      • Daniel Tebar

        Thanks for the succinct answer Mac. In my 20-plus year career, only the best companies take the time to do actual design. Also, according to what I have read, it looks like the French, and Germans do software the right way. Do you share this perception?

      • Restore Palestine

        SMSgt Mac

        Can you please answer the intelligent question beyond the ken of the Anti-JSF crowd?

  • Brian

    Bugs aside, the biggest problem is still one of false expectation. Helmet Mounted Sights can work *very* well when kept simple & in the real-time domain where initial target / missile cueing information is simply overlaid – but that’s not how the JSF works. It works by creating a “virtual reality middleman” inside the helmet which “represents” the outside world, and the pilot interacts with the middleman, the supposed “advantage” of which is to allow the pilot to “see through the canopy” (unnecessary if the jet had decent visibility in the first place), etc, but the disadvantage is that it’s no longer in real-time – it’s massively “laggy” and plagued with stuttering / micro-stuttering, sensor lag & input delays.

    “The JSF doesn’t need to dogfight or see out of the canopy because it’s so clever and will rely completely on EO-DAS + HMS” is already looking very shaky given the “not so clever” continuous targeting issues, latency and jitter experienced with both the EO-DAS and HMS. Sometimes IRST contacts have “dropped off” for no reason, other times it’s taken far longer for it to recognize an incoming large IR target as a threat. The onboard computer is not powerful enough and is already plagued with over-heating issues (sometimes locking up completely). It cannot be replaced without major cost and yet more delays.

    And the AESA radar + super-computer + A2G targeting pod electronics all squeezed into the nose-cone is like sticking a 20kw heater in there making it highly unstealthy to anyone with a decent FLIR / IRST…

    What the F35 pilot sees can be over 1 second behind real-time (huge in any dogfight). This is what the “latency” issues are all about and they are nowhere near as easy to fix as simply upping the GHz on the CPU. It would have been FAR easier to give the jet decent all-round visibility and build a simpler but more reliable HMS that doesn’t try and “stitch” a panoramic thermal display spanning multiple cameras, but rather simply overlay directional cues & estimated range of incoming aircraft / SAM’s.

    Weapons and data-link software issues will probably be fixed, but I absolutely would not like to be on the team that has to “fix” the unfixable latency on the HMS which is an inevitable side-effect of its core-design. Processing vs latency – ask any real-time audio recording studio engineer…

    • blight_

      The problem is that LM is reinventing the wheel. The Israelis have HMD, as did the Soviets. We are puttering about with “new” HMD…and we should do so as a separate technology demonstrator program.

      Almost every doodad on the -35 is going to be researched and developed de novo, then bundled into the F-35 program. Pursuing parallel development of every program, and putting subcontractors in charge of their devices ensures maximum autonomy and perhaps better funding and cost controls, perhaps at the cost of inter-operability and an extended integration stage.

    • citanon

      I don’t understand why would it take a second to stick together 6 images when a modern GPU can render and display 4K scenes at >60fps with 10s of ms of delay?

      Image processing and graphics display is a perfectly parallelizable problem. Unless the noise artifacts processing is so onerous that they have to use dozens and dozens of passes, or the software is badly architectured, it seems that they should absolutely be able throw more silicon at it to get lower latencies.

      Once they get the software optimized, the DAS will probably work like a dream.

      • blight_

        I question if a milspec GPU is even available. It’s probable they are using a CPU to do a GPU’s work, which makes me remember the ’80s-90s and “Software Graphics”.

      • oblatt1

        One you have seen some of the military comms protocols you would no longer wonder.

      • blight_

        “The F-22A is […] the first design to fall foul of processing chip evolution outrunning the system’s development cycle, and the sheer complexity of the software creating major delays to production in its own right. The recently redesigned CIP 2000 configuration uses up to 66 COTS based Motorola/IBM PowerPC RISC (ie Apple Mac compatible) and Intel i960MX processor chips and is aimed at cost reduction and supportability, with a follow on upgrade planned to further increase computing power. Since the ‘G4’ variant, PowerPC chips typically include an embedded ‘Altivec’ short vector processor which is exceptionally well suited to signal processing tasks, as found in radar, comms and EW processing.

        The JSF is built around an evolution of the F-22A model, but much more complex in implementation due to the additional, and extensive, electro-optical suite and digital ‘soft’ cockpit. Its liquid cooled Integrated Core Processors (ICP) are intended to be a cheaper equivalent to the F-22A CIP, relying to a greater extent on COTS packaging technology. Like the F-22A, the JSF is expected to use FC-AE replacing the originally planned IEEE SCI/RT (a commercial flop) in the JAST Pave Pace model, supplemented by IEEE 1394b Firewire bussing (also used in Apple computers) in the Vehicle Management System (VMS). For SDD, the Mercury RACE++ Powerstream processor will be used for signal processing and I/O processing functions – this is a 9U VME format packaged multiprocessor, built around PowerPC RISC processors – essentially a bigger and faster cousin to the 6U VME packaged PowerPC processors now being used in F-15E, F/A-18E/F and F-111C Block C-4.

        The core avionic system, centred in the ICP and its software, will present some significant development risks. While VME packaged PowerPC hardware is now widely used, it has not been used on the massive scale of the JSF to date. The large number of interconnects, density of hardware, and the demanding thermal cycling and vibration environment has the potential to produce reliability problems, especially of the intermittent variety, in the ICP subsystem. This may not become statistically obvious until a good number of systems are operationally deployed – cyclic wearout problems in printed circuit boards and connectors often resemble the behaviour of airframe fatigue damage and will not manifest until some number of cycles is accrued. The F-22A’s Milspec hardened SEM-E packaged system was reported to have had a number of hardware reliability problems, initially misdiagnosed as software faults – the more complex and softer COTS derived ICP has the potential to do the same on a very much larger scale.”

        Trying to figure out the hardware for HMD at the moment, but it doesn’t look good.

  • Juramentado
  • Juramentado
  • citanon

    Are you sure it’s 10 BILLION lines of code, not 10 MILLION? Because I’ve never heard of any company putting together even 1 BILLION lines of code.

    Windows, for example, has about 50 MILLION lines of code…..

    • William_C1

      Yeah that should be million as opposed to billion. If the F-35 had 10 billion lines of code I would expect it to be smarter than all of Washington DC.

      • blight_

        It would also be the buggiest program ever.

        Does it also count subcontractor code or just LM’s?

      • Restore Palestine

        A well-written program of 10 lines is smarter than all of Washington DC.


      100 dollars per line of code. Works out.

      • Dfens

        Lockheed and Boeing both bill 8 engineering hour per line of code. Since their labor rates are about $300/hour, well, you can figure out the rest.

  • Restore Palestine

    Still 200 dollars cheaper than what lawyers cost (per line of written legal document)

  • Tom

    We had Microsoft products for a long time, and was always rebooting (Bill Gates)
    We retired and bought MAC products (they work). maybe the engineers building the most expensive jets in the world need to take notice.