May 2009 – Peter Cavill Interview
Peter Cavill is General Manager, Military/Aerospace Products at GE Intelligent Platforms.
Here, Peter talks about the military/aerospace market and the trends and challenges he sees.
It is unfortunately true that what drives the military/aerospace business is crisis. That can be a war, such as we’re seeing in Iraq and Afghanistan, or it can be terrorism, for example. The fact is that there are many places in the world where there is significant instability – an instability which may lead to war. Beyond that, we see ethnic cleansing, guerrilla warfare, widening access to nuclear capability and the proliferation of biological weapons. Trafficking in drugs creates tensions that can escalate. The mean time between crises seems to be decreasing.
Armed forces in many countries are not only actively engaged around the world, but also on constant stand-by in case it is necessary for them to intervene – and the equipment and systems they will deploy need to be up to date, reliable and in an advanced state of readiness. At the same time, equipment and systems that have been damaged or destroyed need to be repaired or replaced. The US government sees it as imperative that it maintains – and, ideally, extends - its military technology advantage, which means that budgets for defense are remaining solid. The expended spend for 2008 is $672 billion, and the forecast for 2009 is that the US will spend around $711 billion – that’s an increase of around 8%. Meanwhile, we’re also seeing increasing international demand for improved intelligence and defense capabilities as countries prepare themselves for possible conflict – as well as for their roles in international peace keeping. Countries outside the US will, collectively, spend around the same as the US.
Various market researchers say that they believe that it will decline in the coming years, flattening off at a run rate between $500 billion and $600 billion between 2011 and 2018. However, it’s important to know how those spending figures are arrived at. The researchers divide their forecast into two. One part is described as the base budget, and that is forecast to remain flat at just under $500 billion from this year through 2018. The delta spend is made up of the ‘supplemental forecast’, which has grown steadily since 1998 and reaches its peak of an incremental $196 billion above base forecast in 2009. This supplemental spend can be considered to be expenditure that is driven by events – such as the wars in Iraq and Afghanistan – that is over and above the ongoing investment in defense capability. The researchers appear to be assuming that there will be a reduction in worldwide crises from next year – and that is reflected in their forecast. To us, it seems that a substantial reduction from current total spending levels in the coming few years is unlikely given world events today, and given the need for the US armed forces to spend money on upgrading and refitting equipment used in the war zones.
Perhaps the most notable trend is that, as military programs become larger, more sophisticated and more complex, the prime contractors who supply the armed forces increasingly have to focus on the high level, strategic issues. The impact of this is that they can no longer invest time, effort and money in putting together the individual subsystems that go to make up the complete systems – so they’re turning to suppliers and asking “Can you do this for us?” It is becoming the case that suppliers can no longer expect to just provide products that a prime contractor or OEM will integrate, test and qualify – suppliers are having to create a complete in-house systems capability that will, in all likelihood, integrate not only their own products but products from third party vendors. This change in emphasis is tending to favor the larger suppliers with the broader product ranges, the greater in-house resource and the additional expertise and experience – suppliers like GE Intelligent Platforms.
Not necessarily. Those who specialize in niche markets will undoubtedly continue to thrive. For some, partnership agreements will be the way forward. But it’s hard to see a positive outlook for a small company that isn’t highly differentiated in its product offering. It will almost certainly become a target for – or seek - acquisition.
We have certainly seen consolidation in the number of broad line suppliers of embedded computing solutions to the military/aerospace industry. Something like 58% of the market is served by no more than five major suppliers – and GE Intelligent Platforms is the largest of those. For sure, this simplifies procurement for the major prime contractors – they have, in effect, a ‘one stop shop’ for the majority of their requirements, and when all products come from a single supplier, integration and interoperability issues become easier to manage.
We’re definitely seeing a great deal of focus on upgrading legacy systems – of which the armed forces have, inevitably, a growing number. Many of these legacy systems are based on the VMEbus architecture, and military customers today are almost spoiled for choice – whether to continue with VMEbus, or whether to migrate to VXS or VPX. It’s possible to add significant performance to older systems at relatively little cost. If there is to be any reduction in the growth of the US defense budget, we expect it to affect those programs that are at the forefront of technology: money will continue to be found for retrofit programs to maintain today’s capability.
We’re also seeing increasing focus on miniaturization – making systems and subsystems smaller, yet without sacrificing processing and bandwidth capability. That focus is being driven by a major shift towards unmanned vehicles – whether ground, air, sea or undersea – and giving those vehicles more and more autonomy. The challenge is to put the maximum possible amount of computing horsepower in the smallest possible space, yet protecting it from harsh environments and keeping it cool. That’s one of the reasons we’re so confident in the future of 3U VPX – because it has demonstrated that it’s ideal for that type of application. There is also now more emphasis on ‘man packs’ – sophisticated computer systems that can be easily carried around by soldiers on the battlefield.
By far the largest consumer of funds is what is called C4ISR – that stands for Command, Control, Communications, Computers, Intelligence, Surveillance and Reconnaissance. In other words, the focus is on information gathering, information interpretation and information distribution. A significant investment is also being made on radar and sonar – technologies designed, as much as anything, to protect the armed forces’ assets. Mission computing is also receiving substantial investment. It is defined as the subsystem, often mounted in a vehicle, which captures sensor data and interprets it either for action by a human crew or for autonomous action – and it is forecast to receive only slightly less funding than radar and sonar. Perhaps of most interest, though, is what is being called embedded training. This recognizes that troops often face prolonged periods of ‘down time’ in a war zone – and this down time can be put to productive use if the onboard systems in, for example, a tank can also be used for training. The emphasis is on using the identical controls in a ‘live’ environment with action being simulated as realistically as possible.
What we can say with certainty is that systems are becoming more complex. The computing power at the heart of the system needs to be substantial in order to accept the input from a digital processing subsystem (which has, in turn, processed the input from the A/D converters attached to a range of sensors), to manage ‘real world’ I/O, to drive the graphics subsystem which forms the human decision support interface to the system, and that manages communications across the battlefield.
One of the biggest challenges is the mismatch between the electronic components that the commercial market requires, and the components that the military/aerospace market requires. There was a time when the military drove the demand for electronics – but now it represents only a very small fraction of the demand. The figure is less than 0.1%. The problem arises from the fact that commercial applications need the very latest and greatest technology, and that time to market, and time in production before phase out, are incredibly short. Military programs, on the other hand, are designed and deployed over a much longer period of time. The time from design to production, for example, may mean that a component that was state of the art at the outset has been superseded – perhaps even gone end of life – before production starts. Production itself is likely to be over a period of years, rather than the period of months which is often the case for consumer electronics. And where consumer electronics may be in use for months, or perhaps a couple of years, military equipment can remain in use for decades. A commercial product, from initial concept design to withdrawal from the market, will probably be less than two years: for a military product, that timescale is likely to be forty years or more.
We’re also seeing shortening lifetimes for many application-specific components such as graphics processors, Ethernet switches and so on. The impact of this is that manufacturers like GE Intelligent Platforms are having to invest heavily in programs that protect our customers from, or mitigate the effects of, component obsolescence. In fact, this is something we’ve been doing for many years through our ‘Product Lifecycle Management’ program: the rest of the industry has yet to catch up.
The situation has been made more difficult by the RoHS directive – the European Reduction of Hazardous Substances legislation that requires that lead, and other potentially harmful materials, no longer be included in manufacturing processes. This has had a substantial impact on the electronics industry because of the historic use of lead in solder. Manufacturers are being forced to discontinue the production of parts that contain lead – and this is increasing the pace of obsolescence. That means increased inventory costs as manufacturers are forced to make last time buys of components – and the shortage of components is leading to a rise in counterfeit components, forcing manufacturers to invest time, effort and money in detecting and preventing these. Moreover: some defense OEMs will not accept components that use lead-free solder, while others will only accept components that are fully RoHS-compliant.
It certainly is. Historically, the embedded market – and particularly the military/aerospace embedded market – has been a major user of processors based on the PowerPC architecture. It’s estimated that around 65% of all the VMEbus and CompactPCI boards in use in defense applications use this architecture. The problem is, the Freescale 8641D is the end of the road for the E600 core: future designs from Freescale will be based on the E500 core. The good news is, the E500 core dissipates less power: that’s becoming an increasingly important consideration as we try and fit more computing into smaller spaces. The bad news is, it also offers lower performance – at a time when the demand for processing performance has never been greater.
It looked as if the proposed 1682M processor from PA Semi would provide a good way forward, and we were planning to base some new designs on their technology. However, the acquisition of PA Semi by Apple means that the 1682M is very unlikely to ever become a commercially-available component.
The question then arises: can Intel and/or AMD fill the gap? Certainly, Intel’s commitment to the embedded market has grown in recent times, and the company’s roadmaps show future generations of higher performance, lower power processors, and long term support for these – something that’s critical for the embedded military/aerospace market. Ideally, though, they’d consume less power: it may take some kind of breakthrough in cooling technology for Intel silicon to become a really viable option. AMD represents an interesting option too, given the company’s historically greater focus on the embedded market.
The big question in all this, though, is: what happens to the huge investment in legacy PowerPC code? No-one really knows the answer yet, but it could be a hard question that the industry needs to face up to.
The great thing about VMEbus is that it has provided a reliable standard architecture for the embedded computing industry for more than a quarter of a century. It’s been able to do that by constantly reinventing itself, taking advantage of new technologies and responding to new applications and environments. There is no question that systems based on VMEbus will continue to be sold and installed for many years. However, it seems likely that those systems will largely be deployed in programs that already have a substantial commitment to VMEbus: there will be some new programs that will use it, but these are likely to become increasingly few.
The reason for this is that, as good a job as VMEbus has done in improving the performance it delivers, alternatives are becoming available that leverage the VMEbus ecosystem and provide a painless way forward – but that can deliver substantially more performance. VXS is one of those architectures – but I’m thinking more of VPX. Both are certainly gaining traction in the market – the MEADS missile radar uses GE Intelligent Platforms’s quad-8641 VXS boards in a new multi-board digital signal processing system, for example – but VPX, especially in its 3U form, is what’s really catching the eye.
The reason for that is that 3U VPX delivers outstanding performance from a compact board. Historically, military/aerospace applications that were space-constrained had to live with the performance compromises of VMEbus and CompactPCI in their 3U forms. 3U VPX delivers the optimal balance of board functionality and compactness, with excellent I/O and communications capabilities – and it is inherently easier to cool, of course, than a 6U board. We’re very excited about the 3U VPX opportunity, and the product announcements we’ve made over the last year and that we expect to make in the future reflect that excitement.
The other standard that’s trying to position itself as being appropriate for military embedded computing applications is MicroTCA. Thus far, it’s been applied mostly in telecommunications applications, but it unquestionably has some characteristics – it’s high performance, scalable, compact and lends itself well to fault tolerant systems – that are making it attractive to other industries. There’s also a plan to create rugged MicroTCA – something that would be vital for success in the military/aerospace market.
At GE Intelligent Platforms, we have an extensive array of MicroTCA products and substantial in-house expertise with it. If that’s what the defense market wants, we’re ideally placed to provide it – much better placed than our competitors. I’m not so sure, though, that military/aerospace customers are really ready to embrace another standard unless it offers truly compelling advantages over what they already have – and I’m not convinced that MicroTCA has those compelling advantages.
No, it isn’t. The COTS paradigm is still a valid one.
To understand where we are today, it’s important to look back at 1995, when US Senator William Perry originally mandated the use of COTS for military programs. He was, in part, responding to the growing frustration of the American taxpayer that huge amounts of money appeared to be being wasted on developing military-specific versions of commonplace items, like hammers and ashtrays. He was also very conscious of the fact that the US spend on military electronics threatened to get out of control. He wanted to focus the minds of the armed forces, and the companies that supplied them, on the need to save money – and he believed that the way to do that was to force them to acquire commercially available products wherever possible.
His thinking was that this would allow the armed forces to take advantage of the more rapid take-up of new commercial technologies. He believed that the huge investment in research and development could be better carried by commercial suppliers. He could see that competitive forces, and economies of scale, would tend to drive product prices down. He could see how the armed forces could leverage a continuing flow of new products that would enhance the performance of existing products. And he believed that moving the responsibility for the long term support of products from the military and its suppliers to commercial suppliers could deliver extensive savings.
Thus it was that COTS was born. COTS was, in 1995, a fact – it was about commercially available products being bought off the shelf. But it was also about a principle – the requirement to save money.
What has happened since 1995? Back then, military computing – in fact, computing in general – was, relatively, much simpler and more straightforward. Since that time, though, military applications have become substantially more diverse and more sophisticated. New technologies – in processing, in memory, in communications, in graphics – have allowed military imaginations to envisage applications that were once unthinkable. And at the same time, the military has accumulated a vast quantity of legacy systems – many of which it continues to refresh with new technologies.
With this huge diversity, the chances of any one supplier having exactly the right product at exactly the right time are slim. It’s even less likely that a customer will find more than one – so the chances of competitive bidding are minimal. So what is the customer to do? Compromise the application? That’s unlikely. Typically, he’ll search around all the COTS vendors to find the products that come closest to what he wants. Then, he’ll issue a request for proposal to those vendors, describing a COTS product that exactly fits his needs – and he’ll consider only the bids that are technically compliant.
The COTS vendors, in turn, will work out how they can make the modifications to their standard product in order to comply. They will, in effect, be bidding custom solutions – but based on an existing design.
This situation is fairly typical. Does that mean that COTS no longer exists? No, it doesn’t. If you think about what COTS was designed to achieve, those goals are still being achieved. The customer is leveraging the latest technology. He is getting competitive bids, thus bringing the price down. The engineering time, effort and expense to develop the product has been absorbed by the vendor. A roadmap of future technology refresh opportunities exists. And the vendor retains the responsibility for obsolescence mitigation and long term support. The customer is achieving everything that William Perry set out to achieve when COTS solutions were first mandated.
We expect that it will remain a strong, stable market, driven on the one hand by international instability and on the other by the need to maintain or attain technology superiority. Vendors like GE Intelligent Platforms will continue to play a vital role, because we insulate the OEM or prime contractor – and thus the end customer, the armed forces – from shifts in technology, from rapid changes in component technology and supply, from the need to manage obsolescence and from the requirement to deal with the integration issues that arise with emerging standards. It’s a challenging market, but we know that success will be driven by win-win relationships between ourselves, our customers and their customers.