As in previous years, the Navy League's 2008 Sea, Air, Space Symposium was a paean to the central fact of the ongoing computer age, Moore's Law. Formulated in 1965, it states that the unit price of computing power halves every 18 months. At least up to now, that has usually meant that computer power doubles every 18 months. Over the 20- or 30-year life of a ship or airplane, that results in an immense difference between what computers could do at the beginning and what they can do when the platform is modernized or retired. If you attend the show often enough over the years, the impact of Moore's Law becomes obvious.
At one time the commercial world did not need (or realize it needed) enormous computing power. Military computers were on the cutting edge. Although that cutting edge kept moving, it took many generations of Moore's Law for the commercial world to catch up. That the military procurement cycle was grossly outpaced by Moore's Law did not really matter. The computers built around 1960 for the first versions of the Naval Tactical Data System (NTDS) were still being used, in somewhat improved form, two decades later. A standard computer conceived a few years later (UYK-7) was still in service in the 1990s, although by then it was often superseded by the next-generation UYK-43.
Now you have to look hard to find any such machines. Once the commercial sector found that it needed massive computing power, its standard machines fed on Moore's Law. The period of transition was painful. Captains comment ruefully that the laptops carried on board by junior crew members have more computing power than all of the ship's standard military machines.
Limited power, moreover, had real consequences. That was partly because so much now depends on the picture used as the basis for decision-making. As the Navy moves more and more toward operations in complex areas like littorals, the picture becomes similarly complex, and it takes a lot more computer power to create and display it. If this seems abstract, remember that one contributing cause of the 1988 shootdown of an Iranian airliner by the USS Vincennes (CG-49) was that the ship's UYK-7 display computer could show only the centerline of the corridor through which civilian aircraft were supposed to fly. The Iranian Airbus was off that centerline, but inside the corridor. Those in the vessel's combat direction center knew that their picture was incomplete, but as tension mounted they naturally concentrated on the picture, not on caveats in the backs of their minds.
That off-centerline airplane was potentially threatening, so the Vincennes's captain asked for more information. Due to an entirely separate error (an unintentional merger of two Link 11 nets, and sloppy net practice on the part of other ships), the answers he got convinced him that the airplane was diving to attack and he decided to fire first. Almost 300 innocent people died. No computer can repeal Murphy's Law, but for decades we have usually fought (at least in the Navy) on the basis of pictures we assemble using our combat direction systems. The better the picture, the better the basis for combat decision-making.
Out with the Old . . .
For some years the challenge has been to design naval systems that can exploit the flood of new, commercially-based technology without requiring ships be torn apart every 18 months or so. The submarine force led this trend with its ARCI (Acoustic Rapid Capability Insertion) program. ARCI began when the submariners realized they could not keep buying more and more powerful or massive sonar systems, partly because they could not keep buying larger submarines. They discovered that they were barely touching the potential of existing sensors. Better signal processing could buy the acoustic capability they wanted at a reasonable price, hence the "A" in ARCI.
Much of the trick was also to buy enough communication capacity, between sonar and processor, to exploit expected processing improvements. Thirty years is 20 Moore's Law generations, a factor of a million in computing power. At one time ships were wired for NTDS data rates, which meant about 2,000 bits per second. A factor of a million would raise that to two billion bits per second. At the very least, the wiring has to follow essentially commercial standards, Ethernet being a good example.
Now the submariners talk of buying "state of the practice" computers, not at the cutting edge, but at a stage at which they are proven and reliable. From their point of view, which is probably universal, a given computer is useful through several generations of Moore's Law, because when it is bought existing software does not fully use its capacity. Thus they run parallel programs of new software and (more slowly) new hardware.
ARCI was what would now be called an open-architecture program, because whatever software it sponsored had to run not only on that year's computer, but also on computers to be bought two or three years later (and, if the project was to be affordable, on computers to be bought generations later). Another way to say this is that since the end of the mil-spec computer world, defense has had no choice but to ride the wave of commercial computer development. The commercial world is many times the size of the defense world. We in the defense world do not even exercise enough power to nudge it in a desired direction.
This is all wonderful, but many of the devices that should be wired to those high-speed connections were designed in the NTDS era. So open architecture for existing ships or even for existing weapons and sensors must somehow connect devices that like very slow data rates to ultra-high-speed wiring. One company at the show demonstrated how such challenges can be met. When the German Navy decided to deploy to the Middle East, it wanted to install the new MIASS decoy system. Conceived for the latest German warships, MIASS is designed for installation in the Ethernet used in open-architecture command systems. The ships scheduled for deployment had the German equivalent of NTDS. A small U.S. company, Sabtech, found a way to provide a gateway between the ships' NTDS networks and the Ethernet-adapted MIASS. The device is now fitted to all German frigates and mine countermeasures ships.
As with many digital devices, it was developed in a matter of weeks, because it uses both existing hardware and software concepts. According to the company, the U.S. Navy is now evaluating it for the reverse, adapting existing equipment to the Ethernet the Navy uses in open-architecture versions of systems such as Aegis and the SQQ-89 ASW system. For example, the company's computer cards can turn the standard SLQ-32 countermeasures system into a node on an Ethernet. SLQ-32 was conceived as part of an NTDS system, so its language is NTDS; the card translates that into data that can go via an Ethernet.
Other companies showed open architecture versions of existing sensors and other systems, such as the big SPS-48 radar. The submariners have already demonstrated that Moore's Law can spectacularly improve sensor capability; for some years now the same ideas have applied to many radars.
Hidden Benefit?
Remarkably, the show indicated how little this potential is currently understood. Underwater, applying Moore's Law has pushed back attempts by our potential enemies to make their submarines stealthier. Before ARCI, there was a definite feeling in the submarine community that passive sonar was reaching its limits. The choices seemed stark: either buy larger submarines that could accommodate larger arrays, adopt radical new array technologies, or find some way to go active without compromising our own submarines' stealth. ARCI showed that none of these dire paths had to be pursued; what existing sensors already collected could be subjected to far better processing.
Anti-radar stealth is the radar equivalent of submarine silencing. In neither case can the platform gain much stealth after it has been designed, because so much of its capability is inherent in its shape or in its most basic internal arrangements (such as rafting, for a submarine). ARCI shows that the capability of the sensor trying to defeat that stealth is far less rigidly fixed. In the case of radar, Moore's Law can also support sensor netting, which makes it possible to exploit the fact that radar cross-section generally cannot be reduced equally for all aspects: a stealthy airplane will be fleetingly visible to each of several radars. Combining their detections, as in the current Cooperative Engagement Capability (CEC), produces a viable target track.
As it happens, Saab in Sweden has tried exactly this approach. It developed its own equivalent to CEC. The question then is how to exploit the resulting picture. If the target is really stealthy, then it is unlikely that an approaching missile can detect it. Instead, the missile has to be aimed at an intercept point predicted on the basis of the track developed collectively by several radars. Ideally it will be very fast, so that it can approach the target as close as possible to the most recent detections. Saab seems to have developed a maneuvering hypersonic weapon (Mach 5) for exactly this purpose. This work was apparently experimental only (the radar system was cancelled), but it shows what would be required—and that stealth may well be a wasting asset.
Much the same can probably be said of stealthy ships like the new Zumwalt (DDG-1000). The ship's specifications are now about a decade old. Her degree of stealth involved enormous expensive effort. She will probably enter service about 15 years—ten Moore's Law generations—after the specifications were framed. Signal processors will be a thousand times faster. At the same time netting will become even easier than it is right now and it is already far easier than in, say, 1998. Will our enemies then be so stupid as not to follow this sort of logic? How much is stealth worth in a Moore's Law world?