Modern battlefields require far greater local data processing and connectivity than battlefields of the past. These needs are driven by the increase in the use of Inescapable Data collection devices that produce abundant real-time and critical battle and intelligence information. Just as business is being transformed by a plethora of new data sources, so too are military decision-making processes being transformed. Battlefield operators now have access to much "wider" (sources and types of) data that they can factor in to decisions that must be made locally as they integrate the distributed data with myriad other streams of knowledge. "The rate of increase of the data being generated is greater than that of Moore's law (doubling every 18 months), so there is a need for more processing and a need for large systems," describes Jay Bertelli, CEO of Mercury Computer Systems, supplier of massive embedded processing systems. "In the military, commanders want to keep the data 'in theater.' The data streams off of an unmanned aerial vehicle (UAV), but they need to process it nearby and send it back to someone who can act on it, which drives a need for super-computerstyle processing power in very constrained and harsh spaces, in addition to highly connected networks." Barry Isenstein, general manager and vice president of the Defense Electronics Group of Mercury, explains why battlefield processing requirements are so extreme:
What's a teraflop? Military processing typically requires a great deal of signal processing that uses computerized mathematical calculations known as floating-point operations, or flops. Years ago, the performance of a super-computer system was measured in megaflopsone million floating-point operations per second. Today, systems that perform in the gigaflop range (1,000 x more than a megaflop) are needed for most of the sophisticated signal processing operations. The need for teraflop-scale processing power is becoming common as well (an additional 1,000 x in performance or a million times more performance than a megaflop, if you can imagine). All this computer power is being squeezed into ever-smaller spaces. According to Isenstein, "There is a major Lockheed Martin program, the Joint Common Missile program. Its aim is to replace a lot of the current air-to-air and air-to-ground missiles used by the navy and army. The processing space allotted to us within the missile is extremely tight, with a similarly tight budget for heat dissipation." The processing power, although nowhere near a teraflop, is still staggering considering that it is embedded in a missile, which is essentially a single-use product. In real time, the system has to process massive streams of data from many onboard sensors as well as onboard radar and two or three other classified sources of real-time intelligence...in a missile. In addition, targets are not always as stationary as a power plant, for example. Some of the more high-value targets are movable and might be in motion while the missile is in flight. "These missiles have to be very intelligent; you want to hit the enemy missile launchers and not the school next door," adds Isenstein. Accurate target assessment just cannot be done with only one sensor consuming just one stream of data. For example, the missile's onboard radarpossibly the most modern Synthetic Aperture Radar (SAR)could determine that a moving object is a "vehicle," but it could not determine whether the vehicle is an armored car or a school bus. Additional data from other sources is needed to more finely tune the image. For example, adding in a hyperspectral image of the area could help to detect the kind of exhaust fumes (e.g., diesel) emanating from the vehicle, which would narrow down the possibilities as to the type of vehicle. Other sensor data, such as detection of whether the vehicle itself is emitting data, would narrow the list of possible vehicle types even further (because school buses do not emit military signals). In the past, military planners did not have real-time access to so many different sources of data and certainly never had the capacity to process the data into useful information in real time and in such tight spaces (e.g., as within the body of a missile). The ability to coordinate and fuse sensor data is now critical to success in the new battle zones. "The data coming off of a Global Hawk UAV requires a tremendous amount of interpretation," says Isenstein. "In order to dispatch an F16 to drop a payload, commanders need trained personnel who can do the data interpretationa very rare skill. The human is arguably the slowest link in the chain because of the amount of data to sift through. The more processing we can do in the air, the easier the understanding of the data." Analysis of historical data and trained computer programs that distill out the salient information can help to alleviate the problem of the shortage of trained personnel. As the military continues its never-ending quest to compress battlefield reaction time, some missiles will be designed with "loitering" capabilities. Loitering means that a missile can be fired at no target at all. It simply flies in holding patterns until digital imaging, radar, and other data sources point to a possible mission. In such cases, the missile's capability to process more data locally and derive intelligence from that data in seconds is critical. Network-centric warfare is at the heart of the military's new real-time theater operations. Prophet Ground and Future Combat Systems are new programs driving communication, intelligence, and surveillance down to ground-level vehicles. Isenstein continues:
The general trend is to move more processing closer and closer to the soldiers in the field. Therefore, the need is for more intelligence "around the edges" of the battlefield network; commanders can no longer afford the informational latency introduced when field units have to reach back to some distant central command center for directions. As importantly, field units currently do not have the communications bandwidth to send all the raw data back to central command. First, it must be decomposed and digested locally. "They can't make reliable targeting decisions with just one sensor's data. Is it a bus or an armored vehicle? Before you blow it up, you need to coordinate sensor fusion in the field," Isenstein adds. In the 1990s, a commercial technology used for tracking trucking-company vehicles was co-opted for military use. FBCB2 (Force XXI Battle Command, Brigade-and-Below) is a system used to support battlefield tactical missions. Specifically, FBCB2 provides real-time "situational awareness" for the commander, staff, and soldiers. In addition, it uses graphical displays and detailed target identification to provide a common picture of the battlespace. No more commanders and troops yelling coordinates to each other over the radio. The Iraq War exploited this new technology heavily; military networks were able to show battlefield conditions in real time. FBCB2 is a mobile system of networked computers, radios, satellites, and software. Data from a wide variety of sources, including GPS, is blended together to form detailed knowledge maps of a combat situation, including landscape and positions of soldiers and targets. The knowledge is made available to all levels of the command chain, from field combatants to commanders back at central command. FBCB2 leverages Inescapable Data sources and its impact cannot be understated. It empowers the battlefield decision-making process through massive data correlation and fusing. It also breaks down the traditional military knowledge realms by allowing knowledge and decision making to flow up and down the chain of command. Operations of all branches of military service are developing a huge dependency on key Inescapable Data themes: real-time information enabled by the use of Inescapable Data collection devicesboth on the ground and airbornelocally processed into information with tightly compacted super-computer power, and coupled with many other streams of information to derive a new battlefield advantage.
|