Commercial computer servers help Navy ply the ocean's depths for hostile submarines

The Advanced Rapid COTS Insertion program first targeted towed-array processing aboard attack submarines, and now is expanding even to the world's most extensive sonar network—the Integrated Undersea Surveillance System.

By John Keller

The most stealthy and deadly weapon in the world is the submarine. It can operate quietly and unobserved throughout most of the world's oceans to threaten land targets with cruise and ballistic missiles, ships and submarines with torpedoes, and national secrets with untimely disclosure. The submarine also carries with it the potent capability of conducting sensitive reconnaissance and even inserting commando soldiers where they can do the most damage. The mere threat of a submarine's presence can slow merchant ship traffic to a trickle and severely disrupt world economies.

It stands to reason, then, that U.S. and allied interests place a special mandate on the ability to detect any submarine, at any time, anywhere in the world. That is certainly a daunting order, especially given that advances in diesel-electric technology are yielding ever-more-quiet submarines, that a growing number of the world's navies—friendly and hostile—operate submarines, and that submarine weaponry is growing ever more affordable and lethal.

The task of monitoring worldwide submarine traffic is enormous, and common sense would hold that the U.S. experts and their allies are throwing the most advanced application-specific sonar processing equipment ever developed at the problem. Nothing, however, could be farther from the truth.

Over the past decade U.S. Navy sonar systems on hunter submarines, aboard surface vessels, and in shore-based sonar-processing centers have evolved from expensive application-specific processing equipment to common commercially available gear that offers stronger performance at substantially less cost than sonar systems of years past.

It is safe to say, in fact, that except for extremely specialized niche applications, the era of dedicated digital signal processing (DSP) hardware for U.S. Navy sonar applications is rapidly drawing to a close. Today's sonar systems, which use commercially available general-purpose processors and operating systems, represent an accelerating trend towards increased use of commercial off-the-shelf (COTS) sonar-processing equipment that is beginning to reach into almost every sonar system in the U.S. inventory.

The IUSS saw its beginnings in the early 1950s with the development of arrays of ocean floor-based hydrophones tied to land-based signal-processing sites. This system, which first was installed near the Bahamas in 1952, is called the Sound Surveillance System, better known as SOSUS.

The first SOSUS arrays, which ultimately were installed in strategically important basins in the Atlantic, Pacific, and other ocean areas, alerted Navy officials when Soviet submarines attempted to enter the Caribbean, Gulf of Mexico, or Hudson Bay. The idea was to detect the sounds of submarines, approximate their positions by triangulating among several different SOSUS arrays, and then deploy submarines, surface ships, or submarine-hunting aircraft to pinpoint their locations and maintain contact.

One primary goal of SOSUS was to develop the capability to alert Navy leaders when Soviet submarines were passing through the wide strait between Greenland, Iceland, and the United Kingdom—known as the GIUK gap—to identify Soviet submarines in transit between naval bases on the Barents Sea and the open Atlantic. A decade of listening and experiments passed with few, if any, dramatic results. But then things started to happen quickly.

For SOSUS, 1962 was a banner year as the system's hydrophone arrays and signal-processing centers in North Carolina and Barbados first detected the unmistakable sounds of Soviet diesel and nuclear submarines operating in the Atlantic. That same year, SOSUS developers hit pay dirt when the Barbados site detected the faint sounds of a Soviet nuclear submarine as it passed through the GIUK gap more than 2,000 miles away. SOSUS, the grandfather of the IUSS, had proven its worth.

SOSUS achieved new milestones in 1968, when the system's hydrophones and sonar analysts helped locate the sites of a sunken Soviet Golf-class ballistic missile submarine near Hawaii, and the site of the American submarine USS Scorpion, which sank southwest of the Azores.

SOSUS reigned supreme as the primary U.S. fixed-site anti-submarine warfare passive sonar system for three decades until a new strategic mobile sonar system called SURTASS joined it in the early 1980s. SURTASS, short for Surveillance Towed-Array Sensor System, operates from Transportation Auxiliary Ocean Surveillance ships (T-AGOS). It uses a mile-long towed-array passive sonar to detect submarines at long distances, and transmits its sonar data via satellite to acoustic processing sites on shore.

Researchers today are developing the Low-Frequency Active (LFA) sonar system for SURTASS to detect quiet submarines in shallow water. The LFA system, which is to use extremely loud low-frequency sound pulses to ferret out submarines hiding in noisy coastal waters, could be deployed early this decade if it overcomes challenges from environmentalists who claim the LFA poses a threat to marine mammals and other sea life.

By 1985 Navy officials coined the name Integrated Undersea Surveillance System, and placed SOSUS and SURTASS under the umbrella of the IUSS program. Also in the mid 1980s Navy researchers began turning to optical-fiber technology to bolster their ocean-floor sonar arrays. This direction gave rise to the Fixed Distributed System (FDS) and its successor, the Fixed Distributed System-COTS (FDS-C), which was developed in the mid 1990s.

The FDS and FDS-C arrays placed acoustic sensors on the ocean bottom, connected by optical fiber, and arranged in the shape of a large multi-prong fork or rake. This was a fundamental advantage over the older SOSUS, which placed hydrophones along only one wire. With the FDS multi-prong approach, technicians were able not only to detect and locate submarines, but also determine their range and bearing.

The latest development in the FDS program is to move the system to all-optical components—fiber interconnects as well as optical sensors. Engineers from the Northrop Grumman Corp. Navigation Systems Division in Woodland Hills, Calif., won a contract in July 2001 to use optical sensors, all-optical telemetry, and COTS components to the greatest extent practical to convert the FDS-C system to all-optical underwater components. With these advances in optical fixed-site sonar systems, Navy officials have placed the venerable SOSUS arrays on standby mode. That means that SOSUS data is always available, but is not always monitored.

One of the newest sensor arrays in the IUSS family is the Advanced Deployable System, otherwise known as ADS. The ADS is an outgrowth of the FDS program, except that it is designed to be portable and deployable quickly in ocean areas where it is most needed, such as the Persian Gulf, Red Sea, and Arabian Sea. The ADS is to beef-up undersea surveillance in shallow-water coastal areas where hostile submarines pose the most severe near-term threats.

The SDS and its companion Shore Signal and Information Processing Segment (SDS SSIPS), not only are helping Navy officials closely monitor submarine traffic in important areas throughout the world, but also offer the ability to automate a broad number of surveillance tasks also is helping Navy leaders consolidate shore facilities to reduce the number of shore sites and personnel. Together, the SDS, SOSUS, SURTASS, FDS-C, and ADS are called the Fixed Surveillance System, otherwise known as the FSS.

Signal processing for the IUSS has undergone several fundamental upgrades throughout the lifetime of the systems. Processing has evolved from simply having human sonar experts listen to sounds from the SOSUS hydrophone arrays, to complicated application-specific sonar-processing computers based on the Intel i860 DSP chip and the Motorola PowerPC microprocessor. IUSS signal processors also have evolved from military-specific systems to more generic open-systems architectures with as much COTS hardware and software as possible.

Today, however, the IUSS signal processing effort falls under the aegis of a much larger initiative in the U.S. Navy to standardize on one sonar signal-processing architecture for land-based processing sites as well as for deployed submarines, surface ships, and even submarine-hunting aircraft. What describes that overarching project to proliferate one COTS-based sonar signal processing architecture throughout the Navy is the Acoustics Rapid COTS Insertion program—better known as ARCI.

The ARCI evolved from a previous program called the Navy Common Acoustic Processor, or NCAP. The ARCI's original intent was to create a standard COTS-based sonar signal processor to replace the ageing AN/BSY-1, AN/BQQ-5, and AN/BQQ-6 sonar processors aboard U.S. Los Angeles-class attack submarines (SSN 688) and Ohio-class ballistic missile submarines (SSBN 726). It was to tap into the latest commercially developed digital processing technology to reduce acquisition costs, use the most powerful processors available, and reduce maintenance costs. Since the beginning of this decade the value of ARCI has grown such that now it is going Navywide.

Since its first deployment in 1997, the ARCI program's evolution has been in four phases—two that are completed, one that is in progress, and another that lies in the future, explains David Burgess, DSR's corporate vice president. "It was done in that fashion to get all the capabilities we could deliver early to regain the U.S. submarine acoustic advantage," he says.

The first iteration of the ARCI architecture in 1997 revolved around the VME backplane databus and single-board multiprocessing computers from Mercury Computer Systems Inc. of Chelmsford, Mass. Each of these Mercury single-board digital signal processors had four Intel i860 general-purpose microprocessors. The Intel i860 was designed as a general-purpose processor but found wide popularity as a DSP, much like today's Motorola PowerPC AltiVec microprocessor.

The first-phase ARCI architecture primarily was for processing sonar data from the TB-29 thin-line towed array on the 688-class attack boats. The TB-29 sonar, which represents substantial acoustic advantages over the old TB-23 thin-line array, requires ARCI-caliber processing to exercise its full capability, Burgess says. Part of ARCI phase 1 involved converting the old AN/BQQ-5E and Combat Control System (CCS) Mk 2 software from the Navy-specific CMS-2 programming language to C and C++, yet using the same algorithms, for the new ARCI hardware.

The next phase of the ARCI program kicked off in 1998 by adding processing power and user-friendly displays, Burgess says. This phase switched out the Mercury i860-based single-board DSPs and substituted Mercury G-2 Quad PowerPC VME boards based on the 200 MHz Motorola 2604 PowerPC microprocessor. Phase-two ARCI also featured the VX Works real-time Operating System from Wind River Systems in Alameda, Calif., and the Mercury Raceway high-speed interconnect for multiprocessing data exchange. That architecture used two boards for general-purpose processing and six to 11 boards for DSP work, Burgess says.

The third phase of the ARCI evolution is happening now, and may represent one of the most fundamental changes so far in the history of the program. Previous ARCI iterations relied on open-systems-yet-specialized collections of single-board computers, backplane enclosures, and software operating systems, the latest version uses off-the-shelf complete computer servers. The old approach was off-the-shelf at the component level, while the latest approach is off-the-shelf at the subsystem level; it is the difference between buying groceries and cooking dinner at home, or having the same meal catered.

The latest ARCI architecture revolves around the 8U Compaq ProLiant 8500 computer server, which is 14 inches high in a standard computer rack (one "U" is 1 3/4 inches high). The multiprocessor server has eight Intel Pentium III Xeon microprocessors, as much as 16 gigabytes of 100 MHz SDRAM memory, and runs the Linux operating system from Red Hat Software in Raleigh, N.C. "We are using commodity things now, not DOD-specialty things," says Jack Gellen, the ARCI program manager at Lockheed Martin NE&SS.

Mainstream commercial computers have become fast and reliable enough to process even the Navy's most complex signal-processing algorithms; systems designers no longer need to consider specialized or custom-designed processors for almost any sonar application, Gellen says. "As soon as mainstream commodity COTS is fast enough and dense enough for our application, we will use it," he says.

"There is a point where the technology reaches where your program is," Gellen continues. "In sonar, we are pretty much there."

Designers originally conceived the ARCI architecture with submarine deployment in mind, yet extending its architecture to IUSS sonar processing is not a huge technological leap. "Acoustics is acoustics," says DSR's Burgess. Application software for the IUSS is common, but not identical, to submarine sonar-processing code, Gellen says. Commonality between the two applications he estimates at about 80 percent. He cautions, however, "Surveillance and submarines have different sensors, so there are differences."

Each 2U server will have two 2.2 GHz Intel Xeon microprocessors with Hyper-Threading (HT) technology, Burgess says. Each server is an independent computer, with its own power supply, cooling fan, memory, and copy of the software operating system. Data will pass quickly between each 2U server over either Gigabit Ethernet or Fibre Channel, depending on the application, he says.

"For towed array we are using Gigabit Ethernet, and on some applications the implementation may be in Fibre Channel," Burgess says. "The deciding factor may be the cost to implement. We are looking for the most gigaflops per dollar within our heat-dissipation capabilities of our cabinets." Concerning the next generation of high-speed stitched-fabric serial interconnects, Lockheed Martin's Burgess says ARCI engineers have not made a decision. Instead, he says, ARCI designers will seek the most capability at the most reasonable cost.

These 2U servers are similar to the new generation of so-called "blade servers" from well-known suppliers such as Compaq, Hewlett-Packard, Dell Computer, Burgess says, but quickly points out that ARCI is not using blades because this new technology lacks industry standards. "We are not going into blade servers, and we are not going proprietary," Burgess says. "We try to make sure we stay with the commercial marketplace."

Burgess says he doubts that ARCI engineers would seek to mix and match 2U servers from different manufacturers in the same rack because ARCI designers will seek to obtain volume discounts from the 2U manufacturers.

These new sensors, developers say, will be able to detect and track some of the world's quietest submarines through minute phase shifts of light.

U.S. Navy leaders call this new technology the All Optical Underwater Segment — or AO-UWS. Like their electronic predecessors in the Integrated Undersea Surveillance System (IUSS), these optical sensors are to be deployed in strategic ocean areas that either funnel heavy submarine traffic, or where pinpointing hostile submarines is crucial.

The AO-UWS could go on line as early as 2004, says Jim Andersen, director of business development for fiber optic acoustic systems at Northrop Grumman Navigation Systems.

Northrop Grumman engineers are designing the AO-UWS optical sensor array under terms of an $8.9 million contract from the U.S. Space and Naval Warfare Systems Command (SPAWAR) in San Diego. The 24-month development program will culminate in an at-sea demonstration of the system sometime in 2003.

Among the chief advantages of underwater optical sensors are sensitivity and reliability, Andersen says.

The sensor array and signal-transmission media will be all-optical components — manufactured from plastic or silica, which is not susceptible to the corrosive influences of saltwater, Andersen says. Electronic components, on the other hand, can suffer corrosion or short-circuits in seawater, which present the Navy with a persistent maintenance headache.

"The advantage is there is no electronics in the water, or on the 'wet end', so the stuff we put in the water is very reliable," Andersen explains. "All the electronics is on shore or on a ship, which can be mounted in COTS [commercial off-the-shelf] versions; you don't have to package electronics for the water, and you can rely much more on standard COTS electronics."

The AO-UWS works by connecting arrays of optical sensors with optical fiber. These interconnected arrays, in turn, connect via optical fiber with signal-processing gear on shore, aboard ships, or inside submarines.

Two subcontractors are working with Northrop Grumman to develop the AO-UWS — MariPro Inc. of Goleta, Calif., which is concentrating on mechanical wet-end junctions, terminations, and packaging; and Digital System Resources (DSR) Inc. of Fairfax, Va., which concentrates on interfacing to existing Navy sonar processing systems.

Northrop Grumman experts also are using optical sonar sensors on submarines such as the future Virginia-class new attack submarine to enhance submarine-detection capabilities, and also to reduce maintenance tasks.