By Stephen A. Schuster
When U.S. Navy personnel search out mines at sea, they rely on sophisticated mine-hunting sonars that provide accurate detection and classification information. Representing a wealth of accumulated knowledge, these modern sonars vastly improve mine-hunting capabilities from previous-generation systems.
Ultimately, however, the success of a mine-hunting mission depends on the sonar operator`s ability to analyze the sonar-generated data. He must be able to identify real targets and distinguish them from bottom features and other non-threatening forms.
For the U.S. Navy, training sonar operators to perform this difficult task effectively is crucial. Navy instructors must hone the operator`s mine-hunting capabilities across the full range of possible situations, using real-life practice environments. And with today`s shrinking defense budgets, they have to provide this training in the most cost-effective way.
Today, the U.S. Navy is achieving this objective for its AN/SQQ-32 mine countermeasure undersea sonar systems using a state-of-the-art shore-based trainer (SBT). Developed by Raytheon Co. Electrical Systems operation in Portsmouth, R.I. The SBT synthesizes the full range of signals encountered in real, at-sea operations and feeds these signals to the AN/SQQ-32, in real time. The SBT also addresses the budget concerns of the Department of Defense`s Program Management Ships 407 Procurement Office by reducing at-sea training, a costly component of the overall training program.
Mercury multicomputers
"We knew that developing the trainer would be a challenge," says Sudha Reese, technical director of Raytheon`s SBT project. "We were familiar with the AN/SQQ-32 because it was built by our Portsmouth, R.I., operation. The AN/SQQ-32 incorporates very high-frequency classification sonars; much higher frequency than what is used in submarine sonars. That means that the SBT has to feed the sonars with very high-fidelity signals, in real time. And that requires exceptionally high processing power - in excess of 2 gigaflops."
The Raytheon SBT development team met this challenge with a system design incorporating embedded multicomputer systems from Mercury Computer Systems of Chelmsford, Mass. These multicomputers provide the real-time processing speeds required to generate the high-fidelity sonar images.
Contributing to the decision to use Mercury multicomputers was Raytheon`s past success with using Mercury products in the AN/BSY-2 Team Trainer/Modular Acoustic Simulation System and other demanding signal processing applications.
Detection and classification
Mine-hunting sonars detect and classify mines moored beneath the surface and laid on the ocean bottom by capturing and processing signals representing the physical appearance of the mine`s echo and the shape of its shadow. Echo mode is used primarily for objects that are clear of the bottom or in conditions of minimal reverberation. Shadow mode is most effective in bottom conditions where reverberation is normally large. The absence of reverberation, which is caused by the shadow of the mine, provides the best classification clue - with the shape and sharpness of the shadow acting as the key characteristics.
"Using the Mercury multicomputers, our AN/SQQ-32 SBT can synthesize all of these signals at equally high fidelity," explains Reese. "We can create a real-life training environment that accurately represents what the sonar sees in the water. And training is more controllable because the operator of the SBT can influence the mixture of events that are synthesized."
The U.S. Navy also uses the SBT in its product improvement program to test enhanced AN/SQQ-32 products as they are developed. Because the signals generated by the SBT can be controlled, the operator can ensure that all aspects of a system can be exercised during system test.
Configuration and design
Raytheon`s SBT is a Unix-based digital signal processing (DSP) system incorporating eight Mercury embedded multicomputers in VME racks. Each multicomputer is configured with four Intel i860 RISC processors, providing 320 megaflops of processing power per board.
Optimized for 80 percent loading, the eight multicomputers deliver a total processing capability of approximately 2.3 gigaflops at this load level. The SBT injects signals into the AN/SQQ-32 minehunting sonars using a custom interface capable of data rates of 44 Mbit/s.
The DSP functions performed by the Mercury multicomputers incorporate many complex calculations. The DSP Inverse Beamformer, for example, implements fast Fourier transform computations using Mercury`s Enhanced Scientific Application Library. The complex signal generator implements complex convolutions to generate classification images. Because all of these calculations are implemented at the high sample rates associated with high-frequency sonars, the DSP is computationally very intense. The Mercury multicomputers satisfy these demanding, compute-intensive requirements.
"This is the first trainer available for the AN/SQQ-32," Reese points out. "It performs well and has been extremely well received. In fact, the project has been so successful that the U.S. Navy is basing the DSP functions of its larger on-board trainer, the AN/SSQ-94, on our Mercury multicomputer design."
Currently under development by the Naval Surface Warfare Center/Coastal System Station in Panama City, Fla., the on-board trainer incorporates a full range of functions, including the AN/SNN-2 Precise Integration Navigation System, the AN/SYQ-13 Navigation/Command and Control System, and the AN/SLQ-48 Mine Neutralization System.
The Navy is using Mercury`s Race series, a new breed of compact multicomputers based on the ANSI-standard Raceway Interlink switched-fabric interconnect. Raceway provides a sophisticated concurrent-transaction crossbar architecture that supports the enhanced signal-processing performance and I/O data rates required by the AN/SSQ-94.
RACE multicomputers provide a heterogeneous hardware and software environment, supporting a variety of different processor types within a single platform context. This allows extremely rapid insertion of new technology into deployed systems with little or no down time.
"Our customer, the Navy, was very pleased with the performance of the trainer at Techeval - the final delivery test," Reese says. "Initially we were concerned that the computational intensity of the application would be difficult to satisfy. However, through the efforts of our development team and the capabilities of the Mercury embedded systems, we delivered the SBT on time and within budget. We integrated and tested the system in one month, which is a considerable achievement for a project of this magnitude. Today, the SBT provides operational training and support for all AN/SQQ-32 systems."
Stephen A. Schuster is president of the Rainier Corp. PR firm in Princeton, Mass. Schuster holds BSEE and MBA degrees from Northeastern University and has more than a dozen years of engineering design, marketing, and management experience. He has held positions of Senior Design Engineer, Product Marketing Manager, Industry Marketing Manager, Director of Marketing, General Manager, and President. His phone number is 508-464-5302, and e-mail is [email protected]. For more information on Mercury multicomputers, e-mail [email protected].