Monday, July 1, 2002
AV-8B: Open Systems Pioneer
NavAir and Boeing are confident that a truly open systems approach to avionics, using commercial software and hardware satandards, will cut lifecycle costs. The proof of the pudding is a Harrrier upgrade.
Six years ago the Naval Air Systems Command (NavAir) started a program to replace the mission and stores management computers on the AV-8B combat VSTOL (vertical/short takeoff and landing) aircraft with a new mission systems computer (MSC) and warfare management computer (WMC) in order to add new munitions and growth capability into the future. Armed with high-level Pentagon support, the program was to show the benefits of an open systems approach to avionics, based on commercial software and hardware standards. The Open Systems Core Avionics Requirement (OSCAR) program is nearing completion. Will the approach significantly reduce lifecycle costs?
Boeing’s Bold Stroke
Boeing, the prime contractor on OSCAR, began working on open systems concepts in 1995 under its Bold Stroke initiative. Bold Stroke aimed to develop a repository of operational flight program (OFP) software that different aircraft programs could draw from, explains Don Winter, director of open systems architecture at Boeing’s Phantom Works Division. "The AV-8B was an ideal launch platform. There was no low-cost, evolutionary option open to [the U.S. Marine Corps]," he adds. The USMC was faced with an obsolete mission computer, at capacity, with software that was not portable to any other environment.
The mission systems computer runs the airplane, performing navigation and ballistics calculations and other functions. The WMC controls the weapons and countermeasures and drops the ordnance. Both computers employ the following:
C++ programming language,
VxWorks commercial real-time operating system (RTOS),
Commercial software development tools,
VMEbus motherboards, and
Processor cards using commercial PowerPC (PPC) chips.
"The weapons computer employs the same open systems and modularity concepts as the mission computer," says Vince Higbee, Boeing’s AV-8B program manager.
The open systems approach hasn’t been easy. The learning curve for designing, coding and testing software was steeper, and the effort has taken longer than NavAir had planned. (Originally OSCAR was to have been released to the fleet in 2000.) Developers have upgraded OSCAR’s MSC processor three times to match the demands of the software architecture. But future upgrades and maintenance should be easier, and NavAir now has modular, well-documented software, something it didn’t have with the predecessor box’s assembly language code. "OSCAR is a revolution, not an evolution–revolutions are an investment in the future," says Gena Bulleri-Browne, OSCAR integrated program team lead with NavAir, PMA-257.
OSCAR is a tri-national program, involving the Spanish and Italian navies, as well as the USMC. The international partners will be adding the Advanced Medium Range Air-to-Air Missile (AMRAAM). With an AV-8B squadron each, the Spanish and Italians are expected to complete fielding OSCAR-equipped aircraft by June 2004. The U.S. airplanes will add the ability to carry the 1,000-pound Joint Direct Attack Munition (JDAM). The project was in the final stages of developmental test in May 2002 and was slated to begin operational tests at China Lake, Calif., in August. The upgrade, if not accelerated, will be fielded to the U.S. fleet of about 129 airplanes over a period from 2003 to 2009. (The USMC has about 140 aircraft but expects some number of them to be retired between now and 2009.) The service plans to field its first squadron of OSCAR-equipped aircraft in late 2003.
Bold Stroke has become a major activity at Boeing, as the company applies its principles and methodologies, pioneered on the AV-8B, to upgrade programs for the F/A-18E/F, F-15E, and T-45 trainer. The F/A-18E/F and F-15E upgrades will use software architectures–in different instantiations–modeled on that developed for the AV-8B, Winter says. The F/A-18E/F, T-45 and AV-8B mission processors also will share some applications modules. While the F-15E’s operational flight program uses a different software language from the other three aircraft–Ada vs. C++– Boeing has demonstrated the viability of a "hybrid" OFP for possible future use. Boeing has invested upwards of $100 million in Bold Stroke.
The F/A-18E/F, T-45 and AV-8B will use variants of the computer developed by Boeing under the Advanced Mission Computer and Displays (AMC&D) program, which provides a faster processor and more scalable hardware architecture than that developed in OSCAR. The system design, which also offers display processing, will allow "platforms like AV-8B and T-45, as well as potentially others, to pick and choose the modules they need or want, thanks to the open systems concept," says Bulleri-Browne. (The WMC does not use the AMC&D box.)
The publicly stated goals for Bold Stroke, as projected into the future, include a 50-percent reduction in flyaway costs, a 60-percent reduction in development costs, and a 60-percent reduction in operations and support (O&S) costs, compared with historical cost levels. Data from early adopters and extrapolations from test exercises suggest that these goals can be achieved, Boeing says. NavAir agrees that O&S costs could decrease by 60 percent–more if the program is accelerated. The command attributes the projected savings to a combination of the AV-8B’s improved hardware and software architecture, enhanced development processes and increased hardware reliability.
NavAir officials estimates that development and flyaway costs will decrease by 25 percent and 5 percent, respectively. NavAir says its flyaway cost savings estimate is lower than Boeing’s because the mission computers are being procured in relatively small quantities, compared to the predecessor box. Still, "the estimated cost savings are satisfactory," Bulleri-Browne says.
Boeing’s software architecture is designed to minimize the ripple effects of software modifications on application code and to limit software perturbations from hardware changes. The company set out with Bold Stroke to reduce traditionally huge software lifecycle support costs, says Higbee. "The ‘push’ was software; hardware was ‘pulled,’" he says. This imperative led to the use of a higher-order software language (C++), object-oriented design techniques, and commercially available software development tools.
Today’s programmers are more familiar with higher-order languages than with the AV-8B’s legacy assembly language. The move to a higher-order language also made it possible to transfer programmers from project to project, to meet peak load requirements, without lengthy learning curves.
With the AV-8B MSC and the WMC, NavAir moved to commercially available hardware, such as PowerPC and VMEbus. The current MSC chip increases processing speeds from 1 MHz (the legacy computer) to 400 MHz, Higbee says. (The processing overhead associated with the software architecture, however, prevents a real gain of that magnitude.) Memory increased from 256 Kbytes of electrically erasable, programmable read-only memory (EEPROM) and 64 Kbytes of random access memory (RAM) to 32 Mbytes of EEPROM and 64 Mbytes of RAM, plus 128 Kbytes of nonvolatile RAM.
The original mission computer, the proprietary AYK-14, had gotten down to a less than 4 percent processing margin and a less than 3 percent memory margin. The new mission computer increases those figures to 34 percent and 49 percent, respectively. The WMC allows for roughly similar processor and memory margins.
Boeing tested the architecture’s ability to insulate software applications from hardware change. In one case, its engineers substituted a new version of the processing chip for the existing version, because of throughput demands, and measured the "upward effects." The result, according to Winter: "Not one line of code had to be changed at the application level. Something like five lines of code had to be changed in the infrastructure services layer.
In a second case, Boeing added a new com interface, requiring new applications code. The company measured the effect on other applications modules and the ISL. According to Winter, the change affected less than 5 percent of the applications software–a 20-to-1 reduction in the amount of code retest.
Bold Stroke Benefits
Software/Software Isolation. Boeing pioneered the application of object-oriented design techniques to flight software, allowing the isolation of applications components from each other, so that changes to one component don’t ripple over into another component, necessitating major retest and requalification. The cost of requalifying the entire software load "used to be the biggest impediment to fielding new capabilities," says Boeing Phantom Works’ Don Winter.
Software/Hardware Isolation. The software architecture’s infrastructure services layer (ISL) enables most of the software/hardware isolation. The application components, for example, require "no awareness to be built into them of where they’re running or how many processors are at the bottom [of the stack]," Winter says.
During the hardware rollover from the Motorola 603e to the Motorola G3 chip, for example, "some level of regression testing and tweaking of code [was required], but it was fairly easy to do," says Gena Bulleri-Browne, of NavAir.
Code Streamlining. The ISL layer handles numerous "shared functions," such as timing and "replication"–information storage and retrieval–required by application components. This means that services do not have to be implemented multiple times in each component.
Software Portability. Use of the Portable Operating System Interface for Unix, or Posix, a standard subset of Unix functions for real-time, embedded systems, enables the applications modules and ISL to be ported to other real-time operating systems. This frees the AV-8B from dependence on a single operating system vendor. Posix also allows operational flight program (OFP) software to be ported to desktop operating systems, so the OFP can be tested at the developer’s desktop, eliminating most errors before later test stages.
The mission computer software also can be ported to flight simulator s without a major additional software development effort to replicate the OFP’s behavior in a different language and operating system environment.
Open Systems: Next Step?
Now that flight computers are employing high-speed chips, it’s possible–during unused processing cycles–to transfer target imagery and information over data link, using the ultimate commercial technology, Internet browsers and protocols. A fighter aircraft could become a client in a network of the battlefield of the future, says Don Winter, director of open systems architecture at Boeing’s Phantom Works Division. The Air Force asked Boeing to demonstrate the concept under the Weapon System Open Architecture (WSOA) program.
"Dynamic, adaptive resource management allows us to guarantee that ‘hard’ real-time things get done," Winter says. But a high-speed chip’s unused capacity can be employed "to do things, on an as-available basis, like download an image or browse the network for data about a target." Boeing anticipates using real-time Java software when that standard matures. Real-time Java is a version of the popular Internet code that allows small downloaded programs to operate across a network on any hardware.
Boeing is preparing to fly a demo this summer, in which a command and control (C2) aircraft–in this case a B737 testbed aircraft–will transmit data to an F-15E technology demonstrator over the Link 16 data link. "We’ve buried the Internet TCP/IP protocol in the military data link and are sending TCP/IP packets back and forth," Winter explains. The scenario for demonstration: An F-15 launches against a preplanned target. The C2 platform receives information regarding a "pop-up" target via a Joint Tactical Terminal (JTT), annotates the image, and transmits it to the F-15.
Today such a scenario plays out differently. The C2 aircraft sends the target location over Link 16 and then talks through the new mission over voice communications. The new technology could cut the time and increase the clarity of the information exchange.
The technology allows an F-15 weapon system operator and an offboard planner "to collaboratively plan a new mission en route," Winter says. Essentially, "we’re staging a ‘net meeting’ between a planning node and the aft cockpit." The demonstration will showcase capabilities that can be added via open systems technology.
Link 16 was chosen because it is what the warfighters use, even though it is slow, about the speed of a dial-up connection. "We’re hoping to download an image in about two minutes," Winter says. The exercise will employ the user-defined message format available in the Link 16 standard.
The WSOA program has been well received, as it pushes military "hot buttons" like time-critical targeting and situational awareness, Winter says. The first phase will culminate in a flight demonstration of a two-node network.
Although the demo won’t use Link 16’s new J16 "image exchange" message type, because of the time lag between the WSOA and J16 development programs, a possible second phase of WSOA would switch to the new standard. That phase would involve more participants, simulated if not real, Winter says.
Led by the Air Force Research Lab/Information Directorate, with Boeing as the prime, WSOA also is supported by the Defense Advanced Research Projects Agency (DARPA), the Open System Joint Task Force (OSJTF), the Computer Resources Support Improvement program office and the Army’s JTT/Common Integrated Broadcast Service-Module program.
The F-15E also is upgrading its display processor, sharing some of the same software underpinnings and hardware components as the AV-8B mission systems computer, such as general-purpose processing modules using PowerPC chips. The F-15E upgrade consolidates two boxes–mission computer and display processor–into one called the Advanced Display Core Processor (ADCP).
The F-15’s operational flight program (OFP) is constructed primarily with Ada code–formerly required for military systems–rather than C++, the language of Boeing’s open systems, Bold Stroke initiative and of AV-8B, F/A-18E/F and T-45 upgrade programs. But Boeing has demonstrated the viability of "hybrid" OFP code, says Don Winter, director of open systems architecture at Boeing’s Phantom Works Division. The F-15 "could accommodate commonality with the other three aircraft in the future–even mixed with legacy Ada," he says.
The ADCP program also will take advantage of the commercial world’s OpenGL standard, which "opened up a wealth of commercial [development] tools, Winter says. Boeing, for example, has chosen to use the VAPS tool by Virtual Prototypes to design the look and feel of displays on a workstation and then generate actual display processing code that runs in accordance with the OpenGL interface.
Before, if you wanted to change the color of a triangle on a display, you needed to get changes made in the processor manufacturer’s low-level code, Winter says, typically a six-month exercise. With the new prototyping tool, however, "You can make the change on a workstation, literally push the autocode [button], and generate the new software for the aircraft target system."
SCAR focuses on the replacement of two AV-8B computers. However, other upgrade activities for the AV-8B also are in the works. The AV-8B Remanufacture program, for example, will convert 74 U.S. and five Spanish "day attack" aircraft to radar-equipped aircraft. The U.S. Marine Corps also intends to upgrade to full-color, flat-panel displays from the current, obsolete color cathode ray tubes (CRTs), beginning in 2003, and to upgrade current digital map capability with the Tactical Aircraft Moving Map Capability (TAMMAC) in 2006. TAMMAC already is deployed on other aircraft.