By Robert Inkol
The development of EW systems is fundamentally different than that of communications and most other RF systems. For example, a communications system is self-contained in the sense that the transmitter and receiver subsystems are designed to work with each other in a selected target communications scenario. The designer can even use matched filter techniques in the receiver to take full advantage of prior knowledge of the signal waveform. He or she also has complete control over the various design trade-offs and access to knowledge of government regulations, commercially available technology and competing products (i.e.. market research).
In contrast an EW system must operate against threat systems for which limited knowledge exists. Even if we have the benefit of access to ELINT data, threat systems, particularly those based on software defined radio concepts, may have reserve or war modes which are largely unknown. Moreover, defence electronic systems can have service lifetimes of decades. As a result, threat technologies and operational scenarios of concern are likely to change substantially over their lifetimes. The problem is further exacerbated by the need to define equipment configurations for major platforms well before those platforms enter service.
The CANEWS 2 (Canadian Electronic Warfare System) was intended to deliver the technology for a mid-life of the CANEWS naval radar ES system deployed on the Halifax class frigates. CANEWS was basically the “productionized” version of the radar ES test bed developed under the Zander program at the Communications Research Centre (CRC). Led by Alan Sewards, the CRC work successfully demonstrated the use of a Data General Nova minicomputer to process the stream of digitized pulse data produced by an Instantaneous Frequency Measurement (IFM) receiver, a notable innovation at the time.
Developed by Westinghouse under contract, the latter employed a pair
of AN/UYK-505 general purpose minicomputers augmented by specialized hardware.
The user interface displayed a polar plot displaying angle-of-arrival and
received signal power with the highest signal power corresponding to the
center of the plot. Measured radio frequencies were shown with text.
The AN/UYK-505 was a variant of the US Navy AN/UYK-20 standard computer.
Based on early 1970s technology, this was a 16-bit minicomputer reportedly
capable of 0.45 million instructions per second (MIPS)
 |
| Simplified CANEWS block diagram. Note that AN/SLQ-501 was the formal designation assigned to CANEWS under the Joint Electronics Type Designation System (JETEDS). (Graphic provided by Mike Gale) |
The manufacturer of the IFM receiver, MEL Defence Systems Ltd., was encouraged to set up a subsidiary in Canada to manufacture and support CANEWS for the Halifax class frigates and other naval programs.
CANEWS entered operational service in the early 1990s. Although CANEWS was relatively advanced at the time of its inception, it was not without limitations. Processing resource constraints limited CANEWS pulse train analysis for a single radar signal at a time, during which the pulse data for other radars coexisting in the signal environment would be discarded. While this was a clever idea for keeping the processing load manageable with the available technology, it was an obvious shortcoming in a dense signal environment where the timely detection of critical threat signals was essential. Moreover, it was expected that the more sophisticated processing algorithms needed to improve the reliability of the signal analysis processing were possible if additional computational resources could be made available.
By the mid-1980s, the computer technology employed by CANEWS was vastly outperformed by newly developed 32-bit microprocessors, such as the Motorola 68020. Using a real-time multiprocessor operating system, such as Harmony, one could construct powerful multiprocessing systems in a modular fashion using single board computers (SBCs) based on the VME bus standard.
Moreover, the VME subsystem bus (VSB) supported in some SBC implementations could be used for distributing pulse data from the IFM receiver to the SBCs. Additional improvements in performance could be achieved by using Application Specific Integrated Circuits (ASICs) to perform simple preprocessing algorithms on the stream of pulse data generated by the EW receiver. Object oriented programming (OOP) languages supporting a modular and structured approach to software design would improve software development productivity and the reliability of the developed software. Finally, it was hoped to ease the task of the operator by improving the user interface and exploiting Expert System techniques for analyzing signals.
DREO proposals based on the aforementioned concepts were selected in preference to less ambitious proposals produced by MEL with DREO funding. Based on the DREO concepts, the CANEWS 2, a Major Development Program project for the CANEWS mid-life upgrade was launched in the early 1990s. As originally envisaged, CANEWS 2 comprised the original CANEWS receiver, an add-on receiver to extend frequency coverage into the low millimeter wave (MMW) region, an auxiliary receiver to mitigate some of the fundamental limitations of the CANEWS IFM receiver, and a new modular digital processor based on the DREO hardware/software concepts.
Unfortunately, the project organization and planning were flawed from
the outset. A major issue concerned the complexity of managing the
diverse array of organizations involved:
- Chief of Research and Development (predecessor organization to
DRDC)
- Director of Maritime Combat Systems (DMCS)
- 3 industrial partners nominally organized as a consortium:
- Lockheed Martin (which had acquired MEL)
- Xwave Solutions
- COMDEV (subsequently acquired by Honeywell)
- Public Works
There was also a special relationship with DY-4, the manufacturer of the VME single board computer modules.
In practice, the individual industrial partners had their own interests and did not always co-operate with other players. Classification issues further impeded communications. In keeping with the defence R&D management philosophy at the time, it was optimistically assumed that the required technology had already been developed and that individual scientists should only be involved on an ad hoc basis to assist industry in integrating the various components. DMCS had the project management authority, but was hardly in a position to provide technical leadership, co-ordinate the efforts of the various organizations involved, or bring about the timely identification and resolution of issues. There was a DREO Scientific Authority, but the prevailing philosophy was that this position was primarily of an advisory nature. Moreover, several of the key people involved in leading the preceding AMEP project had left DREO. Given that the project participants did not share a common vision, had gaps in their understanding and lacked effective communications channels, it was inevitable that there would be “unknown knowns” and “unknown unknowns” and that the ability to quickly assess and respond to new information would be limited.
Perhaps an even bigger issue was that the project time lines were stretched out to accommodate the departmental financial constraints that reflected the reduced priority enjoyed by defence following the dissolution of the Soviet Union. Consequently, many of the technology choices that had been identified under the preceding AMEP project became obsolescent and difficult choices had to be made whether the cost and effort involved in their replacement could be justified under the circumstances.
The Motorola 680X0 processor family was a good choice in the 1980s, but by the early 1990s, the offerings of other vendors had demonstrated decisive advantages in performance and development potential, the Intel Pentium being a prominent example. The Smalltalk computer language, advocated by DREO and some SKL software developers, although well suited for prototyping and exploring system concepts, never really gained broad market acceptance and had performance overheads that could be problematic in a hard real-time system. Delays also contributed to the overheads and inefficiencies that inevitably result from personnel turnover.
The increased costs resulting from the aforementioned constraints contributed to the scaling back of the original system concept. One of the first dominos to fall was the auxiliary receiver. The acousto-optic receiver technology explored by DRDC was at first sight very promising. It had the ability to perform spectrum analysis over bandwidths in the order of a GHz in real-time, Thus appeared to solve the simultaneous signal problem. However, there were performance shortcomings that proved difficult to resolve. Research programs along similar lines in Allied countries also fell short. DREO had also initiated very promising work on receiver concepts based on the use of fast analog-to-digital converters and digital signal processing techniques. However, this work was too immature for consideration. Receiver technology offered by US vendors was considered, but not followed up on. In the end, cost was almost certainly a major concern and there may have also been hopes that more sophisticated software-based processing could still provide an adequate overall system performance.
At this point, the inadequate understanding of the weaknesses of the CANEWS receiver was certainly a serious issue; otherwise more attention would have been devoted to exploring its improvement or replacement.
Of critical concern, the expected improvement in the robustness of the pulse train analysis processing was proving challenging. The resolution and accuracy with which signal parameters of individual pulses could be measured was proving to be a serious weakness. In particular, the bearing measurements provided by the CANEWS receiver displayed a large variation during ship-borne operation.
Evidently, the placement of the CANEWS antenna system resulted in the reception of reflected signals that significantly degraded the bearing accuracy. Normally, one would like to rely heavily on bearing for deinterleaving trains of pulses as this is a parameter that is physically meaningful and, unlike parameters such as frequency, cannot be varied by the threat system. The discovery of the “unknown unknowns” associated with the receiver behavior resulted in a need to focus on improving the pulse train deinterleaving algorithms. Some success was achieved in this endeavor.
Up to this point, the CANEWS receiver had always been treated as a black box. It was considered as proprietary technology by Lockheed and nobody outside the company had more than a superficial idea of how it worked. A data recorder for capturing pulse data from the CANEWS receiver was developed, but it is not known to the extent to which useful signal data was captured and analyzed. In hindsight, it would have been better if such data had been acquired at an early stage and used to assess the actual performance potential of the target system configuration.
The prototype millimeter wave receiver developed by COMDEV to provide a basic signal detection and direction finding capability fell by the wayside, probably a result of financial constraints.
By the late 1990s, the system implementation had evolved to use a PC running the Windows NT operating system to support the user interface and other system features. The much increased performance available with the later Intel CPUs and graphics processing hardware supported high quality color graphics, significantly improving on the CANEWS performance in this respect. The VME multiprocessor was updated with the VxWorks real-time operating system and slightly updated SBCs, probably using 68040 CPUs. The Smalltalk language was retained.
A final sea trial with a CANEWS 2 system on HMCS Ville de Quebec was carried out in April 2001. Some performance improvement over CANEWS was noted, but this was considered insufficient to justify continued development and procurement and the project was formally closed out in the following year.
The end result was a major disappointment for those involved. Although the technology and capabilities needed to succeed seemed to exist, the organizational and communication problems that, in retrospect, existed from the start could not be overcome.
CANEWS was to remain in service without fundamental changes through the first decade of the 21st century.
References
[1] Cdr. Roger Cyr, “Standard Naval Computers,” Maritime Engineering
Journal, Jan/Apr 1980.
[2] B. Barry and J. Altoft, “Using objects to design and build radar
ESM systems,” Object-oriented Programming, Systems, Languages, and Applications
Conference (OOPSLA), December 1987.
[3] R. Inkol, C. Kunkel and R. Rowbotham, “An Advanced Associative
Comparator Integrated Circuit,” Midwest Symposium on Circuits and Systems,
August 1990.
[4] P. W. East, “Fifty Years of Instantaneous Frequency Measurement,”
IET Radar, Sonar and Navigation, February 2012.
Acknowledgements
The author thanks Pierre Yansouni, Brian Barry, Jim Lee, Jim Morris and Barbara Ford for their helpful comments and suggestions. He also notes that illuminating discussions with Mike Gale provided useful contextual knowledge. Mike also located the CANEWS block diagram.