Advanced compressive receiver techniques.
Passive engagement is a widely sought goal, but even when that is achieved the need for communications will not be obviated. Hence, we have seen the ongoing development of communications systems - worldwide - that offer a low probability of intercept (LPI) and a measure of jam resistance for the users.
One technique that has been developed for LPI communications that creates special problems for the noncooperative interceptor is random frequency hopping during communications. The SINCGARS radio in the US and the Jaguar radio in the UK are examples of such frequency-hopping systems. The extent to which frequency hopping communications is becoming common has been emphasized by news reports that commercial versions of the Jaguar radio are in the hands of South American drug dealers.
The best means to intercept frequency hoppers (FH) is widely debated in the community. It is not our intent to review that debate, so we will state our opinion without defense. We believe that the compressive receiver offers the best available technology for intercepting present-day hoppers when real-time processing is required in a space-constrained, tactical airborne environment. For larger installations, other types of receivers are viable; in the future it is anticipated that digital receiver technology will meet the combined challenge of speed and packaging for many applications.
The technological challenge to intercept an FH signal is simply to achieve a high probability of intercept. This challenge leads to specific goals for parameters such as sweep speed, sensitivity, dynamic range and low spurious signals.
In the following section we show how to develop compressive receiver requirements from the system engineer's point of view. Later we present some of the design details of a compressive receiver that could intercept many of the modem FH radios.
There are two basic needs to be fulfilled to intercept any signal successfully. First, the signal must be present within the detection band of the intercept receiver; second, the signal-to-noise ratio must be adequate for the detection process to be successful. The first need is significant when the signal of interest is a frequency hopper that occupies a small portion of the RF spectrum for a short time. Figure 1 illustrates this situation graphically. A typical hopper may use 25 kHz of a 100-MHz band for as little as 2 ms before hopping to another frequency in the band. The hop period is normally fixed, while the sequence of hop frequencies is pseudo-random. The information bandwidth usually does not exceed the hop resolution, for ECCM reasons.
The compressive receiver (CRx) requirements can be inferred from the signal characteristics. By way of review, Figure 2 is the block diagram of a compressive receiver. The swept local oscillator (LO) expands the signal over a band equal to the width of the sweep. The dispersive delay line (DDL), with a delay versus frequency slope that is inverse to the sweep, compresses the expanded signal, producing pulses at times and amplitudes proportional to the spectral components of the input signal. A logarithmically detected display of the output would appear like that of a typical spectrum analyzer display.
Using the typical signal previously described for illustration, the CRx must sweep 100 MHz (B.sub.s) in 2 ms (T. sub. s) to guarantee that the signal will appear in the CRx no matter what the frequency is during that time. This implies a sweep rate of 50 MHz/ms. If we wish to distinguish each hop - it is important to recognize that a signal is a hopper - then the CRx must have a resolution of 25 kHz or less. The resolution will be determined by the dispersive delay line's dispersion and weighting function. An unweighted line would result in a resolution equal to the inverse of the dispersion time (T.sup.D)- Common weighting (used to reduce the side-lobe responses of the delay line) results in approximately a 50% decrease in resolution. Therefore, to achieve 25-kHz weighted resolution an unweighted resolution of 16.7 kHz would be needed, which would result from a dispersion of 60 [micro s]. (Resolution | [1/60 x 10.sup.6] x 1.5.)
It is necessary to calculate the bandwidth required of the dispersive delay line so that it can be determined that the line can be built and what integration gain will be realized. The delay line bandwidth (B.sup.D) is the product of the sweep rate and the dispersion, which for our example results in a bandwidth of 3 MHz. A 3 MHz by 60[micro s] line is well within the realm of producibility.
The processing gain in decibels (10 log [T.sup.D][B.sup.D]) is 10 log (60 x 10.sub.6) (3 x 10.sub.6) or 23 dB. This might not be acceptable, because the intended receiver of the signal will have a processing gain of 36 dB on the basis of non-coherent bandwidth reduction alone (36 dB = 10 log [100 X 10[sub.6]/ 25 x 10.sub.3]), which implies the interceptor will be at a severe range disadvantage. Processing gain can be increased by choosing a DDL with greater bandwidth and sweeping the line faster. Table 1 lists the characteristics of the ITT CRx which would easily fulfill the receiver requirement to intercept this signal. In fact, this receiver could intercept a signal hopping at a 2-kHz rate.
The process used to develop the CRx requirements in the previous example can be described in general terms. Table 2 lists the steps in the procedure to be followed. Questions of sensitivity and signal level can be answered using the normal methods of receiver system design and are not repeated here.
ADVANCED RECEIVER DESIGNS
Compressive receivers create Fourier transformations of time domain input signals for frequency domain signal processing. Dispersive delay lines (DDL) are used as convolvers (which manipulate the signals to make them easier to process) in performing the Fourier transformations. The transformation to the frequency domain can also be implemented by the digital Fast-Fourier transform (FFT). However, practical digital FFT processing is limited to very narrow bandwidths at present. As technology advances, processing bandwidths will widen. A brief discussion of analog implementations follows.
There are two basic analog approaches used to produce the Fourier transform: the convolve-multiply-convolve (CMC) and the multiply-convolve-multiply (MCM). These are show in Figure 3 a and b. The complex exponential inputs to the mixers represent chirp waveforms (linearly increasing or decreasing as a function of time). Both configurations are functionally identical in that they can produce the same outputs.
Provided a chirp with sufficient time-bandwidth (TB) product can be produced, the CMC configuration generally offers the best overall performance for C[sup.3]I applications.
In a CMC configuration, the first convolver linearly time-skews the frequency components of wideband signals to counter the time-skewing that occurs in the second convolver. This nulling effect minimizes phase distortion in the compressed output signal. If phase information is not important, as in ESM applications, then the first convolver can be eliminated for reduced cost and complexity.
Chirp Generation Techniques
The classic method of generating the chirped (swept) LO signal is by impulsing a dispersive delay line (DDL) having a TB directly related to that of the convolving DDL. The DDL spreads the spectrum of the impulse in time, producing a linear frequency sweep (chirp) as a function of time. For the CMC configuration, the TB of the frequency-chirped LO signal must be four times that of the convolver and the magnitude of the frequency verses time slopes must be identical. Figure 4 depicts a CMC Fourier transformer being driven by three possible types of chirp LO generators.
The impulsed LO uses two DDLs in series followed by a frequency doubler to produce a constant-amplitude frequency-chirped (swept) LO signal having a 4TB = 2T2B characteristic. For very fast-sweep (> 200 MHz/[micro s]) applications, as in radar, the impulsed DDL technique is the only practical means of generating chirps. The chirps, however, will have fairly high noise due to DDL losses and will not perfectly match the slope of the convolver. This limits receiver dynamic range to 40 or 50 dB and decreases frequency resolution.
ITT Avionics has developed two active sweep (chirp) generation techniques ("b." and "c." in Figure 4). The first uses a swept fractional-division (FD) frequency synthesizer which provides very linear, long-duration and very wide sweeps at rates up to approximately 10 MHz/[micro s] along with very low noise outputs. The sweep slope can be digitally adjusted to match the DDL convolver almost perfectly. The swept FD synthesizer contains high-speed digital processing, including gallium arsenide (GaAs) ICs to step-tune a voltage-controlled oscillator (VCO) in fine increments. A wideband phase-lock loop provides closed-loop smoothing of the tuning voltage to the VCO for a smooth frequency-versus-time curve. The FD synthesized chirp generator LO enables compressive receivers to have 60- to 80-dB dynamic range, 10- to 30-kHz resolution and 1 GHz bandwidths.
The swept Direct-Digital frequency Synthesizer (DDS) is the newest active chirp generation technique developed at ITT. It can provide very linear long or short duration sweeps over moderate bandwidths to 250 MHz). Sweep rates up to 200 MHz/[micro s] are possible. As with the FD frequency synthesizer, noise outputs are low and the sweep slope can be finely adjusted for a near perfect match to the DDL convolver. Although the output is discretely stepped, steps can be sufficiently fine to allow good resolution in the receiver. Wideband applications (> 100 MHz) require the use of GaAs logic and GaAs digital-to-analog converters (DAC). The main advantages of the swept DDS chirp generator are small size, low recurring cost and high sweep rate capability. Dispersive Delay Lines
A DDL provides a delay to RF signals which either linearly increases or decreases with frequency. DDLs are commonly rated by their time bandwidth (TB) product which represents the processing gain of the device. The bandwidth (B) is the operational bandwidth of the device and time 7) is the differential delay between the band edges, called the dispersion. There are several types of lines to choose from, including electromagnetic, magnetostatic wave, surface acoustic wave SAW) and acoustic charge transport. Electromagnetic dispersive delay lines are bulky, but they are capable of operating up to a few gigahertz with TB products in the order of 1,000. Magnetostatic wave dispersive delay lines use a yttrium iron garnet (YIG) crystal to produce a propagation delay that increases with frequency. They can operate up to a few gigahertz with TB products in the order of 500.
SAW dispersive delay lines are fabricated on a piezoelectric substrate such as quartz or lithium niobate. The input electrical signal is converted into an acoustic wave via an input transducer, processed in the SAW dispersive delay line and then converted back into an electrical signal at the output. Because of the slow velocity of surface waves, SAW delay lines are orders of magnitude smaller than EM lines. SAW DDLs can be made to produce delays that either increase or decrease with frequency. TB products of 4,000 have been achieved. Depending on the configuration, SAW DDLS can operate from a few megahertz to a few gigahertz with from 15 to 50 dB of insertion loss. In general, the wider the bandwidth, the greater will be the loss. Acoustic charge transport
ACT) delay lines use both surface acoustic wave propagation and charge transportation techniques. Gallium arsenide, which is both piezoelectric and a semiconductor, is used for these lines. In operation, a sinusoidal clocking signal is applied to an input transducer that launches a sinusoidal surface acoustic wave in the substrate. This in turn induces a matching pattern of sinusoidal potential gradients in the substrate. This potential gradient wave propagates in unison with the surface acoustic wave across the substrate. A signal input electrode located in the path of these potential gradient waves then deposits charge proportional to the magnitude of the input signal in the depressions of the waves. In this fashion, charge packets representing samples of the input signal are carried toward the output sense electrode, and are delayed by the interval of time required for the acoustic wave to traverse this distance.
The principle difference between ACT and SAW lines is that with ACT lines the surface wave induced by the clocking signal functions as the carrier of the signal information rather than as a representation of the signal itself. Since the clock transducer does not have to handle a wide signal bandwidth, as a SAW DDL does, it can be built with many fingers to improve its gain.
Acoustic charge transport is an emerging technology and considerable advances are expected to be made over the next few years. Current ACT DDL devices are limited to a TB product of about 300, which restricts available processing gain to about 25 dB. Present devices have rather short time dispersion, approximately 1.5 [micro s], which limits output frequency resolution to about 700 khz. The short time dispersion also necessitates use of high-rate chirp generators with slopes of up to 200 Mhz/[micro. s]. Linear (two tone) dynamic range is presently about 50 dB. Finally, ACT frequency range is presently limited to about a 200-MHz bandwidth due to the upper frequency limit of the SAW channel.
At present the primary advantage of ACT DDLs is their small size (compared to other technologies). If coarser resolution and narrower bandwidths are acceptable, then ACT should be considered.
RECEIVERS IN DEVELOPMENT
ITT Avionics has developed a SAW-based compressive receiver RF processor that uses the swept FD frequency synthesizer discussed earlier to produce the chirp linearity and accuracy necessary for its 25-kHz resolution while scanning at a 1 MHz/[micro.s] rate. The scan rate and start and stop frequencies of the receiver are all digitally programmable via the swept synthesizer. The receiver scans a 480-MHz band and has an input sensitivity of -120 dBm. With advanced digital GaAs technology, FD frequency synthesizers can now be swept over a 1-GHz band, which would allow doubling the bandwidth of the receiver.
Figure 5 shows a simplified block diagram of this dual-conversion receiver. The DDL (60 [micro.s] x 60 MHz) used is of the up-chirp (delay increasing with frequency) type with the synthesized LO sweeping from low to high. With digital reprogramming, the swept LO can provide a down chirp as well to accommodate a down-chirp DDL.
Performance of this linear-swept receiver is provided in Table 3.
Acoustic Charge Transport (ACT)
ITT is currently researching and developing programmable ACT-based receivers with internal funding. To date the company has developed a design concept for an ACT dispersive delay line (DDL) based compressive receiver RF processor. The ACT DDL device selected utilizes 256 taps with 5-bit weighting and can be programmed as an up or down chirp device. A direct digital synthesizer (DDS) is planned to provide scan rates of 70 MHz/[micro.s] for a receiver resolution of 700 kHz. An 18-dB processing gain results in a sensitivity of -90 dBm. Current ACT devices limit bandwidth to about 50 MHz and dynamic range to about 50 dB.
The DDS consists of two cascaded digital phase accumulators followed by a sine map ROM and a digital-to-analog converter (DAC). Each accumulator produces a phase ramp. The ROM maps the digital phase ramp from the second accumulator into a code which is converted to a sinewave at the DAC output. The cascade produces a perfectly linear frequency sweep. The DDS provides complete flexibility in setting upper and lower sweep limits and scan rates.
Gallium arsenide technology advancements are rapidly improving the capabilities of ACT and DDS implementations. Improvements in bandwidth and dynamic range together with the small size and flexibility provided by these devices offer important performance improvements for future EW systems.
The compressive receiver is an essential system component for intercepting frequency hopping signals. Modem digital synthesizer techniques combined with acoustic wave technology will assure the compressive receiver's superiority in the future. As acoustic charge transport devices mature, they will be instrumental in reducing the size of compressive receivers. We believe a compressive receiver approaching the size of a pocket calculator and capable of intercepting any frequency hopper now in the field is within reach.
Raymond A. Luther is a technical consultant and manager of the ESM Systems Group in the Advanced EW Systems Dept. of ITT Avionics. He holds a BS in science and an MSEE from Stevens Institute of Technology.
William J. Tanis is a senior technical consultant and manager of the Advanced Receiver and Synthesizer Development Group in the Design Technology Dept. of ITT Avionics. He has a BSEE from Fairleigh Dickenson University and an MSEE from the New Jersey Institute of Technology.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||includes related article on radio eavesdropping|
|Author:||Luther, Raymond A.; Tanis, William J.; Hardy, Stephen M.|
|Publication:||Journal of Electronic Defense|
|Date:||Jul 1, 1990|
|Previous Article:||Defense contractors adjust to survive the 1990s.|
|Next Article:||EW conversions and definitions.|