Radar is quite simple in concept, but it becomes highly complex in actual implementation. Many decades passed between the time it was first envisioned and when useful working models emerged, which happened just in time to save England from total defeat in the 1940s.

When it comes to testing radar, the situation has become easier thanks to the development of arbitrary waveform generators (AWGs) with bandwidths high enough to generate radar frequencies carrier frequencies to 20 GHz (beyond Ku-band frequencies). One complicating factor is the need to emulate multiple-antenna radar systems based on phased-array antennas—or, more recently, multiple-input, multiple-output (MIMO) antenna architectures. Such setups demand radar test signal sources able to generate multiple signals with tightly controlled timing and phase alignments.

The simplistic notion is that the radar transmitting antenna sends out a beam that bounces off an object. The returning beam is received, and its direction and distance is determined by measuring the round-trip time and (in the case of a rotating antenna) position vector. But a working radar involves concepts that go far beyond that crude explanation.

Complex signal processing and extensive machine learning are part of the picture. In a high-noise environment, realizing an acceptable signal-to-noise ratio is a high-tech undertaking.

Radar provides information not only concerning direction and distance from the observer but also, by measuring the Doppler effect, speed relative to the observer. This data can be combined with rate of vector change to ascertain absolute speed.

When radar signals hit an object, they may reflect straight back to the receiver, off at an angle, or they may scatter or be absorbed. Electrically conductive objects — such as metals, seawater and wet ground — are highly reflective. This is a function not only of the material that comprises the object, but also of the radar signal frequency. Clouds and rain should be transparent for some applications such as aircraft collision avoidance, and they should be relatively opaque to show up in weather radar, so selecting the best frequency is part of the design process.

Radar receivers are typically in the same location as the transmitters, often sharing the same antenna via a circulator. The return signal is, of course, far weaker than the transmitted signal, usually requiring a high degree of amplification and signal processing.

When two materials have differing dielectric or diamagnetic constants, electromagnetic waves will reflect or scatter at the boundary, depending upon its texture and also the wavelength of the radar signal. If the radar wavelength is considerably shorter than the size of the object, it will bounce back or off at an angle. If the wavelength is longer than the size of the target, the object may not be visible at the receiver. Low-frequency radar usually detects objects indirectly, by means of resonances or Rayleigh scattering. With the passage of time, radar has come to rely on higher frequency signals, which require more energy and are more difficult to generate, but provide greater resolution.

Polarization is an important aspect of the radar wave as transmitted, and also as received, which are sometimes different due to the effect the target has on the beam.

The electric and magnetic field oscillations remain perpendicular to one another and to the direction of propagation but they may or may not rotate with the direction of propagation as an axis. In linear polarization, the fields oscillate in a single, unchanging plane. In circular or elliptical oscillation, the fields rotate at an unchanging rate. If the fields rotate to the right with respect to the direction of propagation, it is right circular polarization, and if they rotate to the left, it is left circular polarization.

A transmitted radar signal may exhibit vertical, horizontal, linear or circular polarization. For weather radar, circular polarization is avoided because it renders rain transparent. Linear polarization reveals metal surfaces. Random polarization is appropriate for granular or rocky surfaces. It is used in navigational radar where it is important to image the ground surface.

In addition to the transmission, polarization is affected by the reflective surface of the target. An irregular face will fragment the beam and cause random polarization, which can be detected at the receiver and displayed for the operator.

The maximum range of a radar beam is affected by the medium through which it travels. Vacuum has the least attenuation but there is some dispersal even with a perfect parabolic transmission antenna, due to wave/particle self-interference. Additionally, a line-of-sight signal path cannot be maintained beyond a certain distance because of the curvature of the earth.

Pulse repetition frequency is also a factor. It determines the maximum non-ambiguous range, which depends upon the distance the pulse can travel and return before the next pulse leaves the transmission antenna. Other factors determining range are environmental conditions, size of the target, radar sensitivity and power of the return signal.

Signal detection requires a signal that is greater than the noise floor by the signal-to-noise ratio. Noise can be an internal phenomenon or its source may be external. Either way, its cause is thermal. Powerful radar installations may require active cooling to lower the noise floor so distant objects can be located and displayed. The intensity of external noise depends upon meteorological conditions at the object. When the radar antenna is pointed upward, where the air is usually cooler, there is less noise.

Heterodyne frequency downconversion is widely used in radar installations, and there are significant benefits. For one thing, there is less flicker noise caused by electron motion in amplifiers because of the lower intermediate frequency. Another benefit in radar installations is that the bandwidth is lower at the lower frequency, increasing resolution.

Thermal noise is given by:
N_T = k_B x T x B
Where N_T is thermal noise, _B is Boltzman’s constant, T is absolute temperature, and B is bandwidth.

Distance resolution is by far the foremost parameter of interest in most radar installations. An obvious way to acquire this information is to aim a short pulse of electromagnetic radiation at the target and measure the round-trip time interval. The distance between the transmitting antenna and object of interest is one half the product of transmitted plus reflected signal times, multiplied by the speed of the signal. An inherent difficulty is that the transmitted electromagnetic radiation travels at close to the speed of light, the exact figure depending upon the density of the medium and the frequency of the transmission. Modern high-speed electronics deals successfully with this issue.

Frequency modulation is also used in ascertaining the distance to the target and speed it is moving toward or away from the target. This is more accurate than timing the extremely fast signal with its small transit time. Frequency differences between two signals can be accurately determined by observing how they interact. (This was done in the 19th century Michelson-Morley experiment.)

Radar signals can be described by two broad categories: pulsed RF and continuous-wave (CW) signals. Pulsed RF signals consist of periodic bursts of an RF carrier, possibly modulated via amplitude modulation (AM), frequency modulation (FM), or phase modulation (PM). The rate at which pulses are generated is referred to as the pulse repetition frequency (PRF), while the period of the pulses is called the pulse repetition interval (PRI).

In CW radars, the RF signal is continuous and the range of the radar system is established through time markers carried on by a transmitted signal. The use of FM is the typical way to measure the distance to target because the instantaneous frequency detected from an illuminated target depends on distance.

Pulsed RF radar transmissions may be fixed or variable frequency using various frequency-hopping patterns. These patterns are complex and tough to predict. The carrier frequency may also change for each transmitted pulse.

Pulse compression techniques can boost radar range by transmitting longer pulses for higher average power. Echo processing at the receiver augments spatial resolution by compressing the pulse through correlation or dispersion processing. The two main pulse compression methodologies are FM chirp, which consists of fast frequency sweeps that can be linear frequency modulation (LFM) or nonlinear frequency modulation (NLFM); and phase modulation, in which each pulse is composed of a series of shorter pulses where the carrier phase is controlled by a low-autocorrelation binary sequence of symbols.

AWGs generate radar signals using three different basic methods: baseband generation, intermediate-frequency (IF) signal generation, and direct RF signal generation. The oldest is baseband generation, where an AWG generates a time-domain signal applied to an RF modulator. For complicated digital modulation or fast frequency sweeps (FM chirp), the AWG generates two baseband signals that drive a quadrature modulator controlling both the amplitude and the phase of the carrier instantaneously.

IF generation involves an AWG generating a modulated signal at a relatively low carrier frequency. The signal then goes directly to a signal-processing function block in the receiver or transmitter. To get the final RF/microwave frequency, a frequency upconverter may be used.

If the AWG output frequency is high enough, it can generate the modulated carrier at the final RF/microwave frequency. Direct carrier generation requires excellent AWG sampling clock jitter performance because jitter translates directly to phase noise in the generated carriers.

For most baseband and IF signals, an AWG with a sample rate of a few gigasamples per second can provide the necessary output frequencies. Here the modulator or frequency upconverter will limit the modulation bandwidth of the final RF/microwave signal.

In addition, wideband quadrature modulation is sensitive to in-phase/quadrature (I/Q) amplitude and phase imbalance or quadrature errors. So it can be tricky to produce sufficiently good signals using the baseband or IF radar signal generation approach.

Direct signal generation requires an extremely fast AWG sampling at least 2.5 times higher than the maximum frequency component of the signal to be generated. The latest generation of AWGs can do this beyond Ku-band frequencies (12 to 18 GHz). (Note that current long-range radar systems for autonomous vehicles operate in the 77 GHz band.)

An AWG can generate both undistorted or intentionally distorted signals. The use of intentional distortion can compensate for distortion from external effects or components (connectors, cables, and so forth) and can help improve amplitude flatness or group delay of a test setup.

It is good idea to set the AWG sampling rate well above the minimum Nyquist frequency requirement. Such oversampling can give a flatter frequency response, better image rejection, less quantization noise, and lower pulse-to-pulse jitter. The price paid for oversampling is a need for more memory – thus long record lengths are important for high-speed AWGs. Sampling rates for direct RF generation must exceed those for baseband signal generation, so a given record length translates into a shorter time window for direct RF generation than for the baseband version. The record length is key for realisticcally emulating complex radar systems incorporating staggered pulse sequences, frequency-hopping patterns, or time-varying echos caused by moving targets or antennas that vibrate.

AWG outputs that mimic radar system signals can be helpful in numerous ways. For example, they may be the only way of verifying circuit functions during the design of radar subsystems, when the whole radar isn’t functional. They can also help characterize system behavior under actual operating conditions. A wideband AWG, for example, can simulate a cluttered open-air signal environment containing wide- and narrowband chirp signals, narrowband signals, CW signals, and frequency-hopped radar signals.

Older-generation AWGs suffered from relatively low sampling rates and poor spurious-free dynamic range (SFDR), which limited the generation of output signals for radar tested to a few gigahertz.

Some applications, such as MIMO radar generation, require multiple test signal channels. The channels must also be synchronized, so they must share the same sampling clock and be time-aligned. Any timing difference among channels or channel-to-channel jitter will degrade the radar test signal.

An AWG can generate a continuous radar signal by cycling the contents of the waveform memory through the DAC. But the signal must not become garbled during the wrap-around event. That leads to some special considerations. For one thing, the waveform memory must store an integer number of pulse repetition intervals. Otherwise, the pulse timing will be either longer or shorter than required every time the waveform cycles. To generate coherent radar signals, the AWG record length and sampling rate must be set so the resulting time window is an exact multiple of the carrier frequency period.