DETECTION OF RADAR SIGNALS

IN NOISE

Introduction

Matched filter receiver

Response characteristics and derivation

Correlation function and cross correlation receiver

Efficiency of Non-matched filters

Matched filter with non-white noise

 • The two basic operations performed by radar are ( 1) detection of the presence of
reflecting objects, and (2) extraction of information from the received waveform to
obtain such target data as position, velocity, and perhaps size.
• In this chapter some aspects of the problem of detecting radar signals in the
presence of noise will be considered. Noise ultimately limits the capability of any
radar.
Matched-Filter Receiver
• A network whose frequency-response function maximizes the output peak-signal-
to- mean-noise (power) ratio is called a matched filter. This criterion, or its
equivalent, is used for the design of almost all radar receivers.
• The frequency-response function, denoted H(f), expresses the relative amplitude
and phase of the output of a network with respect to the input when the input is a
pure sinusoid.
• The magnitude |H(f)|of the frequency-response function is the receiver amplitude
pass band characteristic.
• If the bandwidth of the receiver pass band is wide compared with that occupied by
the signal energy, extraneous noise is introduced by the excess bandwidth which
lowers the output signal-to-noise ratio. On the other hand, if the receiver bandwidth
is narrower than the bandwidth occupied by the signal, the noise energy is reduced
along with a considerable part of the signal energy.

• The net result is again a lowered signal-to-noise ratio. Thus there is an optimum
bandwidth at which the signal-to-noise ratio is a maximum. This is well known to
the radar receiver designer.
• The rule of thumb quoted in pulse radar practice is that the receiver bandwidth B
should be approximately equal to the reciprocal of the pulse width τ. This is a
reasonable approximation for pulse radars with conventional superheterodyne
receivers. It is not generally valid for other waveforms.
• The exact specification of the optimum receiver characteristic involves the
frequency- response function and the shape of the received waveform.
• The receiver frequency-response function, is assumed to apply from the antenna
terminals to the output of the IF amplifier.
• The second detector and video portion of the well designed radar superheterodyne
receiver will have negligible effect on the output signal-to- noise ratio if the
receiver is designed as a matched filter. Narrow banding is most conveniently
accomplished in the IF.
• The bandwidths of the RF and mixer stages of the normal superheterodyne receiver
are usually large compared with the IF bandwidth. Therefore the frequency-
response function of the portion of the receiver included between the antenna
terminals to the output of the IF amplifier is taken to be that of the IF amplifier
alone.
• Thus we need only obtain the frequency-response function that maximizes the
signal-to-noise ratio at the output of the IF. The IF amplifier may be considered as a
filter with gain.

• For a received waveform s(t) with a given ratio of signal energy E to noise energy
No (or noise power per hertz of bandwidth), North showed that the frequency-
response function of the linear, time-invariant filter which maximizes the output
peak-signal-to-mean- noise (power) ratio for a fixed input signal-to-noise (energy)
ratio is

• The noise that accompanies the signal is assumed to be stationary and to have a
uniform spectrum (white noise). It need not be Gaussian.

• The filter whose frequency-response function is given by Eq. above has been called
the North filter, the conjugate filter, or more usually the matched filter. It has also
been called the Fourier transform criterion.

• The frequency-response function of the matched filter is the conjugate of the

spectrum of the received waveform except for the phase shift exp (- j2Πft1). This
phase shift varies uniformly with frequency. Its effect is to cause a constant time
delay.
• The frequency spectrum of the received signal may be written as an amplitude
spectrum |S(f)|and a phase spectrum exp [- jφs (f )].
• The matched- filter frequency-response function may similarly be written in terms
of its amplitude and phase spectra |H(f)| and exp [- jφm(f)]. Ignoring the constant
Ga, Eq. above for the matched filter may then be written as

• Thus the amplitude spectrum of the matched filter is the same as the amplitude
spectrum of the signal, but the phase spectrum of the matched filter is the negative
of the phase spectrum of the signal plus a phase shift proportional to frequency.
• The matched filter may also be specified by its impulse response h(t), which is the
inverse Fourier transform of the frequency-response function.

• Physically, the impulse response is the output of the filter as a function of time
when the input is an impulse (delta function).

• A rather interesting result is that the impulse response of the matched filter is the
image of the received waveform; that is, it is the same as the received signal run
backward in time starting from the fixed time t1.
• Figure 1 shows a received waveform s(t) and the impulse response h(t) of its
matched filter. The impulse response of the filter, if it is to be realizable, is not
defined for t< 0. (One cannot have any response before the impulse is applied.)
Therefore we must always have t < t1.

• This is equivalent to the condition placed on the transfer function H(f) that there be
a phase shift exp (-j2Πft1). However, for the sake of convenience, the impulse
response of the matched filter is sometimes written simply as s(- t).

Fig.1 (a) Received waveform s(t); (b) impulse response h(t) of the matched filter.
 Derivation of the matched-filter characteristic
• The frequency-response function of the matched filter has been derived by a
number of authors using either the calculus of variations or the Schwartz inequality.
We shall derive the matched-filter frequency-response function using the Schwartz
inequality.
• We wish to show that the frequency-response function of the linear, time-invariant
filter which maximizes the output peak-signal-to-mean-noise ratio is

• When the input noise is stationary and white (uniform spectral density). The ratio
we wish to maximize is

• Where |so(t)| max = maximum value of output signal voltage and N = mean noise
power at receiver output. The ratio Rf is not quite the same as the signal-to-noise
ratio which has been considered in the radar equation.

• The output voltage of a filter with frequency-response function H(f) is

• Where S(f) is the Fourier transform of the input (received) signal. The mean output
noise power is

• Where No is the input noise power per unit bandwidth. The factor appears before
the integral because the limits extend from - ∞ to +∞, whereas No is defined as the
noise power per cycle of bandwidth over positive values only. Assuming that the
maximum value of |so(t)2| occurs at time t = t1, the ratio Rf becomes

Schwartz's inequality states that if P and Q are two complex functions, then
• The equality sign applies when P = kQ, where k is a constant. Letting

• We get, on applying the Schwartz inequality to the numerator of Eq. earlier, we get

• The frequency-response function which maximizes the peak-signal-to-mean-noise

ratio Rf may be obtained by noting that the equality sign in Eq. applies when P =
kQ, or

Where the constant k has been set equal to 1/Ga.

 The matched filter and the correlation
function
• The output of the matched filter may be shown to be proportional to the input signal
cross-correlated with a replica of the transmitted signal, except for the time delay
t1.
• The cross correlation function R(t) of two signals y(λ) and s(λ), each of finite
duration, is defined as

• The output yo(t) of a filter with impulse response h(t) when the input is yin(t) = s(t) + n(t) is

• If the filter is a matched filter, then h(λ) = s(t1 - λ) and Eq. above becomes

• Thus the matched filter forms the cross correlation between the received signal
corrupted by noise and a replica of the transmitted signal.

• The replica of the transmitted signal is "built in" to the matched filter via the
frequency-response function.

• If the input signal yin (t) were the same as the signal s(t) for which the matched
filter was designed (that is, the noise is assumed negligible), the output would be
the autocorrelation function.
 Cross correlation receiver(correlation detection)

• Equation above describes the output of the matched filter as the cross correlation
between the input signal and a delayed replica of the transmitted signal.
• This implies that the matched-filter receiver can be replaced by a cross-correlation
receiver that performs the same mathematical operation as shown in Fig.

Fig: Block diagram of a cross-correlation receiver

• The input signal y (t) is multiplied by a delayed replica of the transmitted signal
s(t - Tr), and the product is passed through a low-pass filter to perform the
integration.

• The cross-correlation receiver of above Fig tests for the presence of a target at only
a single time delay Tr. Targets at other time delays, or ranges, might be found by
varying Tr. However, this requires a longer search time.

• The cross-correlation receiver and the matched-filter receiver are equivalent

mathematically, the choice as to which one to use in a particular radar application is
determined by which is more practical to implement.

• The matched-filter receiver, or an approximation, has been generally preferred in

the vast majority of applications.