BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a transmission signal
generating unit and a radar transmission device using the same.

2. Description of Related Art
Since the utilization of radio waves is increasing, narrowing
the frequency band of a radar signal is strongly demanded to efficiently use frequencies
between various devices.

In addition, suppressing the spurious components of the
radar signal as low as possible has become an international issue. The development
of technology to narrow the frequency band and to suppress the spurious components
enables to operate the radars within a narrower frequency band. This will contribute
to solve the shortage of frequencies.

As a solution for these problems, a low-power pulse compression
radar is put to practical use.

Japanese Patent Publication Laid-open No. H4-357485
shows a radar signal processing device employing a pulse compression method.
This radar signal processing device transmits a chirp signal (liner FM modulated
signal) as a transmission signal to a relatively moving target, receives a signal
reflected by the moving target as a received signal, then extracts Doppler components
due to the movement of the target from the received signal, and detects the moving
target based on the Doppler components.

In this pulse compression method, a modulated long pulse
is transmitted and after the reception a short pulse is obtained with its signal
to noise ratio (SNR) improved by a pulse compression filter suitable for the modulated
long pulse. This method has many advantages such as the extension of detection range,
the achievement of high range-resolution ability, and the reduction of interference
signals. Thus, the method is applied to many radars.

SUMMARY OF THE INVENTION
In the above pulse compression type radar, a chirp signal
or a phase code modulated signal is used as a transmission signal. These signals
have low side-lobes after pulse compression, but their spectrum widths are wide
and many spurious components are included.

In order to suppress the spurious components, a tapering
is applied to edge parts of the waveform of a transmission signal, however, the
more the spurious components are suppressed, the lower the level of the signal becomes.

Although the above trade-off relationship cannot be solved
completely, it seems to be possible to show the limit of narrowing the frequency
band of a transmission signal while maintaining the signal level for a spurious
level to be low. However, there is no conventional method that shows the limit and
therefore the feasible limit of the performance has been unknown.

An object of the present invention is to provide a transmission
signal generating unit and a radar transmission device using the same which enable
to suppress spurious components of a transmission signal and achieve the maximum
signal level of a center frequency of the transmission signal.

To achieve the above described object, the transmission
signal generating unit of the present invention comprises a window function calculator
that generates a window function that makes all frequencies without a center frequency
of an input signal and its adjacent frequencies zero and makes a signal to noise
ratio of the center frequency maximum; and a transmission signal generator that
generates a transmission signal whose amplitude is modulated based on the window
function generated by the window function calculator.

BRIEF DESCRIPTION OF THE DRAWINGS

- Fig. 1 shows the schema of a radar device employing a transmission signal generating
unit according to an embodiment of the present invention.
- Fig. 2 shows the structure of the transmission signal generating unit according
to the above embodiment.
- Fig. 3 shows effective data for the transmission signal generating unit according
to the above embodiment.
- Fig. 4 shows the structure of a radar transmission device having the transmission
signal generating unit according to the above embodiment.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT
There will be below explained a transmission signal generating
unit according to an embodiment of the present invention in detail with reference
to several figures.

Fig. 1 shows the schema of a radar device employing the
transmission signal generating unit according to the embodiment of the present invention.

The radar device comprises a transmission signal generating
unit 10, a D/A converter 11, a local oscillator 12, a transmitting side mixer 13,
a transmission signal amplifier 14, a circulator 15, an antenna 16, a received signal
amplifier 17, a receiving side mixer 18, an A/D converter 19, a pulse compressor
20, a frequency analyzer 21, and a target detector 22.

The transmission signal generating unit 10 generates a
digital signal (pulse signal) as a transmission signal and transmits it to the D/A
converter 11. The D/A converter 11 converts the transmission signal transmitted
by the transmission signal generating unit 10 to an analog signal and transmits
it to the transmitting side mixer 13. The local oscillator 12 generates a local
signal having a local frequency and transmits it to the transmitting side mixer
13 and the receiving side mixer 18. The transmitting side mixer 13 mixes the transmission
signal transmitted by the D/A converter 11 and the local signal transmitted by the
local oscillator 12 to obtain a radio frequency signal and transmits it to the transmission
signal amplifier 14.

The transmission signal amplifier 14 amplifies the radio
frequency signal transmitted by the transmitting side mixer 13 to a predetermined
signal level and transmits it to the circulator 15. The circulator 15 switches between
the first operation that outputs the radio frequency signal transmitted by the transmission
signal amplifier 14 to the antenna 16 and the second operation that outputs a received
signal received by the antenna 16 to the received signal amplifier 17.

The antenna 16, such as an array antenna, transmits the
radio frequency signal, transmitted by the transmission signal amplifier 14 through
the circulator 15, toward a target. Also, the antenna 16 receives a reflected wave
from the target and then transmits it to the circulator 15 as a received signal.

The received signal amplifier 17 amplifies the received
signal, received from the antenna 16 through the circulator 15, with a low noise
and transmits it to the receiving side mixer 18. The receiving side mixer 18 converts
the received signal received from the received signal amplifier 17 to an intermediate
frequency signal (IF signal) by mixing the received signal and the local signal
received from the local oscillator 12 and transmits it to the A/D converter 19.
The A/D converter 19 converts the IF signal transmitted by the receiving side mixer
18 to a digital signal and transmits it to the pulse compressor 20.

The frequency analyzer 21 performs Fourier transformation
on a signal compressed by the pulse compressor 20 to transform data from time-domain
to frequency-domain. Then, the received signal is decomposed to detect the relative
speed of the target. The target detector 22 extracts Doppler components from the
decomposed components, which represent the speed components of the target, to detect
the target.

Next, there is explained the detail of the transmission
signal generating unit 10 according to the embodiment of the present invention.

Fig. 2 shows the structure of the transmission signal generating
unit 10 in detail. The transmission signal generating unit 10 comprises a window
function calculator 31 and a transmission signal generator 32.

The window function calculator 31 generates a window function
H that makes all frequencies without a center frequency of an input signal (phase-modulated
rectangular pulse) and its adjacent frequencies zero and that makes the SNR of the
center frequency maximum, and transmits the generated window function H to the transmission
signal generator 32. The detail of the window function calculator 31 will be explained
later.

The transmission signal generator 32 generates a transmission
signal by modulating the amplitude of the input signal using the window function
H transmitted by the window function calculator 31.

However, there is explained a transmission signal generating
method, in particular, how to calculate the window function H in the window function
calculator 31.

[Method of generating spurious-free transmission signal]
There is shown how to calculate a window function H that
theoretically makes a filter with loss minimum under a constraint condition to make
a transmission signal spurious free, below called a "spurious free condition".

Let W be a weight vector corresponding to the sampled data
of a transmission pulse,
$$\mathbf{W}=\left[{w}_{1},\phantom{\rule{1em}{0ex}},{w}_{2},\phantom{\rule{1em}{0ex}},\cdots ,\phantom{\rule{1em}{0ex}},{w}_{{N}_{\mathit{1}}}\right]$$

where the subscript "*N*_{f}
" denotes all sampling numbers of the transmission pulse in an aperture time.

Further, let y be a spectrum pattern vector expressing
the frequency spectrum of these data,
$$\mathbf{y}=\left[{y}_{1},\phantom{\rule{1em}{0ex}},{y}_{2},\phantom{\rule{1em}{0ex}},\cdots ,\phantom{\rule{1em}{0ex}},{y}_{{N}_{\mathit{1}}}\right]$$

This spectrum pattern vector y comprises outputs at respective frequencies (discrete
sample points) on a frequency space.

Then, we can describe a relationship between the weight
vector W and the spectrum pattern vector y as
$${\mathbf{y}}^{T}={\mathbf{QW}}^{T},$$
$$\mathbf{Q}=\left(\begin{array}{ccc}{q}_{11}& \cdots & {q}_{1{N}_{f}}\\ \vdots & \ddots & \vdots \\ {q}_{{N}_{\mathit{f}}1}& \cdots & {q}_{{N}_{f}{N}_{f}}\end{array}\right),$$
$${q}_{\mathit{nk}}={e}^{-j\frac{2\mathit{\&pgr;}}{{N}_{\mathit{f}}}(n-1)\left(k,-,1\right)}$$

where "Q" represents a fast Fourier transform matrix (FFT matrix) and
*n,k=1,2,...,N*_{f}. It is noted here that the subscript "*N*_{f}
" defined above also represents the number of FFT points, and the superscript
"*T*" represents transpose.

The inverse matrix (IFFT matrix) of the FFT matrix (4)
is calculated as
$$\hat{\mathbf{Q}}=\frac{1}{{N}_{f}}\mathbf{Q}*$$

where "*" denotes complex conjugate.

It is noted that the convolution of the spectrum pattern
calculated by (3) and that of an input pulse is a spectrum pattern to be observed.

Now, let us suppose the width of the transmission pulse
satisfying a predetermined basic performance, such as range resolution ability,
as effective data, and suppose that the effective data is in the central area of
the weight vector W as shown in Fig. 3.

Then, a weight vector W_{m} is expressed as
$$\begin{array}{l}{\mathbf{W}}_{m}^{T}={\mathbf{uW}}^{T}\\ \phantom{\rule{5em}{0ex}}={\mathit{\&agr;}}^{*}\mathbf{u}\{{\left({\mathbf{u}}^{T},,{\mathbf{u}}^{*}\right)}^{-1}{\}}^{T}{\mathbf{u}}^{T*}{\mathbf{S}}^{T*}\end{array}$$
that makes outputs from a spurious-frequency area zero and the SNR of the center
frequency maximum. Here "S" is a steering vector showing the center frequency.

Thus, a window function H except for a constant term is
represented by
$$\mathbf{H}={\mathbf{u}}^{*}\left\{{\left({\mathbf{u}}^{T},,{\mathbf{u}}^{*}\right)}^{-1}\right\}{\mathbf{u}}^{T},$$
$$\mathbf{u}=\hat{\mathbf{Q}}{\mathbf{Q}}_{m}\hat{\mathbf{Q}}{\mathbf{Q}}_{s}$$
that is the weight of an aperture excluding the steering vector S.

For reference, the calculation processes from (7) to (9)
is described in detail below.

(1) Method of generating spurious-free filter
When the data outside the effective data are made zero
as shown in Fig.3, a weight vector is represented by
$${\mathbf{W}}_{s}=\left[\underset{m}{\underset{\u23df}{0\cdots 0}},\phantom{\rule{1em}{0ex}},{w}_{m+1},\cdot ,\cdot ,\cdot ,\cdot ,\cdot ,\cdot ,{w}_{{N}_{f}-m},\phantom{\rule{1em}{0ex}},\underset{m}{\underset{\u23df}{0\cdots 0}},\phantom{\rule{1em}{0ex}}\right]\mathrm{.}$$

Moreover, a frequency vector that allows only the outputs on frequency sample points
within a range from the position "a main lobe - *N*_{x}" to the position
"the main lobe + *N*_{x}" and makes outputs on the other frequency
sample points zero is represented by
$${\mathbf{y}}_{m}=\lfloor 0\cdots 0\phantom{\rule{1em}{0ex}}{y}_{K-{N}_{x}}\cdots \phantom{\rule{1em}{0ex}}{y}_{K}\phantom{\rule{1em}{0ex}}\cdots \phantom{\rule{1em}{0ex}}{y}_{K-{N}_{x}}\phantom{\rule{1em}{0ex}}0\cdots 0\rfloor \mathrm{.}$$

Here "K" is the number representing the main lobe (i.e. center frequency) of a frequency
filter to be observed, and therefore the other frequency sample points are supposed
as side-lobes.

Then, we can describe a relationship between the weight
vector W_{s} and the frequency vector y_{m} as
$${\mathbf{y}}_{m}^{T}={\mathbf{Q}}_{m}{\mathbf{W}}_{s}^{T},$$
$${\mathbf{Q}}_{m}=\left[\begin{array}{ccc}0& \cdots & 0\\ \vdots & \ddots & \vdots \\ 0& \cdots & 0\\ {q}_{k-{N}_{f}.1}& \cdots & {q}_{k-{N}_{f}\mathrm{.}{N}_{f}}\\ \vdots & \phantom{\rule{1em}{0ex}}& \vdots \\ {q}_{k-{N}_{f}.1}& \cdots & {q}_{k-{N}_{f}\mathrm{.}{N}_{f}}\\ 0& \cdots & 0\\ \vdots & \ddots & \vdots \\ 0& \cdots & 0\end{array}\right]\mathrm{.}$$

As described above, since the frequency vector y_{m}
shows a spectrum pattern where all side-lobes without the main lobe and its neighborhood
are made zero. When the weight vector W=W_{m} that satisfies
$${\mathbf{y}}_{m}^{T}={\mathbf{QW}}^{T}$$
is the weighted vector to be intended. Here (14) is obtained by substituting y_{m}
into (3). When (12) is set as a constraint condition for (14), there is obtained
$${\mathbf{y}}_{m}^{T}={\mathbf{QW}}_{m}^{T}={\mathbf{Q}}_{m}{\mathbf{W}}_{s}^{T}\mathrm{.}$$
Accordingly, the weight vector W_{m} to be obtained is represented by
$${\mathbf{W}}_{m}^{T}=\hat{\mathbf{Q}}{\mathbf{Q}}_{m}={\mathbf{W}}_{s}^{T}\mathrm{.}$$

(2) Method of maximizing filter output
"SNR" is defined as the ratio (unit: dB) of noise to an
output signal, and the SNR of the present case is represented by
$$\mathit{SNR}=\frac{\left({\mathbf{SW}}_{m}^{T}\right){\left({\mathbf{SW}}_{m}^{T}\right)}^{T*}}{{{\mathbf{W}}_{m}\mathbf{W}}_{m}^{T*}}$$

where "S" is a vector that shows the series of sample values of the input signal
corresponding to the center frequency of the frequency filter, and is written as
$$\mathbf{S}=\lfloor 1\phantom{\rule{1em}{0ex}}{e}^{-\mathit{j\&phgr;}}{\phantom{\rule{1em}{0ex}}e}^{-\mathit{j}\mathrm{2}\mathit{\&phgr;}}\cdots \phantom{\rule{1em}{0ex}}{e}^{-\mathit{j}\left({N}_{\mathit{f}},-,1\right)\mathit{\&phgr;}}\rfloor ,$$
$$\mathit{\&phgr;}=-2\mathit{\&pgr;}\frac{K-1}{{N}_{f}}\mathrm{.}$$

Under the above described constraint condition (side-lobe
free condition), the following identity is introduced to obtain the weight vector
W_{m} which makes the SNR represented by (17) maximum,
$${\mathbf{QW}}_{s}^{T}={\mathbf{Q}}_{s}{\mathbf{W}}^{T}$$

where
$${\mathbf{Q}}_{s}=\begin{array}{ccc}\underset{m}{\begin{array}{c}\underset{\u23df}{\begin{array}{c}[\begin{array}{ccc}0& \cdots & 0\\ \vdots & \phantom{\rule{1em}{0ex}}& \vdots \\ \vdots & \ddots & \vdots \\ \vdots & \phantom{\rule{1em}{0ex}}& \vdots \\ 0& \cdots & 0\end{array}\end{array}}\end{array}}& \underset{N}{\begin{array}{c}\underset{\u23df}{\begin{array}{ccc}{q}_{1,m+1}& \cdots & {q}_{1,{N}_{f}-1}\\ \vdots & \phantom{\rule{1em}{0ex}}& \vdots \\ \vdots & \phantom{\rule{1em}{0ex}}& \vdots \\ \vdots & \phantom{\rule{1em}{0ex}}& \vdots \\ {q}_{{N}_{f},m+1}& \cdots & {q}_{{N}_{f},{N}_{f-1}}\end{array}}\end{array}}& \underset{m}{\begin{array}{c}\underset{\u23df}{\begin{array}{c}\begin{array}{ccc}0& \cdots & 0\\ \vdots & \phantom{\rule{1em}{0ex}}& \vdots \\ \vdots & \ddots & \vdots \\ \vdots & \phantom{\rule{1em}{0ex}}& \vdots \\ 0& \cdots & 0\end{array}]\end{array}}\end{array}}\end{array}$$

Note here that (20) is rewritten as
$${\mathbf{W}}_{s}^{T}=\hat{\mathbf{Q}}{\mathbf{Q}}_{s}{\mathbf{W}}^{T}\mathrm{.}$$

Using (22), (16) can be rewritten as
$$\begin{array}{ll}{\mathbf{QW}}_{m}^{T}& =\hat{\mathbf{Q}}{\mathbf{Q}}_{m}{\mathbf{W}}_{s}^{T}\\ \phantom{\rule{1em}{0ex}}& =\hat{\mathbf{Q}}{\mathbf{Q}}_{m}\hat{\mathbf{Q}}{\mathbf{Q}}_{S}{\mathbf{W}}^{T}\\ \phantom{\rule{1em}{0ex}}& ={\mathbf{uW}}^{T}\end{array}$$

where
$$\mathbf{u}=\hat{\mathbf{Q}}{\mathbf{Q}}_{m}\hat{\mathbf{Q}}{\mathbf{Q}}_{s}\mathrm{.}$$

Further, the denominator (noise output) of (17) is rewritten as
$$\begin{array}{ll}{{\mathbf{W}}_{m}\mathbf{W}}_{m}^{T*}& =({\mathbf{uW}}^{T}{)}^{T}({\mathbf{uW}}^{T}{)}^{*}\\ \phantom{\rule{1em}{0ex}}& ={\mathbf{Wu}}^{T}{\mathbf{u}}^{*}{\mathbf{W}}^{T*}\\ \phantom{\rule{1em}{0ex}}& ={\mathbf{WzW}}^{T*}\end{array}$$

where
$$\mathbf{z}={\mathbf{u}}^{T}{\mathbf{u}}^{*}\mathrm{.}$$

Likewise, the numerator (signal output) of (17) is rewritten as
$$\begin{array}{l}\left({\mathbf{S\; W}}_{m}^{T}\right){\left({\mathbf{S\; W}}_{m}^{T}\right)}^{T*}=\left({\mathbf{S\; uW}}^{T}\right)({\mathbf{S\; uW}}^{T}{)}^{T*}\\ \phantom{\rule{24em}{0ex}}=\left({\mathbf{v\; W}}^{T}\right)({\mathbf{v\; W}}^{T}{)}^{T*}\end{array}$$

where
$$\mathbf{v}\equiv \mathbf{Su}\mathrm{.}$$

Thus, using the re-defined vectors
$$\mathbf{F}\equiv {\mathbf{vz}}^{-\frac{1}{2}*},\phantom{\rule{1em}{0ex}}\mathbf{G}\equiv {{\mathbf{W}}^{*}\mathit{z}}^{\frac{1}{2}*}$$
and Schwarts's inequality for arbitrary vectors F and G
$$\left({\mathbf{FG}}^{T*}\right){\left({\mathbf{FG}}^{T*}\right)}^{T*}\leqq \left({\mathbf{FF}}^{T*}\right)\left({\mathbf{GG}}^{T*}\right),$$
(17) can be rewritten as
$$\begin{array}{ll}\mathit{SNR}& =\frac{\left({\mathbf{V\; W}}_{m}^{T}\right){\left(\mathbf{v},\phantom{\rule{1em}{0ex}},{\mathbf{W}}^{T}\right)}^{T*}}{{{\mathbf{W}}_{m}\mathbf{W}}_{m}^{T*}}\\ \phantom{\rule{1em}{0ex}}& =\frac{\left({\mathbf{F}\mathbf{\cdot}\mathbf{G}}^{T*}\right){\left({\mathbf{F}\mathbf{\cdot}\mathbf{G}}^{T*}\right)}^{T*}}{{\mathbf{G}}^{*}{\mathbf{G}}^{T}}\\ \phantom{\rule{1em}{0ex}}& \le \frac{\left({\mathbf{F}\mathbf{\cdot}\mathbf{F}}^{T*}\right)\left({\mathbf{G}\mathbf{\cdot}\mathbf{G}}^{T*}\right)}{{\mathbf{G}}^{*}{\mathbf{G}}^{T}}\\ \phantom{\rule{1em}{0ex}}& ={\mathbf{F}\mathbf{\cdot}\mathbf{F}}^{T*}\\ \phantom{\rule{1em}{0ex}}& ={\mathbf{v\; z}}^{-\frac{1}{2}*}\cdot {\mathbf{z}}^{-\frac{1}{2}*}{\mathbf{v}}^{T*}\\ \phantom{\rule{1em}{0ex}}& ={\mathbf{v\; z}}^{-1*}{\mathbf{v}}^{T*}\mathrm{.}\end{array}$$

When the equality is established in (31), the SNR takes
a maximum value. Then, the condition for the equality is given as
$$\mathbf{G}=\mathit{\&agr;}\mathbf{F}$$

where &agr; is a constant.

(3) Derivation of side-lobe-free filter coefficient
Substituting (29) into (32), we get
$$\begin{array}{l}\mathbf{G}=\mathit{\&agr;}\mathbf{F}\\ \leftrightarrow {\mathbf{W}}^{*}{\mathbf{z}}^{\frac{1}{2}*}=\mathit{\&agr;}\phantom{\rule{1em}{0ex}}{\mathbf{v\; z}}^{-\frac{1}{2}*}\\ \leftrightarrow {\mathbf{W}}^{*}=\mathit{\&agr;}\phantom{\rule{1em}{0ex}}{\mathbf{v\; z}}^{-1*}\\ \leftrightarrow \mathbf{W}={\mathit{\&agr;}}^{*}{{\mathbf{v}}^{*}\mathbf{z}}^{-1}={\mathit{\&agr;}}^{*}{\mathbf{S}}^{*}{\mathbf{u}}^{*}({\mathbf{u}}^{T}{\mathbf{u}}^{*}{)}^{-1}\mathrm{.}\end{array}$$
Substituting (33) into (16), we can obtain the weight vector W_{m} to be
intended as
$$\begin{array}{l}{\mathbf{W}}_{m}^{T}={\mathbf{uW}}^{T}\\ \phantom{\rule{6em}{0ex}}={\mathit{\&agr;}}^{*}\mathbf{u}\{{\left({\mathbf{u}}^{T},,{\mathbf{u}}^{*}\right)}^{-1}{\}}^{T}{\mathbf{u}}^{T*}{\mathbf{S}}^{T*}\\ \phantom{\rule{6em}{0ex}}={\mathit{\&agr;}}^{*}{\mathbf{H}}^{T}{\mathbf{S}}^{T*},\phantom{\rule{7em}{0ex}}\cdots \phantom{\rule{3em}{0ex}}\left(34\right)\\ \therefore {\mathbf{W}}_{m}={\mathit{\&agr;}}^{*}{\mathbf{S}}^{*}\mathbf{H}\mathbf{,}\phantom{\rule{2em}{0ex}}\mathbf{H}={\mathbf{u}}^{*}({\mathbf{u}}^{T}{\mathbf{u}}^{*}{)}^{-1}{\mathbf{u}}^{T}\mathrm{.}\phantom{\rule{6em}{0ex}}\cdots \phantom{\rule{4em}{0ex}}\left(35\right)\end{array}$$

This window function H has a filter band width corresponding
to the number of effective data set initially and makes the SNR maximum under the
side-lobe free condition. It is clear that above calculations do not use any convergence
method.

By using the window function H obtained above it is possible
to generate a transmission signal where the SNR of the center frequency of the input
signal is made maximum and the spurious components of the input signal are reduced.
That is, according to the transmission signal generating unit 10, since the signal
loss of the center frequency is made minimum, the signal level can be ensured and
the frequency band can be narrowed.

In the above example, the waveform of the input signal
formed with the data number *N*_{f}
including a predetermined center frequency is defined as an original waveform
and further the original waveform is defined as the steering vector S. Then the
window function H is applied to generate the weighted vector W corresponding to
the transmission signal. It is however possible to store the window function H that
is pre-calculated in the above steps in a memory unit (not shown).

It is also possible to use a signal with a predetermined
frequency as the original waveform and also to use a frequency-modulated waveform,
such as a chirp signal, as the original waveform. In addition, it is also possible
to transmit continuously or intermittently a plurality of phase-modulated pulses
with a waveform whose amplitude is modulated using the above window function.

The transmission signal generating unit 10 according to
the present embodiment comprises: the window function calculator 31 that calculates
a window function that makes all frequencies without a center frequency of an input
signal and its adjacent frequencies zero and makes the SNR of the center frequency
maximum; and the transmission signal generator 32 that generates a transmission
signal whose amplitude is modulated in a shape of an envelope curve.

This enables to generate the transmission signal where
the spurious components are reduced and the signal level of the center frequency
is made maximum.

When the direct generation of the transmission signal is
difficult, it is also possible to make a required center frequency by frequency-converting
the transmission signal from the transmission signal generating unit 10 to a signal
with a higher frequency.

Fig. 4 shows the structure of a radar transmission device
which is applied with the transmission signal generating unit according to the present
embodiment.

A radar transmission device 40 comprises an intermediate
frequency signal (IF signal) generating unit 10a as the transmission signal generating
unit 10 in Fig. 1, a local signal generator 34 (same as the local oscillator 12)
that generates a local signal, a frequency converter 33, and a high-frequency signal
transmitter 35. The frequency converter 33 frequency-converts (up-convert) an output
signal from the IF signal generating unit 10a, using the local signal, to a frequency
signal having a higher frequency than that of the output signal. The high-frequency
signal transmitter 35 transmits the frequency signal frequency-converted by the
frequency converter 33.

The radar transmission device 40 is applied with the transmission
signal generating unit 10a according to the present embodiment. Thus, it generates
a transmission signal by modulating the amplitude of an input signal based on a
window function that makes, for the input signal, all of the outer frequencies excluding
a center frequency and its adjacent frequencies zero and at the same time makes
the SNR of the center frequency maximum. It is therefore possible to suppress spurious
components and to make the signal level of the center frequency maximum. Such radar
transmission device is applicable to transmission units of radar systems, and so
on.

This application is based upon the
Japanese Patent Applications No. 2006-145733, filed on May 25, 2006
, the entire content of which is incorporated by reference herein.