Aerospace and Electronic Systems Magazine May 2017 - 6

Collision Avoidance Radar System for the Bullet Train
simultaneously. The dynamic range and sensitivity to the targets
have driven the design for the radar analog front end that contains
a receiver, a transmitter, and a reference generator. The digital section is composed of a personal computer (PC) and four printed
circuit boards, including a signal generator board (waveform generator), a digital acquisition board, a signal processing board, and
a control board. The data acquisition board samples and digitizes
the intermediate frequency (IF) signal from the front end. Then,
the signal processing board accomplishes range-Doppler processing, detection, and localization of targets. After that, data of targets with information of range, velocity, and azimuth are sent to
the PC that is linked to the signal processing board through a local area network for a future process in other applications, such
as target classification and recognition. The radar control board
coordinates the working of the whole system by sending control
orders to the other three boards.

CONFORMAL ANTENNAS AND SPARSE NONUNIFORM
LINEAR ARRAY
The multichannel receiver in the CA radar is composed of eight
independent linear channels. Each receiving channel is connected
to an antenna, which permits the use of array processing techniques with the linear channel in the direction-of-arrival (DOA)
estimation of the targets. The array elements in the receive array
are designed to be conformal antennas so as not to increase the
aerodynamic drag of the bullet train. The receive antenna system,
a half-circle array of eight conformal antennas, is designed to be
placed on the head of the bullet train. Figure 3a shows the sketch
map of the designed installation.
According to [17]-[20], with the help of a linear interpolation procedure, the problem of DOA estimation of targets with
the designed half-circle array can be directly transformed into a
direction-finding problem in a simple virtual linear array with
the same aperture. For simplicity in the preliminary stage, we
use a linear array (shown in Figure 3b) in the experiment instead of the designed circle array, which will be developed in
our future work.
In order to suppress the grating lobes, which will cause angle
ambiguity in the receive beampattern, SNLA is used instead of
SULA. For the purpose of reducing the peak sidelobe level of the
SNLA, a genetic algorithm [21]-[23] and a simulated annealing
algorithm [24], [25] can be used to find the optimum array element positions under the optimization constraints of the number of
elements, the aperture, and the minimum element spacing. In this
article, a genetic algorithm is used to find the optimized element
positions. Suppose the aperture of the SNLA is L and there are N
array elements in the SNLA. Let di, 1 ≤ i ≤ N, denote the position
of the ith element. In order to keep the aperture, we set d1 = 0, dN
= L. With minimum element spacing dc, the optimization problem
[23] can be expressed as follows:
min

d i − d j ≥ d c > 0,

PSLL = f ( d 2 , d 3 ,..., d N −1 ) s.t. d1 = 0, d N = L,
i, j ∈ Z , 1 ≤ j < i ≤ N


Figure 3.

(a) Sketch map of the designed installation of conformal antennas on the
bullet train. (b) Equivalent experimental linear array.

(1)

Figure 4.

Normalized beampattern of the receive array for two array configurations.

IEEE A&E SYSTEMS MAGAZINE

MAY 2017

Table of Contents for the Digital Edition of Aerospace and Electronic Systems Magazine May 2017