Aerospace and Electronic Systems Magazine March 2018 - 25

Ji et al.

Figure 5.
R-T data.

Figure 7.

R-T and D-T image of two closing targets. (a) R-T image at the 111
Doppler bin. (b) D-T image at the 50 range bin.

Figure 6.

D-T image at the 65 range bin.

Therefore, the R-D-T data can be used to track moving targets by integrating detection and tracking. The integration method
makes full use of the location and echo amplitude information of the
targets and avoids the association process of target points; traces exist in the traditional detection before the tracking process. Thus, the
integration method can significantly improve tracking performance
for HFSWR, making it particularly suitable for compact HFSWR.

INTEGRATION METHOD PROCESS
STEP 1: CONSTRUCTION OF R-D-T DATA
To construct R-D-T data, some parameters should be set flexibly
when considering a maneuvering target, such as the coherent integration time of each frame TCIT, the interval between each frame
ΔT, and the number of frames in each batch N. For most moving
targets, TCIT can be set between 150 and 300 s to guarantee a sufficient velocity resolution and ΔT can be set as 60 s. For a maneuvering target, values of both TCIT and ΔT should be appropriately
reduced. Each batch R-D-T can take 6-10 min to ensure that the
real target and false target can be distinguished. To reduce the time
interval of detection results, two consecutive batches of R-D-T
data can be overlapped.
Some data preprocessing is also required, including clutter
suppression and predetection using a low threshold. Clutter suppression is used mainly to suppress the first-order sea clutter and
radio frequency interference. Predetection with a low threshold is
used to maintain the area target information as far as possible for
weak signals so that integrated detection can be effectively perMARCH 2018

formed. Noise and clutter can cause the appearance of false targets,
but these can be eliminated in subsequent processing.

STEP 2: INTEGRATION TRACKING WITHOUT AZIMUTH
INFORMATION
Based on the constructed R-D-T data, the integration of tracking
and detection can be realized by employing a tracking-before-detection method. Here, a dynamic programming method was used
[12]. For the target motion and measurement models, the target
state vector X(k) at time k is defined in R-D coordinates as

(

)

X k = rk , rk , vk , vk 

(3)

where rk, vk and rk, vk are the range and velocity components along
the range and radial velocity directions, respectively, and (·)T is the
transpose operator. For a moving target, rk = vk. Then, the target
T
state vector reduces to X k = rk , vk , vk  . According to the relationships among range, velocity, and acceleration, the state-update
equation can be derived as

(

X k +1 = FX k

)

(4)

1 T T 2 / 2 


T  is the state transform matrix, whereas T
where F = 0 1
0 0
bk 

is the time interleaved between sampling times k − 1 and k. Here,
bk is the acceleration ratio, and bk = 1 when the target is moving at
constant velocity or accelerating.

IEEE A&E SYSTEMS MAGAZINE

Table of Contents for the Digital Edition of Aerospace and Electronic Systems Magazine March 2018