J. Electromagn. Eng. Sci Search

CLOSE


J. Electromagn. Eng. Sci > Volume 26(4); 2026 > Article
Park, Seo, and Choi: Least-Squares Range Measurement of FM Ranging Radar for Maneuvering Target

Abstract

This paper introduces a least squares (LS)-based method designed to enhance the accuracy of range measurements in frequency modulated interrupted continuous wave (FMICW) pulse-Doppler radars, specifically addressing the challenge of accurately tracking highly maneuvering targets. Conventional FM ranging methods assume constant velocity within dwell periods, leading to significant biases in range estimation when targets experience rapid acceleration or jerk. Our proposed approach incorporates acceleration and jerk parameters into the radar measurement model, utilizing multiple FMICW bursts with varied slopes to jointly estimate range, velocity, and higher-order motion parameters. Theoretical analyses, supported by the calculation of the Cramér-Rao bound, demonstrate the method’s superior estimation accuracy and optimality. Simulation and flight test results using an X-band AESA radar system validate the method’s effectiveness, showing a substantial reduction in range bias and enabling earlier and more reliable target detection compared to conventional methods.

I. Introduction

Accurate range measurement and reliable tracking of highly maneuvering targets, such as modern fighter aircraft, pose significant challenges in radar signal processing. Traditional radar techniques often assume that target velocity remain nearly constant within a dwell time, an assumption that becomes problematic in long-range radar applications or when tracking targets undergoing rapid maneuvers [1–4]. In particular, conventional FM ranging-based radars, which utilize frequency modulated interrupted continuous wave (FMICW) waveforms, typically rely on the assumption of constant range and velocity during dwell periods [5–7]. However, as target dynamics increase in complexity, this assumption becomes invalid, leading to significant range estimation biases and compromised detection To address these challenges, recent studies have explored advanced waveform designs and estimation techniques tailored for handling high-speed and accelerating targets. FMICW radars employ waveforms with varying FM slopes to resolve range ambiguities and improve target discrimination, especially in clutter-free airborne radar scenarios [7, 8]. Nonetheless, despite these advances, existing FM ranging methods remain vulnerable to performance degradation when targets exhibit significant acceleration or jerk, as these motion parameters induce errors into beat frequency-based range estimations.
In this paper, we propose a least squares (LS)-based approach for range measurement in FMICW pulse-Doppler radars, explicitly accounting for target acceleration within dwell times. By extending the conventional model to incorporate acceleration and jerk terms into the estimation framework, our method leverages multiple bursts to jointly estimate range, velocity, and higher-order motion parameters. We demonstrate theoretically, through the derivation of the Cramér-Rao bound (CRB), that our proposed approach yields improved accuracy and robustness compared to traditional methods. Additionally, simulation and flight test results, obtained using an X-band active electronically scanned array (AESA) radar system, validate the practical effectiveness of our approach. These evaluations illustrate that the proposed method significantly reduces range bias under maneuvering target conditions, enabling earlier detection and more stable tracking performance.
It is important to position this work within the broader context of radar signal processing. The scope of this paper is focused on the ‘measurement generation’ stage, which aims to produce a high-fidelity state estimate (range, velocity, and acceleration) from the beat frequency measurements acquired within a single Dwell in search phase. This contrasts with ‘tracking filtering’ techniques (such as Kalman or particle filters), which operate at a higher level by processing a sequence of such measurements over multiple Dwells to maintain a target’s track. The LS-based method presented here is therefore not an alternative to tracking filters. Instead, it provides high-quality, unbiased measurements as input to a tracking filter. The primary contribution of this paper is solving the significant measurement bias inherent in conventional FM ranging methods when observing maneuvering targets, a step that is essential for any subsequent tracking filter to perform effectively.
This paper is organized as follows: Section II reviews the conventional range measurement methods using FM ranging radar and examines the effects of target acceleration. Section III describes the proposed LS-based approach for range estimation. Section IV presents the results of range measurements obtained through simulations and flight tests. Finally, Section V concludes the paper.

II. Conventional FM Ranging Radar and Effect of Acceleration

Conventional pulse-Doppler radar systems initially obtain ambiguous range and velocity measurements. This ambiguity arises from fundamental limitations of the pulse waveform parameters: the maximum unambiguous range is limited by the pulse repetition interval (PRI), causing targets beyond this limit to appear at folded ranges, while the maximum unambiguous velocity is limited by the pulse repetition frequency (PRF), resulting in Doppler frequency aliasing. Such ambiguities are unavoidable in long-range or high-speed scenarios [9–11]. These ambiguities are then resolved using information from multiple bursts with different pulse repetition frequencies (PRF), often employing methods such as the Chinese Remainder Theorem [9]. In contrast, FM ranging-based pulse-Doppler radar utilizes an FMICW waveform, where the center frequency varies over time. This radar employs a high PRF (HPRF) waveform, detecting ambiguous ranges while simultaneously resolving these ambiguities using the range derived from the beat frequency, which represents the frequency difference between the transmitted and received waveforms for each burst. This method is frequently employed in aircraft radar systems, where long-range detection in clutter-free environments is required.
Fig. 1 illustrates the conceptual diagram of FM ranging based on FMICW waveforms. The radar utilizes three different slopes for the transmitted center frequency, as shown by the blue dotted lines. The common choice of FM slopes is no-FM, FM-up, FM-down [10], but any combination of FM slopes that satisfy specific conditions, such as clutter spreading [8] and range ambiguity [5], can be utilized. The FMICW waveform is transmitted by sampling the pulse signals based on the blue FM slopes. In this representation, the center frequency of each pulse (solid line) is treated as a single discrete sample of the overall FM slope (dotted line). The received signal, incorporating a time delay td and Doppler frequency fd, is received with a center frequency indicated by the red dotted lines.
This can be expressed mathematically as follows. When the transmitted waveform is stx(t), the received waveform after down-conversion, srx (t), is given by:
(1)
srx(t,τ)=Arrect(t-td(τ)Tp)stx(t-td(τ))×exp(j2πfdτ) exp [jπkfm(τ-td(τ))2],
where t is fast-time, τ is slow-time, Ar is complex amplitude of received signal, td (τ) is time delay of the return signal at τ, Tp is pulse width, fc is center frequency, kfm is frequency modulation slope.
Assuming the pulse compression result is modeled as a sinc function and the received waveform is de-chirped with respect to the transmitted waveform, the resulting signal after pulse compression and de-chirp, spc (t, τ) and sdc (t, τ), can be respectively expressed as:
(2)
spc(t,τ)≈Apc sinc[Bp(t-td(τ))]×exp(j2πfdτ) exp[jπkfm(τ-td(τ))2],
(3)
sdc(t,τ)≈Adc sinc[Bp(t-td(τ))]×exp(j2πfdτ) exp[-j2πkfmtd(τ)τ],
where Bp is pulse bandwidth, Apc, Adc is complex amplitude of pulse compressed/de-chirped signal, respectively. Note that, the first component represented as a sinc function, causes range migration, which can be compensated within the burst [1, 12].
The beat frequency can be derived by differentiating phase term of Eq. (3). The beat frequency of the i-th burst, fb,i, can be expressed as [11]:
(4)
fb,i=-2viλ-ki2ric0
where ki is frequency modulation slope of i-th burst, ri, vi is range and radial velocity of the target of i-th burst, c0 is speed of light, λ is wavelength.
When beat frequency measurements from two bursts with different FM slopes are available, the radar can measure the range for a single target. However, in the presence of multiple targets or false detections caused by noise, several ghosts detection may appear [10, 11]. To address this, FM ranging-based radars can use three or more bursts with different FM slopes in a single dwell for multi-target detection. When there are three or more bursts, the radar clusters detection signals originating from the same target using combinations by identifying combinations that satisfy the following Eq. (5) for i-th, i-th, and l-th bursts. Of course, in practical applications, measurement errors are taken into account, and clustering is performed when the difference between the left-hand side and right-hand side is below a certain threshold.
(5)
kifb,j-kjfb,iki-kj=kjfb,l-klfb,jkj-kl.
Conventional FM ranging-based radars assume that the range and velocity remain constant within a dwell (ri ≈ rj, vi ≈ vj ). Under this assumption, the range can be simply calculated as
(6)
ri=c02fb,i-fb,jkj-ki.
With this expression, range can be calculated from any pair of bursts within a single dwell. Without loss of generality, range is calculated from the first and second bursts (i = 2, i = 1) in this chapter.
This constant-velocity assumption is valid when the target’s maneuver is sufficiently slow, causing negligible movement within the dwell time. However, if the target’s acceleration is high or changes significantly during the dwell time, this can lead to two problems in the conventional method. Firstly, since the Doppler shifts vary rapidly across bursts, the left-hand and right-hand values in Eq. (5) for the same target signal may differ. Consequently, the threshold for the difference between the left and right sides of Eq. (5) must be increased to enable target detection. Furthermore, even if signals from the same target are clustered by increasing the threshold, FM ranging measures range based on the beat frequency. As a result, the frequency variation caused by acceleration leads to range errors. Considering the target’s acceleration, v2 = v1 + ∫ a(t)dt. Substituting this into the Eq. (6) yields the following range error term:
(7)
ɛR=-∫a(t)dtkj-kifc
where fc is center frequency of the radar waveform.
This error can become significant when the target’s acceleration is high, and the dwell time is long. For instance, in a X-band radar system with a center frequency of 10 GHz, a difference in FM slope between bursts of 10 MHz/s, a target line-of-sight acceleration of 10 m/s2, and a burst time of 100 ms, the error is calculated to be approximately 1 km. Therefore, for long-range radars with extended dwell times or radars targeting high-speed maneuvering objects such as fighter jets, it is necessary to account for acceleration when measuring the range.

III. Least Square Range Measurement for Accelerated Target

This chapter describes a proposed range measurement method based on least squares. To begin with the problem statement, it is assumed that the target’s acceleration, a0, remains constant within the dwell time. When assuming constant acceleration, at least three measurements from different bursts are required. At first, the scenario with three bursts is considered; however, the same method can be applied when there are more than three bursts. The derivation for the case assuming four bursts is described in the latter part of this chapter.
When the time interval between successive bursts are uniformly given as T, the range and velocity of the i-th burst, ri and vi, can be expressed based on the range and velocity of the first burst, r1 and v1, as follows:
(8)
ri=r1+v1(i-1)T+12a0[(i-1)T]2
(9)
vi=v1+a0(i-1)T
By substituting Eq. (8), (9) into Eq. (4), the beat frequency measurements matrix F can be expressed as shown in Eq. (10) below:
(10a)
F=HaX+ɛ
(10b)
F=(fb,1fb,2fb,3)
(10c)
X=(r1v1a)
(10d)
Ha=(-2k1c0-2λ0-2k2c0-2k2Tc0-2λ-k2T2c0-2Tλ-2k3c0-4k3Tc0-2λ-4k3T2c0-4Tλ)
Here, the added ɛ component represents the measurement noise of the beat frequency, which follows Gaussian noise with zero mean and variance σ2(i.e. ɛi~N(0, σ2)). Since the beat frequency is measured independently from each burst, the noise components are assumed to be independent and identically distributed (i.i.d.).
When the measurements follow such a linear model, the least squares solution is the best linear unbiased estimator (BLUE) by the Gauss-Markov theorem. Consequently, the estimates for range, velocity, and acceleration, denoted as X^, are as follows:
(11)
X^=(HaTHa)-1HaTF
Meanwhile, when the target’s maneuvering is very significant, the assumption of constant acceleration may not be sufficient. In such cases, the jerk (time derivative of radial acceleration) component should be considered, and the range can be calculated based on the assumption of constant jerk. With the inclusion of the jerk component, the degrees of freedom (DOF) become 4, requiring beat frequency measurements from at least four bursts. In this case, the range and radial velocity of the i-th burst can be expressed as follows:
(12)
ri=r1+v1(i-1)T+12a1[(i-1)T]2+16J0[(i-1)T]3,
(13)
vi=v1+a1(i-1)T+12J0[(i-1)T]2
Similarly, substituting this into Eq. (4), the beat frequency measurements matrix F is expressed as follows:
(14a)
F=HjX+ɛ
(14b)
F=(fb,1fb,2fb,3fb,4)
(14c)
X=(r1v1a1J0)
(14d)
Hj=(-2k1c0-2λ00-2k2c0-2k2Tc0-2λ-k2T2c0-2Tλ-k2T33c0-T2λ-2k3c0-4k3Tc0-2λ-4k3T2c0-4Tλ-8k3T33c0-4T2λ-2k4c0-6k4Tc0-2λ-9k4T2c0-6Tλ-9k4T3c0-9T2λ)
Then, the state vector estimate X^ can be obtained using Eq. (15), just as in the case of constant acceleration.
(15)
X^=(HjTHj)-1HjTF
In general, FM ranging radars use three bursts to eliminate ghosts while using least number of bursts, making it common to use Ha, which only accounts for acceleration. However, in cases where target detection is required even under significant maneuvering, an additional burst can be allocated to use Hj, which accounts for jerk. Such a design decision will depend on the radar’s requirements, balancing whether to allocate more resources for faster detection and accurate ranging or conserve resources for quicker radar scanning.
Next, to demonstrate the accuracy of the proposed method, the CRB was calculated. The CRB represents the theoretical lower limit on the variance of any unbiased estimator, including the best achievable estimation accuracy under unbiased conditions. According to [13], the CRB for frequency estimation in pulse-Doppler radar is given as follows:
(16)
CRBf=32π2·SNR1·TR2·1Np(Np2-1)
When the measurements are obtained in the LS from as shown in Eqs. (10) and (11), the CRB of measurement X is given as follows [14]:
(17)
CRBX=CRBf(HafHa)-1
This inverse matrix can be mathematically derived, but it is better to be calculated by numerically because of high complexity. For example, if k1, k2, k3 = 0, 10, 5 [MHz/s], T = 100 ms, λ = 0.03 m, (HaTHa)-1 is then:
(18)
(HaTHa)-1=(599.27-0.46821.5607-0.46820.0006-0.00241.5607-0.00240.0163)
CRB of range/velocity/acceleration is diagonal elements of CRBf(HaTHa)-1. For SNR1 = 0.01 (i.e., −20 dB) and Np = 100, TR = 1 ms, CRB of range/velocity/acceleration are (95.43 m)2, (0.06 m/s)2, (0.48 m/s)2, respectively. The CRB for scenarios involving jerk can also be derived in a similar manner, but this derivation is omitted in this paper.
For the conventional ranging method in Eq. (6), CRB of the range CRBR=CRBf c02/2(ki-kj)2. In this case, CRB of range is determined by the FM slope difference between the two bursts. If ki − kj = 10 MHz/s, CRB for the corresponding parameter is (82.65m)2.
While the conventional method uses the information from two bursts to estimate the range, the proposed method uses information from three bursts to jointly estimate range, velocity, and acceleration. As a result, the minimum achievable variance for range estimation may increase compared to the conventional method. Nevertheless, considering that range errors up to half of the PRI can be corrected through ambiguity resolving, as discussed in [5], this limitation has little practical impact. Furthermore, by removing the range bias caused by target acceleration, the proposed method can offer a significant advantage for stable and reliable range estimation.
Regarding real-time implementation, the proposed LS solution X^ = (HTH)−1HTF does not pose a computational bottleneck. The matrix H (for both Ha and Hj models) consists only of pre-determined radar parameters (e.g., ki, T) and is independent of the real-time measurements F. Consequently, the pseudo-inverse matrix, H† = (HTH)−1HT, can be pre-calculated offline and stored. The only real-time operation is a single matrix-vector multiplication, X^ = H†F (e.g., a 3×3 or 4×4 multiplication). This computational load is minimal, negligible compared to the FFTs required to obtain the beat frequencies fb,i, and thus is well-suited for real-time implementation.

IV. Simulation & Flight Test Results

This chapter presents the results of simulations and flight tests to validate the proposed approach. The simulations model a highly maneuvering target with significant acceleration and jerk, demonstrating the detection performance of the conventional FMICW method and the proposed Least Squares-based method. Detection performance was tested using waveforms with three bursts and four bursts to detect targets with acceleration and jerk. In addition, the flight test results describe the detection performance of FMICW waveform-based airborne radar for accelerating targets, comparing the conventional method with the proposed method.
The radar is assumed to be an X-band airborne radar. The simulation scenario for validation under acceleration conditions is as follows: the own aircraft is traveling northward at a speed of 300 m/s, while the target aircraft is traveling southward at 300 m/s. The radial acceleration of the target is assumed to be from 0 m/s2 to 90 m/s2. In this scenario, jerk was assumed to be zero in order to isolate and analyze only the effects of acceleration.
For the simulation, the target’s beat frequencies were first calculated according to Eqs. (4), (12), and (13), and measurement noise corresponding to the given SNR was then added to generate beat frequency measurements for each burst. Using these measurements, the mean range bias and estimation accuracy were evaluated for both the conventional and proposed methods through a Monte Carlo simulation. The number of iterations for the Monte Carlo simulation was set to 108. SNR1 was set to 0.01, the number of pulses Np to 100, the pulse repetition interval TR to 1 ms, and the FM slopes k1, k2, k3 were set to 0, 10, and 5 MHz/s, respectively. In this case, the range measurement results obtained using the conventional method described in Eq. (6) (denoted as linear) and the Least Squares method described in Eq. (10) (denoted as LS) are shown in Fig. 2.
The conventional method exhibited a linear range bias proportional to the target’s acceleration, with the given parameters resulting in approximately 1 km of bias per 10 m/s2 of acceleration. This outcome is consistent with the prediction derived from Eq. (7) in Section II.
Next, in the same scenario, target jerk components of 100, 200, and 300 m/s3 were additionally considered. The resulting range bias measurements are shown in Fig. 3. As illustrated in the figure, when jerk is present, both the conventional method described by Eq. (5) and the LS method based on Ha (Eq. (10)) exhibit increasing range bias proportional to the jerk magnitude. In contrast, the LS method based on Hj (Eq. (14)) maintains zero bias regardless of the jerk value (Note that, the LS method using Hj appears as a single line overlapping at zero, due to its consistent zero bias across all jerk values).
Next, the range accuracy of the proposed method was evaluated. The accuracy of the proposed range estimation method is determined by the precision of beat frequency measurements. Since beat frequency accuracy can vary depending on the radar system, this study assumes that the variance of the measurement is α times larger than the CRB. The values α = 1.5 and α = 3 were used in the experiments. Under these conditions, the sample variance of the range estimates obtained using the conventional method (linear), and the LS methods based on Ha and Hj, are summarized in Table 1.
As anticipated in Section III, the proposed LS method exhibits lower range precision compared to the conventional method. However, it offers the advantage of eliminating bias by simultaneously estimating higher-order terms such as velocity and acceleration.
The selection between the constant acceleration (Ha ) and constant jerk (Hj ) models represents a critical design trade-off between accuracy, precision, and radar resource management. The Hj model is proven to eliminate significant range bias from complex maneuvers (as shown in Fig. 3), but this high accuracy (low bias) comes at the cost of lower precision (higher estimation variance, as seen in Table 1) and increased resource load (four bursts vs. three). The optimal choice is therefore dependent on the specific radar’s mission. For example, a high-performance fire-control radar, where minimizing bias is critical for a weapons-grade track, might be configured to use the Hj model as standard for any target exhibiting maneuvers above a certain threshold. In contrast, a long-range surveillance radar, which must manage resources efficiently across many targets, might be designed to use the Ha model by default, thereby prioritizing lower variance and faster scan updates while accepting a manageable level of bias for non-critical targets.
The flight test results are as follows. The flight test data were collected using an X-band AESA radar mounted on an aircraft, and the target aircraft used for the test was an FA-50. The radar operated at a speed of approximately 300 m/s and an altitude of 10,000 m over the southern coast of South Korea. The target aircraft approached the radar platform from the front while maneuvering at a speed of approximately 300 m/s in a circular motion, generating acceleration about 10–15 m/s2 due to its trajectory. The radar was tested to detect the target aircraft as it approached while performing circular maneuvers. While specific waveform parameters such as PRF and FM slope are omitted for security reason, the test employed typical HPRF parameters. These operational parameters are summarized in Table 2.
As shown in Figs. 4 and 5, the target initially followed a circular trajectory with acceleration and then transitioned to straight and level flight at a constant velocity of approximately - 300 m/s. When using the conventional range measurement method, target detection was achieved only after time point T7, when the target entered the constant velocity trajectory.
Although beat frequencies necessary for FM ranging were measured at [T1, T2, …, T6], tracking was not initiated either because the clustering condition of Eq. (5) was not satisfied or due to the range bias inherent in the conventional method.
Table 3 summarizes the range measurements (R^) and range errors (ɛR) with respect to the true target position between time points T1 and T7, comparing the conventional method (linear) and the proposed method (LS). The estimated velocity (V^) and acceleration (Â) values obtained from the proposed method are also presented. Here, the true target position at time T7 is denoted as R0, and all range values are referenced relative to this point.
At T7, where the target exhibits almost no acceleration, both the conventional and proposed methods show low range bias. However, in the [T1, T6] interval where the target is accelerating, the conventional method exhibits a range bias of approximately 3 to 6 km. In contrast, the proposed method maintains errors within a few hundred meters without significant bias in the range estimation. Note that, as explained in Section II, such levels of error can be mitigated through the ambiguity-resolving process. In this scenario, the use of the proposed method allowed the target to be detected and tracked approximately 8.6 km earlier compared to the conventional method.

V. Conclusion

In this paper, we proposed a LS-based method to improve range measurement accuracy in FMICW pulse-Doppler radars, particularly focusing on scenarios involving highly maneuvering targets. Traditional FM ranging methods, while effective in clutter- free environments, often fail to accurately measure the range of targets experiencing significant acceleration or jerk, resulting in substantial biases and delayed detection. Our proposed approach integrates acceleration and jerk terms into the measurement model, effectively utilizing multiple FMICW bursts with varying slopes to jointly estimate range, velocity, and higher-order motion parameters. Through comprehensive simulations and flight tests conducted with an X-band AESA radar system, we demonstrated that the proposed LS-based method significantly reduces range bias compared to conventional methods. Future research directions include extending this methodology to even more complex target motions and exploring computationally efficient implementations to ensure real-time applicability in airborne radar systems.

Notes

This work was supported by the Agency for Defense Development Grant, funded by the Korean government (No. 274190001).

Fig. 1
FM ranging diagram using FMICW waveform.
jees-2026-4-r-376f1.jpg
Fig. 2
Range measurement bias with target acceleration.
jees-2026-4-r-376f2.jpg
Fig. 3
Range measurement bias with target jerk.
jees-2026-4-r-376f3.jpg
Fig. 4
Flight test scenario.
jees-2026-4-r-376f4.jpg
Fig. 5
Doppler velocity of the target.
jees-2026-4-r-376f5.jpg
Table 1
Range accuracy of conventional/proposed method
Method α = 1.5 α = 3 CRB
Linear 101.23 143.14 82.65
LS (Ha) 116.89 165.33 95.40
LS (Hj) 319.66 452.66 261.32
Table 2
Parameters of experimental radar
Parameter Value
Carrier frequency X-band
Pulse repetition frequency 100–300 kHz
FM slope difference (|ki–kj|) 5–20 MHz/s
Number of bursts 3
Burst interval (T) 50–150 ms
Platform speed ≈300 m/s
Platform altitude ≈10,000 m
Table 3
Range measurements and error of conventional (linear)/proposed (LS) method in flight test
Method Time

T1 T2 T3 T4 T5 T6 T7
Linear
X^ (m) R0+3,866.10 R0+2,310.24 R0+1,928.58 R0+2,459.27 R0+870.62 R0−268.57 R0−310.93
ɛR (m) −4,788.93 −5,967.06 −5,346.17 −3,760.86 −4,052.55 −3,779.64 −310.93
LS (proposed)
X^ (m) R0+8,592.51 R0+7,747.90 R0+7,418.70 R0+6,590.44 R0+4,839.50 R0+3,497.70 R0−9.48
ɛR (m) −62.52 −529.40 143.95 370.30 −83.67 −13.37 −9.48
V^ (m/s) −99.96 −113.57 −149.05 −182.84 −216.75 −246.48 −283.51
V^ (m/s2) −13.75 −15.86 −16.00 −12.01 −11.41 −10.78 −0.70

References

1. K. Jin, T. Lai, Y. Wang, G. Li, and Y. Zhao, "Coherent integration for radar high-speed maneuvering target based on frequency-domain second-order phase difference," Electronics, vol. 8, no. 3, article no. 287, 2019. https://doi.org/10.3390/electronics8030287
crossref
2. D. Kirkland, "Imaging moving targets using the second-order keystone transform," IET Radar, Sonar & Navigation, vol. 5, no. 8, pp. 902–910, 2011. https://doi.org/10.1049/ietrsn.2010.0304
crossref
3. R. Rytel-Andrianik, "On the ambiguity function for accelerating target in FMCW radar," Journal of Telecommunications and Information Technology, vol. 27, no. 1, pp. 40–44, 2007. https://doi.org/10.26636/jtit.2007.1.792
crossref
4. J. Xu, J. Yu, Y. N. Peng, and X. G. Xia, "Radon-Fourier transform for radar target detection, I: generalized Doppler filter bank," IEEE Transactions on Aerospace and Electronic Systems, vol. 47, no. 2, pp. 1186–1202, 2011. https://doi.org/10.1109/TAES.2011.5751251
crossref
5. C. Shi, J. Zhao, and X. Lu, "Research on linear-carrier FM method of high-PRF ranging for airborne AESA," In: Proceedings of the 2009 IET International Radar Conference; Guilin, China. 2009, pp 1–3. https://doi.org/10.1049/cp.2009.0297
crossref
6. L. Rejfek, Z. Mosna, J. Urbar, and P. K. Knizova, "System for automatic detection and analysis of targets in FMICW radar signal," Journal of Electrical Engineering, vol. 67, no. 1, pp. 36–41, 2016. https://doi.org/10.1515/jee-2016-0005
crossref
7. J. Jung, J. Chun, and B. Jeon, "Multi-target detection in FMICW radar using machine learning," The Journal of Korean Institute of Electromagnetic Engineering and Science, vol. 34, no. 7, pp. 561–569, 2023. https://doi.org/10.5515/KJKIEES.2023.34.7.561
crossref
8. J. H. Yoon and Y. Kim, "Clutter rejection for FM ranging airborne radar," Journal of Electromagnetic Engineering and Science, vol. 23, no. 6, pp. 455–460, 2023. https://doi.org/10.26866/jees.2023.6.r.190
crossref
9. C. Alabaster, Pulse Doppler Radar: Principles, Technology, Applications. Edison, NJ: SciTech Publishing, 2012. https://doi.org/10.1049/SBRA024E

10. M. I. Skolnik, Radar Handbook. 3rd ed. New York, NY: McGraw-Hill, 2008.

11. G. W. Stimson, H. Griffiths, C. Baker, and D. Adamy, Stimson’s Introduction to Airborne Radar. 3rd ed. Edison, NJ: SciTech, 2014.

12. D. Zhu, Y. Li, and Z. Zhu, "A keystone transform without interpolation for SAR ground moving-target imaging," IEEE Geoscience and Remote Sensing Letters, vol. 4, no. 1, pp. 18–22, 2007. https://doi.org/10.1109/LGRS.2006.882147
crossref
13. A. Dogandzic and A. Nehorai, "Cramer-Rao bounds for estimating range, velocity, and direction with an active array," IEEE Transactions on Signal Processing, vol. 49, no. 6, pp. 1122–1137, 2001. https://doi.org/10.1109/78.923295
crossref
14. S. M. Kay, Fundamentals of Statistical Signal Processing, Volume I. Estimation Theory. Upper Saddle River, NJ: Prentice Hall, 1993.

Biography

jees-2026-4-r-376i1.jpg
Sungyeong Park, https://orcid.org/0000-0003-1510-2475 received his B.S and M.S degrees in electrical engineering from Korea Advanced Institute of Science and Technology (KAIST), Daejeon, Republic of Korea, in 2014 and 2016, respectively. Since 2016, he has been a senior researcher at the Agency for Defense Development (ADD), Daejeon, Korea. His research interests include radar signal processing and modern radar detection theory.

Biography

jees-2026-4-r-376i2.jpg
Jeongjik Seo, https://orcid.org/0000-0002-2907-085X received the B.S. degree in Electronic Engineering from Kyungpook National University, Daegu, and the M.S. degree in Electrical Engineering from the Korea Advanced Institute of Science and Technology (KAIST), Daejeon, Republic of Korea, in 2013 and 2015, respectively. Since 2016, he has been a senior researcher at the Agency for Defense Development (ADD), Daejeon, Korea. His research interests include radar tracking and radar data processing.

Biography

jees-2026-4-r-376i3.jpg
Seonghee Choi, https://orcid.org/0009-0008-9016-1700 received the Ph.D. degree in electrical engineering from Hanyang University in 2017. Since 2002, he has been a principal researcher at the Agency for Defense Development (ADD). His research interests include radar data processing and target tracking theory.
TOOLS
Share :
Facebook Twitter Linked In Google+
METRICS Graph View
  • 0 Crossref
  • 0 Scopus
  • 1,280 View
  • 75 Download
Related articles in JEES

ABOUT
ARTICLE CATEGORY

Browse all articles >

BROWSE ARTICLES
AUTHOR INFORMATION
Editorial Office
101-1101 Lotte Castle President, 109 Mapo-daero, Mapo-gu, Seoul 04146, Republic of Korea
Tel: +82-2-337-9666    Fax: +82-2-711-7550    E-mail: admin-jees@kiees.or.kr                

Copyright © 2026 by The Korean Institute of Electromagnetic Engineering and Science.

Developed in M2PI

Close layer
prev next