1.

Abstract
A radar system is a sytem that is designed for collecting data on distant targets through sending
electromagnetic wave and analyzing received signal. It can be single target or multiple targets. Originally
this system was developed for aircraft detection, but after certain periods the radar systems find diverse
applications in military, law enforcement, space exploration, environmental sensing, navigation, and
security. Given the rising concerns about highway safety and efficient transportation globally, radar
technology is utilized to enhance vehicle tracking, providing real-time and accurate data for proactive
traffic management. The objectives include determining vehicle speeds, optimizing scanning angles,
enhancing range estimation, reducing false alarms, and augmenting the number of detected vehicles.

2. Introduction
Radar serves as a radiolocation system employing radio waves to ascertain the distance, angle, and radial
velocity of objects in relation to a specific site. Its applications range from detecting and tracking aircraft,
ships, and missiles to mapping weather formations and terrain. Comprising a transmitter, transmitting
antenna, receiving antenna (often shared), and a processor, the system reflects radio waves off objects,
providing crucial information about their locations and speeds.
Though radar system was initially developed for military purposes, but now the applications of radar are
vast and varied, encompassing air and terrestrial traffic control, radar astronomy, air-defense and anti-
missile systems, marine radars for navigation, aircraft anti-collision systems, meteorological monitoring,
and more. Doppler effect in radar system , is used to determine the speed and direction of the objects
being detected. The principle of the Doppler is based on the frequency shifting of radar waves when they
interact with moving objects.
In the realm of radar technology, multipath effect is a phenomenon where radar signals take multiple
paths to reach the target and return to the receiver. This can occur when signals are reflected, scattered,
or refracted by various objects or surfaces in the environment, resulting in multiple delayed echoes
reaching the radar receiver. In a typical radar scenario, the emitted electromagnetic waves interact with
objects in the environment, and their reflections contribute to the received signal. When these reflections
follow multiple paths due to obstacles, terrain, or atmospheric conditions, it leads to multipath radar
reflection.
Off late, the sound highway and the sound transportation are now a common issue all over the world.
Highway faces horrible accidents due to adverse weather like storm, rain, fog and strong sunlight etc.
Traffic accidents lead a significant danger, resulting in harm and fatalities to passengers, pedestrians, and
damage to vehicles and road infrastructure and their frequency is on the rise. Bangladesh, in particular,
has a high rate of road accidents, despite numerous efforts by authorities to reduce them. A significant
portion of serious accidents results from exceeding speed limits or driving at inappropriate speeds.
Vehicle speed plays a pivotal role in accident severity, emphasizing the importance of setting appropriate
speed limits for road safety. A fast and highly accurate tracking system can be utilized to guess the vehicle
movement statistics from video data, including the current speed of cars on the road. This information
empowers officers to oversee traffic conditions effectively.
3. Literature Review

Literature review on highway vehicle tracking with multipath radar reflections:

Existing Achievements Research gaps
Methods
Orthographic (i) Real-time traffic monitoring with (i) Low-cost auxiliary hardware for
approximation accurate location classification of 46 low-level processing
method cars and 7 light vans. (ii) 3-D reasoning carried out
(G. D. Sullivan (ii) Demonstrated high performance automatically off-line at set-up time
et al., 1996) [6] with minimal misclassification. (iii) Balancing computing performance
(iii)Evaluation of hypotheses within and computational costs for
small 2-D regions. efficiency
Multi-sensor (i) Enhanced detection range, reliability, (i) Impact of multiple model
fusion method and accuracy of object tracking. estimation on optimal longitudinal
(R. Mobus et al., (ii) Implementation of multiple model control.
2004) [7] filtering for capturing dynamics of
maneuvering objects.

MIMO STAP (i) Efficient multipath clutter mitigation (i) Adapting to dynamic clutter
approach (ii) Boost target visibility by 40.7 dB, (ii) Optimize computational
(V. F. Mecca et unmasking obscured targets complexity for large antenna arrays
al., 2006) [8]
Scanning laser (i) System combines linear feature (i) Tracking moving objects from a
range finder extraction, tracking, and motion moving vehicle is challenging.
approach (R. A. evaluator. (ii) Object shape change can be
MacLachlan et (ii) Low false alarm rate in collision misinterpreted as motion.
al., 2006) [9] warning system (1 per 150 moving (iii) Spurious velocities on fixed
tracks) objects can cause false collision
(iii)Minimizing velocity errors (σ to 0.13 warnings.
m/s).
Vehicle (i) Ability to handle high traffic (i) Difficulty in tracking individual
segmentation situations and highly compressed vehicles in heavy traffic due to
method videos. occlusion and shadow.
(G. Jun et al., (ii) High detection accuracy (97.3% in (ii) Noisy scenes in traffic monitoring
2008) [10] Mecidiyekoy, 86.0% in Halic) videos due to weather conditions
(iii)Implemented background and video compression.
subtraction, occlusion detection, and
motion analysis
Kalman filter (i) Separate methods for occlusion and (i) Inter-object occlusion
method merge are applied to handle (ii) Occlusion of objects by
(X. Li et al., confusing situations. background obstacles
2010) [11] (ii) Establish object motion model for (iii) Splits and merges during tracking
fast tracking.
Laser sensor and (i) Estimation of location, orientation, (i) Underestimation of elliptical target
extended size, and type of extended targets. size .
Kalman filter (ii) Rectangular and elliptical target (ii) Target loss during occlusion.
method detection. (iii) No ground truth for comparison.
(K. Granström et
al., 2011) [12]
Gaussian (i) Variable detection for occlusion (i) Infeasibility of implementing the
Mixture PHD adaptability. exact filter due to processing all
filter method (ii) Practical implementation of the PHD possible measurement set
(K. Granström et filter using Gaussian mixtures for partitions.
al., 2012) [13] extended targets. (ii) Spatially close targets and target
occlusion need suitable remedies.
Time Reversal (i) Beam compression for finer azimuth (i) Develop AI to identify targets from
ISAR method resolution TR-ISAR images.
(I. de Arriba- (ii) Multipath modeling and mitigation (ii) High-resolution images with
Ruiz et al., (iii)Pre-processing algorithm for target numerous scatterers may require
2013) [14] location estimation computational optimization.
Truncated (i) Implementation of TSVD for image (i) Linearized PO model neglects
singular value reconstruction and mitigating multiple scattering phenomena.
decomposition ghosting artifacts caused by (ii) Mutual interactions among objects
(TSVD) multipath reflections. and creeping waves are neglected.
inversion (ii) Overcoming nonlinearity by (iii)The linear inverse problem is ill-
method resorting to the PO approximation. posed and requires regularization.
(G. Gennarelli et
al., 2014) [15]
Cropping (i) Single-camera vehicle detection of (i) Complex videos cause high false
method 87.7% accuracy with image positives (70.03% accuracy).
(T. Kumar et al., processing. (ii) Opposing traffic reduces tracking
2016) [16] (ii) Minimization of false positive accuracy.
detections on road edges by (iii)Diverse lighting conditions need
cropping. testing.
(iii)92.2% tracking accuracy based on
frame movement.
Multi-sensor (i) Tracks unknown number of targets (i) Distributed variants for
method with multiple sensors. decentralized sensor networks.
(F. Meyer et al., (ii) Low complexity, scales quadratically (ii) Further optimization for real-time
2016) [17] with targets, linearly with performance in resource-
sensors/measurements. constrained systems
(iii)Outperforms PHD, cardinalized
PHD, and multi-Bernoulli filters.
Machine- (i) 91% accuracy in differentiating real (i) Address identification of static
Learning-Based and ghost detections. ghost detections.
Ghost Detection (ii) Achieves 85% success with a feature (ii) Align with UDA and ISO standards
Identification subset for limited hardware. for sensor assessment.
(R. Prophet et
al., 2019) [18]
 Optimal (i) Dual HRA schemes for (i) Extend HRA to denser or cluttered
heterogeneous asynchronous multi-target tracking environments.
resource in heterogeneous radar networks. (ii) Incorporate non-linear resource
allocation (ii) Schemes optimize resource constraints and analyze robustness
method (J. Yan utilization for better tracking under various conditions.
et al., 2020) [19] accuracy with limited resources.

Appearance (i) Two feature extraction methods: (i) Deep features need improvement
model and visual traditional (58.9% accuracy) vs. deep for better tracking (lower accuracy
object tracking learning (15.9%). than traditional).
method (ii) Efficient bipartite graph approach for (ii) Analyze performance under
(F. Bafghi et al., optimized tracking. various lighting and weather
2020) [20] (iii)Handles occlusions with conditions.
hypothetical nodes. (iii)Reduce computational cost for real-
time applications.

4. Existing Solution:

5. Future Work:
• Various trackers may be used to track the vehicles
• To observe the various tracker effects on performance improvement.
• Detection number can be increased more.
• Position, velocity and angle error can be reduced
• Accuracy can be increased by adopting LMS, LCMV or Robust algorithms into the existing
system.

6. References:
[1] G. D. Sullivan, K. D. Bake, A. D. Worrall, C. I. Attwood and P. R. Remagnino, "Model-based
vehicle detection and classification using orthographic approximation," in Image and vision
computing, United Kingdom, 1996, pp. 649-654.
[2] R. Mobus and U. Kolbe, "Multi-target multi-object tracking, sensor fusion of radar and infrared,"
IEEE Intelligent Vehicles Symposium, 2004, Parma, Italy, 2004, pp. 732-737.
[3] R. A. MacLachlan and C. Mertz, "Tracking of moving objects from a moving vehicle using a
scanning laser range finder," in IEEE intelligent transportation systems conference, Canada, 2006.
[4] V. F. Mecca, D. Ramakrishnan, and J. L. Krolik, “MIMO radar space-time adaptive processing
for multipath clutter mitigation,” in Fourth IEEE Workshop on Sensor Array and Multichannel
Processing, pp. 249– 253, 2006.
[5] G. Jun, J. K. Aggarwal and M. G¨okmen, "Tracking and segmentation of highway vehicles in
cluttered and crowded scenes," in IEEE Workshop on Applications of Computer Vision
Computing, (WACV), Copper, Colorado, 2008.
[6] X. Li, K. Wang, W. Wang and Y. Li, "A multiple object tracking method using Kalman filter,"
The 2010 IEEE International Conference on Information and Automation, Harbin, China, 2010,
pp. 1862-1866.
[7] K. Granström, C. Lundquis and U. Orguner, "Tracking rectangular and elliptical extended targets
using laser measurements," in 14th international conference on information fusion, Chicago,
USA, 2011.
[8] K. Granström, C. Lundquist and U. Orguner, "Extended target tracking using a Gaussian Mixture
PHD filter," IEEE transactions on aerospace and electronic systems, vol. 48, no. 4, pp. 3268-
3286, 2012.
[9] I. de Arriba-Ruiz, F. Pérez-Martínez, and J. M. Muñoz-Ferreras, “Time-reversal-based multipath
mitigation technique for ISAR images,” IEEE transactions on geoscience and remote sensing,
vol. 51, no. 5, pp. 3119– 3138, 2013.
[10] G. Gennarelli and F. Soldovieri, “Multipath ghosts in radar imaging: Physical insight and
mitigation strategies,” IEEE Journal of Selected Topics in Applied Earth Observations and
Remote Sensing, vol. 8, no. 3, pp. 1078–1086, 2014.
[11] T. Kumar and D. S. Kushwaha, "An efficient approach for detection and speed estimation of
moving vehicles," in Twelfth international multi-conference on information processing-2016
(IMCIP-2016), India, 2016.
[12] F. Meyer, P. Braca, P. Willett, and F. Hlawatsch, “A scalable algorithm for tracking an unknown
number of targets using multiple sensors,” IEEE Trans. Signal Process., vol. 65, no. 13, pp. 3478–
3493, Jul. 2017.
[13] R. Prophet, J. Martinez, J.-C. F. Michel, R. Ebelt, I. Weber, and M. Vossiek, “Instantaneous ghost
detection identification in automotive scenarios,” in IEEE Radar Conference (RadarConf), 2019.
[14] J. Yan, W. Pu, S. Zhou, H. Liu, and M. S. Greco, “Optimal resource allocation for asynchronous
multiple targets tracking in heterogeneous radar networks,” IEEE Trans. Signal Process., vol. 68,
pp. 4055–4068, 2020.
[15] F. Bafghi and B. Shoushtarian, "Multiple-vehicle tracking in the highway using appearance
model and visual object tracking," in The 11th iranian and the first international conference on
machine vision and image processing, Isfahan, Iran, 2020.