Micro-Doppler based Target Classification
ABSTARCT :
Description: The world today has bought on a need to pay increased attention to safety and security issues, for example, search and rescue operations, surveillance, and protection of critical infrastructure. These tasks are often labour intensive and potentially dangerous. This provides an incentive to create systems that aid operators to gain situational awareness. In this regard, unmanned aerial vehicles, pose a significant threat to privacy and security.
To understand and assess this threat, classification between different drone models and types is required. One way in which this has been demonstrated experimentally is through this use of micro-Doppler information from radars. Normally birds and drones are often confused and there is need for a method which clarifies their corresponding class.
Background: Due to substantial increase in the number of affordable drones in the consumer market and their regrettable misuse, there is a need for efficient technology to detect drones in airspace. Drone and birds both include micro-Doppler signatures due to their propeller blade rotation and wing beats respectively.
These distinctive features can then be used to differentiate a drone from a bird, along with studying them separately. Detailed Description: Classification of drones and non-drones using micro-Doppler signatures captured from Radar as a sensor.
Moving parts of an object produce modulated Doppler components called micro-Doppler. The modulated Doppler signature is presented as added components to the Doppler signature of the drone body.
Due to rotating blades, frequency modulated components, which are quite revealing, are produced. To observe the time varying micro-Doppler’s, received FMCW should be processed by Joint time-frequency analysis.
A Conventional air surveillance radar system(operating usually at L-Band or S-Band) can rely on the radar cross section(RCS) of an aircraft for detection, but this may not always provide reliable detection, but this may not always provide reliable detection in case of drones.
Even if a dedicated system is built to be sensitive enough to detect small object like a drone, just RCS information is not adequate. Birds have similar physical size to drones and also will fly at similar altitude and speeds.
A reliable drone detection radar system must have the capability to discriminate between a drone and a bird. And hence micro-Doppler signature analysis is the key parameter for classification.
Expected Solution: Input: The micro-Doppler signatures of drones and bird Output: Class of the object(Drone or Bird)
EXISTING SYSTEM :
In many scenarios, the sub-components of a target exhibit micro-motions such as rotation, vibration, tumbling, and coning, in addition to the target’s “bulk motion”. Examples of such micro-motions include rotating blades of a helicopter or multirotor, rotating blades of propeller of a fixed wing aircraft, swinging arms and legs of a pedestrian, and rotation of wind turbines.
The micro-motions induce Doppler modulations on the received signal known as micro-Doppler effect [4]. Micro-Doppler effect was first introduced in coherent laser detection and ranging systems [5]. Due to these modulations, sidebands are generated about the target’s Doppler frequency shift.
The micro-Doppler signature is the characteristic to a particular target’s micro-motion and hence it can be exploited for classification of different targets [4]. Figure 1 shows the simulated micro-Doppler signature of a running human with a radial speed of 2 m/s [6].
DISADVANTAGE :
Complexity of Signal Processing: Micro-Doppler signals can be complex and require sophisticated algorithms for processing. This complexity can lead to longer computation times and increased power consumption.
Environmental Interference: Micro-Doppler signatures can be affected by environmental factors, such as clutter, noise, and multipath propagation. This can degrade classification performance and lead to false positives or negatives.
Environmental Interference: Micro-Doppler signatures can be affected by environmental factors, such as clutter, noise, and multipath propagation. This can degrade classification performance and lead to false positives or negatives.
Limited Range: The effectiveness of micro-Doppler techniques may diminish at longer ranges. The Doppler shift becomes less pronounced with distance, which can complicate the classification of targets that are farther away.
Calibration Requirements: Accurate classification may require careful calibration of the radar system and the environment, which can be time-consuming and may require specialized knowledge.
PROPOSED SYSTEM :
For the detection of the micro-Doppler signatures from the received signal, covariance based stransform based algorithm has been shown in Fig.1 and explained below in stepwise manner:
Firstly the signal has been transmitted from the radar sensor and is received back to the receiver of the Radar sensor.
Then, covariance based s-transform has been applied to plot the time-frequency representation or spectrograms of the received signal.
And then spectrograms of the detected and existing micro-Doppler signatures have been compared. If both are matched then target is detected else target is not detected.
ADVANTAGE :
High Sensitivity to Small Movements: Micro-Doppler analysis is highly sensitive to even small, nuanced motions of targets. This capability allows for the identification of distinct motion patterns that are unique to different types of objects.
Enhanced Target Discrimination: By analyzing the micro-Doppler signatures, it is possible to differentiate between targets that may appear similar in conventional radar returns. This allows for better classification of various objects, including humans, vehicles, and animals.
Non-Destructive Testing: Micro-Doppler techniques can be applied in non-invasive ways, allowing for classification without physical contact or interaction with the target.
Rich Information Content: The micro-Doppler effect provides additional information about the target's motion, including rotational and translational dynamics. This extra data can significantly improve classification performance.
Real-Time Processing: With advancements in hardware and algorithms, micro-Doppler techniques can be implemented in real time, enabling immediate classification and response.
|