A systematic introduction to the theory, development and latest research results of radar data processing technology
* Presents both classical theory and development methods of radar data processing
* Provides state-of-the-art research results, including data processing for modern style radars, and tracking performance evaluation theory
* Includes coverage of performance evaluation, registration algorithm for Radar network, data processing of passive radar, pulse Doppler radar, and phased array radar
* Has applications for those engaged in information engineering, radar engineering, electronic countermeasures, infrared techniques, sonar techniques, and military command
1.1 Aim and Significance of Radar Data Processing
Generally, a modern radar system consists of two important components: a signal processor and a data processor. The signal processor is used for target detection (i.e., the suppression of undesirable signals produced by ground or sea surface clutter, meteorological factors, radio frequency interference, noise sources, and man-made interference) [1-3]. When the video output signal, after signal processing and constant false alarm rate (CFAR) detection fusion, exceeds a certain detection threshold, it can be determined that a target has been discovered. Then, the discovered target signal will be transmitted to the data recording device, where the space position, amplitude value, radial velocity, and other characteristic parameters of the target are recorded, usually by computers. The measurement output from the data recording device needs to be processed in the data processor, which associates, tracks, filters, smooths, and predicts the obtained measurement data - such as the target position (radial distance, azimuth, and pitch angle) and the motion parameters [4-6] - for the effective suppression of random errors occurring during the measurement, estimation of the trajectory and related motion parameters (velocity and acceleration, etc.) of the target in the control area, prediction of the target's position at the next moment, and formation of a steady target track, so that highly accurate real-time tracking is realized [7-9].
In terms of the level at which radar echo signals are processed, radar signal processing is usually viewed as the primary processing of the information detected by the radar unit. It is done at each radar station, with information obtained from the same radar and the same scanning period and distance unit, with the aim of extracting useful target information from clutter, noise, and various active and passive jamming backgrounds. Radar data processing is usually viewed as secondary processing of the radar information [10-13]. Making use of information from the same radar, but with different scanning periods and distance units, it can be done both at each independent radar station and at the information processing center or system command center of the radar network. Data fusion of multiple radars can be viewed as a third or tertiary processing of the radar information, which is usually done at the information processing center. Specifically, the information the processing center receives is the measurement from the primary processing or the track from the secondary processing (usually called the local track) by multiple radars, and the track after fusion (called the global track or system track). The function of the secondary processing of radar information, based on the primary processing, is to filter and track several targets, and estimate the targets' motion parameters and characteristic parameters. Secondary processing is done strictly after primary processing, while there is no strict time limit between secondary and tertiary processing. The third level of processing is the expansion and extension of secondary processing, which is mainly reflected in space and dimension.
1.2 Basic Concepts in Radar Data Processing
The input to the radar data processing unit is the measurement from the front, which is the object of data processing, while the output is the track formed after data processing is conducted. Generally, functional modules of radar data processing include measurement pretreatment, track initiation and termination, and data association and tracking. A wave gate must be set up between the association and the tracking process, and their relationship is shown in the block diagram in Figure 1.1. The content and related concepts of the functional modules of radar data processing are briefly discussed as follows.
Figure 1.1 Radar data processing relation diagram
Measurements, also called observations, refer to noise-corrupted observations related to the state of a target . The measurements are not usually raw data points, but the output from the data recording device after signal processing. Measurements can be divided, according to whether they are associated with the known target track, into free measurements and correlated measurements. Free measurements are spots that are not correlated with the known target track, while correlated measurements are spots that are correlated with the known target track.
1.2.2 Measurement Preprocessing
Although modern radar adopts many signal processing technologies, there will always be a small proportion of clutter/interference signals left out. To relieve the computers doing the follow-up processing job from a heavy burden, prevent computers from saturation, and improve system performance, the measurement given by the primary processing needs to be preprocessed, which is called "measurement preprocessing": the preprocessing of secondary processing of radar information. The preprocessing is a precondition of correct processing of radar data, since an effective measurement data processing method can actually help yield twice the result with half the effort, with the target tracking accuracy improved while the computational complexity of the target tracking is reduced. The measurement preprocessing technology mainly involves system error registration, time synchronization, space alignment, outlier rejection, and saturation prevention.
220.127.116.11 System Error Registration
The measurement data from radars contains two types of error. One is random error, resulting from the interior noise of the measurement system. Random error may vary with each measurement, and may be eliminated to some extent by increasing the frequency of measurement and minimizing its variance in the statistical sense by means of methods like filtering. The other is system error, resulting from measurement environments, antennas, servo systems, and such non-calibration factors in the data correction process as the position error of radar stations and the zero deviation of altimeters. System error is complex, slowly varying, and non-random, and can be viewed as an unknown variable in a relatively long period of time. As indicated by the findings in Ref. , when the ratio of system errors to random errors is greater than or equal to 1, the effect of distributed track fusion and centralized measurement fusion deteriorates markedly, and at this point system errors must be corrected.
18.104.22.168 Time Synchronization
Owing to the possible difference in each radar's power-on time and sampling rate, the target measurement data recorded by data recording devices may be asynchronous. Therefore, these observation data must be synchronized in multiple-radar data processing. Usually, the sampling moment of a radar is set as the benchmark for the time of other radars.
22.214.171.124 Space Alignment
Space alignment is the process of unifying the coordinate origin, coordinate axis direction, etc. of the data from the radar stations in different places, so as to bring the measurement data from several radars into a unified reference framework, paving the way for the follow-up radar data processing.
126.96.36.199 Outlier Rejection
Outlier rejection is the process of removing the obviously abnormal values from radar measurement data.
188.8.131.52 Saturation Prevention
Saturation prevention mainly deals with saturation in the following two cases.
- In the design of a data processing system, there is a limit to the number of target data. However, in a real system, saturation occurs when the data to be processed exceed the processing capacity.
- The time used to process data is limited. Saturation occurs when the number of measurements, or batches of targets, reaches a certain extent. In this case, the processing of the data from one observation has to be interrupted before the processor starts to deal with the next batch of data.
1.2.3 Data Association
In the single-target, clutter-free environment, where there is only one measurement in the target-related wave gate, only tracking is involved. Under multi-target circumstances, where a single measurement falls in the intersection area of several wave gates or several measurements fall in the related wave gate of a single target, data association is involved. For instance, suppose two target tracks have been established before the radar's nth scanning, and two echoes are detected in the nth scanning, are the echoes from two new targets or from the two established tracks at that time? If they are from the two established tracks at that time, then in what way can the echoes resulting from the two scans and the two tracks be correctly paired? The answer involves data association, the establishment of the relationship between the radar measurements at a given moment and the measurements (or tracks) at other moments, to check whether these measurements originate from the processing of the same target (or to ensure a correct process of measurement-and-track pairing).
Data association, also called "data correlation" or "measurement correlation," is a crucial issue in radar data processing. False data association could pair the target with a false velocity, which could result in the...