Method for Processing Signals in a Direction-Finding System
A method for processing signals in a direction-finder system, including detection of pulses in the received signals, estimation of characterizing pulse parameters, sorting the pulses according to the characterizing pulse parameters, de-interleaving the received pulses into pulses coming from at least one emitter, determining emitter parameters characterizing said emitter, and comparing said emitter parameters with emitter parameters from known emitters in order to recognize and identify the emitter.
The present invention discusses a novel method for processing signals in a direction-finder system for identifying and localising radio frequency emitters.
Radio frequency emitters (radars, satellite uplink stations, cell-phone base stations, relay links) can be detected, analysed, and geo-referenced from a remote observation platform. This is achieved using a sensor with an antenna system for detecting the radiation, connected to a receiver and processing system. These systems can be deployed from satellites, aircraft, UAVs, ships vehicles or mounted in masts.
Typical solutions employ radio receiver systems operating in the frequency bands 1 through 12 GHz. These systems employ multiple receiving antennas and multiple receivers to derive a course direction to the emitters.
A direction-finder system installed in a ground observation platform will cover a large area, i.e. it will “see” a large part of the ground below. This means that the system will receive a large numbers of radio signals; up to several million signals may be received simultaneously. The signals originate from navigation radars, which are the main target for the direction-finding system, but may also originate from “false” sources, such as echo/reflection signals from the same radars, other radio transmitters, ground clutter, etc. The false signals will often overlap the desired signals. The large numbers of signals received poses a problem for the processing unit in the direction-finding system. The processing unit has the task of filtering out the wanted signals and find the true bearing to the sources.
SUMMARY OF THE INVENTIONAn object of the present invention is to provide a system and method for processing the signals received by a direction-finding system, which solves the problem outlined above.
Thus, it is an object to provide a system and method as mentioned above, in which a target emitter appearing in a dense traffic scenario may be identified and its position determined with a high accuracy.
Another object is to provide a system and method which may process a large number of signals simultaneously with low probability of collision in the signal pipe-line.
These objects are achieved in a system and method as claimed in the appended patent claims.
In particular the invention relates to a method for processing signals in a direction-finder, including detection of pulses in the received signals, estimation of characterizing pulse parameters, sorting the pulses according to the characterizing pulse parameters, de-interleaving the received pulses into pulses coming from at least one emitter, determining emitter parameters characterizing said emitter, and comparing said emitter parameters with emitter parameters from known emitters in order to recognize and identify the emitter.
Advantageous embodiments of the invention appear from the following dependent claims.
BRIEF DESCRIPTION OF THE DRAWINGSThe invention will now be described in detail in reference to the appended drawings, in which:
Initially, we will give an overview of the direction-finder system as a background for the present invention. As shown in
Briefly, the antenna arrangement includes four antenna panels mounted in a 2×2 relationship, as well as an omni-directional guard antenna that may be mounted at the center of the antenna unit.
The signals from the receiver section are entered into the following processing unit, which includes a digitizer with two stacked A/D converters 24. The design of the A/D converters is further discussed in co-pending patent application with title “A digitizer”.
The Signal processing gate array 25 receives signal from the digitiser at a constant rate, and performs the crucial tasks of detection and data reduction in addition to basic pulse measurement.
The microprocessor system 26 receives pulse information at a scenario dependent rate, and performs all secondary measurements and higher level algorithms.
The processing unit 3 with the A/D converters 24, gate arrays 25 and processor 26, has as its task to filter out all false signal and classify, identify and find the direction to all radar sources observed within it area of sight. This is done in a multi-stage process as illustrated in
-
- 1. Pulse detection sorting wanted information from noise
- 2. Pulse parameter estimation, which is performed on each detected pulse. The estimated parameters may include:
- direction of arrival
- carrier frequency
- pulse width pulse amplitude
- 3. De-interleaving. De-interleaving is the process of identifying which of the incoming pulses that come from each of the observed emitters, thus sorting all pulses coming from the same emitter into one (or more) pulse trains.
- 4. Emitter analysis and emitter recognition. Emitter analysis means that an improved set of pulse parameters are determined, now sorted per emitter, forming a “fingerprint” characterizing each emitter. The emitters are then identified by comparison with known signatures stored in a database
Detection
The first process in the inventive direction-finding system is to detect the received pulses. Pulse detection is performed on the continuous stream of samples, and is essentially a data reduction process. Only samples that contain signal information are sent further down the chain. In its simplest implementation, this may be done by comparison with a fixed trigger value and keeping pulses above that value. Pulses below the threshold are regarded as noise. The pulses are subsequently transformed into the frequency domain.
However, from an SNR point of view, the detection should be performed on narrow-band signals (e.g. after FFT) to maximize Pd (probability of detection) and minimize Pfa (probability of false alarm). To achieve this, a continuous 8 point real to complex Fourier transform (producing 4 frequency channels) is used to implement multiple receiver channels. Detection is implemented as a simple square-law detector. The detection of a pulse triggers storing of actual samples and further processing.
Pulse Parameter Estimation
Pulse parameter estimation is subsequently performed on the complex samples:
Direction of arrival, which is the first parameter identified, may be calculated directly from phase comparison of the signals arriving at the different antenna panels.
Carrier frequency may be calculated for each antenna panel individually, and improved by combining the results from the different panels prior to de-interleaving. This parameter is easily calculated to an accuracy of 1 MHz.
Pulse width and pulse amplitude are error-prone parameters in a noisy environment, due to the reliance on peak measurement. Some improvements may be achieved by optimum curve fitting to a “standard” pulse shape (e.g. rectangular or Gaussian), but both parameters will still have quite a large variance. Most radars within an area will have approximately the same pulse width due to similar range setting of their radars. In addition, the amplitude will vary from pulse to pulse due to antenna motion at the emitter (antenna rotation and ship roll).
Time of arrival may not be used directly in a de-interleaving process. The sorting criteria will be pulse repetition interval. Since navigation radars both stagger and jitters this parameter, only a limited number of emitters may be discriminated using PRI (even more so, since most of the radars within an area will use the same average PRI—due to similar range setting of their radars).
It is preferred to measure pulse frequency and phase on the frequency domain data produced in the detection process, as this improves the measurement accuracy.
Prior to amplitude and pulse width measurement a continuous 4 point inverse complex Fourier Transform is performed. The reason for this is to generate a complex time series and thereby minimise problems with signal and sampling frequency intermodulation. Pulse amplitude and pulse width is measured on this complex time series.
Histogram Building
When the parameters above have been found, the parameters/data for each pulse is entered into a 3D histogram based on direction of arrival (2D) and frequency. Subsequently, pulse trains are extracted from the histogram, largest peak in histogram first. While the data are entered into the histogram as a serial stream of data from unsorted pulses, each peak in the histogram should in most cases represent one and only one emitter source, and the extracted data will thus consist of a serial stream of data pertaining to one emitter, ie a train of pulses emitted by and received from the source in question.
The pulse trains are then de-interleaved, whereupon emitter parameters characterizing each emitter are calculated based on all pulses in the resulting pulse trains.
De-Interleaving
A few de-interleaving methods are used with ESM-systems, including:
Pulse Repetition Interval (PRI) based de-interleaving: This method is mainly used when constant-PRI is expected from the emitters, as is the case with current pulse-Doppler radars. The method is simple, and the demands on the receiving system are not very high. The method is mainly used with crystal receivers where only pulse times, amplitude and pulse are available. In complex scenarios with a high number of emitters and in scenarios with emitters that jitter and stagger PRI, this method fails.
Frequency based de-interleaving: This method is mainly used when constant-frequency emitters are expected, as is the case with current pulse-Doppler radars and typical magnetron-based radars. The method requires a more advanced receiver, capable of determining carrier frequency. In a dense environment with many radars operating close in frequency this method fails.
Direction of arrival based de-interleaving: This method requires that the receiver system is capable of determining the direction of arrival of each received pulse. Thus, some kind of mono-pulse antenna arrangement is required (rotating directional antenna is not sufficient). No assumptions need to be made of the emitter characteristics. In a dense environment, where a number of emitters are located within a resolution cell of the ESM-receiver, this method fails.
Most functional ESM-systems base their de-interleaving process on a combination of the above methods, and they often use further parameter analysis in order to qualify the de-interleaving process.
The inventive system is constructed as a highly accurate mono-pulse receiver system. In addition, carrier frequency may be measured with reasonable accuracy, as can most of the basic pulse parameters. Current navigation radars are based on magnetron transmitters and therefore use constant carrier frequency. In order to reduce interference between multiple radars, the carrier frequency varies from radar set to radar set even within the same production series. To reduce interference even more, some kind of variation of the PRI (jitter or stagger) is employed on most radars.
Based on the characteristics of the navigation radars, the inventive method employs a combination of direction of arrival and frequency based de-interleaving as a first sorting criterion.
In some cases two or more emitters will be so close in carrier frequency that DOA or frequency alone cannot resolve them. In order to increase the quality further, a second stage of de-interleaving is performed on the candidate emitter pulse trains. Due to the fact that there is a high (intended) variation in PRI pattern (stagger and jitter) from radar set to radar set, advanced PRI-analysis may be used to qualify the de-interleaving process. Similarly, emitter antenna analysis may be used to qualify the de-interleaving process as can pulse width analysis.
The last stage in the algorithm is not directly connected to de-interleaving, but is essential to limit the number of anticipated emitters in costal waters and to be able to calculate position estimates. Duplicate emitters may arise from land echoes when the emitter antenna is pointing away from the satellite, and the positions of these duplicates will be the land area where the signal is reflected off. If the reflection point is more than a resolution cell removed from the emitter, the reflection point will be identified as a separate emitter by the previous stages. Therefore, as each emitter is extracted from the histogram the emitter parameters must be calculated and compared to the previously extracted emitters. All emitter parameters, excluding DOA, should be compared in order to identify duplicate emitters. Most often, the real emitter will have the strongest signal, but not always. Therefore the selection should also take into account variance in all parameters (including DOA). The real emitter will be the one with lowest variance in all parameters. Beware, though, that the real emitter is not always seen in coastal waters as it may be hidden by terrain.
Emitter Analysis
When the various emitters present has been resolved in the de-interleaving process, several pulses from the same emitter will be stacked (integrated) in order to improve the data set. Subsequently, the processor will perform another emitter analysis determining an improved set of parameters characterizing the emitter. This includes updating the estimates for the pulse parameters (frequency, pulse width) and improving localisation estimates based on all available pulses. In addition, antenna parameters (beam width, rotational speed, polarisation) and PRI parameters (PRI and jitter) may now be estimated. Based on this information, initial attempts on emitter classification (determining radar make and type) may be made.
Emitter Recognition
Emitter identification is the last step in the processing, and is based on all the previously measured parameters, in addition to the actual pulse waveform. The following parameters are used in an attempt to uniquely identify the radar and thereby the vessel:
-
- Frequency
- Pulse width
- Antenna parameters (beam width, rotational speed and polarisation)
- PRI parameters (PRI and jitter)
- Modulation on pulse (MOP)
In order to perform MOP measurement, the complex waveform generated in the pulse parameter estimation process is necessary. The MOP measurement may be based on a single pulse within the pulse train or on all pulses.
This improved set of parameters form a signature of the emitter. The signature is compared with signatures from known emitters stored in a database, in order to find the identity of the emitter.
Scenario
In the following evaluation a simplified scenario is used as a basis,
Based on this scenario, the number of vessels within the main antenna lobe (130 NM×500 NM) will be approximately 40-50.
In addition, the maximum number of emitters seen from the direction finder platform at one time is assumed to be limited to 1000.
The average pulse rate received from a single emitter is 10 pulses/s (<3.2 pulses/s for larger radars typical of commercial vessels, ≈16 pulses/s for the smallest radars typical for small recreational vessels).
Based on the characteristics of the navigation radars, the de-interleaving should be based on a combination of direction of arrival and frequency as a first sorting criterion. The direction of arrival criterion will resolve all vessels in the grid (50 NM grid distance vs. 5 NM resolution).
Regarding the cluster of five vessels within the cluster cells, the DOA analysis will not discriminate these. Assuming a conservative estimate of the frequency spread, a uniform distribution of carrier frequency over 60 MHz, and a frequency resolution of 1 MHz in the carrier frequency determination in the receiver, there is a 16% probability that two emitters will fall into the same histogram cell.
Thus, the first stage of the recommended de-interleaving process will resolve more than 90% of all vessels in the scenario. In order to increase the quality further, a second stage of de-interleaving is performed on the candidate emitter pulse trains. This stage is based on detailed PRI analysis, analysis of pulse width and antenna analysis.
Claims
1-14. (canceled)
15. A method for processing signals in a direction-finder system, characterized in that the method includes the following steps:
- detection of pulses in the received signals,
- estimation of characterizing pulse parameters,
- sorting the pulses according to the characterizing pulse parameters,
- de-interleaving the received pulses into pulses coming from at least one emitter,
- determining emitter parameters characterizing said emitter, and
- comparing said emitter parameters with emitter parameters from known emitters in order to recognize and identify the emitter.
16. A method as claimed in claim 15, characterized in that the pulse detection step includes comparing the received signals with a threshold level.
17. A method as claimed in claim 15, characterized in that the pulse detection step includes filtering the received signals into a number of frequency limited channels and performing a square-law detection of the signals in each channel.
18. A method as claimed in claim 15, characterized in that the step of estimation characterizing pulse parameters includes:
- determining direction of arrival, and/or
- determining carrier frequency, and/or
- determining pulse width, and/or
- determining pulse amplitude.
19. A method as claimed in claim 18, characterized in that direction of arrival is estimated by comparing the phase of signals arriving from a number of antenna panels, each antenna panel being connected to a different receiving channel.
20. A method as claimed in claim 18, characterized in that pulse width and pulse amplitude are estimated by optimum curve fitting to a rectangular or Gaussian pulse shape.
21. A method as claimed in claim 15, characterized in that the sorting step includes entering the received pulses into a 3D histogram based on said characterizing pulse parameters and extract pulse train in order of diminishing amplitude starting with the highest amplitude.
22. A method as claimed in claim 15, characterized in that the de-interleaving includes analysis of direction of arrival and/or carrier frequency.
23. A method as claimed in claim 22, characterized in that the de-interleaving includes analysis of the additional parameters pulse repetition interval parameters and/or pulse width and/or antenna parameters.
24. A method as claimed in claim 15, characterized in that said step of determining emitter parameters includes to integrate pulses originating from an emitter, and determining carrier frequency and/or pulse width and/or emitter location and/or antenna parameters and/or pulse repetition interval parameters.
25. A method as claimed in claim 15, characterized in that said recognition step includes to perform an initial emitter classification in radar make and type based on carrier frequency and/or pulse width and/or antenna parameters and/or pulse repetition interval parameters.
26. A method as claimed in claim 15, characterized in said recognition step includes to compare the emitter characterizing parameters carrier frequency and/or pulse width and/or antenna parameters and/or pulse repetition interval parameters and/or modulation on pulse with corresponding parameters for known emitters, said parameters for known emitters being stored in a database.
27. A method as claimed in claim 23, characterized in that said antenna parameters includes beam width and/or rotational speed and/or polarisation.
28. A method as claimed in claim 23, characterized in that said pulse repetition interval parameters includes pulse repetition interval and/or stagger patterns.
Type: Application
Filed: Dec 30, 2004
Publication Date: Sep 27, 2007
Inventors: Jens Fredrik Hjelmstad (Lillestrom), Per Atle Valand (Oslo)
Application Number: 11/568,955
International Classification: G01S 7/02 (20060101);
