Experimental Seaborne Passive Radar

sensors

Article Experimental Seaborne Passive Radar

Gustaw Mazurek * , Krzysztof Kulpa , Mateusz Malanowski * and Aleksander Droszcz

Faculty of Electronics and Information Technology, Warsaw University of Technology, Nowowiejska 15/19 Street, 00-665 Warsaw, Poland; [email protected] (K.K.); [email protected] (A.D.) * Correspondence: [email protected] (G.M.); [email protected] (M.M.)

Abstract: Passive bistatic radar does not emit energy by itself but relies on the energy emitted by illuminators of opportunity, such as radio or television transmitters. Ground-based passive radars are relatively well-developed, as numerous demonstrators and operational systems are being built. Passive radar on a moving platform, however, is a relatively new ﬁeld. In this paper, an experimental seaborne passive radar system is presented. The radar uses digital radio (DAB) and digital television (DVB-T) for target detection. Results of clutter analysis are presented, as well as detections of real-life targets.

Keywords: passive bistatic radar; passive coherent location; maritime; seaborne

1. Introduction Passive coherent location (PCL) radar, also known as passive bistatic radar, relies on the signals emitted by illuminators of opportunity, such as radio or television transmitters [1]. The radar is equipped with at least two receiving channels. One channel is connected to an antenna that is pointing towards the transmitter. In this way, the reference Citation: Mazurek, G.; Kulpa, K.; signal is received. The second channel is connected to an antenna that is used to receive the Malanowski, M.; Droszcz, A. echo signal. By correlating the reference and echo signals, one can estimate the time delay Experimental Seaborne Passive Radar. and frequency shift. The system can localize the target in the Cartesian coordinates by Sensors 2021, 21, 2171. combining the delay and frequency shift measurements from multiple transmitter–receiver https://doi.org/10.3390/ pairs. As a result, passive radar can be effectively used for the tracking of moving targets. s21062171 Most passive radars are stationary ground-based systems. The technology is relatively mature and well-understood. One of the challenges in passive radar technology is installing Academic Editor: Michail Antoniou passive radar on moving platforms. This includes airplanes [2–6], ground vehicles [7], and ships [8,9]. Received: 19 February 2021 In this paper, we focus on seaborne passive radar, where the system was installed Accepted: 17 March 2021 on a ship. First, the passive radar system is presented. Next, the processing algorithms Published: 20 March 2021 are introduced. Finally, the real-life clutter analysis and the results of the experiments on the detection of airborne targets using digital radio (DAB) and digital television (DVB-T) Publisher’s Note: MDPI stays neutral are shown. The measurement campaign in which we collected the data was the NATO with regard to jurisdictional claims in published maps and institutional affil- APART-GAS 2019 (active passive radar trials ground-based, airborne, seaborne) trials, iations. which occurred in Poland on the Baltic Sea in September 2019 [10–13]. The trial involved numerous cooperative targets, such as aircrafts and ships, and various radar sensors, including passive and active systems either stationary or installed onboard different moving platforms. The paper’s novelty is the presentation of real-life clutter characteristics during dif- Copyright: © 2021 by the authors. ferent sea states and the detection of cooperative and noncooperative airborne targets by Licensee MDPI, Basel, Switzerland. seaborne passive radar, which confirms its usefulness. This article is an open access article distributed under the terms and 2. Materials and Methods conditions of the Creative Commons 2.1. System Setup and Signal Acquisition Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ A block diagram of the considered PCL receiver is depicted in Figure1. The hardware 4.0/). setup consists of two dual-channel software-defined radio (SDR) receivers (RSPduo [14])

Sensors 2021, 21, 2171. https://doi.org/10.3390/s21062171 https://www.mdpi.com/journal/sensors Sensors 2021, 21, x FOR PEER REVIEW 2 of 18

Sensors 2021, 21, 2171 2 of 19

A block diagram of the considered PCL receiver is depicted in Figure 1. The hardware setup consists of two dual-channel software-defined radio (SDR) receivers (RSPduo [14]) connectedconnected toto aa host host PC, PC, a a receiving receiving antenna antenna array, array, and and a a training training signal signal generator. generator. All All tunerstuners inin thethe receivers receivers areare tuned tuned to to the the same same frequency frequency of of a a DAB DAB or or DVB-T DVB-T broadcast broadcast stationstation selected selected for for the the illumination illumination of of a radara radar scene. scene. The The RSPduo RSPduo devices devices are are connected connected to theto the host host PC PC via via distinct distinct USB USB ports ports used used for fo datar data transfer transfer and and power power supply. supply. The The training training signalsignal generator generator is is controlled controlled and and powered powered from from another another USB USB port. port. It is It responsible is responsible for the for periodicalthe periodical generation generation of pilot of pilot signals signals that arethat later are later employed employed in the in synchronization the synchronization of the of channelsthe channels [15]. [15]. The pilotThe pilot signals signals are combined are combin withed RFwith signals RF signals from thefrom receiving the receiving antennas an- usingtennas a setusing of directionala set of directional couplers. couplers. After filtering, After filtering, downconversion, downconversion, and digitalization and digitaliza- in thetion RSPduo in the RSPduo devices, devices, the received the received complex complex baseband baseband signals are signals finally are stored finally in thestored SSD in drivethe SSD of the drive host of PC the for host further PC for offline further processing. offline processing.

FigureFigure 1. 1.Four-channel Four-channel receiver receiver used used in in the the experiments. experiments.

TheThe receiving receiving antenna antenna array array installed installed for for PCL PC experimentsL experiments on the on portthe sideport ofside the of ship the isship shown is shown in Figures in Figures2 and 32 withand 3 the with proviso the provis thato only that partonly of part the of antennas the antennas were were used used in thein the experiments experiments described described in this in this paper. paper. Four Four commercial-grade commercial-grade VHF VHF antennas antennas (Yagi–Uda, (Yagi– 800Uda,× 800100 × 100750 × mm,750 mm, 4-element, 4-element, 6 ... 68 … dBi 8 dBi gain, gain, vertical vertical polarization) polarization) were were employed employed to acquireto acquire RF RF signals signals from from the the DAB DAB frequency frequency band; band; three three of of them them in in the the surveillance surveillance path path andand oneone inin thethe reference path—installed in in the the opposite opposite direction. direction. Another Another set set of offour four an- antennastennas (dipole, (dipole, 200 200 × ×700700 × 100× 100 mm, mm, 2 dBi 2 dBi gain, gain, horizontal horizontal polarization) polarization) was was installed installed for forDVB-T DVB-T reception. reception. An An additional additional vertically vertically polarized polarized VHF VHF antenna was installedinstalled onon the the frontfront of of the the ship ship as as a a reference, reference, as as shown shown in in Figure Figure3. 3. The The surveillance surveillance antenna antenna set set and and the the referencereference antenna antenna were were selected selected individually individually for for each each signal signal acquisition, acquisition, depending depending on the on experiment’sthe experiment’s scenario. scenario. AA DAB DAB transmitter transmitter (218.64 (218.64 MHz, MHz, 1.536 1.536 MHz MHz bandwidth, bandwidth, 3.2 3.2 kW kW ERP, ERP, vertical vertical polariza- polar- ◦ 0 00 ◦ 0 00 tion,ization, 54 12 54°12’15”15 N, 16 N,13 16°13’42”42 E) and E) aand DVB-T a DVB-T transmitter transmitter (191.5 (191.5 MHz, MHz, 7.6 MHz 7.6 bandwidth,MHz band- ◦ 0 00 ◦ 0 00 15width, kW ERP, 15 kW horizontal ERP, horizontal polarization, polarization, 54 00 15 54°00’15”N, 16 44 N,27 16°44’27”E) were selectedE) were asselected the sources as the ofsources illumination, of illumination, both located both near located Koszalin, near Poland Koszalin, (ca. Poland 15 km from (ca. the15 km coast). from Those the trans-coast). mittersThose transmitters were chosen were because chosen of their because vicinity of th toeir the vicinity measurement to the measurement area. An interesting area. An fact in- isteresting that DVB-T fact transmittersis that DVB-T usually transmitters operate usually in the UHF operate band in (approximately the UHF band (approximately 400–800 MHz). In400–800 Poland, MHz). however, In Poland, some DVB-Thowever, transmitters some DVB-T also transmitters operate in thealso VHF operate band in (aroundthe VHF 200band MHz), (around which 200 provided MHz), which an interesting provided andan interesting unusual opportunity and unusual for opportunity measurement. for Between consecutive signal acquisition sessions, the PCL receiver was alternately tuned to either the DAB or DVB-T channel. In each case, a set of antennas with polarization consistent with the illuminating transmitter was employed.

Sensors 2021, 21, x FORSensors PEER 2021 REVIEW, 21, x FOR PEER REVIEW 3 of 18 3 of 18

Sensors 2021, 21, 2171measurement. Betweenmeasurement. consecutive Between signal consecutive acquisition signal sessions, acquisition the PCL sessions, receiver the was PCL al- receiver was3 of 19al- ternately tuned ternatelyto either thetuned DAB to oreither DVB-T the DABchannel. or DVB-T In each ch case,annel. a set In ofeach antennas case, a withset of antennas with polarization consistentpolarization with theconsistent illuminating with the transmitter illuminating was transmitteremployed. was employed.

Figure 2. Receiving antenna array installed on the side of the ship. Figure 2. ReceivingFigure antenna 2. Receiving array installed antenna on array the side installed of the on ship. the side of the ship.

(a) (a) (b) (b)

Figure 3. TopFigure view 3. of Top the view antennaFigure of the array 3. antenna Top (a) andview array additional of the (a) antennaand reference additional array antennas ( areference) and onadditional frontantennas of shipreference on (frontb) (some an oftennas ship parts (b on) of front the picture of ship (b) (some parts of the picture are masked for security and privacy reasons). are masked(some for security parts of and the privacy picture reasons). are masked for security and privacy reasons).

The RF signals ThewereThe RF RFdownconverted signals signals were were downconverted downconverted and digitized in and and two digitized digitized RSPduo in devicesin two two RSPduo RSPduo (shown devices devicesin (shown (shown in in Figure 4), whichFigureFigure are compact4 ),4), which which dual-tuner are are compact compact SDR dual-tuner dual-tuner receivers SDRfrom SDR receivers SDRplayreceivers from [14],from SDRplayintended SDRplay [for 14[14],], intended intended for for general purposegeneralgeneral SDR applications purpose purpose SDR SDR in applications theapplications frequency inin range thethe frequencyfrequency from 1 kHz rangerange up fromfromto 2 GHz 11 kHzkHz [16]. upup toto 22 GHz GHz [ 16[16].]. The sampling rateTheThe was sampling sampling set to rateF rates = was2 wasMHz, set set towhich toF sFs= = allows 2 2 MHz, MHz, it which whichto receive allows allows the it itwhole to to receive receive bandwidth the the whole whole bandwidth bandwidth of the DAB signal.ofof the theIn DABthe DAB case signal. signal. of the In In DVB-T the the case case sign of of theal, the only DVB-T DVB-T a part signal, sign ofal, the only only bandwidth a a part part of of theis the acquired bandwidth bandwidth is is acquired acquired in this way. Theinin remaining this this way. way. The out-of-bandThe remaining remaining comp out-of-band out-of-bandonents are components comp filteredonents out are throughare ﬁltered filtered the out out RSPduo through through the the RSPduo RSPduo tuner, thanks totuner,tuner, internal thanks thanks oversampling to to internal internal and oversampling oversampling anti-aliasing and and filters. anti-aliasing anti-aliasing ﬁlters. filters.

Sensors 2021, 21, 2171 4 of 19 Sensors 2021, 21, x FOR PEER REVIEW 4 of 18

FigureFigure 4. 4.RSPduo: RSPduo: dual-tuner dual-tuner SDR SDR (software-deﬁned (software-defined radio) radio) receiver. receiver.

EachEach RSPduo RSPduo device device was was connected connected to to the the host host PC PC via via a a High-Speed High-Speed USB USB 2.0 2.0 interface interface (also(also used used as as the the power power supply) supply) and and contained contained two two complete complete RF RF frontends. frontends. Each Each frontend frontend includedincluded a a bank bank of of bandpass bandpass filters, filters, variable-gain variable-gain LNA LNA (low-noise (low-noise amplifiers), amplifiers), and optional and op- notchtional filters. notch Thefilters. ADC The (analog-to-digital ADC (analog-to-digital converter) converter) had 14-bit had performance 14-bit performance (for a 2 (for MSPS a 2 sampleMSPS sample rate), and rate), the and sensitivity the sensitivity was around was around 0.1 µV[ 0.116]. μV The [16]. receiver’s The receiver’s performance, perfor- combinedmance, combined with the with completeness the completeness of the RF of frontend, the RF frontend, the small the size small of the size device, of the and device, the lowand power the low consumption, power consumption, determined determined the choice the of choice the device of the for device our experiments.for our experiments. TwoTwo or or moremore RSPduo devices can can be be synchr synchronizedonized by by connecting connecting their their dedicated dedicated ref- referenceerence OUT/IN OUT/IN ports, ports, so soit is it possible is possible to obtain to obtain the the same same sampling sampling rates rates and and carrier carrier fre- frequenciesquencies in inmultiple multiple RF RF tuners. tuners. However, However, providing providing repeatable, repeatable, coherent coherent multichannel multichan- nelsignal signal acquisition acquisition requires requires additional additional hardware hardware and andsoftware software modules modules [15], [such15], suchthe train- the traininging signal signal generator generator (see (see Figure Figure 1). 1). DueDue to to the the lack lack of of a a commoncommon trigger trigger signal, signal, the the coherent coherent acquisition acquisition from from two two or or more more RSPduoRSPduo devices devices becomes becomes a a challenge challenge [ 17[17].]. Each Each of of the the devices devices starts starts signal signal reception reception with with anan unknownunknown delaydelay (because(because ofof sequentialsequential initializationinitialization inin thethe software)software) andand a a random random phasephase shift shift due due to to the the uncontrolled uncontrolled initial initial conditions conditions of of PLLs PLLs (phase-locked (phase-locked loops) loops) in in the the downconverters.downconverters. ToTo solvesolve thisthis problem,problem, aa methodmethod forfor estimating estimating phase phase shifts shifts and and time time delaysdelays introduced introduced in in the the receivers receivers has has been been developed developed and and implemented implemented [ 15[15]] based based on on the the Sensors 2021, 21, x FOR PEER REVIEWmatched filtering of signals with superimposed pilot sequences from the training signal5 of 18 matched filtering of signals with superimposed pilot sequences from the training signal generator,generator, whose whose architecture architecture is is shown shown in in Figure Figure5. 5. The pilot sequence consisted of three repetitions of a complete PRN (pseudo-random) sequence, generated in an MCU (microcontroller) that modulates the carrier signal obtained from the frequency synthesizer followed by an attenuator. The carrier frequency (Fc) was programmed in accordance with the frequency of the broadcast transmitter used as the illumination source, and it was generated only during the short periods of pilot signal emission to avoid interferences with the received signals. BPSK (bipolar phase-shift keying) modulation was performed in a dual-balanced mixer. The modulated signal was then passed to a simple power divider to obtain its four copies that are finally fed into the directional couplers (see Figure 1.) The pilot signal was injected periodically in every 5 s.

FigureFigure 5. 5.Internal Internal structure structure of of the the training training signal signal generator. generator.

TheThe pilot acquired sequence signals consisted were stored of three on repetitions an SSD in ofthe a host complete PC, and PRN were (pseudo-random) later processed sequence,offline in generated the MATLAB in an MCUenvironment. (microcontroller) During thatthe preprocessing modulates the carrierstage, signalthe pilot obtained signals fromwere the extracted frequency with synthesizer matched filters followed and byused an to attenuator. calibrate the The phase carrier offsets frequency and time (Fc) delays was programmedbetween the received in accordance signals. with The the details frequency of the training of the broadcast signal properties, transmitter generation, used as the and illuminationprocessing have source, been and disclosed it was generated in [15]. The only experimental during the hardware short periods setup of that pilot was signal used during the experiments (without a host machine, which was a typical notebook) is presented in Figure 6.

Figure 6. Four-channel receiver with the training signal generator and the directional couplers.

2.2. Signal Processing The surveillance antennas were connected to channels 1, 2, 3 of the receiver and the reference antenna was connected to channel 4. After compensating for the phase shifts and time delays, the signals from the reference channel x4[n] and the selected surveillance channel x1[n] were processed using the PCL algorithm shown in Figure 7.

Figure 7. Simplified block diagram of the PCL (passive coherent location) processing algorithm.

Sensors 2021,, 21,, xx FORFOR PEERPEER REVIEWREVIEW 5 of 18

Sensors 2021, 21, 2171 5 of 19

emission to avoid interferences with the received signals. BPSK (bipolar phase-shift keying) modulation was performed in a dual-balanced mixer. The modulated signal was then passed to a simple power divider to obtain its four copies that are finally fed into the Figure 5. Internal structure of the training signal generator. directional couplers (see Figure1.) The pilot signal was injected periodically in every 5 s. The acquired signals were stored on an SSD in the host PC, and were later processed The acquired signals were stored on an SSD inin thethe hosthost PC,PC, andand werewere laterlater processedprocessed offline in the MATLAB environment. During the preprocessing stage, the pilot signals offline in the MATLAB environment. During the preprocessing stage, the pilot signals were extracted with matched filters and used to calibrate the phase offsets and time delays were extracted with matched filters and used to calibratecalibrate thethe phasephase offsetsoffsets andand timetime delaysdelays between the received signals. The details of the training signal properties, generation, between the received signals. The details of the training signal properties, generation, and and processing have been disclosed in [15]. The experimental hardware setup that was processing have been disclosed in [15]. The experimental hardware setup that was used used during the experiments (without a host machine, which was a typical notebook) is during the experiments (without a host machine, which was a typical notebook) is pre- presented in Figure6. sented in Figure 6.

FigureFigure 6. 6.Four-channel Four-channel receiver receiver with with the the training training signal signal generator generator and and the the directional directional couplers. couplers.

2.2.2.2. Signal Signal Processing Processing TheThe surveillance surveillance antennas antennas were were connected connected to to channels channels 1, 1, 2, 2, 3 3 of of the the receiver receiver and and the the referencereference antenna antenna was was connected connected to to channel channel 4. 4. After AfterAfter compensating compensatingcompensatingfor forfor the thethe phase phasephase shifts shiftsshifts and time delays, the signals from the reference channel x [n] and the selected surveillance and time delays, the signals from the reference channel 4x4[n] and the selected surveillance channel x [n] were processed using the PCL algorithm shown in Figure7. channel x11[n] were processed using the PCL algorithm shown in Figure 7.

Figure 7. Simplified block diagram of the PCL (passive coherent location) processing algorithm. FigureFigure 7. 7.Simpliﬁed Simplified block block diagram diagram of of the the PCL PCL (passive (passive coherent coherent location) location) processing processing algorithm. algorithm.

The remaining channels x2[n], x3[n] were intended to be processed by a more advanced version of the PCL algorithm supported by adaptive beamforming. The signal processing pipeline was composed of a clutter-removal ﬁlter (described with details in [1,18–22])

and a cross-ambiguity function calculator [1,23]. The default integration time processing algorithm was 262 ms. The ﬁnal results of the processing are a series of 2-D plots S[v,R] showing signals reﬂected from the objects in the bistatic velocity/range (v/R) plane.

2.3. Compensation of Phase Fluctuations Caused by Ship’s Motion We must note that, during the storm, the ship swayed intensely in the rough waves and changed, to a small extent, its orientation relative to the illuminating transmitter. As SensorsSensors 2021 2021, 21,, x21 FOR, x FOR PEER PEER REVIEW REVIEW 6 of6 18 of 18

TheThe remaining remaining channels channels x2[ nx],2[ nx],3[ nx]3 [nwere] were intended intended to beto beprocessed processed by bya morea more ad- ad- vancedvanced version version of theof the PCL PCL algorithm algorithm supported supported by byadaptive adaptive beamforming. beamforming. The The signal signal processingprocessing pipeline pipeline was was composed composed of aof clutter-removal a clutter-removal filter filter (described (described with with details details in in [1,18–22])[1,18–22]) and and a cross-ambiguity a cross-ambiguity function function calculator calculator [1,23]. [1,23]. The The default default integration integration time time processingprocessing algorithm algorithm was was 262 262 ms ms. The. The final final results results of theof the processing processing are are a series a series of 2-Dof 2-D plotsplots S[v S,R[v] ,Rshowing] showing signals signals reflected reflected from from the the objects objects in thein the bistatic bistatic velocity/range velocity/range (v/ R(v)/ R) plane.plane.

Sensors 2021, 21, 2171 2.3.2.3. Compensation Compensation of Phase of Phase Fluctuations Fluctuations Caused Caused by Ship’sby Ship’s Motion Motion 6 of 19 WeWe must must note note that, that, during during the the storm, storm, the the ship ship swayed swayed intensely intensely in thein the rough rough waves waves andand changed, changed, to ato small a small extent, extent, its itsorientation orientation relative relative to theto the illuminating illuminating transmitter. transmitter. As As the reference antenna was located at the front and the surveillance antennas were placed thethe referencereference antennaantenna waswas locatedlocated atat thethe frontfront andand thethe surveillancesurveillance antennasantennas werewere placedplaced on the port side of the ship (see Figure 3), the distance between these antennas was several onon thethe portport sideside ofof thethe shipship (see(see FigureFigure3 3),), the the distance distance between between these these antennas antennas was was several several meters.meters.meters. As As Asa result, aa result,result, both bothboth the thethe phase phasephase and andand amplitu amplitudeamplitude deof oftheof thethe direct directdirect signal signalsignal in inthein thethe reference referencereference and andand surveillancesurveillancesurveillance antennas antennas antennas depended depended depended on on onthe the the vessel’s vessel’s pitchpitch pitch andand and roll, roll, roll, and and theand the phase the phase phase offset offset offset between be- be- tweenthetween referencethe the reference reference signal signal andsignal and its and crosstalk its itscrosstalk crosstalk in the in measurement thein the measurement measurement signal signal wassignal was changing was changing changing during dur- dur- the ingintegrationing the the integration integration time. time. On time. the On otherOn the the other hand, other hand, the hand, clutter the th cluttere removalclutter removal removal ﬁlter filter requires filter requires requires a constant a constant a constant phase phaserelationshipphase relationship relationship between between between these these signals these signals signals during during during a single a single a single block’s block’s block’s coherent coherent coherent processing, processing, processing, such such as such in as stationaryinas stationaryin stationary PCL PCL installation. PCL installation. installation. Figure Figure Figure8 shows 8 shows 8 shows the examplethe the example example plots plots of plots the of discussed theof the discussed discussed phase phase offset, phase offset,estimatedoffset, estimated estimated with with 2 with ms 2 timems 2 ms time steps time steps during steps during during one one integration one integration integration period—in period—in period—in the the cases the cases cases of calmof calmof seascalm seasandseas and the and the storm. the storm. storm.

(a) (a) (b)( b)

FigureFigureFigure 8. Surveillance/reference 8.8. Surveillance/referenceSurveillance/reference signals’ signals’ phase phase difference difference during during calm calm seas seas (a) (anda)) andand the the thestorm stormstorm (b). (( bb).).

TheTheThe phase phasephase difference differencedifference between betweenbetween the thethe reference referencereference signal signalsignal and andand its itsitscrosstalk crosstalkcrosstalk fluctuated fluctuatedfluctuated in in bothin bothboth casescasescases due duedue to totheto thethe ship’s ship’sship’s movement. movement.movement. However, However,However, during duringduring the thethe storm, storm,storm, these thesethese fluctuations fluctuationsfluctuations were werewere foundfoundfound to tobeto bemuchbe muchmuch more moremore intense. intense.intense. Our OurOur experime experimentsexperimentsnts with withwith PCL PCLPCL signal signalsignal processing processingprocessing have havehave shown shownshown thatthatthat these thesethese fluctuations fluctuations fluctuations severely severely severely degrade degrade degrade the the the clutter clutter clutter removal removal removal filter’s filter’s filter’s efficiency efficiency efficiency in thein the casecase case of of theof the long long integration integration period. period. period. In In suchIn such such a acase, case,a case, because because because of of theof the the loss loss loss of of coherency,of coherency, coherency, it itis it isalmost is almost almost impossibleimpossibleimpossible to to filterto filter filter out out out the the the direct direct direct path path path inte interference interferencerference component component andand and obtainobtain obtain clear clear clear PCL PCL PCL images. im- im- ages.Toages. addressTo Toaddress address this problem,this this problem, problem, a more a advancedmore a more advanced advanced signal processingsignal signal processing processing algorithm algorithm hasalgorithm been has developed has been been developedwithdeveloped the structurewith with the the structure presented structure presented in presented Figure 9in. Figurein Figure 9. 9.

FigureFigureFigure 9. PCL 9.9. PCLPCL processing processingprocessing algorithm algorithmalgorithm modified modiﬁedmodified for for forlong longlong integrationintegration integration periods.periods. periods.

spline interpolation on the entire integration period. After that, both signals are divided into 200 overlapping blocks of equal length. To improve immunity against time-variant signal properties, each of these blocks is processed in a separate instance of the clutter removal filter (described previously). Finally, the filters’ output signals are merged into the entire integration period and processed in a cross-ambiguity function calculator. This approach allows for practically removing the direct path interference component in severe time-varying phase offsets resulting from the rapid movement of the PCL platform. We employed such a modified PCL processing algorithm to obtain clutter images shown in the following section. Sensors 2021, 21, x FOR PEER REVIEW 7 of 18

Firstly, the phase offset between the reference signal and its crosstalk is estimated by computing a cross-correlation function on subsequent parts of the integration period (block size was 4000 samples, which corresponds to 2 ms). According to the obtained es- timates, the phase offset of the measurement signal is then corrected in a phase equalizer, with spline interpolation on the entire integration period. After that, both signals are divided into 200 overlapping blocks of equal length. To improve immunity against time- variant signal properties, each of these blocks is processed in a separate instance of the clutter removal filter (described previously). Finally, the filters’ output signals are merged into the entire integration period and processed in a cross-ambiguity function calculator. This approach allows for practically removing the direct path interference component in severe time-varying phase offsets resulting from the rapid movement of the PCL platform. We employed such a modified PCL processing algorithm to obtain clutter images shown

Sensors 2021, 21, 2171 in the following section. 7 of 19

3. Results of Clutter Structure Analysis 3. ResultsDue of to Clutter the training Structure signals’ Analysis presence in the acquired signal, bright flashes occur in theDue whole to the S[ trainingv,R] plane, signals’ repeatable presence in with the acquired Tr period, signal, as shown bright flashes in Figure occur 10. in the Since the training wholesignalS[ vstructure,R] plane, repeatableand its time with ofT emissionr period, as were shown known in Figure in 10advance,. Since the this training interference may be signalremoved structure using and an its timeadaptive of emission filtering were knownalgorithm, in advance, for which this interference development may be has been left for removed using an adaptive filtering algorithm, for which development has been left for futurefuture work. work. Besides Besides these these short periodsshort periods of interference, of inte afterrference, filtering after out thefiltering clutter, theout the clutter, the cross-ambiguitycross-ambiguity plots plots can reveal can the reveal reflections the fromreflections moving objectsfrom inmoving the operating objects range in the operating ofrange the system, of the as system, demonstrated as demonstrated in Section4. in Section 4.

FigureFigure 10. 10.Sudden Sudden flash flash effect causedeffect caused by the interference by the interfer from anence injected from training an injected signal. training signal. Clutter Structure Variations Clutter Structure Variations In moving PCL platforms, the residual clutter structure and intensity change as the platformIn ismoving moving andPCL changing platforms, its orientation the residual [18–21 clu]. Intter Figure structure 11, we can and see intensity that the change as the shipplatform is turning is moving over a port, and which changing changes its the orientation direction of the [18–21]. antenna In array Figure with 11, respect we can see that the toship the illuminatingis turning transmitter.over a port, This which change changes entailed a the drop direction in the clutter of intensitythe antenna by 15 array dB with respect (the same color scale is used on three plots). In all cases, only the direct leakage signal was Sensors 2021, 21, x FOR PEER REVIEW 8 of 18 removedto the illuminating by an adaptive transmitter. lattice clutter cancelerThis change [22] with entailed the length a drop of filtration in the truncated clutter intensity by 15 todB the (the first same two range color cells. scale is used on three plots). In all cases, only the direct leakage signal was removed by an adaptive lattice clutter canceler [22] with the length of filtration truncated to the first two range cells.

Figure 11.Figure Change 11. Change of clutter of clutter intensity intensity caused caused by turning turning of platformof platform with antennawith antenna array. array.

The residual clutter structure in marine PCL installation also depends on the sea

Thestate residual as determined clutter by structure weather conditions. in marine The PCL clutter installation image observed also depends during calm on the sea state as determinedseas is depicted by weather in Figure conditions.12, with cross-sections The clu fortter different image bistatic observed ranges during (R). For calm this seas is de- pictedexperiment, in Figure the12, DAB with transmitter cross-sections (218.64 MHz)for different was selected bistatic for illumination, ranges (R and). For the this experiment, integrationthe DAB transmitter time was extended (218.64 to 5 MHz) s to improve was theselected resolution for inillumination, Doppler frequency and andthe integration time was extended to 5 s to improve the resolution in Doppler frequency and bistatic velocity. The short parts (52 ms) of the signal with interference from the training sequence were excluded from each integrated block’s processing to obtain clear images.

(a) (b) Figure 12. Clutter image during calm seas (a) and its cross-sections for selected bistatic ranges (b).

With the increased integration time, we can see artifacts repeating in the bistatic velocity axis with a ca. 15 m/s period. These artifacts result from the periodical structure of the DAB signal observed in the time domain (i.e., the transmission frame duration of 96 ms in mode I [24] with the NULL symbol at the beginning of each frame). A method of mitigating this effect may be developed in further work to improve the image’s quality. Moreover, the clutter becomes Doppler-spread (up to 4 … 5 m/s) around v = 0 line due to the PCL platform’s motion. A more distorted image of the residual clutter was obtained when PCL signal acquisition was performed during a storm (see Figure 13). The Doppler spread slightly increased for the nearest bistatic distances (e.g., R = 450 m). Additional regions of increased noise level (e.g., at R = 2 km, 4 km, 6 … 9 km) may result from signal reflections from heavy rainfall and rough waves at sea. In these regions, the Doppler spread can reach up to 10 … 20 m/s. The detection of departing CASA aircraft is also clearly visible around R = 9 km (the red rectangle). The echo is blurry in several range cells because of the long integration period in relation to the target’s speed. This aircraft detection proves that the modified version of the algorithm shown in Figure 9 can still detect moving targets.

Sensors 2021, 21, x FOR PEER REVIEW 8 of 18

Figure 11. Change of clutter intensity caused by turning of platform with antenna array.

The residual clutter structure in marine PCL installation also depends on the sea state as determined by weather conditions. The clutter image observed during calm seas is de- Sensors 2021, 21, 2171 picted in Figure 12, with cross-sections for different bistatic ranges (R). For this experi-8 of 19 ment, the DAB transmitter (218.64 MHz) was selected for illumination, and the integration time was extended to 5 s to improve the resolution in Doppler frequency and bistatic velocity.bistatic The velocity. short parts The short(52 ms) parts of the (52 signal ms) of wi theth signal interference with interference from the training from the sequence training weresequence excluded were from excluded each fromintegrated each integratedblock’s processing block’s processing to obtain clear to obtain images. clear images.

(a) (b)

FigureFigure 12. 12. ClutterClutter image image during during calm calm seas seas (a (a) )and and its its cross-sections cross-sections for for selected selected bistatic bistatic ranges ranges (b (b).).

WithWith the the increased increased integration integration time, time, we we can can see seeartifacts artifacts repeating repeating in the in bistatic the bistatic velocityvelocity axis axis with with a ca. a ca.15 m/s 15 m/s period. period. These These artifacts artifacts result result from from the theperiodical periodical structure structure of theof theDAB DAB signal signal observed observed in the in thetime time domain domain (i.e., (i.e., the thetransmission transmission frame frame duration duration of 96 of ms96 in ms mode in mode I [24] I [24with] with the theNULL NULL symbol symbol at the at thebeginning beginning of each of each frame). frame). A method A method of mitigatingof mitigating this this effect effect may may be bedeveloped developed in infurther further work work to toimprove improve the the image’s image’s quality. quality. Moreover,Moreover, the the clutter clutter becomes becomes Doppler Doppler-spread-spread (up toto 44 ...… 55 m/s) m/s) around around vv == 00 line line due due to to thethe PCL PCL platform’s platform’s motion. motion. AA more more distorted distorted image image of of the the residual residual cl clutterutter was obtained when PCL signalsignal acquisi-acqui- sitiontion waswas performed performed during during a storma storm (see (see Figure Figure 13). 13). The The Doppler Doppler spread spread slightly slightly increased in- creasedfor the for nearest the nearest bistatic bistatic distances distances (e.g., R (e.g.,= 450 R m).= 450 Additional m). Additional regions regions of increased of increased noise noiselevel level (e.g., (e.g., at R =at 2 R km, = 2 4km, km, 4 6 km,... 96 km)… 9 maykm) resultmay result from signalfrom signal reﬂections reflections from heavyfrom heavyrainfall rainfall and rough and rough waves waves at sea. at In sea. these In regions, these regions, the Doppler the Doppler spread canspread reach can up reach to 10 up... to20 10 m/s. … 20 The m/s. detection The detection of departing of departing CASA CASA aircraft aircraft is also is clearly also clearly visible visible around aroundR = 9 km R Sensors 2021, 21, x FOR PEER REVIEW 9 of 18 =(the 9 km red (the rectangle). red rectangle). The echo The is blurryecho is in blurry several in range several cells range because cells of because the long of integration the long integrationperiod inrelation period in to relation the target’s to the speed. target’s This speed. aircraft This detection aircraft detection proves that proves themodiﬁed that the modifiedversion of version the algorithm of the algorithm shown in shown Figure in9 can Figure still 9 detect can still moving detect targets. moving targets.

(a) (b)

FigureFigure 13.13. ClutterClutter imageimage duringduring thethe stormstorm ((aa)) andand itsits cross-sectionscross-sections forfor selected selected bistatic bistatic ranges ranges ( b(b).).

4. Results of Moving Target Detection A group of objects took part as cooperative targets in the APART-GAS campaign, such as civil and military aircraft, boats, and navy ships. In the following experiments, the integration time was set to 262 ms to facilitate the detection of fast-moving targets, unless noted otherwise. Target positions shown in maps have been derived from GPS logs and a separate ADS-B radio receiver.

4.1. Scenario 1: CASA C-295 and Fishing Boat (DVB-T Illumination) In Figure 14, we can see a situation map with the military ship (carrying the PCL receiver) marked with a red square, a cooperative CASA C-295 aircraft [25] passing around 10 km from the ship (green mark), a cooperative fishing boat in the close surroundings (yellow mark), and an accidentally passing cruise plane SAS756 (white square). The boresight directions of the surveillance (S) and reference (R) antennas have been illus- trated with triangles of different colors.

Figure 14. Situation map with a cooperative CASA C-295 aircraft and a nearby fishing boat.

A signal from the DVB-T (191.5 MHz) transmitter was selected as a source of illumination with receiving antennas with horizontal polarization and the reference antenna from the side of the ship. After filtering out the clutter, the echoes from the CASA aircraft can be seen as a dot moving consequently in the bistatic v/R plane (around R = 9 km), marked with red rectangles in Figure 15 (the consecutive v/R maps are separated by 4 s). The fishing boat’s echoes from the close surrounding cannot be separated from the clutter image due to the negligible Doppler frequency. The echoes from the cruise plane SAS756 that was moving more than 20 km northeast from the ship are not visible on the plots.

Sensors 2021, 21, x FOR PEER REVIEW 9 of 18

Sensors 2021, 21, 2171 (a) (b) 9 of 19 Figure 13. Clutter image during the storm (a) and its cross-sections for selected bistatic ranges (b).

4.4. Results Results of of Moving Moving Target Target Detection Detection AA group group of of objects objects took took part part as cooperativeas cooperative targets targets in the in APART-GAS the APART- campaign,GAS campaign, such assuch civil as and civil military and military aircraft, aircraft, boats, boats, and and navy navy ships. ships. In In the the following following experiments, experiments, the the integrationintegration time time was was set set to to 262 262 ms ms to to facilitate facilitate the the detection detection of of fast-moving fast-moving targets, targets, unless unless notednoted otherwise. otherwise. Target Target positions positions shown shown in in maps maps have have been been derived derived from from GPS GPS logs logs and and a a separateseparate ADS-B ADS-B radio radio receiver. receiver.

4.1.4.1. Scenario Scenario 1: 1: CASA CASA C-295 C-295 and and Fishing Fishing Boat Boat (DVB-T (DVB-T Illumination) Illumination) InIn Figure Figure 14 14,, we we can can see see a a situation situation map map with with the the military military ship ship (carrying (carrying the the PCL PCL receiver)receiver) marked marked with with a red a red square, square, a cooperative a cooperative CASA CASA C-295 aircraft C-295 aircraft[25] passing [25] around passing 10around km from 10 km the from ship the (green ship mark),(green mark), a cooperative a cooperative ﬁshing fishing boat in boat the in close the close surroundings surround- (yellowings (yellow mark), mark), and an and accidentally an accidentally passing passing cruise cruise plane SAS756plane SAS (white756 (white square). square). The bore- The sightboresight directions directions of the of surveillance the surveillance (S) and (S) reference and reference (R) antennas (R) antennas have beenhave illustratedbeen illus- withtrated triangles with triangles of different of different colors. colors.

FigureFigure 14. 14.Situation Situation map map with with a a cooperative cooperative CASA CASA C-295 C-295 aircraft aircraft and and a a nearby nearby ﬁshing fishing boat. boat.

AA signal signal fromfrom thethe DVB-TDVB-T (191.5 MHz) transmitter transmitter was was selected selected as as a asource source of of illumi- illu- minationnation with with receiving receiving antennas antennas with with horizontal polarization andand thethe referencereference antenna antenna fromfrom the the side side of of the the ship. ship. After After ﬁltering filtering out out the the clutter, clutter, the the echoes echoes from from the the CASA CASA aircraft aircraft cancan be be seen seen as as a a dot dot moving moving consequently consequently in in the the bistatic bistaticv /vR/Rplane plane (around(aroundR R= = 99 km), km), markedmarked with with red red rectangles rectangles in in Figure Figure 15 15 (the (the consecutive consecutivev /v/RRmaps maps areare separatedseparated byby 44 s). s). TheThe ﬁshing fishing boat’s boat’s echoes echoes from from the the close close surrounding surrounding cannot cannot be be separated separated from from the the clutter clutter imageimage due due to to the the negligible negligible Doppler Doppler frequency. frequency. The The echoes echoes from from the the cruise cruise plane plane SAS756 SAS756 thatthat was was moving moving more more than than 20 20 km km northeast northeast from from the the ship ship are are not not visible visible on on the the plots. plots.

4.2. Scenario 2: CASA C-295 and Fishing Boat (DAB Illumination) In Figure 16, we can see another situation map with the cooperative CASA C-295 aircraft passing on the starboard side of the ship, parallel to the coastline, between the illuminating transmitter and the PCL receiver, as well as the cooperative ﬁshing boat around 8 km away from the ship. A signal from the DAB (218.64 MHz) transmitter has been selected for illumination with vertically polarized antennas and the reference antenna at the front of the ship. Sensors 2021, 21, x FOR PEER REVIEW 10 of 18 Sensors 2021, 21, x FOR PEER REVIEW 10 of 18

4.2. Scenario 2: CASA C-295 and Fishing Boat (DAB Illumination) 4.2. Scenario 2: CASA C-295 and Fishing Boat (DAB Illumination) In Figure 16, we can see another situation map with the cooperative CASA C-295 aircraft passing onIn the F igurestarboard 16, we side can of see the another ship, parallel situation to themap coastline, with the between cooperative the CASA C-295 illuminating transmitteraircraft passing and theon thePCL starboard receiver, sideas well of the as ship,the cooperative parallel to thefishing coastline, boat between the around 8 km awayilluminating from the transmitter ship. A signal and thefrom PCL the receiver,DAB (218.64 as well MHz) as transmitter the cooperative has fishing boat Sensors 2021, 21, 2171 10 of 19 been selected arofor undillumination 8 km away with from vertic theally ship. polarized A signal antennas from the andDAB the (218.64 reference MHz) an- transmitter has tenna at the frontbeen of selected the ship. for illumination with vertically polarized antennas and the reference antenna at the front of the ship.

FigureFigure 15. 15. DetectionDetection results results of of CASA CASA C-295 C-295 (subsequen (subsequentt time time stamps), stamps), scene scene from from Figure Figure 14. 14 . Figure 15. Detection results of CASA C-295 (subsequent time stamps), scene from Figure 14.

Figure 16. Situation map with a cooperative CASA C-295 aircraft and a distant fishing boat. Figure 16. Situation map with a cooperative CASA C-295 aircraft and a distant fishing boat. Figure 16. Situation map with a cooperative CASA C-295 aircraft and a distant fishing boat. The CASA aircraft’s detection results can be seen in Figure 17 as a dot moving in the bistatic v/R plane (fromTheThe CASA CASA3 to 1 aircraft’skm), aircraft marked’s detection detection with a results resultsred rectangle can can be be seen (theseen inconsecutive in Figure Figure 17 17 asv /asR a amaps dot dot moving moving in in the the are separatedbistatic bybistatic 10 s).v /v The/RRplane plane intensity (from(from of 3 3 to thisto 1 1 km), echokm),marked increasesmarked with with as the a a red red aircraft rectangle rectangle gets (the closer(the consecutive consecutive to the v/ vR/Rmaps maps PCL receiver.are Theare separated separated detection by byfalls 10 10 s).into s). The The the intensity intensityclutter area of of this this(v = echo echo0) as increases increasesthe aircraft as as theis the flying aircraft aircraft above gets gets closer closer to to the the the transmitter–receiverPCLPCL receiver. receiver. baseline. The The detection detection The echoes falls falls intofrom into the thethe clutter clutterfishing area area boat ( v(v =cannot = 0) 0) as as the bethe aircraftextracted aircraft is is flying flying above above from the plotsthe thesince transmitter–receiver transm they itterare lost–receiver in the baseline. baseline.clutter. TheThe echoesechoes fromfrom the the fishing fishing boat boat cannot cannot be be extracted extracted fromfrom the the plots plots since since they they are are lost lost in in the the clutter. clutter.

Sensors 2021, 21, x FOR PEER REVIEW 11 of 18 Sensors 2021, 21,Sensors x FOR PEER2021, 21REVIEW, 2171 11 of 18 11 of 19

Figure 17. Detection results of the CASA C-295 aircraft (subsequent time stamps, scene in Figure FigureFigure 17. 17.Detection Detection results results of theof the CASA CASA C-295 C-295 aircraft aircraft (subsequent (subsequent time time stamps, stamps, scene scene in Figurein Figure 16). 16). 16). 4.3. Scenario 3: PC-12, Fishing Boat, CASA C-295, F-16 (DVB-T Illumination) 4.3. Scenario 3: 4.3.PC-12, Scenario Fishing 3: PCBoat,-12, CASA Fishing C-295, Boat ,F-16 CASA (DVB-T C-295, Illumination) F-16 (DVB-T Illumination) In Figure 18, we have shown a situation map with the cooperative PILATUS PC-12 In Figure 18, weIn Figurehave shown 18, we a havesituation shown map a situationwith the mapcooperative with the PILATUS cooperative PC-12 PILATUS PC-12 (SUI812(SUI812 [ 26[26])]) aircraft aircraft passing passing in in front front of of the the ship, ship, the the ﬁshing fishing boat boat around around 7 7 km km from from the the (SUI812 [26]) front,aircraft and passing the CASA in front aircraft of the moving ship, the away fishing at the boat back around with an7 km F-16 from jet ﬁghter.the A signal front, and the front,CASA and aircraft the CASA moving aircraft away movingat the back away with at thean F-16back jetwith fighter. an F -A16 signaljet fighter. A signal fromfrom the the DVB-T DVB-T (191.5 (191.5 MHz) MHz) transmitter transmitter was was selected selected for for illumination illumination with with horizontally horizontally from the DVB-Tpolarized (191.5 receivingMHz) transmitter antennas was and selected the reference for illumination antenna from with the horizontally side of the ship. polarized receivingpolarized antennas receiving and theantennas reference and antenna the reference from theantenna side offrom the theship. side of the ship.

Figure 18. SituationFigureFigure map 18. 18. Situationwith Situation the cooperative map map with with the theaircraft cooperative cooperative and fishing aircraft aircraft boat. and and ﬁshing fishing boat. boat.

Detection resultsDetectionDetection for this resultsresults situation forfor are this depicted situation in are are Figure depicted depicted 19. The in in Figure echoes Figure 19. from 19 The. Thethe echoes PC- echoes from from the thePC- 12 aircraft flyingPC-1212 nearlyaircraft aircraft the flying PCL ﬂying nearly receiver nearly the (ca. thePCL 5 PCL km)receiver receiver are clearly (ca. (ca. 5 km) seen 5 km) are in clearlyarethe clearlybistatic seenseen v in/R the plane in bistatic the bistatic v/R planev/R around R = 7.5planearound km, aroundwith R =the 7.5R bistatic =km, 7.5 with km, velocity the with bistatic theturning bistatic velocity from velocity positive turning turning tofrom negative. frompositive positive The to detec-negative. to negative. The detec- The tion was not detectionpossibletion was duringwas not notpossible a possible period during duringfrom a 17 period a0 periodto 177 from froms when 170 170 to the to 177 177 bistatic s s when when velocity the bistaticbistatic was velocityvelocity was was close to zero closeandclose the to to zeroecho zero and wasand the lostthe echo echoin the was was clutter. lost lost in inDetection the the clutter. clutter. of Detectionthe Detection fishing of boatof the the cannot ﬁshing fishing be boat boat cannot cannot be be

Sensors 2021, 21, x FOR PEER REVIEW 12 of 18 Sensors 2021, 21, 2171 12 of 19

Sensors 2021, 21extracted, x FOR PEER from REVIEW the v/R plots, and the distant F-16 jet fighter and CASA airplane are also 12 of 18 extracted from the v/R plots, and the distant F-16 jet ﬁghter and CASA airplane are also not visible in thisnot configuration. visible in this conﬁguration.

extracted from the v/R plots, and the distant F-16 jet fighter and CASA airplane are also not visible in this configuration.

Figure 19.Figure Detection 19. Detection results results of the of SUI812/PILATUS the SUI812/PILATUS PC-12 PC-12 aircraft aircraft (scene (scene in in Figure Figure 18). 18). Figure 19. Detection results of the SUI812/PILATUS PC-12 aircraft (scene in Figure 18). 4.4. Scenario 4: PC-12, Fishing Boat, F-16 (DVB-T Illumination) 4.4. Scenario 4: PC-12,4.4. Scenario Fishing 4: Boat, PC-12, F-16 Fishing (DVB-T Boat ,Illumination) F-16 (DVB-T Illumination) A situation map with the PILATUS PC-12 aircraft (SUI812), two F-16 jet fighters flying A situation map Awith situation the PILATUS map with PC-12 the PILATUS aircraft PC (SUI812),-12 aircraft two (SUI812), F-16 jet two fighters F-16 jet fly- fighters fly- near the ship, and the nearby fishing boat is shown in Figure 20 (PCL receiver setup was ing near the ship,ing and near the the nearby ship, andfishing the nearbyboat is fishing shown boat in Figureis shown 20 in (PCL Figure receiver 20 (PCL setup receiver setup identical as previously.) was identical as previously.)was identical as previously.)

FigureFigure 20. 20.Situation Situation map map of of the the PILATUS PILATUS PC-12 PC-12 and and two two F-16 F- 16 jet jet ﬁghters. fighters.

Figure 20. Situation mapDetection of the PILATUS results forPC-12 this and situation two F-16 are depictedjet fighters. in Figure 21. The echoes from the PC- 12 aircraft departing the PCL receiver are visible in the bistatic v/R plane around R = 9 km Detection resultsand moving for this to situation v = 0, R = are 9.5 depicted km. At the in same Figure time, 21. we The can echoes see rapidly from themoving PC- detection 12 aircraft departingresults the from PCL the receiver two F- 16are fighters. visible These in the echoes bistatic can v /moveR plane as fast around as 200 R m/s = 9 (in km the bistatic and moving to v = 0, R = 9.5 km. At the same time, we can see rapidly moving detection results from the two F-16 fighters. These echoes can move as fast as 200 m/s (in the bistatic

Sensors 2021, 21, 2171 13 of 19

Detection results for this situation are depicted in Figure 21. The echoes from the Sensors 2021, 21, x FOR PEER REVIEW 13 of 18 PC-12 aircraft departing the PCL receiver are visible in the bistatic v/R plane around R = 9 km and moving to v = 0, R = 9.5 km. At the same time, we can see rapidly moving detection results from the two F-16 ﬁghters. These echoes can move as fast as 200 m/s (in range) and quicklythe bistatic and unpredictably range) and quickly change and their unpredictably bistatic velocity, change unlike their bistaticthe CASA velocity, C- unlike the 295 or PC-12 CASAairplanes. C-295 or PC-12 airplanes.

FigureFigure 21. 21.Detection Detection results results of of the the PILATUS PILATUS PC-12 PC-12 and and two tw F-16o F-16 jet jet ﬁghters fighters (scene (scene from from Figure Figure 20 20).).

4.5. Cruse Airplane4.5. Cruse THY9 Airplane (DAB Illumination) THY9 (DAB Illumination) A more in-depthA more analysis in-depth of the analysis acquired of signals the acquired shows that signals it is showsalso possible that it to is de- also possible to tect cruise airplanesdetect cruise flying airplanes accidentally flying near accidentally the PCL receiver near the installed PCL receiver on the installed military on the military ship. Figure 22ship. shows Figure a situation 22 shows with a situation the ship withsailing the northeast ship sailing slowly northeast and the slowly THY9 and [27] the THY9 [27] cruise airplanecruise (Boeing airplane B789 (BoeingDreamliner) B789 flying Dreamliner) northwest, flying ca. northwest, 10 km from ca. the 10 kmship. from The the ship. The DAB transmitterDAB (218.64 transmitter MHz) (218.64 was selected MHz) wasfor illumination, selected for illumination,and the vertically and thepolarized vertically polarized antennas withantennas the front with reference the front antenna reference were antenna employed. were Detection employed. results Detection of this results cruise of this cruise flight are shownflight inare Figure shown 23, inwith Figure a clear 23, dot with traveling a clear dot in a traveling bistatic R in = a 8 bistatic km areaR and= 8 km area and decreasing bistaticdecreasing velocity. bistatic During velocity. a period During of 77 a period… 86 s, ofthe 77 echo... 86 from s, the the echo airplane from is the airplane is lost in the clutterlost inarea the due clutter to low area bistatic due to velocity. low bistatic This velocity. moment This correlates moment very correlates well with very well with the THY9 markthe crossing THY9 mark on the crossing transmitter– on thereceiver transmitter–receiver baseline on the baseline map, onas shown the map, in as shown in Figure 22. Figure 22.

Figure 22. Situation map for the THY9 airplane (Boeing Dreamliner).

Sensors 2021, 21, x FOR PEER REVIEW 13 of 18

range) and quickly and unpredictably change their bistatic velocity, unlike the CASA C- 295 or PC-12 airplanes.

Figure 21. Detection results of the PILATUS PC-12 and two F-16 jet fighters (scene from Figure 20).

4.5. Cruse Airplane THY9 (DAB Illumination) A more in-depth analysis of the acquired signals shows that it is also possible to detect cruise airplanes flying accidentally near the PCL receiver installed on the military ship. Figure 22 shows a situation with the ship sailing northeast slowly and the THY9 [27] cruise airplane (Boeing B789 Dreamliner) flying northwest, ca. 10 km from the ship. The DAB transmitter (218.64 MHz) was selected for illumination, and the vertically polarized antennas with the front reference antenna were employed. Detection results of this cruise flight are shown in Figure 23, with a clear dot traveling in a bistatic R = 8 km area and decreasing bistatic velocity. During a period of 77 … 86 s, the echo from the airplane is Sensors 2021, 21, 2171 lost in the clutter area due to low bistatic velocity. This moment correlates very well14 ofwith 19 the THY9 mark crossing on the transmitter–receiver baseline on the map, as shown in Figure 22.

Sensors 2021, 21, x FOR PEER REVIEW 14 of 18

Figure 22. Situation map for the THY9 airplane (Boeing Dreamliner). Figure 22. Situation map for the THY9 airplane (Boeing Dreamliner).

Figure 23. DetectionFigure 23. resultsDetection of the results THY9 of airplane the THY9 (subsequent airplane (subsequent time stamps). time stamps).

4.6. Cruse Airplane4.6. THY79K Cruse Airplane (DAB THY79K Illumination) (DAB Illumination) In Figure 24, weIn can Figure see a24 very, we similar can see asitu veryation similar with situation another with cruise another airplane, cruise THY79K airplane, THY79K (Boeing Dreamliner(Boeing [28]), Dreamliner flying six [28 minutes]), ﬂying sixlater minutes in the later same in theair samecorridor, air corridor, and the and coop- the coopera- erative CASA airplanetive CASA (marked airplane with (marked a green with sq auare) green square)flying at ﬂying the atcoastline, the coastline, around around 5 km 5 km from from the ship (PCLthe shipreceiver (PCL setup receiver was setup identical was identical as previously). as previously). In the detection plot (Figure 25), there are clearly visible echoes from the THY79K airplane that was moving on a track located at 7 … 10 km of the bistatic range. The shape and location of this track are very similar to that of THY9 shown in Figure 23. The detection of THY79K is not visible from 454 to 467 s when it falls into the clutter area (bistatic v = 0), and this moment correlates with the THY79K mark passing the transmitter–receiver baseline (see Figure 24). Detection of the departing CASA airplane is also visible as a bright point traveling steadily from 6 to 10 km of the bistatic range. It goes beyond the visible area (10 km) after 455 s.

Figure 24. Situation map for THY79K airplane (Boeing Dreamliner) and CASA C-295.

Sensors 2021, 21, x FOR PEER REVIEW 14 of 18

Figure 23. Detection results of the THY9 airplane (subsequent time stamps).

4.6. Cruse Airplane THY79K (DAB Illumination) In Figure 24, we can see a very similar situation with another cruise airplane, THY79K (Boeing Dreamliner [28]), flying six minutes later in the same air corridor, and the cooperative CASA airplane (marked with a green square) flying at the coastline, around 5 km from the ship (PCL receiver setup was identical as previously). In the detection plot (Figure 25), there are clearly visible echoes from the THY79K airplane that was moving on a track located at 7 … 10 km of the bistatic range. The shape and location of this track are very similar to that of THY9 shown in Figure 23. The detection of THY79K is not visible from 454 to 467 s when it falls into the clutter area (bistatic v = 0), and this moment correlates with the THY79K mark passing the transmitter–receiver Sensors 2021, 21, 2171 baseline (see Figure 24). Detection of the departing CASA airplane is also visible15 of as 19 a bright point traveling steadily from 6 to 10 km of the bistatic range. It goes beyond the visible area (10 km) after 455 s.

FigureFigure 24. 24.Situation Situation map map for for THY79K THY79K airplane airplane (Boeing (Boeing Dreamliner) Dreamliner) and and CASA CASA C-295. C-295.

In the detection plot (Figure 25), there are clearly visible echoes from the THY79K airplane that was moving on a track located at 7 ... 10 km of the bistatic range. The shape and location of this track are very similar to that of THY9 shown in Figure 23. The detection of THY79K is not visible from 454 to 467 s when it falls into the clutter area (bistatic v = 0), and this moment correlates with the THY79K mark passing the transmitter–receiver baseline (see Figure 24). Detection of the departing CASA airplane is also visible as a bright Sensors 2021, 21, x FOR PEER REVIEW 15 of 18 point traveling steadily from 6 to 10 km of the bistatic range. It goes beyond the visible area (10 km) after 455 s.

FigureFigure 25. 25. DetectionDetection results results of ofthe the THY79K THY79K and and CASA CASA airplanes airplanes (subsequent (subsequent time time stamps). stamps).

4.7. Cruse Airplane4.7. Cruse LOT4CG Airplane (DVB-T LOT4CG Illumination) (DVB-T Illumination) In Figure 26, Inwe Figure can see 26 a, situation we can see map a situation with the map airplane with LOT4CG the airplane [29] LOT4CG (EMBRAER [29] (EMBRAER 175) entering175) over entering the land, over passing the land, the ship passing with thethe shipPCL withreceiver, the PCL and receiver,later the andillumi- later the illuminating DVB-T transmitter (191.5 MHz). The horizontally polarized antennas with the side reference antenna were employed for signal acquisition.

Figure 26. Situation map for the LOT4CG airplane (EMBRAER 175).

This flight was detected in the bistatic v/R plane (see Figure 27) as a very weak signal, visible only during a short time (around 29 s). It must be noted that the cross-section area of this airplane is smaller than in the case of Boeing B789, which yields a lower energy of reflected signals. After doubling the integration time to 524 ms, this aircraft’s echo became stronger and visible for about 33 s longer, as shown in Figure 28.

Sensors 2021, 21, x FOR PEER REVIEW 15 of 18

Figure 25. Detection results of the THY79K and CASA airplanes (subsequent time stamps). Sensors 2021, 21, 2171 16 of 19 4.7. Cruse Airplane LOT4CG (DVB-T Illumination) In Figure 26, we can see a situation map with the airplane LOT4CG [29] (EMBRAER 175) entering over the land, passing the ship with the PCL receiver, and later the illuminating DVB-T transmitter (191.5 MHz). The horizontally polarized antennas with the side nating DVB-T transmitter (191.5 MHz). The horizontally polarized antennas with the side reference antenna were employed for signal acquisition. reference antenna were employed for signal acquisition.

Figure 26. Situation map for the LOT4CG airplane (EMBRAER 175). Figure 26. Situation map for the LOT4CG airplane (EMBRAER 175).

ThisThis ﬂight flight was was detected detected in in the the bistatic bistaticv /vR/Rplane plane (see (see Figure Figure 27 27)) as as a a very very weak weak signal, signal, visiblevisible only only during during a a short short time time (around (around 29 29 s). s). It It must must be be noted noted that that the the cross-section cross-section area area ofof this this airplane airplane is is smaller smaller than than in in the the case case of of Boeing Boeing B789, B789, which which yields yields a a lower lower energy energy of of Sensors 2021, 21, x FOR PEER REVIEW 16 of 18 reﬂectedreflected signals. signals. After After doubling doubling the the integration integration time time to to 524 524 ms, ms, this this aircraft’s aircraft’s echo echo became became strongerstronger and and visible visible for for about about 33 33 s s longer, longer, as as shown shown in in Figure Figure 28 28..

Figure 27. DetectionFigure 27. resultsDetection of the results LOT4CG of the airplane LOT4CG (scene airplane in Figure (scene 26). in Figure 26).

Figure 28. LOT4CG detection visible with integration time increased to 524 ms (scene in Figure 26).

5. Conclusions

Sensors 2021, 21, x FOR PEER REVIEW 16 of 18

Sensors 2021, 21, 2171 17 of 19

Figure 27. Detection results of the LOT4CG airplane (scene in Figure 26).

FigureFigure 28. 28.LOT4CG LOT4CG detection detection visible visible with with integration integration time time increased increased to 524to 524 ms ms (scene (scene in Figurein Figure 26). 26). 5. Conclusions 5. Conclusions The presented results clearly show that seaborne passive radar using nearby DAB or DVB-T transmitters can detect airborne targets of different kinds. This allows the system to be used for monitoring the airspace over coastal areas. Such a system can be relatively inexpensive, as it can be built using commercial off-the-shelf VHF antennas, general- purpose SDR receivers, and a computer. Better detection performance can be obtained with more advanced equipment (e.g., more sensitive receivers, higher-gain antennas) and more sophisticated signal processing (e.g., integrating signals from more receiving channels and extending the integration time, which necessitates the use of range/Doppler migration correction techniques). Maritime targets (such as the ﬁshing boat) were not detected in the conducted experiment. This resulted from the fact that the bistatic velocity of the target was small and the target echo was inseparable from the clutter, which was spread in the velocity dimension. To detect slow targets, such as other ships or boats, longer integration times should be employed, which will increase the velocity resolution. In addition, more advanced clutter removal techniques may be employed, such as STAP (space-time adaptive processing), which provides higher selectivity for unwanted signal attenuation. In general, the problem of clutter removal in seaborne platforms is much more complicated than in the case of stationary ground-based radars. This complication results from the fact that the platform is moving, and an additional Doppler shift is introduced. The clutter characteristics will change depending on the geometry of the transmitter, clutter, and radar, which should be taken into account when the algorithm for clutter cancellation is designed.

Author Contributions: Conceptualization, G.M. and K.K.; methodology, K.K.; software, G.M., K.K.; validation, K.K. and M.M.; formal analysis, K.K.; investigation, G.M.; resources, K.K. and A.D.; data curation, A.D.; writing—original draft preparation, G.M. and M.M.; writing—review and editing, G.M.; visualization, G.M. and M.M.; supervision, K.K. and M.M.; project administration, K.K.; funding acquisition, K.K. All authors have read and agreed to the published version of the manuscript. Sensors 2021, 21, 2171 18 of 19

Funding: This research received no external funding. Institutional Review Board Statement: Not applicable. Informed Consent Statement: Not applicable. Data Availability Statement: Data sharing is not applicable to this article. Conﬂicts of Interest: The authors declare no conﬂict of interest.

References 1. Malanowski, M. Signal Processing for Passive Bistatic Radar; Artech House: Boston, MA, USA; London, UK, 2019; ISBN 9781630816629. 2. Brown, J.; Woodbridge, K.; Griffiths, H.; Stove, A.; Watts, S. Passive bistatic radar experiments from an airborne platform. IEEE Aerosp. Electron. Syst. Mag. 2012, 27, 50–55. [CrossRef] 3. Dawidowicz, B.; Kulpa, K.S.; Malanowski, M. Suppression of the ground clutter in airborne PCL radar using DPCA technique. In Proceedings of the 2009 European Radar Conference (EuRAD), Rome, Italy, 30 September–2 October 2009; pp. 306–309. 4. Dawidowicz, B.; Kulpa, K.S.; Malanowski, M.; Misiurewicz, J.; Samczynski, P.; Smolarczyk, M. DPCA Detection of Moving Targets in Airborne Passive Radar. IEEE Trans. Aerosp. Electron. Syst. 2012, 48, 1347–1357. [CrossRef] 5. Wojaczek, P.; Summers, A.; Cristallini, D.; Walterscheid, I.; Lombardo, P.; Colone, F. Results of Airborne PCL Under CCI Conditions Using DVB-T Illuminators of Opportunity. In Proceedings of the 2018 International Conference on Radar (RADAR), Brisbane, Australia, 27–31 August 2018; pp. 1–6. 6. Wojaczek, P.; Colone, F.; Cristallini, D.; Lombardo, P. Reciprocal-Filter-Based STAP for Passive Radar on Moving Platforms. IEEE Trans. Aerosp. Electron. Syst. 2019, 55, 967–988. [CrossRef] 7. Dawidowicz, B.; Kulpa, K.S. Experimental results from PCL radar on moving platform. In Proceedings of the 2008 International Radar Symposium, Wroclaw, Poland, 21–23 May 2008; pp. 1–4. 8. Ummenhofer, M.; Schell, J.; Heckenbach, J.; Kuschel, H.; O’Hagan, D.W. Doppler estimation for DVB-T based Passive Radar systems on moving maritime platforms. In Proceedings of the 2015 IEEE Radar Conference (RadarCon), Arlington, VA, USA, 10–15 May 2015; pp. 1687–1691. 9. Palmer, J.; Summers, A.; Ummenhofer, M.; Bournaka, G.; Palumbo, S.; Cristallini, D.; Kuschel, H. Receiver platform motion estimation using terrestrial broadcast transmitters for passive radar. In Proceedings of the 2015 IEEE Radar Conference, Johannesburg, South Africa, 27–30 October 2015; pp. 151–155. 10. Płotka, M.; Abratkiewicz, K.; Malanowski, M.; Samczyński,P.; Kulpa, K. The Use of the Reassignment Technique in the Time-Frequency Analysis Applied in VHF-Based Passive Forward Scattering Radar. Sensors 2020, 20, 3434. [CrossRef][PubMed] 11. Płotka, M.; Malanowski, M.; Samczyński,P.; Kulpa, K.; Abratkiewicz, K. Passive Bistatic Radar Based on VHF DVB-T Signal. In Proceedings of the 2020 IEEE International Radar Conference (RADAR), Washington, DC, USA, 28–30 April 2020; pp. 596–600. 12. Radecki, K.; Samczyński,P.; Gromek, D.; Kulpa, K. The use of Barycentric BPA for passive SAR imaging. In Proceedings of the 2020 21st International Radar Symposium (IRS), Warsaw, Poland, 5–8 October 2020; pp. 236–239. [CrossRef] 13. Duk, V.; Wojaczek, P.; Rosenberg, L.; Cristallini, D.; O’Hagan, D.W. Airborne Passive Radar Detection for the APART-GAS Trial. In Proceedings of the 2020 IEEE Radar Conference (RadarConf20), Florence, Italy, 21–25 September 2020; pp. 1–6. 14. RSPduo Dual Tuner 14-Bit SDR. Available online: https://www.sdrplay.com/wp-content/uploads/2018/05/RSPduoDatasheetV0 .6.pdf (accessed on 14 December 2020). 15. Mazurek, G.; Rytel-Andrianik, R. Pilot-based calibration of dual-tuner SDR receivers. In Proceedings of the 2020 28th European Signal Processing Conference (EUSIPCO), Amsterdam, The Netherlands, 18–21 January 2021; pp. 1971–1975. 16. RSPduo General Specifications. Available online: https://www.sdrplay.com/wp-content/uploads/2018/06/RSPDuo-Technical- Information-R1P1.pdf (accessed on 18 February 2021). 17. Mazurek, G. Lightweight SDR Platform for Passive Coherent Location Applications. In Proceedings of the 2019 Signal Processing Symposium (SPSympo), Krakow, Poland, 17–19 September 2019; pp. 39–44. 18. Kulpa, K.; Dawidowicz, B.; Maslikowski, L.; Gajo, Z. Single channel clutter cancelation in mobile PCL. In Proceedings of the 2018 22nd International Microwave and Radar Conference (MIKON), Poznan, Poland, 14–17 May 2018; pp. 593–594. 19. Kulpa, K.; B ˛aczyk,M.; Misiurewicz, J.; Malanowski, M.; Gromek, D. Limits of Ground Clutter CLEAN Based Cancelation in Mobile PCL Radar. In Proceedings of the 2018 19th International Radar Symposium (IRS), Bonn, Germany, 20–22 June 2018; pp. 1–7. 20. Kulpa, K.; B ˛aczyk,M.; Malanowski, M.; Misiurewicz, J.; J˛edrzejewski,K.; Gajo, Z.; Ma´slikowski,Ł. CLEAN Removal of Ground Clutter in Mobile Passive Radar. In Proceedings of the 2019 IEEE Radar Conference (RadarConf), Boston, MA, USA, 22–26 April 2019; pp. 1–4. 21. Kulpa, K. Simple Sea Clutter Canceller for Noise Radar. In Proceedings of the 2006 International Radar Symposium (IRS), Krakow, Poland, 24–26 May 2006; pp. 1–4. 22. Kulpa, K. Adaptive Clutter Rejection in Bi-static CW Radar. In Proceedings of the 2004 International Radar Symposium (IRS), Warsaw, Poland, 19–21 May 2004; pp. 61–68. 23. Kulpa, K.; Misiurewicz, J.; Malanowski, M.; Samczyński, P.; Smolarczyk, M. Recent developments in passive radars. In Proceedings of the Military Sensors 2009 Conference, London, UK, 18–20 November 2009. Sensors 2021, 21, 2171 19 of 19

24. Hoeg, W.; Lauterbach, T. Digital Audio Broadcasting: Principles and Applications of Digital Radio, 2nd ed.; John Wiley & Sons: Chichester, UK, 2003; ISBN 0-470-85013-2. 25. EADS CASA C-295. Available online: https://en.wikipedia.org/wiki/EADS_CASA_C-295 (accessed on 14 December 2020). 26. Swiss Air Force 812 (SUI812). Available online: https://flightaware.com/live/flight/SUI812 (accessed on 31 January 2021). 27. Turkish Airlines 9 (THY9). Available online: https://flightaware.com/live/flight/THY9 (accessed on 31 January 2021). 28. Turkish Airlines 79K (THY79K). Available online: https://flightaware.com/live/flight/THY79K (accessed on 31 January 2021). 29. LOT Polish Airlines 4CG (LOT4CG). Available online: https://flightaware.com/live/flight/LOT4CG (accessed on 31 January 2021).