Dipartimento INFOCOM – Roma, 21 Ottobre 2008 Dottorato di ricerca in Telerilevamento – XXII ciclo Relazione dell’attività di ricerca svolta durante il II anno Carlo Bongioanni Multichannel Passive Radar: signal processing and experimental prototype development Tutor: Prof. Pierfrancesco Lombardo Outline Short summary of first year activity PBR processing scheme Wide-band prototypes development INFOCOM acquisition campaigns Experimental results validation: ATC data Multi-Frequency detection schemes Performance assessment of the Multi-Frequency approach A new acquisition device: ICS-554B Conclusions and future works Publications and conferences Carlo Bongioanni – Dottorato di ricerca in Telerilevamento (XXII ciclo) Relazione dell’attività di ricerca svolta durante il II anno 2 PBR Processing scheme REFERENCE CHANNEL Ref SURVEILLANCE CHANNEL DISTURBANCE CANCELLATION Surv Surv’ Ref 2D CROSSCORRELATION CFAR THRESHOLD Carlo Bongioanni – Dottorato di ricerca in Telerilevamento (XXII ciclo) Relazione dell’attività di ricerca svolta durante il II anno 3 Time varying characteristics of the received FM radio signals Instantaneous characteristics of the transmitted waveform Both range resolution and sidelobe level depend on the instantaneous transmitted waveform on the different FM radio channels (program content) REFERENCE CHANNEL SURVEILLANCE CHANNEL MULTIPATH REMOVAL DISTURBANCE CANCELLATION 2D CROSSCORRELATION CFAR THRESHOLD Instantaneous characteristics of the e.m. environment The instantaneous characteristics of the FM channels also depend on the presence/absence of multipath effects (propagation) Co- and inter- channel interferences can increase the system noise floor thus reducing the achievable detection performance Carlo Bongioanni – Dottorato di ricerca in Telerilevamento (XXII ciclo) Relazione dell’attività di ricerca svolta durante il II anno 4 Experimental results on the single FM channels Temporal sequence of three plots on the same Range-Velocity map 93.8 MHz 89.8 MHz -400 -400 -300 -300 15 15 410 -200 425 -100 361 331 392 100 429 316 200 315 20 40 60 80 100 120 Relative Bistatic Range [Km] 140 92.1 MHz 160 180 100 429 316 422 20 40 60 80 100 120 Relative Bistatic Range [Km] 413 425 -100 361 49 364 337 396 0 302 331 392 100 429 316 402 428 314 200 300 422 40 80 100 120 Relative Bistatic Range [Km] 361 140 160 180 200 297 364337 302 331 392 100 429 316 400 0 400 428 314 40 80 100 120 Relative Bistatic Range [Km] 361 140 160 180 200 Carlo Bongioanni – Dottorato di ricerca in Telerilevamento (XXII ciclo) Relazione dell’attività di ricerca svolta durante il II anno 297 364 337 396 0 302 331 392 100 429 316 400 0 428 314 315 287 300 10.05.42 10.06.05 10.06.29 60 425 200 422 196 413 -100 402 315 287 20 410 49 396 0 300 10.05.42 10.06.05 10.06.29 60 425 200 287 200 406 196 413 -100 402 315 410 -200 406 297 Bistatic Velocity [m/s] 49 180 15 -200 196 Bistatic Velocity [m/s] 410 160 -300 15 406 140 94.3 MHz 400 -300 20 10.05.42 10.06.05 10.06.29 -400 15 -200 315 287 -400 -300 428 314 91.3 MHz 400 -400 Bistatic Velocity [m/s] 302 400 0 200 364 337 331 392 300 10.05.42 10.06.05 10.06.29 297 396 0 200 422 400 0 361 402 287 300 425 -100 428 314 196 413 297 364 337 302 402 400 0 49 396 0 406 196 413 Bistatic Velocity [m/s] Bistatic Velocity [m/s] 49 410 -200 406 422 20 40 10.05.42 10.06.05 10.06.29 60 80 100 120 Relative Bistatic Range [Km] 140 160 180 5 200 Multi-Frequency detection schemes (I) zn,m n=0,..,N-1, m=0,..,M-1 R.V. z with pz(z) and Dz(z) 0 1 n N-1 REFERENCE CHANNEL SURVEILLANCE CHANNEL xm=f(z0,m, z1,m, …, zN-1,m) m=0,..,M-1 R.V. x with px(x) and Dx(x) Ref DISTURBANCE CANCELLATION Surv Surv’ Ref 2D CROSSCORRELATION MULTI-FREQUENCY INTEGRATION CFAR THRESHOLD Doppler Shift Carlo Bongioanni – Dottorato di ricerca in Telerilevamento (XXII ciclo) Relazione dell’attività di ricerca svolta durante il II anno 6 Multi-Frequency detection schemes (II) The availability of different FM radio channels can collect a huge amount and a large diversity of information Integration strategies Decentralized (“L out of N” logic) Centralized Linear: xm = ∑z zn,m n=0,..,N-1, m=0,..,M-1 R.V. z with pz(z) and Dz(z) n N −1 SUM 0 1 n, m m = 0,.., M − 1 N-1 n =0 Centralized Non-Linear: MAX xm = max{z n, m }n = 0,.., N −1 m = 0,.., M − 1 MIN xm = min {z n, m }n = 0,.., N −1 m = 0,.., M − 1 xm=f(z0,m, z1,m, …, zN-1,m) m=0,..,M-1 R.V. x with px(x) and Dx(x) Slightly different integration times required Noise floor normalization before integration Carlo Bongioanni – Dottorato di ricerca in Telerilevamento (XXII ciclo) Relazione dell’attività di ricerca svolta durante il II anno 7 Multi-Frequency detection schemes (III) Theoretical Pfa expressions for the considered integration strategies: CA-CFAR detection scheme C E N T R A L I Z E D N −1 SUM MAX P fa = ∑ n =0 M NM + n − 1 n G SUM −CA + M NM + n G SUM −CA M N iM Pfa = ∑ N ( −1) n +1 ∑ N ( −1) i +1 n i nG MAX −CA + iM n =1 i =1 N M Pfa = GMIN −CA + M MIN M M N DECENTRALIZED n i Pfa = ∑ N PfaSC (1 − PfaSC ) N −i i i = L Carlo Bongioanni – Dottorato di ricerca in Telerilevamento (XXII ciclo) Relazione dell’attività di ricerca svolta durante il II anno PfaSC M = GSC −CA + M M 8 Multi-Frequency detection schemes (IV) Theoretical Pfa expressions for the considered integration strategies: GO-CFAR detection scheme C E N T R A L I Z E D SUM M Pfa = ∑ m =0 N −1 N −1 M (− 1)m m ∑ k0 =0 N MAX MIN ∑ k1 =0 N −1 L ∑ km −1 =0 Pfa = ∑ N ( −1) n +1 n n =1 N GSUM (N − 1 + k0 + k1 + L + k m−1 )! −GO (N − 1)!k 0 !k1!Lk m−1! (GSUM −GO + m) N +k0 +k1 +L+km−1 NM ( −1) i +1 ∑ NM i nG MAX −GO + i i i =1 M m Pfa = ∑ M ( −1) m+1 m G MIN −GO + m m=1 N DECENTRALIZED i Pfa = ∑ N PfaSC (1 − PfaSC ) N −i i i = L M PfaSC Carlo Bongioanni – Dottorato di ricerca in Telerilevamento (XXII ciclo) Relazione dell’attività di ricerca svolta durante il II anno m = ∑ M ( −1) m +1 m GSC −GO + m m =1 9 Multi-Frequency detection schemes (V) Theoretical Pfa expressions for the considered integration strategies: SO-CFAR detection scheme C E N T R A L I Z E D SUM MAX N −1 N −1 P fa = 1 − ∑ ∑ N −1 L k0 = 0 k1 =0 N M −1 N ( i +1)−1 Pfa = MN ∑ ∑ ∑ k =1 i =0 l =0 1 N M − 1 N (i + 1) − 1 ( −1) k +i +l +1 k i l kG MAX −SO + l Pfa = MIN N DECENTRALIZED N G SUM ( N − 1 + k 0 + k1 + L + k M −1 )! − SO N (N − 1)! k 0 ! k1!L k M −1! (G SUM −GO + M ) −1+k0 +k1 +L+k M −1 k M −1 =0 ∑ M G MIN − SO + M Pfa = ∑ N PfaSC (1 − PfaSC ) N −i i i = L i Carlo Bongioanni – Dottorato di ricerca in Telerilevamento (XXII ciclo) Relazione dell’attività di ricerca svolta durante il II anno PfaSC = M G SC − SO + M 10 Wide-Band Prototypes development Reference Antenna Dual channel 1st Nyquist 2nd Nyquist A/D Converter FM Band Direct RF sampling 0 Fs/2 Surveillance Antenna Fs (88 ÷ 108 MHz) External Clock Source RF Amp. IF Amp. Down Conv MIXER RF Low-pass Dual channel Filter A/D Converter IF Signal down-conversion & baseband sampling LO RF IF RF Amp. Down Conv IF Amp. Low-pass Filter MIXER Carlo Bongioanni – Dottorato di ricerca in Telerilevamento (XXII ciclo) Relazione dell’attività di ricerca svolta durante il II anno 11 INFOCOM Acquisition campaigns RX site: Riva di Traiano (Civitavecchia, ~70km north of Rome) Reference antenna steered toward Monte Argentario (pointing north-west) Surveillance antenna pointed at about 180° (standard arrival and departure routes to and from Leonardo Da Vinci airport in Fiumicino) Selection of a limited number of FM radio channels 300 data files covering 2 hours Single acquisition duration: about 1 sec Temporal spacing between acquisitions: about 23 sec Live ATC registrations available Carlo Bongioanni – Dottorato di ricerca in Telerilevamento (XXII ciclo) Relazione dell’attività di ricerca svolta durante il II anno 12 Experimental results validation: ATC data Example: 17 May 2007 acquisition Single channel (93.8MHz) Multi-Frequency integration: 89.8, 92.1, 93.8MHz (nominal Pfa = 10-4) Comparison between detection results and live air traffic control registrations Performance evaluation for the implemented system: Pd e Pfa Carlo Bongioanni – Dottorato di ricerca in Telerilevamento (XXII ciclo) Relazione dell’attività di ricerca svolta durante il II anno 13 Performance assessment of the MF approach (I) Extensive performance evaluation for the Multi-Frequency approaches 5 different MF configurations considered: MFC2: 89.8 + 93.8 MHz MFC3: 89.8 + 93.8 + 92.1 MHz MFC4a: 89.8 + 93.8 + 92.1 + 91.3 MHz MFC4b: 89.8 + 93.8 + 92.1 + 94.3 MHz MFC5: 89.8 + 93.8 + 92.1 + 91.3 + 94.3 MHz ΔR1: 5 ÷ 35 km ΔR2: 35 ÷ 65 km ΔR3: 65 ÷ 95 km CA-CFAR and GO-CFAR: obtained False Alarm Rates are largely comparable with the nominal Pfa value (10-3) SO-CFAR: limited robustness when applied against real data Carlo Bongioanni – Dottorato di ricerca in Telerilevamento (XXII ciclo) Relazione dell’attività di ricerca svolta durante il II anno 14 Performance assessment of the MF approach (II) 1 0.9 0.9 0.8 0.8 0.7 0.7 0.6 0.6 Detection Rate 1 0.5 0.4 0.4 0.2 0.2 ∆R1 ∆R2 ∆R3 0.1 0 0.5 0.3 0.3 SFC1 SFC2 SFC3 SFC4 SFC5 MFC2 MFC3 Single/Multi-Frequency Configurations 1 CA-CFAR MFC4a MFC4b MFC5 ∆R1 ∆R2 ∆R3 0.1 0 SFC1 SFC2 SFC3 SFC4 SFC5 MFC2 MFC3 Single/Multi-Frequency Configurations ∆R1 ∆R2 ∆R3 0.9 MFC4a MFC4b MFC5 GO-CFAR 0.8 0.7 Detection Rate Detection Rate Comparison between single frequency channels (SFC1-5) and the decentralized integration strategies (Pfa = 10-3) 0.6 0.5 0.4 0.3 0.2 SO-CFAR 0.1 0 SFC1 SFC2 SFC3 SFC4 SFC5 MFC2 MFC3 Single/Multi-Frequency Configurations MFC4a MFC4b MFC5 Carlo Bongioanni – Dottorato di ricerca in Telerilevamento (XXII ciclo) Relazione dell’attività di ricerca svolta durante il II anno 15 Performance assessment of the MF approach (III) False alarm rates obtained with the different centralized integration strategies in conjunction with the different CFAR detection schemes (nominal Pfa = 10-3) CA-CFAR and GO-CFAR: proper control of the Pfa with all the considered integration schemes SO-CFAR: false alarm rate comparable with the nominal one only in few cases MAX approach (both with CA and GO-CFAR) yields slightly higher false alarm rates at short ranges: disturbance residuals in Range-Velocity map Carlo Bongioanni – Dottorato di ricerca in Telerilevamento (XXII ciclo) Relazione dell’attività di ricerca svolta durante il II anno 16 Performance assessment of the MF approach (IV) Comparison among the different centralized integration strategies (Pfa = 10-3) LINEAR MAX MIN 1 1 1 0.9 0.9 0.9 0.8 0.8 0.8 0.7 0.7 0.7 0.6 0.5 0.4 Detection Rate Detection Rate Detection Rate CA-CFAR ∆R1 0.6 0.5 0.4 0.6 0.5 0.4 0.3 0.3 0.3 0.2 0.2 0.2 0.1 0.1 0.1 0 0 MFC2 MFC3 MFC4aMFC4b MFC5 Multi-Frequency Configurations LINEAR 0 MFC2 MFC3 MFC4aMFC4b MFC5 Multi-Frequency Configurations ∆R2 ∆R3 MAX MFC2 MFC3 MFC4aMFC4b MFC5 Multi-Frequency Configurations MIN 1 1 1 0.9 0.9 0.9 0.8 0.8 0.8 0.7 0.7 0.7 0.6 0.5 0.4 Detection Rate Detection Rate Detection Rate GO-CFAR ∆R1 0.6 0.5 0.4 0.6 0.5 0.4 0.3 0.3 0.3 0.2 0.2 0.2 0.1 0.1 0.1 0 MFC2 MFC3 MFC4aMFC4b MFC5 Multi-Frequency Configurations 0 MFC2 MFC3 MFC4aMFC4b MFC5 Multi-Frequency Configurations ∆R2 ∆R3 0 MFC2 MFC3 MFC4aMFC4b MFC5 Multi-Frequency Configurations MIN approach always yields worst performance SUM (LINEAR) and MAX integration strategies show similar performance SUM has slight advantage at far range MAX is slightly better than SUM in detection probability when using a small number of integrated channels CA-CFAR seems preferable with respect to GO-CFAR (slight performance improvement) Carlo Bongioanni – Dottorato di ricerca in Telerilevamento (XXII ciclo) Relazione dell’attività di ricerca svolta durante il II anno 17 Performance assessment of the MF approach (V) Detection performance comparison of the best performing integration strategies using a CA-CFAR detection scheme (Pfa = 10-3) ∆R2 ∆R3 1 0.9 0.9 0.9 0.8 0.8 0.8 0.7 0.7 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 SUM MAX DEC SC2 SC1 MFC2 MFC3 MFC4aMFC4b MFC5 Multi-Frequency Configurations Detection Rate 1 Detection Rate Detection Rate ∆R1 1 0.6 0.5 0.4 0.3 0.2 0.1 0 SUM MAX DEC SC2 SC1 MFC2 MFC3 MFC4aMFC4b MFC5 Multi-Frequency Configurations SUM MAX DEC SC2 SC3 0.6 0.5 0.4 0.3 0.2 0.1 0 MFC2 MFC3 MFC4aMFC4b MFC5 Multi-Frequency Configurations Detection capability limited due to the long baseline between the TX and RX site Proposed multi-frequency approaches yield a remarkable performance improvement with respect to the best single-channels case for all the considered surveillance regions Carlo Bongioanni – Dottorato di ricerca in Telerilevamento (XXII ciclo) Relazione dell’attività di ricerca svolta durante il II anno 18 ICS-554B: main features Technical Specifications: ADC outputs are applied to a bank of 4 digital down-converter (DDC) modules (Graychip GC4016 devices) Each GC4016 device includes 4 independent DDC channels 4 analog input channels (4 ADCs) ADC resolution: 14 bits Input impedance: 50Ohm Full scale input: 1.2 Vpk-pk (about 5.5dBm) Input Signal Bandwidth: 2MHz ÷ 200MHz Internal or external clock and trigger Max. Sampling Rate: 100MHz for 4 channels, simultaneous Min. Sampling Rate: 30MHz for 4 channels, simultaneous Internal Sample Clock Oscillator: 100MHz fixed frequency crystal Spurious Free Dynamic Range (SFDR): > 85dB Signal-to-Noise Ratio (SNR): > 71dB User Programmable FPGA (1 million gates) SMA coaxial connectors Simultaneous down-conversion of up to 16 arbitrary signal bands (e.g. 16 FM channels): onboard decimation, software selectable Carlo Bongioanni – Dottorato di ricerca in Telerilevamento (XXII ciclo) Relazione dell’attività di ricerca svolta durante il II anno 19 ICS-554B: block diagram Several operation modes available Examples of acquisition configurations: 2 antennas (2 ADC channels), 16 FM channels (8+8) 4 antennas (4 ADC channels), 16 FM channels (4+4+4+4) … Carlo Bongioanni – Dottorato di ricerca in Telerilevamento (XXII ciclo) Relazione dell’attività di ricerca svolta durante il II anno 20 ICS-554B: GC4016 block diagrams Each GC4016 device can be configured for 4, 2 or 1 output channel Each DDCs can have independent frequency, phase and gain controls Carlo Bongioanni – Dottorato di ricerca in Telerilevamento (XXII ciclo) Relazione dell’attività di ricerca svolta durante il II anno 21 Conclusions and future works (I) The development of wideband experimental prototypes allowed us to apply a multi-frequency approach which enhances the surveillance capabilities of PBR. The proposed approach yields improved performance w.r.t. the single FM radio channels, since it makes the detection robust w.r.t. both the content of transmitted radio broadcast channel and the channel multipath conditions. The performance improvement is paid in terms of computational load In the specific considered case, a significant increase in the number of integrated channels does not yield a further performance improvement (poorer characteristics of additional channels). Further improvement expected by adding good-performing channels (when available). Detection performance is still limited by interferences due to other FM radio transmissions… Carlo Bongioanni – Dottorato di ricerca in Telerilevamento (XXII ciclo) Relazione dell’attività di ricerca svolta durante il II anno 22 Conclusions and future works (II) Different strategies can be considered to improve the detection performance: Automatic selection of the FM radio channels Direct signal-to-Noise Ratio (DNR): this parameter allows to identify high power transmissions and strongly affects theoretical cancellation and detection performance Reference to Surveillance Power Ratio (Front-to-Back ratio) Program content Interference rejection based on the exploitation of different polarizations different FM radio channels transmitted with different polarizations, even from the same transmitter: use of more than two receiving channels Carlo Bongioanni – Dottorato di ricerca in Telerilevamento (XXII ciclo) Relazione dell’attività di ricerca svolta durante il II anno 23 Conclusions and future works (III) Different strategies can be considered to improve the detection performance: Antenna based interference rejection using multiple receiving channels Direction of Arrival (DOA) estimation example: one reference channel and two surveillance channels to perform detection and DOA estimation based on phase differences evaluation (monopulse) DOA estimation with a multi-channel and multi-frequency configuration Track initiation Multi-temporal analysis (used together with multi-frequency approach) Carlo Bongioanni – Dottorato di ricerca in Telerilevamento (XXII ciclo) Relazione dell’attività di ricerca svolta durante il II anno 24 Conclusions and future works (IV) Exploitation of different opportunity waveforms: digital TV signals Frequency range: 52 ÷ 88 and 174 ÷ 230 MHz 470 ÷ 870 MHz VHF UHF Wider bandwidth: 7 ÷ 8 MHz increased range resolution High power transmissions Wide coverage: switch-off of analogical transmissions already started Lazio: II semester of 2009 OFDM modulation Presence of deterministic peaks Carlo Bongioanni – Dottorato di ricerca in Telerilevamento (XXII ciclo) Relazione dell’attività di ricerca svolta durante il II anno 25 Publications and conferences A. Lauri, F. Colone, R. Cardinali, C. Bongioanni, P. Lombardo, “Analysis and emulation of FM radio signals for passive radar”, 2007 IEEE Aerospace Conference, Big Sky (MT), USA, 3-10 March 2007 C. Bongioanni, F. Colone, S. Bernardini, L. Lelli, A. Stavolo, P. Lombardo, “Passive radar prototypes for multifrequency target detection”, Signal Processing Symposium 2007, Jachranka Village (Poland), 24-26 May 2007 F. Colone, P. Lombardo, C. Bongioanni, “Track initiation for FM-based passive radar using multi-frequency and multi-temporal integration”, 5th Multi-National Passive Covert Radar Conference, Shrivenham, Swindon (UK), 13-15 November 2007 P. Lombardo, F. Colone, C. Bongioanni, “Comparison of different approaches for a MultiFrequency FM Based Passive Bistatic Radar”, Sensors and Technology for Defence Against Terrorism, Mannheim, Germany, 22-25 April 2008 P. Lombardo, F. Colone, C. Bongioanni, A. Lauri, T. Bucciarelli, “PBR activity at INFOCOM: adaptive processing techniques and experimental results”, 2008 IEEE Radar Conference, Rome (Italy), 26-30 May 2008 C. Bongioanni, F. Colone, P. Lombardo, “Performance Analysis of a Multi-Frequency FM Based Passive Bistatic Radar ”, 2008 IEEE Radar Conference, Rome (Italy), 26-30 May 2008 (Best student paper award) F. Colone, C. Bongioanni, P. Lombardo, “Multi-Frequency target detection in FM Radio Based Passive Bistatic Radar”, IEEE Transactions on Aerospace and Electronic Systems (attualmente in revisione) Carlo Bongioanni – Dottorato di ricerca in Telerilevamento (XXII ciclo) Relazione dell’attività di ricerca svolta durante il II anno 26