Enhanced processing gain via pulse polarity switching in an RF photonic phase filter

Fast pulse polarity switching is proposed and demonstrated to enhance processing gain in a comb-based radio-frequency photonic phase filter. The polarity switching scheme overcomes previous limits on time bandwidth product and processing gain based on the number of optical frequency comb lines. In an experiment with broadband jamming noise, the peak signal-to-noise ratio of compressed RF output pulses is improved by ~30 dB compared to the input average signal-to-noise ratio. © 2016 Optical Society of America OCIS codes: (060.5625) Radio frequency photonics; (320.5520) Pulse compression. References and links 1. M. I. Skolnik, Radar Handbook (McGraw-Hill, 1990). 2. M. K. Simon, J. K. Omura, R. A. Scholtz, and B. K. Levitt, Spread Spectrum Communications Handbook (McGraw-Hill, 1994). 3. R. Brocato, J. Skinner, G. Wouters, J. Wendt, E. Heller, and J. Blaich, “Ultra-wideband SAW correlator,” IEEE Trans. Ultrason. Ferroelectr. Freq. Control 53(9), 1554–1556 (2006). 4. M. Bolea, J. Mora, B. Ortega, and J. Capmany, “Highly chirped single-bandpass microwave photonic filter with reconfiguration capabilities,” Opt. Express 19(5), 4566–4576 (2011). 5. M. Song, V. Torres-Company, R. Wu, A. J. Metcalf, and A. M. Weiner, “Compression of ultra-long microwave pulses using programmable microwave photonic phase filtering with > 100 complex-coefficient taps,” Opt. Express 22(6), 6329–6338 (2014). 6. H.-J. Kim, A. Rashidinejad, and A. M. Weiner, “Low-loss ultrawideband programmable RF photonic phase filter for spread spectrum pulse compression,” IEEE Trans. Microw. Theory Tech. 62(12), 4173–4187 (2015). 7. M. Li, M. Antonio, L. Sophie, J. P. Yao, and J. Azana, “Reconfigurable and single-shot chirped microwave pulse compression using a time-spectrum convolution system,” in Int. Topical Meeting Microw. Photon. Conf. (IEEE, 2011), pp. 9–12. 8. C. Wang and J. P. Yao, “Chirped microwave pulse compression using a photonic microwave filter with a nonlinear phase response,” IEEE Trans. Microw. Theory Tech. 57(2), 496–504 (2009). 9. E. Hamidi and A. M. Weiner, “Phase-only matched filtering of ultrawideband arbitrary microwave waveforms via optical pulse shaping,” J. Lightwave Technol. 26(15), 2355–2363 (2008). 10. Y. Li, A. Dezfooliyan, and A. M. Weiner, “Photonic synthesis of spread spectrum radio frequency waveforms with arbitrarily long time apertures,” J. Lightwave Technol. 32(20), 3580–3587 (2014). 11. A. Rashidinejad, D. E. Leaird, and A. M. Weiner, “Ultrabroadband radio-frequency arbitrary waveform generation with high-speed phase and amplitude modulation capability,” Opt. Express 23(9), 12265–12273 (2015). 12. H.-J. Kim, D. E. Leaird, A. J. Metcalf, and A. M. Weiner, “Comb-based RF photonic filters based on interferometric configuration and balanced detection,” J. Lightwave Technol. 32(20), 3478–3488 (2014). 13. C. Finot, B. Kibler, L. Provost, and S. Wabnitz, “Beneficial impact of wave-breaking for coherent continuum formation in normally dispersive nonlinear fibers,” J. Opt. Soc. Am. B 25(11), 1938–1948 (2008). 14. V. K. Ingle and J. G. Proakis, Digital Signal Processing Using MATLAB (CENGAGE Learning, 2012). 15. K. N. Madsen, T. D. Gathman, S. Daneshgar, T. C. Oh, J. C. Li, and J. F. Buckwalter, “A High-Linearity, 30 GS/s Track-and-Hold Amplifier and Time Interleaved Sample-and-Hold in an InP-on-CMOS Process,” IEEE J. Solid-State Circuits 50(11), 2692–2702 (2015). 16. S. E. Turner, Jr., R. B. Elder, D. S. Jansen, and D. E. Kotecki, “4-bit adder-accumulator at 41-GHz clock frequency in InP DHBT technology,” IEEE Microw. Wirel. Compon. Lett. 15(3), 144–146 (2005). 17. Y. Li, A. Rashidinejad, J.-M. Wun, D. E. Leaird, J.-W. Shi, and A. M. Weiner, “Photonic generation of W-band arbitrary waveforms with high time-bandwidth products enabling 3.9mm Range Resolution,” Optica 1(6), 446– 454 (2014). 18. E. Hamidi, D. E. Leaird, and A. M. Weiner, “Tunable programmable microwave photonic filters based on an optical frequency comb,” IEEE Trans. Microw. Theory Tech. 58(11), 3269–3278 (2010). Vol. 24, No. 22 | 31 Oct 2016 | OPTICS EXPRESS 25846 #273995 http://dx.doi.org/10.1364/OE.24.025846 Journal © 2016 Received 19 Aug 2016; revised 28 Sep 2016; accepted 10 Oct 2016; published 28 Oct 2016


Introduction
Spread spectrum radio-frequency (RF) systems are well known and widely applied [1,2].In civilian cellular radio, spread spectrum enables multiple-access through its interference suppression property.In defense electronics spread spectrum is valued for its low power spectral density, which imparts covertness, and for its ability to redistribute a narrowband signal over a much larger bandwidth, thereby giving jamming resistance.In chirped radar spread spectrum signals provide low peak power in both time and frequency, important in the typical scenario of a peak-power-limited transmitter, while retaining both high pulse energy (for good detectability) and high bandwidth (which converts to high time and range resolution with a pulse compression receiver).The ability to spread a signal over a large frequency range at low power spectral density may provide an avenue toward operation in a highly congested RF spectrum with reduced interference.
In general, large spectral spreading factor (termed processing gain) is desirable for all spread spectrum applications.From a time domain perspective, the processing gain is closely related to the time-bandwidth product (TBP) of the spread spectrum signal and the achievable pulse compression ratio.
Current spread spectrum systems are typically constrained to RF bandwidths of hundreds of MHz or below.One of the key bottlenecks is the implementation of pulse compression receivers at frequencies above approximately 1 GHz [3].Photonics is fundamentally capable of delivering and processing larger signal bandwidths compared to electronics and with greater signal integrity.Recently schemes based on RF photonics that achieve phase filtering and compression of high frequency spread spectrum signals have been demonstrated [4][5][6][7][8][9].
To date the highest TBP reported was 75 [4].Scaling to substantially higher TBPs is difficult and is limited, for example, by the number of comb lines in RF photonics filtering schemes based on optical frequency combs [5,6] as well as by the complexity provided by optical spectral shaping and filtering devices [4,[7][8][9].In this paper we introduce a novel hybrid spread spectrum scheme that overlays coordinated photonic-electronic processing on top of our previously reported, comb-based, reconfigurable RF photonic phase filtering approach [6].The new scheme preserves the advantages of photonic spectral processing while scaling to essentially unlimited processing gains.We experimentally demonstrate a processing gain of ~103, which allows us to achieve 30 dB signal-to-interference improvement in an experiment with broadband noise jamming.

Hybrid processing concept
The proposed hybrid spread spectrum scheme is illustrated in Fig. 1. Figure 1(a) shows best current practice: a waveform which we call a "chip," with bandwidth B, duration T (a few ns to perhaps tens of ns) and TBP below 100, is generated at the transmitter.The receiver employs an RF photonic phase filter to compress this chip into a pulse with duration ~B−1 , with corresponding peak power enhancement.Such operation has been demonstrated previously, e.g [6].In order now to expand the TBP and processing gain as required, we concatenate a series of N chips into a "spread spectrum frame," in which individual chips with temporal profile v c (t) are phase modulated according to a suitable pseudorandom code under electronic control.The RF waveform w(t) generated at the transmitter is now written where the am are a series of complex amplitudes corresponding to the pseudorandom code.Figure 1(b) depicts the case where the am take on binary phase shift values ± 1; however, other schemes (e.g., quadrature phase shifting) are also possible.Note that pseudorandom phase codes such as PN sequences should be used for the a m in order to preserve the important low power spectral density property of the spread spectrum signal; in contrast, repeating identical copies of the spread spectrum chip (setting all the a m = 1) would lead to undesirable peaking in the RF spectrum.The time aperture of the spread spectrum frame is now NT, and its TBP is NBT, i.e., expanded N times compared to the TBP of a single chip.Such spread spectrum frames composed of concatenated phase switched chips can be generated either by an electronic waveform generator or through optical generation schemes such as [10,11].Now at the receiver, we may consider processing to involve two steps.In the first step, Fig. 1(c), the waveform is again presented to the RF photonic spectral phase filter which achieves compression of the individual chips.This results in a series of RF pulses of duration 1/B, but the polarities remain modulated.At this point we can write the received and partially processed RF waveform r 1 (t) as ( ) ( ) where v comp (t) is the waveform of a single chip after compression.In the second step, Fig. 1(d), the individual chips are modulated by the conjugate of the pseudorandom spreading code, which despreads the spread spectrum frame such that the polarities of the compressed chips are now all the same.At the this stage of processing, the waveform is written (For a pseudorandom phase spreading code, we have the coefficients |a m | 2 = 1.)These two processing steps could be performed in either order or even simultaneously.In our experiment the steps shown in Figs.1(c) and 1(d) are performed simultaneously in the RF photonic phase filter, which is modified as discussed below to accomplish this task.The series of compressed pulses may now be sampled; the additional processing gain is achieved by integrating the sampled signal either in hardware or in digital signal processing.In this paper we use a simple tapped delay line filter implemented via off-line processing, where to obtain the final processed signal r final (t),we compute the simple sum In this way the processing gain attainable with the single chip photonic-assisted compression (BT) is multiplied up by the number of chips (N) to achieve processing gain NBT.The number of chips and the enhancement of the processing gain can potentially be very large.The configuration is similar to that employed in our previous RF photonic phase filtering experiment [6], except for the addition of the phase modulator in the upper arm.The filter transfer function can be expressed as

Comb-based RF photonic phase filter
where P n is the RF amplitude of n th tap produced by beating the nth optical carrier with its corresponding sideband generated at the MZM; ∆ω is the comb spacing; ω RF is the angular RF frequency; ψ 2 is the coefficient for the second-order spectral phase imparted by the dispersive fiber; τ is the delay difference between the two interferometer arms; and the φ n 's are the phases applied to each optical comb line with the pulse shaper.θ(t) ( = πV C (t)/V π ) is the optical phase shift introduced by the phase modulator, where V C (t) is the control voltage to the phase modulator and V π is the half-wave voltage of the phase modulator.The first phase terms indicate that the second-order phase from the dispersive fiber introduces linear group delay between the various filter taps, which enables implementation of a finite impulse response filter.The differential tap delay (T) is Δω•ψ 2 .
The amplitude and phase of the filter taps can be programmed by the pulse shaper inside the interferometer.In the current paper, the optical carrier power is controlled in the pulse shaper to provide taps with a flat amplitude profile, which increases the TBP.Pulse compression action depends on the ability to control the phase of the taps.These may be programmed by applying the desired optical phases (φ n ) in the pulse shaper; interference between carriers from the upper branch of the interferometer and sidebands from the lower branch of the interferometer transfers the optical phases into the electrical domain.For example, a quadratic RF phase response (or linear chirp delay response) of the filter can be achieved by applying a quadratic spectral phase function to the optical carriers, as follows [6]: (6) The pulse compression functionality described thus far is identical to that in [6].The new feature here is that the pulse polarity in the phase filter can be switched by using the optical phase modulator inside the interferometer.For example, when the control voltage to the phase modulator is zero or V π , the filter transfer function can be expressed as As shown in Eq. ( 7), the transfer function with a control voltage of V π has the opposite sign, compared to the filter transfer function with zero control voltage.This indicates that the input waveform polarity of the filter can be changed via the control voltage.In principle, by using appropriate control signals to the phase modulator, other phase factors beyond 0 and π can also be implemented; however, only 0 and π levels will be demonstrated in the experiments that follow.
It is worth noting that polarity switching has recently been demonstrated in the context of photonic-assisted RF arbitrary waveform generation (RF-AWG) [10,11].In one paper the polarity switching overcomes the time aperture limitations usually encountered in photonicsbased RF-AWG.This enables generation of waveforms with nonrepeating features over essentially unlimited time apertures, important to avoid distance ambiguities in ranging experiments [10].In a second paper, a phase modulator and pulse shaper are incorporated together into an interferometer in a configuration similar to that of Fig. 2 to overlay two-level and multi-level phase shift keying onto ultrabroadband RF spread spectrum waveforms [11].
Here for the first time we modify an RF photonic phase filter to receive and process such polarity-switched spread spectrum waveforms.

Experiment and results
A broadband and flat-topped frequency comb is generated in the same setup used in our previous work on phase filtering [6].It consists of an electro-optic frequency comb including one intensity modulator and one phase modulator, a pulse shaper, and a length of highly nonlinear fiber.A seed electro-optic frequency comb with the comb spacing of 12.5 GHz is shaped in the programmable pulse shaper to yield a secant hyperbolic shaped pulse.This input pulse shape to the highly nonlinear fiber results in self-phase modulation-based spectral broadening with an approximately flat-topped shape [13].The broadened comb after a highly nonlinear fiber has ~29 nm optical bandwidth within a 5-dB power variation.The broadened comb shown in Fig. 3(a) is directed to the interferometer shown in Fig. 2. The pulse shaper (Finisar WaveShaper 1000P) in the upper branch of the interferometer selects and shapes 293 lines of the broadened comb to produce constant amplitude filter taps.The phases applied to the comb lines can be reconfigured to provide essentially arbitrary tap-dependent phases; however, in the experiments here we focus on one specific quadratic phase function.The halfwave voltage of the phase modulator (EOSPACE PM-5K1-20) is 2.8 V at 1 GHz.In the lower interferometer arm, the MZM (EOSPACE AZ-0K5-10) has a half-wave voltage of 3V at 1 GHz and an extinction ratio of 20 dB.The total output photocurrent at the output of the BPD (Discovery Semiconductors DSC720-HLPD) is 8.2 mA.The dispersive fiber of the bidirectional fiber configuration is implemented by a dispersion compensating fiber (DCF) with total dispersion of −800 ps/nm.The differential tap delay is 79.3 ps.The filter frequency response is evaluated by a network analyzer (Agilent N5230C).
Figures 3(b) and 3(c) show the frequency response of the phase filter.The pulse shaper is programmed to realize a quadratic phase function corresponding to β = 0.004 (see Eq. ( 6)).As explained in [6], the value of β realized is equal to the sum of the quadratic phase programmed onto the pulse shaper (0.0027) and an additional quadratic phase contribution of 0.0013 arising from the higher order dispersion of the DCF.The 3-dB bandwidth and chirp rate measured for this filter are 4.2 GHz and 206 MHz/ns, respectively.The product of filter bandwidth and delay aperture is equivalent to a TBP of ~86, to our knowledge the highest yet reported for an RF photonic pulse compression filter.The measured filter amplitude and group delay responses agree closely with the responses simulated based on Eq. ( 5); the measured chirp rate agrees closely with the calculated value (202 MHz/ns) based on Eq. ( 4) of ref. 6, using β = 0.004.For pulse compression experiments an input down-chirp waveform is generated from an RF arbitrary waveform generator (Tektronix AWG7122C) with a sampling rate of 24 GS/s.The chirp waveform has an RF bandwidth of 4 GHz at the center frequency of 4 GHz and time aperture of 19.4 ns (TBP = 77.6).The peak-to-average power ratio (PAPR) of the chirp waveform is approximately 5 dB within the time aperture.The guard period between adjacent input chirp waveforms is set equal to the time aperture of an individual chirp, yielding an overall repeat period of 38.8 ns.The pulse polarity control signal is generated from a function generator (Agilent 33250A) synchronized with the RF-AWG.At the output of the filter, a low-noise amplifier and a bandpass filter (with a filter center frequency of 4 GHz and a filter bandwidth of 4 GHz) are used.Then, the output waveforms are recorded by a real-time oscilloscope (Tektronix DSA72004B) with 8 bit analog-to-digital converter and 50 GS/s sampling rate.
Figure 4 shows chirped pulse compression and pulse polarity switching of the input downchirp waveform.With a constant control signal, as shown in Fig. 4(b), all compressed pulses at the filter output have the same positive peaks.In another experiment the control input to the phase modulator is driven by an amplitude V π square wave repeating at 12.9 MHz, which is half the repetition rate of the input RF chirps.As shown in Fig. 4(c), this causes the polarity of the compressed pulses at the output of RF photonic filter to alternate between positive and negative on successive pulses (we emphasize that all the input chirps have the same polarity for the data of Fig. 4).Note that all waveforms are obtained through single-shot measurements because the filter has good noise performance and thus does not require averaging of multiple traces.As expected, the compressed pulses are well-matched to the autocorrelation of the input waveform both with and without polarity control, Figs.4(e) and 4(f).The PAPR of the compressed pulses is approximately 21 dB and is improved by 16 dB compared to the input chirp waveform.The output pulse width defined as the full width at half maximum is ~250 ps, which is the inverse of the RF bandwidth.The compression ratio between input and output RF waveforms is comparable to the TBP of the filter response.Broadband additive noise is superimposed with the input sequence of chirps; the combined chirp sequence and noise are then connected to our RF photonic filter.Figure 5(a) shows the phase-coded down-chirp input sequence as well as the polarity control waveform which will be applied to phase modulator in the RF photonic filter.Here these waveforms are generated from channel 1 and 2 of the RF arbitrary waveform generator, respectively, each with a sampling rate of 12 GS/s.As before, each down-chirp waveform has an RF bandwidth of 4 GHz at the center frequency of 4 GHz and time aperture of 19.4 ns. Figure 5(d) shows an overlay of the individual input chirps with positive and negative polarities, respectively; the π phase shift is clearly evident.The ultrawideband jamming noise, generated by cascaded RF amplifiers and RF filters, is characterized by 10-dB bandwidth of 1.5 GHz at ~5.1 GHz center frequency.The generated chirp sequence and noise are combined by a 3-dB power combiner and directed to the RF photonic phase filter.As shown in Fig. 5(b), with a low input average signal-to-noise ratio (SNR) of 1.6 dB, the down-chirp waveforms are largely obscured at the filter input.At the output of the filter, compressed pulses are clearly observed, with amplitudes well above the noise, Fig. 5(c).Furthermore, because the polarity control input of the RF phase filter compensates the polarity reversals incorporated on the input chirp sequence, the compressed output sequence shown in Fig. 5(c) has only positive polarities.Therefore, at this point the processed signal already has the form of Eq. ( 3).The measured output peak SNRs are ~23 dB.The ratio of the output peak SNR to the input average SNR is 21.4 dB.This is in good agreement with the 21.9 dB SNR improvement, equal to the pulse compression gain (18.9 dB) plus 3 dB, expected for jamming noise limited operation [6].
At this point the individual spread spectrum waveforms (the chips) have been compressed via the RF photonic phase filter into a sequence of isolated peaks.In this first stage of processing, the compression comprises an analog chirp correlation operation implemented via the RF photonic phase filter.In the next stage of processing, we perform correlation at the frame level using a tapped-delay-line filter consisting of delay-lines, weighting, and a summer [1].The tapped delay line filter is implemented via a combination of analog hardware and digital processing.The weighting function is performed in hardware by pulse polarity switching of our RF photonic filter, as has already been described.The remaining operations (delays and summation) are accomplished through off-line digital processing of the signal from the BPD recorded on the real-time oscilloscope according to Eq. ( 4).The receiver is assumed to have accurate knowledge of the spacing T of the waveform chips making up the spread spectrum frame; here the spacing of the input chirp waveforms is 38.8 ns. Figure 5(e) shows a single waveform recorded after the RF photonic phase filter but before subsequent digital processing.After the digital processing, the peak voltage of the compressed pulse shown in Fig. 5(f) is increased by a factor of 10.2.The increase is somewhat lower than the ideal value of 15 which corresponds to the length of the PN sequence.The difference between measured and ideal values is attributed to timing errors.The standard deviation of the timing error in the peaks is ~18.6 ps, which is consistent with the timing jitter of the electronic RF-AWG used to generate the input spread spectrum signal.However, the root-mean-square noise voltage is only increased by a factor of ~3.9.This is very close to the factor of √15 expected upon averaging of uncorrelated noise signals.As a result, the peak SNR value is increased up to 31.4 dB.Instead of the offline tapped-delay line processing, electronic methods with analog circuits with high sampling rates on the order of tens of gigasample-persecond can be used [15,16].Here, the peak voltage values of compressed pulses would be sampled by a sample and hold circuit and then input to a summation circuit.

Discussion and conclusion
We introduce a novel pulse polarity switching technique to enhance the processing gain attainable with ultrawideband RF photonic phase filters.Our demonstration improves on a previously introduced RF photonic filtering configuration based on an optical frequency comb source, an interferometric pulse shaping arrangement, dispersive frequency-to-time conversion, and balanced detection by incorporating a phase modulator into one of the interferometer arms to provide polarity switching capability.Our experiments start with a phase filter with a 3-dB bandwidth of 4.2 GHz configured for compression of frequency modulated input signal with 206 MHz/ns chirp rate.This provides a time-bandwidth product of ~86, comparable to but slightly higher than our previous experiments.By overlaying an additional hybrid signal processing step consisting of polarity switching and off-line digital filtering, processing gain and noise suppression are enhanced for input spread spectrum sequences with matching polarity modulation.In an experiment with strong additive broadband noise, fully analog pulse compression via the RF photonic phase filter delivers 23 dB output SNR, 21.4 dB higher than the 1.6 dB input SNR.For a polarity-switched, length-15 input sequence, our new scheme provides peak output SNR of 31.4 dB, an additional 8.4 dB improvement.
Although the current experiments utilize two-level phase modulation (corresponding to switching of the polarities of spread spectrum chips), our scheme is general and is applicable to arbitrary multi-level phase modulation without any modification of the hardware.Furthermore, in our scheme the enhancement in the SNR is expected to scale linearly with the length of the phase modulation sequence.Although the specific example of a length-15 polarity switching sequence was demonstrated above, fundamentally there is no limit to the length of the phase modulation sequence.Practically, the length of the phase modulation sequence will be constrained by timing jitter within the waveform frame.To reduce such jitter, it may be desirable to utilize photonics-based generation of polarity (or multi-level phase) switched RF spread spectrum input sequences [10,11], as photonic approaches to high bandwidth RF arbitrary waveform generation have been shown to deliver lower RF phase noise than their electronic arbitrary waveform generator counterparts [17].
An important attribute of RF photonic phase filters for spread spectrum pulse compression is that they are asynchronous [9,18], avoiding serious challenges in acquiring synchronism with high bandwidth signals of interest.It is important to note that although our polarity switching scheme is no longer completely asynchronous, synchronism requirements are substantially relaxed compared to traditional time domain spreading and despreading at multi-GHz rates.The first receiver processing step, chip level pulse compression, remains fully asynchronous.This provides an initial processing gain which enhances the desired signal compared to the noise or RF background, facilitating timing acquisition.For the second step, the phase demodulator only needs to synchronize at the frame level (~ns) to initiate further processing gains.Although time alignment at the highest time resolution is eventually desired, the processing gain is expected to degrade gracefully with slight timing misalignments.

Fig. 1 .
Fig. 1.Hybrid spread spectrum scheme employing photonic spectral processing with polarity switching for large processing gain.(a) Chirped pulse compression for a single chip, (b) Polarity-switched chirped waveform sequence, (c) Chirped pulse compression without polarity switching, (d) Chirped pulse compression with polarity switching for despreading the phasecoded waveform.

Figure 2
Figure2shows the configuration of the RF photonic phase filter with pulse polarity switching.The broadband optical frequency comb is directed to the optical interferometer.One arm is sent to an erbium-doped fiber amplifier (EDFA) through an optical phase modulator driven by the control signal and a programmable pulse shaper, which are used to switch the pulse polarity and program the filter taps, respectively; The other passes through a Mach-Zehnder modulator (MZM) driven by the RF input signal of interest.The MZM is biased at the minimum transmission point, which suppresses both the optical carriers and optical noise[12].The outputs of the EDFA and MZM are combined by a 2 × 2 optical coupler.The two outputs of the 2 × 2 optical coupler are connected to the balanced photodetector (BPD) through a bidirectional fiber configuration including a single dispersive fiber and two optical circulators[12].This simultaneously matches the dispersion and the time delay of the paths between the 2 × 2 coupler and the BPD.

Fig. 3 .
Fig. 3. (a) Broadband and flat-topped comb (resolution = 0.01 nm), (b) RF filter amplitude response, and (c) RF group delay response.In (b) and (c), colored solid and black dashed curves show measurement and simulation, respectively.

Fig. 4 .
Fig. 4. Compression of constant polarity chirped inputs with pulse polarity switching in the filter.(a) Sequence of constant polarity input down-chirp waveforms; (b) Sequence of output waveforms with constant polarity control to the filter; (c) Sequence of color-coded output waveforms with alternating polarity control applied to the filter for successive waveforms; the black line shows the control waveform.(d), (e), and (f) are zoom-in views of (a), (b), and (c), respectively.In (e) and (f), black dash lines show the autocorrelation of the input waveform.Now we perform experiments on a sequence of chirped signals which are polarity switched upon generation according to a length-15 pseudo-noise (PN) sequence [14].Broadband additive noise is superimposed with the input sequence of chirps; the combined chirp sequence and noise are then connected to our RF photonic filter.Figure5(a)shows the phase-coded down-chirp input sequence as well as the polarity control waveform which will be applied to phase modulator in the RF photonic filter.Here these waveforms are generated from channel 1 and 2 of the RF arbitrary waveform generator, respectively, each with a sampling rate of 12 GS/s.As before, each down-chirp waveform has an RF bandwidth of 4 GHz at the center frequency of 4 GHz and time aperture of 19.4 ns.Figure5(d)shows an overlay of the individual input chirps with positive and negative polarities, respectively; the π phase shift is clearly evident.The ultrawideband jamming noise, generated by cascaded RF amplifiers and RF filters, is characterized by 10-dB bandwidth of 1.5 GHz at ~5.1 GHz center

Fig. 5 .
Fig. 5. Compression of PN polarity switched chirped inputs in the presence of ultrawideband jamming noise.(a) Color-coded input chirp sequence, without the jamming noise.The red and blue indicate flipped polarity of the down-chirp; black indicates the matching signal which will be applied to the polarity control of the RF photonic phase filter.(b) Input waveform with the jamming noise.(c) Output waveform from the RF photonic phase filter; compression of the individual chirps and polarity compensation are accomplished simultaneously.(d) Overlay of the input chirp waveforms with positive and negative polarities, without jamming noise.(e) Single compressed pulse at the filter output.(f) Compressed pulse with improved SNR after digital tapped-delay line.