A novel remoted instantaneous frequency measurement system using all optical mixing is demonstrated. This system copies an input intensity modulated optical carrier using four wave mixing, delays this copy and then mixes it with the original signal, to produce an output idler tone. The intensity of this output can be used to determine the RF frequency of the input signal. This system is inherently broadband and can be easily scaled beyond 40 GHz while maintaining a DC output which greatly simplifies receiving electronics. The remoted configuration isolates the sensitive and expensive receiver hardware from the signal sources and importantly allows the system to be added to existing microwave photonic implementations without modification of the transmission module.
© 2013 OSA
MODERN electronic warfare self protection systems utilize instantaneous frequency measurement (IFM) to ensure a high probability of intercept when detecting potential threats . Electronic IFMs have been demonstrated operating from 2 to 20 GHz, but can be limited by bandwidth of delay lines and mixers. Microwave photonic (MWP) systems can provide exceptionally broadband delays, summation and mixing of RF signals in the optical domain. IFM systems have been reported using MWP techniques; however, most require broadband photodectors [2–4], which can be expensive and failure prone. This becomes particularly important if multiple IFMs are employed to cover several frequency bands.
We have recently demonstrated photonic IFMs using four wave mixing (FWM) in highly nonlinear fiber (HNLF)  where the amplitude comparison function (ACF) is obtained by measuring the DC power of the idler. It is shown that this idler contains a coherent superposition of differentially delayed modulated optical signals . This system requires only DC detection, significantly reducing implementation cost, complexity and likelihood of failure. However, this advantage is offset by the requirement for two optical carriers at the transmitter. In practical applications it may not be possible to modify the transmitter and thus techniques to remote the receiver from the transmitter should be considered.
In this paper we propose and demonstrate a photonic IFM system which extends the system of  through use of an additional HNLF at the receiver to copy the input signal onto a second optical wavelength. Thus all of the components of the IFM system can be remoted to the receiver, with no specific requirements placed on the transmitter. This modular isolation of the IFM system from the transmitter greatly improves practical potential. It should be noted that a preliminary proof of this concept was presented by us in ; however, the current paper provides an improved implementation, detailed demonstration and rigorous analysis.
The organization of this paper is as follows. Section 2 reviews our previous all optical IFM system. Section 3 then introduces our new remoted all optical IFM system proposal and derives the equations predicting the system output as a function of input frequency. Section 4 describes the remoted system implementation and characterizes its components. Section 5 presents the system demonstration including system response, frequency interpretation and quantification of frequency measurement error. Finally, Section 6 presents discussion and conclusions on system performance, limitations and possible future enhancements.
2. Original all optical IFM
Figure 1 illustrates our previously demonstrated all optical IFM system . At the transmitter, two optical carriers are modulated with the RF signal to be measured. These are then differentially delayed and mixed producing new optical wavelengths (idlers) via FWM . The idlers have characteristics of both original signals. The carrier at the idler wavelengths is the product of the two original carriers, but the signal carried by these idlers is a coherent summation of the two input carriers. Since the sidebands of the input carriers are differentially delayed, a relative phase shift which is proportional to the RF modulating frequency is incurred. Hence, the sidebands at each idler may coherently sum or cancel causing the total optical power at the idler to oscillate with RF frequency enabling frequency measurement.
The system of Fig. 1 has the advantages of ultra-broad bandwidth due to the use of all optical mixing and requires no high speed electronic components since only the total optical power must be measured at the output with no requirement to analyze frequency. Further, despite the output being a coherent superposition of two independently delayed optical carriers; it is also surprisingly stable due to the fact that the idler carrier is the product of the original optical carriers and hence is immune to relative phase variations on these carriers .
A disadvantage of this system is that a non-standard transmitter is required. This limitation is shared by many other proposed photonic IFM system [2–4]. It may be desirable to measure the RF frequency of a photonic signal without requiring alteration or even access to the transmitter. Such a remote IFM is described in the following section.
3. Remoted all optical IFM principle
The proposed ‘remoted’ IFM system of this paper is presented in Fig. 2 . Here a single optical carrier, at frequency ω3, is intensity modulated by the RF signal, at frequency Ω, that we wish to measure. This transmitter is typical of any microwave photonic link and although it has been illustrated as modulated using a Mach-Zehnder modulator, no assumption needs to be made about the specific means of intensity modulation and thus direct modulation of the laser should be equally applicable. The modulated optical field be simply represented as:
After modulation, the optical carrier is transmitted through an arbitrarily long optical fiber link and arrives at the receiver which is independent of the transmitter unit. On entering the receiver, the signal is combined with a strong optical pump at frequency ω4, and they are mixed via FWM in a HNLF to produce copies of the signal at idler frequencies ω2 = 2ω3-ω4 and ω5 = 2ω4-ω3. The idler field at ω5 can be written:
The original signal at ω3 and copied signal at ω5 are isolated by filtering and differentially delayed by time Δt. Assuming this delay is not dispersive, these fields can be described as:.
In principle, the system of Fig. 2 could be used with a standard microwave photonic transmitter. This would be a great advantage over previous implementations. However, for this system to be practical, the cascaded all-optical mixing must provide sufficient gain such that the oscillating output power can be measured above the system noise floor.
4. System characterization
Having conceived an implementation of our novel remoted IFM system, it is now possible to attempt a proof of concept demonstration. However, before attempting such a demonstration, it will be valuable to characterize the system elements systematically in stages to ensure that each stage behaves as required. The component specifications are in the caption of Fig. 2.
4.1 Frequency response of the photonic transmitter
The first characterization step is to establish the frequency response of the transmitted signal modulation index, m3 in Eq. (6). Our previous all optical IFM demonstration , established that the output power versus frequency of the IFM system depends on both the amplitude of the RF signal and the combined frequency dependent responses of the modulator and RF cabling at the system input. Since this response does not include the frequency response of a high speed photodetector and the output RF cabling, it cannot be obtained directly by measuring the photonic link gain. We have reported a technique to determine the frequency dependent modulation index of a transmitter by measuring the link’s DC output voltage at null biasing . Using this technique, for an input RF signal power level of 10 dBm, we determined with m30 = 0.129 and α = 0.136 (GHz-1/2).
4.2 Optical mixing in HNLF1
Having obtained the frequency response of the transmitter modulation index, we next characterize the efficiency and frequency response of FWM in HNLF1 in Fig. 2 to determine m5 in Eq. (6). The system was configured as shown in Fig. 2 and detailed in Section 5. The RF signal at the transmitter was set to 10 dBm power level and frequency Ω = 40 GHz. The signal carrier (ω3) and pump (ω4) were set to λ3 = 1545.3 nm and λ4 = 1546.9 nm with equal power adjusted to provide ~10 dBm at the input to HNLF1. These are named Ch3 and Ch4 to avoid confusion between the frequency and wavelength domains. To characterize the optical mixing in HNLF1, we measured the optical spectrum of signals at input and output of HNLF1 using and optical spectrum analyzer (OSA).
Figure 3(a) shows the measured input spectrum (illustrated in Fig. 2(iii). This spectrum consists of the modulated carrier at Ch3 and the un-modulated pump at Ch4. The RF sidebands of Ch3 are ~28 dB down from the carrier. Figure 3(b) shows the spectrum after passing through HNLF1 (illustrated in Fig. 2(iv). Idlers carrying side-bands are clearly evident. Sidebands are also now evident on the pump at Ch4, which can be attributed to XPM.
Figures 3(c) and 3(d) present the spectra at Ch3 and Ch5 independently isolated using the reconfigurable filter. It is evident that the signals at Ch3 and Ch5 have thus been successfully isolated with unwanted spectral components suppressed by at least 40 dB. The ASE noise has also been reduced by ~30 dB. We expect Ch5 to be a a copy of the signal at Ch3. To verify this expectation, we demodulate Ch3 and Ch5 using high speed photodiodes and then analyze using electrical spectrum analyzer (ESA), as illustrated in inset e) of Fig. 2.
Figure 4 presents the RF power measured at Ch3 and Ch5 for various cases of channel illumination. The green and blue curves present the RF signal power received on Ch3 and Ch5 respectively when both signal (Ch3) and pump (Ch4) are present at the input. The RF powers reduce with increasing frequency and the two RF power measurements exhibit essentially the same frequency response scaled by appropriate mixing gain. This shows that the RF signal at Ch5 was in fact a faithful copy of the output on Ch3.
To enable comparison to a simple microwave photonic link without FWM, the pump (Ch4) was switched off and signal power (Ch3) was increased by 3 dB to maintain 10d Bm input power to HNLF1. The output RF power measured on Ch3 is presented as the red curve in Fig. 4. This is of identical form to the cases when FWM does occur and hence we can be confident that HNLF1 does produce two copies of the input signal on Ch3.
The predicted photonic transmitter frequency response of Section 4.1 is also shown as a solid line in Fig. 4. A fixed gain offset is added to match the prediction to the measurement at Ch3 for the case of illumination of only Ch3. Agreement between the predicted and measured responses is excellent. This confirms that the photonic components and nonlinear mixing in HNLF1 do not alter the frequency response of the photonic transmitter and that the simple model of Section 4.1 can be used up to 40 GHz.
Referring back to Fig. 3(b) and the inset of Fig. 3(c), it can be noted that the carrier to sideband ratios (CSR) of the two idlers at Ch2 and Ch5 are −19 and −22 dB respectively. The CSR at Ch2 is 3 dB higher than that at Ch5. This is expected from optical mixing within HNLF1 . However, the CSR at Ch5 and Ch3 are −22 and −28 dB respectively which signifies that the modulation indices at these two channels are not the same. A 6 dB sideband enhancement of Ch5 relatively to Ch3 has been achieved during wavelength conversion process in HNLF1. This translates to a 3 dB enhancement of modulation index. From Section 4.1, the DC modulation index of Ch3 was m30 = 0.129. Thus, the modulation index at Ch5 should be twice this or m50 = 0.258 which agrees well with the measured RF power of Fig. 4.
4.3 Optical mixing in HNLF2
The final characterization step was to establish that the idlers generated by mixing Ch3 and Ch5 in HNLF2 were measurable at the IFM output. To do this, the system was configured as depicted in Fig. 2 and Filter1 was programmed to provide 100 GHz passbands at both Ch3 and Ch5. Figure 3(e) presents the optical spectrum output from Filter1 and also after reflection from the CFBG to impart a differential delay. Only Ch3 and Ch5 are present in the spectrum with other wavelengths being suppressed. This spectrum was input to HNLF2 and Fig. 3(f) presents the spectrum measured at the output of HNLF2. Idlers have evidently been generated at 1542.1 nm (Ch1) and 1551.7 nm (Ch7). The idler at Ch1 is more intense than Ch7 due to the power imbalance between Ch3 and Ch5 at the input to HNLF2.
An arrayed waveguide grating (AWG) was employed to isolate the various channels at the output of HNLF2. Figure 3(f) shows the spectrum at the output of the AWG which is nominally only the idler at Ch1, as illustrated in Fig. 2(viii). The AWG has suppressed the optical powers in Ch3 and Ch5 by approximately 40 dB, however these channels are still present at the AWG output. This will add a DC offset to the optical power measured at the IFM output and may increase the noise. However, the RF sidebands are clearly visible on Ch1 indicating that their power should be measureable.
This characterization shows that the optical mixing and filtering processes have behaved as expected and it is thus possible to attempt to demonstrate the complete IFM system.
5. Demonstration of remoted nonlinear mixing IFM
Having characterized the elements of the system, it is now possible to demonstrate the remoted frequency measurement operation. Using the setup of Fig. 2, LD1 was set to 1545.3 nm (Ch3). A 10 dBm RF signal was modulated onto Ch3 using a Mach Zehnder modulator and then transmitted to the remoted receiver. At the receiver, the input signal at Ch3 was combined with a local optical pump provided by LD2 at 1546.9 nm (Ch4) using a 3 dB coupler. The combined optical signal was then amplified to 20 mW using EDFA1 to ensure efficient mixing in the HNLF1. The output of HNLF1 consisted of idlers at 1543.7 nm (Ch2) and 1548.5 nm (Ch5). The idler at Ch5 is the phase conjugate copy of the signal at Ch3. Thus, Ch3 and Ch5 represented two copies of the RF signal required for IFM operation . The signals at Ch3 and Ch5 were extracted by programming Filter1 to provide two 100 GHz passbands centered at 1545.3 nm and 1548.5 nm. The filter output was amplified to 80 mW using EDFA2. Ch3 and Ch5 were then differentially delayed by time Δt = 80 ps using the CFBG and then mixed in HLNF2 to create new idler products appearing at 1542.1 nm (Ch1) and 1551.7 nm (Ch7). Ch1 was then isolated using Filter2 (actually an AWG) and its power was measured using a low speed photodetector and a DC voltmeter.
To demonstrate RF frequency measurement, the RF input frequency was scanned from 1-40 GHz in steps of 1 GHz and the DC output voltage was measured as presented in Fig. 5(a) . The output oscillates with a period of 12.5 GHz which is expected from the time delay of Δt ~80 ps. The oscillation amplitude decreases with increasing of frequency due to the frequency response of the photonic link [5,8]. The measured voltage versus frequency oscillates in a predictable manner and thus could indeed be used for frequency measurement.
To interpret this oscillating response as a measurement of frequency, a quantitative prediction of the output voltage as function of input RF frequency is required. The output voltage can be predicted using Eq. (6). This equation requires several system parameters which can be obtained as described in Section 4. To calculate the output DC voltage, it was also necessary to scale Eq. (6) to account for the system frequency invariant response . This scaling factor was obtained by measuring the system output when the RF signal was turned off. This factor was measured to be ~5 V which agrees well with value calculated from the measured optical intensity of Fig. 3(g). The predicted voltage is plotted using solid line in Fig. 5(a). Excellent agreement between the prediction and measurement is evident.
Frequency interpretation of the measured signals was conducted by solving Eq. (6) using the data of Fig. 5(a). Due to the oscillatory nature of Eq. (6), unambiguous interpretation of frequency can only be made within a half period of oscillation – between a peak and a valley of the voltage curve. The voltage curve of Fig. 5(a) consists of 7 bands where unambiguous frequency measurement is possible. Solving Eq. (6) within each of these bands, we obtain the frequency measurement shown in Fig. 5(b). It is evident that the interpreted frequencies are very close to the actual signal frequency indicated by a solid straight line with the slope of unity. The frequency measurement accuracy degrades at the band edges due to reduction of the voltage-frequency slopes. Frequency measurement errors are calculated and presents as insets of Fig. 5(b) for two representative bands Band 1: DC-6 GHz; and Band 6: 30-37 GHz. For both bands, the errors are less than ± 250 MHz which suggests that the system is stable and accurate. These results complete the remoted IFM demonstration.
7. Discussion and conclusions
In conclusion, we have demonstrated a novel concept of photonic instantaneous frequency measurement (IFM) in which we extended the previous demonstrated nonlinear mixing IFM system to enable operation with a conventional, single wavelength intensity modulated photonic transmitter. This system allows the photonic transmitter to be remoted from the IFM receiver protecting the sensitive and expensive receiver components from the harsh environment of the receiver front end. Importantly, this all-optical IFM receiver design could be added to installed microwave photonic links as an upgrade.
The proof of concept was implemented using fiber platform and demonstrated up to 40 GHz with excellent agreement between the predictions and measurements. The interpreted frequency error was within +/− 250 MHz throughout the entire 40 GHz band, demonstrating excellent system stability and noise performance. While electronic IFM solutions are highly sophisticated, it is difficult to scale them to high frequencies. The demonstrated all-optical system is inherently broadband and could be easily scaled to 40 GHz and beyond if required.
The key advancement of this system over the previous demonstration is the copier stage. The demonstrated system used FWM in HNLF for this copying function, however techniques employing SOAs  or gain saturation of DBR lasers  could also be considered.
In order to achieve the demonstrated performance, it was necessary to use high pump powers to achieve sufficient mixing gain in the cascaded HNLFs. The power and spectral distribution must be carefully balanced to minimize the effect of Stimulated Brillouin Scattering (SBS) which can cause system instability . It is also important that the filtering stages effectively isolate the desired channels from the pump and other mixing products. This is particularly true in the copying stage where the copied idler is far weaker than the original signal. It is anticipated that using other wavelength conversion techniques where the converted wavelength is naturally separated from the pump would alleviate this requirement.
It is conceivable that the system could be extended to utilize multiple pumps within the same system to achieve simultaneous parallel IFM functions  and thus allow realization of complex and complete IFM systems. We have shown that such an extension can be achieved with minimal increase in component count and complexity; however, filtering can become complex . This challenge could be alleviated using our recently reported scheme where each channel is labeled with low frequency tones .
A possible multi-channel system with cascaded HNLFs, cascaded gratings, high performance filtering and wavelength routing may appear to be prohibitively complex and hence impractical bulky and expensive. Recent progress in integrated microwave photonics has demonstrated that sophisticated MWP circuits and several compact nonlinear media such as SOAs could practically be integrated into a photonic chip . The design and realization of such an all-optical signal processing chip is currently under investigation.
References and links
1. J. Tsui, Digital techniques for wideband receivers (SciTech Publishing, 2004).
2. L. Nguyen and D. Hunter, “A photonic technique for microwave frequency measurement,” IEEE Photon. Technol. Lett. 18(10), 1188–1190 (2006). [CrossRef]
3. T. Mengual, B. Vidal, and J. Marti, “Photonic RF frequency measurement combining SSB-SC modulation and birefringence,” Opt. Commun. 283(13), 2676–2680 (2010). [CrossRef]
4. S. Pan and J. Yao, “Instantaneous microwave frequency measurement using a photonic microwave filter pair,” IEEE Photon. Technol. Lett. 22(19), 1437–1439 (2010). [CrossRef]
5. L. A. Bui, M. D. Pelusi, T. D. Vo, N. Sarkhosh, H. Emami, B. J. Eggleton, and A. Mitchell, “Instantaneous frequency measurement system using optical mixing in highly nonlinear fiber,” Opt. Express 17(25), 22983–22991 (2009). [CrossRef] [PubMed]
6. L. Bui and A. Mitchell, “Remoted instantaneous frequency measurement system using optical mixing in highly nonlinear fiber”, Australian Conference on Optical Fibre Technology, 5–9 Dec. (2010).
7. G. P. Agrawal, Nonlinear Fiber Optics (Academic, 2007).
8. L. Bui, N. Sarkhosh, and A. Mitchell, “Photonic instantaneous frequency measurement: parallel simultaneous implementations in as single highly nonlinear fiber,” IEEE Photon. J. 3(5), 915–925 (2011). [CrossRef]
9. T. Durhuus, B. Mikkelsen, C. Joergensen, S. Danielsen, and K. Stubkjaer, “All-optical wavelength conversion by semiconductor optical amplifiers,” J. Lightwave Technol. 14(6), 942–954 (1996). [CrossRef]
10. T. Durhuus, R. Pedersen, B. Mikkelsen, K. Stubkjaer, M. Oberg, and S. Nilsson, “Optical wavelength conversion over 18 nm at 2.5 Gb/s by DBR-laser,” IEEE Photon. Technol. Lett. 5(1), 86–88 (1993). [CrossRef]
11. L. Bui and A. Mitchell, “Parallel All-Optical Instantaneous frequency measurement system using channel labeling,” IEEE Photon. Technol. Lett. 24(13), 1118–1120 (2012). [CrossRef]
12. D. Marpaung, C. Roeloffzen, R. Heideman, A. Leinse, S. Sales, and J. Capmany, “ Integrated microwave photonics,” Laser and Photonics Reviews, 1–33 (2013).