Broadband RF front-end using microwave photonics filter

Jingjing Wang; Minghua Chen; Yunhua Liang; Hongwei Chen; Sigang Yang; Shizhong Xie

doi:10.1364/OE.23.000839

1. Introduction

Due to advantages in the aspects of operation bandwidth, transmission loss, and immunity to electromagnetic interference (EMI), microwave photonics (MWP) holds great potentials for enabling the front-end processing in RF receivers including band-pass filtering and down-conversion [1,2]. Typically, down-conversion, which aims at transferring the incoming radio frequency (RF) to the better handled intermediate frequency (IF) signals, is optically realized by employing heterodyne detection or electro-optic mixing [3–5]. Meanwhile, the band-pass filtering for suppressing the strong out-of-band signals usually incorporates optical filters, which simplified the structure, but also degrade the fineness and reconfigurability of the system due to their broad bandwidth and limited tunability [6,7]. An alternative to perform the filtering is to use the optical frequency comb (OFC) based tunable microwave photonics filter (MPF) for narrower bandwidth filtering. Concerning the large number of coherent spectral lines in OFC, which acts as local oscillator (LO), there are also great potentials of this filter to accomplish the down-conversion mixing simultaneously without any external LOs [8–12]. Based on this idea, a scheme of high-quality filtering associated with down-conversion was proposed recently, which takes advantages of a programmable line-by-line phase shaper for flexibility [13]. However, the input RF signal in this kind of scheme must be strictly restricted in a Nyquist zone, which is half of the comb line spacing, to avoid the beating interference between undesired signals and their nearest adjacent comb lines. Moreover, the tunability is limited by the available dispersion control range of the programmable shaper.

In this paper, we propose a broadband RF front-end with an OFC-based MPF. By using two dispersion elements and a variable delay line (VDL), which induces the flexible configuration of the transfer functions of all beat tones, we suppress the interferences from other Nyquist zones to break the strict limitation on the processing bandwidth for the first time as well as enhancing the frequency tunability greatly. In experiments, RF input with wide bandwidth of 4 Nyquist zones could be processed and the center frequency of the pass band signals was widely tuned from 5.5 GHz to 14.5 GHz. Moreover, both processing bandwidth and frequency tunability could be further improved by increasing the free spectral range (FSR) of the filter.

2. Principle

The architecture of our proposed scheme is shown in Fig. 1. RF input with wide bandwidth of tens gigahertz is received through an antenna and modulated onto the spectrum of a coherent OFC, which have been phase reshaped by a dispersion element D1 with group-velocity dispersion (GVD), θ_2, and separated into two branches through an optical splitter. With the assistance of an I/Q modulator, the OFC in top branch goes through a carrier-suppressed single-sideband (CS-SSB) modulation. Meanwhile, the comb spectral in the other branch is time delayed via a VDL. Then these two branches were combined and passes through another dispersion element D₂ with GVD φ₂ before the final detection with the cutoff frequency of 𝛿f/2, where 𝛿f is the frequency spacing between comb lines. Hence, the frequency of the nth comb line, f_n, could be expressed as f_n = f₀ + n∙δf. Besides, the initial phases of the incoming OFC lines are set to be identical.

Fig. 1 The architecture diagram of proposed OFC-based front-end of RF photonic receiver, VDL: variable delay line, PD: photo-detector, LPF: low-pass filter, LNA: low nosie amplifier, CS-SSB modulation: carrier-suppressed single-sideband modulation.

Download Full Size | PDF

We first consider a microwave signal, f_RF, to be CS-SSB modulated onto the coherent OFC, which have been reshaped by the dispersion element D1 with a quadratic phase of θ that modeled as of θ(f) = θ₂∙2π²∙(f-f₀)² [12]. The modulated RF sidebands then pass through the second dispersion element D₂ to introduce another quadratic phase of φ, which modeled as φ(f) = φ₂∙2π²∙(f-f₀)², and arrive at the photo-detector (PD) with the expression of

E_{1} (t) = α \cdot \sum_{n = 0}^{N - 1} e_{n} \cdot e^{- j \cdot [2 π \cdot (f_{n} + f_{R F}) \cdot t + θ (f_{n}) + φ (f_{n} + f_{R F})]}

where α represents the modulation efficiency, N is the number of comb lines and e_n is the complex electric field of the nth comb line’s. Meanwhile, with an additional VDL to introduce a time delay, τ [14], the OFC carriers in the other branch that arrive at the PD can be expressed as

E_{2} (t) = \sum_{m = 0}^{N - 1} e_{m} \cdot e^{- j \cdot [2 π \cdot f_{m} \cdot (t + τ) + θ (f_{m}) + φ (f_{m})]}

After coherent detection, great amount of mixing frequency components would be generated between E₁(t) and E₂(t), and the signal currents could be expressed as

\begin{matrix} i (t) \propto E_{1} (t) \cdot E_{2}^{*} (t) + E_{1}^{*} (t) \cdot E_{2}^{} (t) \\ = α \sum_{n = 0}^{N - 1} \sum_{m = 0}^{N - 1} e_{n} e_{m}^{*} \cdot e^{- j \cdot 2 π \cdot [f_{R F} - (m - n) \cdot δ f] \cdot t} e^{- j \cdot {[θ (f_{n}) + φ (f_{n} + f_{R F}) - θ (f_{m}) - φ (f_{m}) - 2 π \cdot f_{m} \cdot τ]}} + c . c \end{matrix}

where the terms, |E₁(t)|² and |E₂(t)|², are neglected as beat notes between OFC carriers are eliminated by the low-pass filter after PD, whose cutoff frequency is δf /2. In fact, most of frequency components in Eq. (3) would be suppressed after the low-pass filter, only the IF frequency component with the frequency, f_IF, satisfying f_IF = |f_RF-(m-n)∙δf |∈[0,δf /2] could be obtained in the output. After derivation, we can get

f_{I F} = f_{R F} - r \cdot δ f_{} o r_{} f_{I F} = r \cdot δ f - f_{R F}

where

r = m - n = (δ f / 2 + f_{RF}) / δ f

Note that the symbol ‘⌊ ⌋’ in Eq. (5) means the floor function. According to Eq. (3), the transfer functions of f_IF could be derived as

H (f_{I F}) \propto \sum_{n = 0}^{N - 1 - r} (e_{n} e_{n + r}^{*}) \cdot e^{- j {n \cdot 2 π \cdot [f_{I F} - r \cdot f_{D} - f_{τ}] \cdot T}}

or

H (f_{I F}) \propto \sum_{n = 0}^{N - 1 - r} (e_{n}^{*} e_{n + r}) \cdot e^{- j {n \cdot 2 π \cdot [f_{I F} + r \cdot f_{D} + f_{τ}] \cdot T}}

with

f_{D} = δ f \cdot (θ_{2} / φ_{2})

f_{τ} = τ / (2 π \cdot φ_{2})

T = 2 π \cdot δ f \cdot φ_{2}

which is very similar to the periodic filtering transfer function of MPF but with a frequency shift of (r∙f_D + fτ). Note that T represents the differential time delay between taps introduced by the second dispersion element, and the period of the transfer function, FSR, could be expressed as FSR = 1/T.

For a broadband input, it could be derived from Eqs. (4)-(6) that different input RF tones belonging to different Nyquist zones might exit at the same IF in output but experience different filtering responses. To filter out the RF signal of interest, f_0,RF, which would been down-converted to IF domain, and simultaneously suppress the interference signal, we should make a careful design on f_D and f_τ, where f_D is a constant determined by the two opposite dispersion in our system and variable fτ depends on the time delay between two branches [14].

For RF signal of interest with the frequency of f_0,RF, the corresponding IF beat tones within [0,δf /2] should be selected by its transfer function. According to Eq. (6), we get

(f_{0, R F} - r_{0} \cdot δ f) \mod F S R = (r_{0} \cdot f_{D} + f_{τ}) \mod F S R

with

r_{0} = (δ f / 2 + f_{0, R F}) / δ f

Accordingly, to filter out all frequency components within the frequency range [0, f_0,max ] (f_0,max is the maximum realizable center frequency) continuously by tuning the VDL in one direction, we set f_D to satisfy

f_{D} < - β \cdot δ f

where β is slightly larger than 1, that is, the absolute value of f_D is twice larger than the bandwidth of Nyquist zone. So we can simplify Eq. (10) into

f_{0, R F} - r_{0} \cdot δ f = r_{0} \cdot f_{D} + (f_{τ} \mod F S R)

In this way, by linearly tuning the VDL, the center frequency of system could be nearly continuously altered up to f_0,max, which could be derived to be approaching (FSR + r_0,max∙f_D+ r_0,max∙δf) with r_0,max = ⌊(δf/2 + f_0,max)/δf ⌋.

Besides flexibly filtering and down-converting the RF signal of interest, we intent to suppress the IF tones generated from interference signals to achieve a broadband front-end processing. From Eq. (6), this could be realized by suppressing the interference IF tones with their shifted filtering transfer functions. If the pass-band of the filtering transfer function is shifted to be out of the frequency range [0, δf /2], the input signal lying within the corresponding Nyquist zone would be suppressed. Note that for the low-frequency input lying in the 1st Nyquist zone, which refers to the situation of r = 0, the filtering transfer function is totally same as that of a tunable MPF [14]. Thus, to prevent these low frequency components from being reserved in IF band, the center frequency of the first pass band, (f_τ mod FSR), should satisfy

(f_{τ} \mod F S R) \notin [0, δ f / 2]

It could be found that the interference is inevitable if the VDL structure is not included, such as in conventional schemes [13]. In this way, once f_D and f_τ are set to satisfy Eqs. (12) and (14), respectively, any interference signals lying within [0, δf∙⌊(FSR + f_τ mod FSR)/ |f_D| ⌋ ] could be suppressed by comparing the effective shift of the periodic filtering transfer function to δf /2, and the maximum processing bandwidth satisfies

f_{B, \max} \geq δ f \cdot (2 F S R / | f_{D} |)

Thus, in theory, our system is free from the strict limitation on the processing bandwidth of only one Nyquist zone, and, the center frequency of pass-band could be flexibly tuned in a wide band using the VDL. Moreover, since the transfer function of the beat tones is very similar to that of MPF, some characteristics of MPF are also suitable for this system, e.g., the 3-dB bandwidth f_3dB is close to FSR/N and thus could be narrowed by increasing the number of OFC tones, and the main-lobe to side-lobe suppression ratio (MSSR), which could be used to approximate the suppression ratio of the interference, could be further improved by imposing shape factor on OFC spectral [8,9].

We make simulation on the proposed scheme with 25-GHz-bandwidth RF input, where the frequency spacing of comb lines is assumed to be 10 GHz and the RF signal of interest is centered at 12.7 GHz. According to Eqs. (12) and (14), we set f_D and the FSR of system to be −13 GHz and 19 GHz, respectively, and tune f_τ, to be 15.7 GHz. Then, the simulated power envelope of the output signals with 20-tone OFC could be derived from Eq. (6) and is shown in Fig. 2, where lines with different colors represent the power of beating signals for five 5-GHz-width Nyquist zones respectively. It could be observed that only narrow band RF signal centered at 12.7 GHz is down-converted into the IF band and filtered out simultaneously, meanwhile, the interference are all suppressed. As a result, a five-fold increasing for the processing bandwidth could be realized in this case, agreeing with the prediction in Eq. (15).

Fig. 2 The simulated power envelope of the output IF signals for the input RF signals ranging from 0 to 25 GHz.

Download Full Size | PDF

3. Experimental setup and results

Our experimental setup is illustrated in Fig. 3(a). The OFC with the spectrum depicted in Fig. 3(b) was generated by cascaded intensity modulator (IM) and phase modulator (PM) driven by RF signal at 10 GHz, and then phase compensated by programmable shaper to realize the same initial phase, which could generate pulse as shown in the inset of Fig. 3(b). The OFC first propagated through a 50 km SMF with dispersion of 850 ps/nm and then got split into two branches. In the top branch, nearly 30 dB suppression ratio was realized for CS-SSB modulation with an I/Q modulator. In the bottom branch, a VDL was utilized for flexible time tuning. The combined two branches were then led to another propagation with dispersion of −656.8 ps/nm, which was used for realizing a filter with 19-GHz FSR, and finally sent into one PD with the cutoff frequency of 5 GHz.

Fig. 3 (a) The experimental setup for proposed RF front-end, CW: continuous wave, PC: polarization controller, PPS: programmable phase shper, (b) The spectrum (blue) and waveform (left-top, red) of the generated OFC.

Download Full Size | PDF

Firstly, we observed the response of the simultaneous down-conversion and filtering in this scheme. By adjusting VDL to achieve τ = 82 ps, the center frequency was tuned to be at 12.7 GHz. Then the power of the IF output for RF input ranging from 0.5 GHz to 19.5 GHz were shown in Fig. 4(a), where the input signal around 12.7 GHz with the 3-dB bandwidth of 1.2 GHz was selected out and down-converted. Meanwhile the interference beating tones from other Nyquist zones within the range of almost 20 GHz was suppressed, which agreed well with analytical results. Here, the small differences between the experimental and theoretical results in the aspects of bandwidth and the response are caused by the I/Q modulator’s bandwidth of 20 GHz and the imperfect CS-SSB modulation, respectively. Moreover, the suppression ratio of interference achieved was more than 18 dB, which was approximated by the value of MSSR and could be improved further by imposing shape factor on OFC spectrum [9].

Fig. 4 (a) The experimental power envelope of the IF output for RF input ranging from 0.5 GHz to 19.5 GHz, (b) The comparison between suppressed and reserved response for low-frequency RF input at 0.6GHz that obtained with and without the VDL structure

Download Full Size | PDF

Then, the response for low-frequency RF input at 0.6 GHz was picked out to verify the benefits of low-frequency interference suppression from the usage of VDL. We firstly set τ to be the same with that in the measurement for Fig. 4(a). Since the value of f_τ in this occasion was adjusted to realize the filtering at 12.7 GHz, which satisfied Eq. (14), we obtain a suppressed response for the 0.6-GHz input, as depicted with the red line in Fig. 4(b). Then, we adjusted τ to satisfy τ = m∙2π∙FSR∙φ₂ to represent the situation in conventional schemes without VDL, where m is an integer number and the effective time delay between two arms is zero. This obtained response was depicted with the blue line in Fig. 4(b). By comparing these two responses, it could be observed that the achieved suppression ratio exceeded 21 dB.

In addition, the frequency tunability of the experimental system was also tested. By setting τ to be 73 ps, 82 ps and 88 ps, respectively, we obtained three power envelops in IF band, which were accordingly centered at 10.9 GHz, 12.7 GHz and 13.8 GHz, as depicted in Fig. 5(a). For simplicity, we have neglected the interference signal in adjacent Nyquist zones that would be suppressed. Then we tuned the VDL from 44 ps to 92 ps and demonstrated the center frequency tunability ranging from 5.5 to 14.5 GHz. The measured power of selected IF output for RF signals with different frequencies were obtained and shown in Fig. 5(b). As shown in Eq. (13), the tunability can be improved further by increasing the FSR.

Fig. 5 (a) The power envelops in IF band when τ is set to be 73 ps, 82 ps and 88 ps, where the to-be–suppressed interference signal from adjacent Nyquist zones have been neglected. (b) The measured power of output IF signals for the input RF signals from 5.5 GHz to 14.5 GHz with time dealy tuned from 44 ps to 92 ps

Download Full Size | PDF

4. Conclusion

In this paper, a novel RF front-end using OFC-based MPF is proposed. With the assistance of two opposite dispersion elements and one VDL, we break the processing bandwidth limitation and enhance the frequency tunability simultaneously. In experiments, RF signal centered at 12.7 GHz was selected and down-converted to IF band with wide-bandwidth input up to 19.5 GHz, which is approaching the scale of 4 Nyquist zones. Besides, the center frequency of system was successfully tuned from 5.5 GHz to 14.5 GHz.

Acknowledgments

This work is supported by National Program on Key Basic Research Project (973) under Contract 2012CB315703, and NSFC under Contract 61120106001, 61132004 and 61322113, and the Program for New Century Excellent Talents in University (NCET-10-0520), and Tsinghua University Initiative Scientific Research Program.

References and links

1. J. P. Yao, “Microwave photonics,” J. Lightwave Technol. 27(3), 314–335 (2009). [CrossRef]

2. J. Capmany and D. Novak, “Microwave photonics combines two worlds,” Nat. Photonics 1(6), 319–330 (2007). [CrossRef]

3. M. Hossein-Zadeh and K. J. Vahala, “Photonic RF down-converter based on optomechanical oscillation,” IEEE Photon. Technol. Lett. 20(4), 234–236 (2008). [CrossRef]

4. H. Yu, M. Chen, H. Gao, S. Yang, H. Chen, and S. Xie, “RF photonic front-end integrating with local oscillator loop,” Opt. Express 22(4), 3918–3923 (2014). [CrossRef] [PubMed]

5. V. R. Pagán, B. M. Haas, and T. E. Murphy, “Linearized electrooptic microwave downconversion using phase modulation and optical filtering,” Opt. Express 19(2), 883–895 (2011). [CrossRef] [PubMed]

6. A. Agarwal, T. Banwell, and T. K. Woodward, “Optically filtered microwave photonic links for RF signal processing applications,” IEEE J. Lightwave Technol. 29(16), 2394–2401 (2011). [CrossRef]

7. H. Yu, M. Chen, P. Li, S. Yang, H. Chen, and S. Xie, “Silicon-on-insulator narrow-passband filter based on cascaded MZIs incorporating enhanced FSR for downconverting analog photonic links,” Opt. Express 21(6), 6749–6755 (2013). [CrossRef] [PubMed]

8. J. Capmany, B. Ortega, and D. Pastor, “A tutorial on microwave photonic filters,” J. Lightwave Technol. 24(1), 201–229 (2006). [CrossRef]

9. V. R. Supradeepa, C. M. Long, R. Wu, F. Ferdous, E. Hamidi, D. E. Leaird, and A. M. Weiner, “Comb-based radiofrequency photonic filters with rapid tunability and high selectivity,” Nat. Photonics 6(3), 186–194 (2012). [CrossRef]

10. M. Song, V. Torres-Company, and A. M. Weiner, “Noise comparison of RF photonic filters based on coherent and incoherent multiwavelength sources,” IEEE Photon. Technol. Lett. 24(14), 1236–1238 (2012). [CrossRef]

11. J. Wang, M. Chen, H. Chen, S. Yang, and S. Xie, “Large-tap microwave photonics filter based on recirculating frequency shifting loop,” IEEE Photon. Technol. Lett. 26(12), 1219–1222 (2014). [CrossRef]

12. J. Liao, X. Xue, H. Wen, S. Li, X. Zheng, H. Zhang, and B. Zhou, “A spurious frequencies suppression method for optical frequency comb based microwave photonic filter,” Laser Photon. Rev. 7(4), L34–L38 (2013). [CrossRef]

13. V. Torres-Company, D. E. Leaird, and A. M. Weiner, “Simultaneous broadband microwave downconversion and programmable complex filtering by optical frequency comb shaping,” Opt. Lett. 37(19), 3993–3995 (2012). [CrossRef] [PubMed]

14. 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). [CrossRef]

Broadband RF front-end using microwave photonics filter

Abstract

1. Introduction

2. Principle

3. Experimental setup and results

4. Conclusion

Acknowledgments

References and links

Cited By

Figures (5)

Equations (16)

Optics Express