We propose a novel scheme to generate ultra-wideband (UWB) pulse by employing a Sagnac interferometer comprising a phase modulator. This structure performs a dual-input and dual-output intensity modulator (IM), ultimately resulting in the flexibility to select the shape and the polarity of the generated UWB pulse. The experiment results show a good agreement with the theoretical investigation in terms of both pulse profile and spectrum, which conforms to the definition of UWB signals by the U.S. Federal Communications Commission. Furthermore, the proposed scheme is independent of the voltage bias point.
©2007 Optical Society of America
Ultra-wideband (UWB) impulse radio technology plays an important role in the realization of future pervasive and heterogeneous networking due to its major advantages including low cost, immunity to multi-path fading, high data-capacity, and low power-consumption [1, 2]. A regulation established by the U.S. Federal Communications Commission (FCC) and accepted by the industry defines UWB signals to occupy a 10-dB bandwidth of at least 500 MHz or a fractional bandwidth greater than 20% with a power spectral density limit of-41.3 dBm/MHz . The main limitation on a UWB wireless link is the limited propagation distance (typically <10 m) over which the expected high data-rate can be realized. Therefore, UWB-over-fiber systems have emerged to exploit the advantages provided by the optical fiber, where there is a strong demand to generate, modulate and distribute UWB signals directly in the optical domain. On the other hand, it has been reported that Gaussian monocycle and doublet shape are good alternatives in UWB impulse radio systems . Considering the above two considerations, several schemes have been proposed to optically generate monocycle or doublet pulses. Gain saturation effect in an SOA is utilized to generate monocycle pulses . UWB monocycle pulses are obtained by the cascade of an optical phase modulator (PM) and a section of polarization-maintaining fiber (PMF) . Wang et al. proposed an approach to UWB doublet generation using a Mach-Zehnder modulator (MZM) biased at its nonlinear region . A fiber Bragg grating (FBG) is employed as a frequency discriminator to perform phase-to-intensity conversion, finally achieving both monocycle and doublet pulse generation [7, 8]. Wang et al. proposed a method based on cross-gain modulation (XGM) effect in an SOA to generate monocycle UWB pulses . An all-fiber method for UWB pulse generation has been reported based on optical spectral shaping and frequency-to-time conversion .
In this paper, we propose and experimentally demonstrate a flexible scheme using a bias drift-free intensity modulator (IM) to generate UWB pulses with both monocycle and doublet shapes. The required IM has been reported based on a Sagnac interferometer comprising a traveling-wave PM and a nonreciprocal quadrature bias unit . This structure is inherently insensitive to the operating temperature and the direct-current (DC) bias drift. Moreover, the ambient effects can be neglected if the loop is short enough . In virtue of the dual-input and dual-output characteristic of the designed IM, it is simple and convenient to manage the shape and polarity of the generated UWB pulse.
As shown in Fig. 1, the structure of the IM is based on a Sagnac interferometer which is composed of a 3-dB coupler, a traveling-wave PM, a variable fiber delay line (VFDL) and a nonreciprocal quadrature bias unit. By adjusting the VFDL, the PM is optimized at the center of the loop. As a commercial PM is typically manufactured for mono-directional operation, the clockwise (CW) and counterclockwise (CCW) lights in the loop are modulated by the applied electrical signal V(t) with a different efficiency above a frequency related to the transit time τ when they pass through the PM. This effect will lead to a phase difference Δφ(t) with respect to V(t) between the CW and CCW lights, which is given by 
where V π is the forward half-wave voltage of the PM, and ω is the angular frequency of V(t). If the phase difference can be up to π when V(t) varies, the entire setup can act as an IM. Assuming the power splitting ratio of the coupler is ideally 50:50, the transfer function of the proposed IM are given by
where ± represents the transfer functions of transmission (-) and reflection (+) respectively. Φ is the phase shift induced by the nonreciprocal quadrature bias unit which is set to π/2 to guarantee an operation at quadrature. Due to lack of the bias unit composed of a quarter-wave plate between two oppositely-oriented 45° Faraday rotators, two polarization controllers (PC) are inserted inside and outside the loop respectively to achieve the same function . As depicted at the right part of Fig. 1, the transfer functions for transmission and reflection given by Eq. (2) both have a profile which is similar to that of a typical Mach-Zehnder modulator (MZM) and are π out of phase. The interferometer is insensitive to environmental conditions (such as vibration or temperature change) due to the common path of CW and CCW lights in the loop. However, it is sensitive to time-dependent variations of ambient conditions, if the variations occur over the transit time of the light through the loop. The effects do not become a problem as long as the length of the loop is several meters .
The operation principle of the proposed scheme for UWB monocycle and doublet pulse generation is illustrated in Fig. 2. As described in Fig. 1, the designed IM possesses two outputs (named Output1 and Output2 in Fig. 2) with opposite slope of the transfer function when working at quadrature point. When a continuous-wave light is injected into the Sagnac loop and meanwhile an electrical Gaussian pulse is applied to the PM, a positive optical pulse with Gaussian profile will be generated at the transmitted port and at the same time a negative one will appear at the reflected port. If the designed IM is used as a dual-input device with the help of two circulators (i.e. multiple optical carriers at different wavelength are selectively injected into the loop from both ports), a group of optical pulses at all these wavelengths will be simultaneously generated from either Output1 or Output2, as shown in Fig. 2. Note that the two outputs feature a combination of optical pulses with reversed polarity for each wavelength. Consequently, the polarity can be alternated simply by choosing the output port. A dispersive element (e.g. a fiber coil) is used to introduce a suitable differential time delay for each output. UWB pulses with different shapes and polarities can be finally generated after O/E conversion. Three wavelengths labeled λ1, λ2, λ3 in Fig. 2 are sufficient for the UWB pulse generation with arbitrary shape and polarity. In our scheme, one can choose the shape just by whether removing the optical carrier at λ3 from the input port, and vary the polarity by alternating the two output ports. On the other hand, the setup described in Fig. 2 can perform a two-or three-tap microwave photonic filter with positive and negative coefficients. Therefore, the proposed scheme can be explained in another way that UWB pulses are generated by constructing a microwave photonic filter with positive and negative coefficients. This method has been first proposed and experimentally demonstrated by Yao et al [14, 15].
Assuming that the optical pulse can be well approximated by a Gaussian function and the square-law photodetection ideally brings no change to the pulse profile, the detected electrical signal can be viewed as the integration of a set of Gaussian pulses with different polarities.
where Ai, (i=1, 2, 3) are positive values representing the amplitude of the Gaussian pulses which compose the ultimate UWB pulse, Δτ which is given by the expression D·L·Δλ is the differential time delay between adjacent Gaussian pulses by launching these pulses with different wavelengths into a dispersive fiber, D is the fiber dispersion coefficient, L is the fiber length, Δλ is the wavelength spacing of these Gaussian pulses, T FWHM is the full width at half maximum (FWHM) of each Gaussian pulse after fiber propagation, and the sign “±” characterizes the polarity of these Gaussian pulses. These Gaussian pulses will be broadened after fiber propagation due to the fiber dispersion. The profile of each pulse still maintains a Gaussian shape. Although the pulse broadening in the temporal domain will lead to the narrowing of the efficient bandwidth in the frequency domain, this effect is not significant enough to cause the violation of the FCC definition on UWB signals, indicating the feasibility of the proposed scheme. The formula for monocycle shape can be obtained just by eliminating the third term in Eq. (3). The power spectral density (PSD) of s(t) can be obtained by Fourier transform. In theory, two conditions require adjustment in order to optimize the ultimate UWB pulse. One is the mutual relation of Ai, (i=1, 2, 3). The other is Δτ which dominantly determines the profile and spectrum of the ultimate UWB pulse. A Δτ greater than the duration of a Gaussian pulse will result in a complete separation of these Gaussian pulses, which is not favorable. A smaller Δτ value will mean some superposition between adjacent Gaussian pulses in the time domain, thus some power counteraction will happen which will lead to lower power efficiency. A smaller Δτ will also lead to a broader spectrum which may exceed the defined spectrum range by FCC. Hence there is a tradeoff between avoiding more power degradation and achieving a moderate pulse shape. Setting T FWHM=83.3 ps, a series of simulations were carried out to provide the optimal case which is shown in Fig. 3. For the optimal Δτ which equals to 80 ps, all the waveforms show a good profile and all the spectra have a fractional bandwidth of far more than the defined 20% by the FCC. The peak value of the normalized pulse amplitude is moderately less than 1 (higher than 0.8 in Fig. 3), indicating a little power degradation. The FWHM of the monocycle and doublet pulse is 69 ps and 53 ps respectively, and the 10-dB bandwidth is 7.33 GHz and 5.4 GHz.
The experiment setup is described in Fig. 4. As explained in [12, 13], it is essential to place one PC inside the loop and the other PCs respectively following three tunable laser diodes (LD) which emitted three continuous-wave lights at λ1=1558.1 nm, λ2=559.0 nm, and λ3=1559.9 nm. The output power of each LD was carefully adjusted to ensure the desired profile of the UWB pulse. A PM (Avanex IM10-P) was optimized at the center of the loop by tuning the VFDL. The original electrical Gaussian pulse train from a pulse pattern generator (PPG) was amplified to impart up to π shift between CW and CCW lights, and then drove the PM with a fixed pattern “1000 0000 0000 0000” (one “1” per 16 bits) at a bit rate of 12 Gbit/s, indicating that the pulse repetition rate is 750 MHz and the FWHM is about 83 ps. With the help of two circulators, the modulated optical signals were routed out of the loop from two output ports labeled Port1 and Port2 in Fig. 4. The output signals from the two ports were selectively launched to 5-km standard single-mode fiber (SSMF) in order to introduce a time delay of around 80 ps. After O/E conversion by a photodiode (PD), the output electrical signal was analyzed by a 50 GHz oscilloscope and a 26.5 GHz electrical spectrum analyzer (ESA).
The experiment results are summarized in Fig. 5, where the simulated electrical spectra are also plotted in dashed line for the sake of providing an evident comparison. Since the original electrical pulse train from PPG is periodic, the measured electrical spectra are all discrete with a frequency spacing of 750 MHz which is just equal to the repetition rate of generated UWB pulses. The envelope describes the spectrum of a single UWB pulse. The switch depicted in Fig. 4 determines the pulse polarity. By simply shutting off LD3, the pulse shape can be converted from doublet to monocycle. As shown in Fig. 5, the experiment results perform a well agreement with the theoretical prediction in terms of both pulse profile and spectrum. For Port1, the FWHM of the monocycle and doublet pulses are 69 ps and 65 ps respectively. The central frequencies are 4.8 GHz and 4.9 GHz, and the 10-dB bandwidths are about 8.25 GHz and 5.2 GHz respectively. Thus the corresponding fractional bandwidths are 172% and 106%. For Port2, the monocycle and doublet pulses have a FWHM of 73 ps and 60 ps, a central frequency of 4.75 GHz and 4.8 GHz, and a 10-dB bandwidth of 8.1 GHz and 5.2 GHz respectively. The fractional bandwidth is accordingly 171% and 108%. As the generation of either monocycle or doublet UWB pulse depends on the on-off of LD3, an electrical signal with two levels (“1” or “0”) can control the pulse shape by driving a typical IM after LD3. The optical signal modulated by data “1” or “0” is expected to emulate the on-off of LD3. Note that the generated UWB doublet pulses are exactly doublet-like pulses, rather than real ones, since a doublet pulse is strictly achieved by the second derivation of a Gaussian one. Furthermore, there is some undesired energy at low frequencies, which should be removed (e.g. by using a UWB bandpass filter) in order to meet the FCC regulations.
We have proposed and demonstrated a novel scheme to generate UWB pulses using an IM based on Sagnac interferometer. Using two LDs, a pair of polarity-reversed monocycle pulses with a FWHM of about 70 ps and a fractional bandwidth of beyond 170% was experimentally generated. By simply turning on the third LD, a pair of polarity-reversed doublet pulses with a FWHM of about 60 ps and a fractional bandwidth of larger than 100% was also obtained. In our scheme, both the shape and polarity of the generated UWB pulse can be easily selected and controlled, which is flexible and favorable for the future UWB applications. The proposed method features an insensibility to the direct-current (DC) bias drift due to the inherent property of the designed IM. In real applications, a polarization-maintaining structure rather than a PC-based setup is recommended to achieve a more stable operation [11, 13].
This work was partially supported by the National 863 Program of China (2007AA01Z264, 2006AA01Z256), the National Natural Science Foundation of China (60702006 and 60736002), the New Century Excellent Talent Project in Ministry of Education of China (NCET-06-0093), and the 111 Project (B07005).
References and links
1. G. R. Aiello and G. D. Rogerson, “Ultra-wideband wireless systems,” IEEE Microw. Mag. 4, 36–47 (2003). [CrossRef]
2. D. Porcine, P. Research, and W. Hirt, “Ultra-wideband radio technology: potential and challenges ahead,” IEEE Commun. Mag. 41, 66–74 (2003). [CrossRef]
3. X. Chen and S. Kiaei, “Monocycle shapes for ultra wideband system,” in IEEE Int. Symp. Circuits and Systems. 1, 26–29 (2002).
4. J. Dong, X. Zhang, J. Xu, and D. Huang, “All-optical ultra-wideband monocycle generation utilizing gain saturation of a dark return-to-zero signal in a semiconductor optical amplifier,” Opt. Lett. 32, 2158–2160 (2007). [CrossRef] [PubMed]
5. H. Chen, M. Chen, C. Qiu, J. Zhang, and S. Xie, “UWB monocycle pulse generation by optical polarisation time delay method,” Electron. Lett. 43, 542–543 (2007). [CrossRef]
6. Q. Wang and J. Yao, “UWB doublet generation using nonlinearly-biased electro-optic intensity modulator,” Electron. Lett. 42, 1304–1305 (2006). [CrossRef]
7. F. Zeng and J. Yao, “Ultrawideband impulse radio signal generation using a high-speed electrooptic phase modulator and a fiber-bragg-grating-based frequency discriminator,” IEEE Photon. Technol. Lett. 18, 2062–2064 (2006). [CrossRef]
8. J. Dong, X. Zhang, J. Xu, D. Huang, S. Fu, and P. Shum, “Ultrawideband monocycle generation using crossphase modulation in a semiconductor optical amplifier,” Opt. Lett. 32, 1223–1225 (2007). [CrossRef] [PubMed]
9. Q. Wang, F. Zeng, S. Blais, and J. Yao, “Optical ultra-wideband monocycle pulse generation based on crossgain modulation in a semiconductor optical amplifier,” Opt. Lett. 31, 3083–3085 (2006). [CrossRef] [PubMed]
10. C. Wang, F. Zeng, and J. Yao, “All-fiber ultrawideband pulse generation based on spectral shaping and dispersion-induced frequency-to-time conversion,” IEEE Photon. Technol. Lett. 19, 137–139 (2007). [CrossRef]
11. M. L. Dennis, I. N. Duling III, and W. K. Burns, “Inherently bias drift free amplitude modulator,” Electron. Lett. 32, 547–548 (1996). [CrossRef]
12. N. A. Whitaker Jr., R. J. Lustberg, M. C. Gabriel, and H. Avramopoulos, “Low-drift modulator without feedback,” IEEE Photon. Technol. Lett. 4, 855–857 (1992). [CrossRef]
13. E. H. W. Chan and R. A. Minasian, “A new technique for generating negative coefficients in photonic signal processors based on dual-input and dual-output intensity modulator sagnac interferometers,” IEEE Photon. Technol. Lett. 18, 1252–1254 (2006). [CrossRef]
14. J. Yao, F. Zeng, and Q. Wang, “Photonic Generation of Ultrawideband Signals,” J. Lightwave Technol. 25, 3219–3235 (2007). [CrossRef]