SIGNAL processing in the optical domain is considered

1410 JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 23, NO. 3, MARCH 2005 All-Optical Microwave Filters Using Uniform Fiber Bragg Gratings With Identical Reflectivities Fei Zeng, Student Member, IEEE, Student Member, OSA, and Jianping Yao, Senior Member, IEEE, Member, OSA Abstract We propose a novel approach to realizing all-optical microwave filtering using an array of fiber Bragg gratings with identical reflectivities. The filter tapping coefficients are determined by the spectrum profile of a broadband light source and the central wavelengths of the gratings. Since the fiber Bragg gratings have identical reflectivities, the characteristics of the gratings with high uniformity are possible, which simplifies the fiber Bragg grating fabrication and reduces the implementation error. In addition, the spectrum profile of the broadband light source can be controlled using an optical filter, which could be used to control the filter coefficients to suppress the filter sidelobes. A microwave filter using four fiber Bragg gratings is experimentally implemented. Index Terms All-optical microwave filter, broadband source, fiber Bragg grating, high birefringence fiber, Lyot Sagnac loop. I. INTRODUCTION SIGNAL processing in the optical domain is considered a promising technique for many applications such as radar and broadband wireless access networks, thanks to the numerous advantages provided by photonics, such as low loss, small dispersion, light weight, high time-bandwidth product, and immunity to electromagnetic interference [1], [2]. In addition, the capability of processing high-frequency and wide-band signals directly in the optical domain without the need for inefficient and costly intermediate conversions to and from the optical and electrical domains can be of great practical value for potential applications, including the direct interfacing of fiber processors with high-speed optical communication systems and radio-over-fiber networks. For these reasons, there is considerable interest in fiber-optic signal processors for a number of frequency- and time-domain applications, such as filtering, correlation, and Fourier transformation. Different delay line filter configurations have been proposed, in which the tapping elements can be optical couplers [3], Mach Zehnder lattices [4], high dispersion fibers [5], arrayed waveguide gratings (AWGs) [6], or fiber Bragg gratings (FBGs) [7] [9]. Among these tapping elements, FBGs have been considered a good candidate for all-optical microwave filtering because of the numerous advantages provided by FBGs. In FBG-based delay line filters, the tapping intervals can be easily manipulated by controlling the grating spacing; and the tapping weights can also be controlled by varying grating reflectivity during the Manuscript received June 3, 2004; revised August 30, 2004. The authors are with the Microwave Photonics Research Laboratory, School of Information Technology and Engineering, University of Ottawa, Ottawa, ON K1N 6N5, Canada (e-mail: jpyao@site.uottawa.ca). Digital Object Identifier 10.1109/JLT.2004.839981 FBG fabrication process. In addition, the interaction wavelength can be controlled or tuned via changing the grating pitch by applying strain, which can be realized by variable heating, piezoelectric controlling, mechanical controlling, or magnetic field controlling. Furthermore, by using wide-band chirped fiber Bragg gratings (CFBGs) as dispersion elements, more sophisticated tunability can be achieved. All these features make the implementation of wide-band FBG-based all-optical microwave filters with easy reconfigurability and tunability possible. In general, FBG-based all-optical microwave filters can be divided into two categories. In the first category [10], a multiwavelength laser source or a laser array is used. Time delays are achieved when the multiwavelength light is reflected at different locations of a chirp grating or a uniform FBG array. The difficulty in implementing this technique is that the light source is very complicated, and few taps and limited filtering functions can be achieved. The shortest tapping interval of the microwave filter is limited by the coherent lengths of the laser diodes, which have relatively longer coherent lengths than incoherent light source. Although, in principle, coherent processing can be used for all-optical signal processing, it is difficult to implement in practice. Therefore, in all-optical microwave filters the tapped optical signals are combined incoherently, and coherent interference should be avoided. In the second category [11], [12], a single broadband light source is used. Different time delays are realized by using a tapped delay line consisting of an array of uniform FBGs. The tapping weights are determined by the FBGs reflectivities or by controlling the spectrum of the broadband optical source. In this approach, the light source is easy to obtain and more taps can be acquired. Usually light emitting diodes (LEDs) or amplified spontaneous emission (ASE) sources using erbium-doped fiber amplifiers (EDFAs) are used as the broadband sources, which are considered incoherent with very short coherent length. Therefore, there is no limit on the tap intervals of the microwave filters. In [11], Hunter et al. proposed an FBG-based transversal filter with 29 taps. In their setup, a relatively flat broadband source was employed. By properly selecting the reflectivities of the FBGs, a frequency response with reduced sidelobe level was demonstrated. However, the power spectrum of a broadband source is usually not flat as we expect. Then the reflectivities of the FBGs cannot be used as the filter tapping coefficients directly. Two methods can be employed to solve this problem. One is to apply an optical gain equalizer to flatten the power spectrum of the broadband source. But, this will increase the complexity and instability of the whole system. Another method is to introduce a weighting sequence according to the power spectrum incident 0733-8724/$20.00 2005 IEEE

ZENG AND YAO: ALL-OPTICAL MICROWAVE FILTERS USING UNIFORM FIBER BRAGG GRATINGS 1411 Fig. 1. Basic structure of an FBG-based all-optical microwave filter. into the FBG array to calibrate the reflectivities [13]. In practice, the fabricated FBGs with different reflectivities have different reflection bandwidths, lengths, and dispersion characteristics, which will introduce filter implementation errors. To solve this problem, we propose to take the advantage of the unflat power spectrum of a broadband source by synthesizing all-optical microwave filters using the central wavelengths of the FBGs as the filter synthesis parameters. In this approach, the FBGs are designed to have identical reflectivities and the tapping weights are determined by properly choosing the Bragg wavelengths, which makes it possible to simplify the FBG fabrication and reduce implementation errors. Furthermore, different window functions can be achieved by employing a tunable optical filter to change the spectrum of the broadband source; filters with reduced sidelobe levels can be realized. This paper is organized as follows. In Section II, a theoretical model of uniform FBG-based all-optical microwave filters is presented, showing the relationship among the system transfer function, the characteristics of the FBGs, and the broadband source. Based on this model, the filter synthesis parameters are determined and a design example is given. In SectionIII, four FBGs with identical reflectivities are fabricated based on the design example. Experimental implementation of the all-optical microwave filter using the fabricated FBG array is carried out. The effects on the filter frequency response due to the errors in reflectivities and bandwidths of the fabricated FBGs are evaluated. In Section IV, a microwave filter with tunable filter coefficients incorporating a Lyot Sagnac fiber loop is demonstrated. By tuning the polarization states in the Lyot Sagnac loop, the microwave filter with different sidelobe suppression levels is realized. Finally, a conclusion is drawn in Section V. II. THEORY The FBG-based all-optical microwave transversal filter to be discussed here can be interpreted using the grating array structure shown in Fig. 1. The structure consists of an optical circulator and an array of FBGs, but the broadband source and the external electrooptic modulator (EOM) are not shown in the diagram. The input of the FBG array is marked as modulated broadband light with the schematic diagram of its power spectrum shown. The FBG array is composed of a number of FBGs interconnected as the tapping elements of the transversal filter. Each FBG reflects a slice of the modulated light at a specific wavelength, and the reflected time-delayed signals are summed at a photodetector. The power spectrum of the reflected time-delayed optical signals is also shown in Fig. 1. Based on this configuration, the output optical power of the delay-line filter is linearly proportional to the input optical power, provided that there are no nonlinear effects in the system. The entire system can be treated as a linear time-invariant (LTI) system in which the recovered electrical signal is the convolution of the modulating electrical signal with the impulse response of the filter. The transfer function of the microwave filter is derived as follows. First, let us consider a broadband light source with the power spectrum over the working wavelength range shown in Fig. 2. The effective power intensity of the modulated light incident into the FBG array can be written as where is the optical power spectrum density at wavelength, which is within the range is the electrical signal applied to the modulator; and is the modulating index, which is assumed to be a constant for different carrier frequencies and different modulating frequencies. Second, the FBG array is composed of 1 uniform gratings written on a single-mode fiber at different locations. Basically, the coherent interference of partial reflectance within the th FBG creates a bandpass reflection response and stopband transmission response at a central wavelength with a null-to-null bandwidth and a reflectivity. The free spectral range (FSR) of the microwave filter is (1)

1412 JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 23, NO. 3, MARCH 2005 Fig. 2. Optical power spectrum of the modulated light. determined by the minimum spacing between any two adjacent FBGs. The time delay between two adjacent FBGs (the th and the 1th gratings) can be identical or nonuniform, which is expressed as where is the effective refractive index of the single-mode fiber; is the spacing between those two adjacent FBGs; and is the speed of light in free space. Then the FSR is given by (2) FSR (3) The optical power of the modulated light after reflection by the FBG array is (4) When this reflected light is incident on a photodetector, the output electrical signal is Since we are only interested in the ac component, the dc component at the photodetector output is not included in (5). Applying the Fourier transform to both sides of (5), we get the frequency response where with are the effective coefficients. From (6), we can see that the filter design now becomes to find out the optimal solutions of, and, to achieve an optimal approximation of. Usually it is very difficult to obtain the solutions since many synthesis parameters have to be considered simultaneously, especially some of them are tightly associated, i.e., the gratings reflectivities and the bandwidths. As we have mentioned in Section I, gain flattening filters can be applied to flatten the power spectrum of the incident light, say, to make for all the wavelengths. However, this will increase the complexity and instability of the whole system. Even when an ideal gain flattening filter is applied, problems still exist when fabricating (5) (6)

ZENG AND YAO: ALL-OPTICAL MICROWAVE FILTERS USING UNIFORM FIBER BRAGG GRATINGS 1413 the gratings with different characteristics. The problems are demonstrated as follows. For a given desired filter response, specific weighting window needs to be applied to suppress the sidelobes or to reduce the ripples in either the passband or the stopband. Some of the filter coefficients may be required to be less than 0.1, whereas some others are required to be great than 0.9, which means that some FBGs have to be written with very low reflectivities and some others may have much higher reflectivities. These filter design requirements lead to some practical issues to be addressed carefully. First, in real FBG fabrication system, it is difficult to write FBGs with very low reflectivities, i.e., less than 0.1. Although in theory it is possible, the accuracy cannot be guaranteed and high implementation errors will be induced. To write FBGs with different reflectivities and bandwidths, the FBG fabrication system, e.g., the illuminated intensity and time of the ultraviolet (UV) light, the UV beam width, and the annealing parameters, have to be carefully adjusted. Second, to fabricate FBGs with different reflectivities will result in different FBG bandwidths. Since the bandwidths are also important synthesis parameters for an all-optical microwave filter, this effect must be taken into account. The null-to-null reflection bandwidth of a uniform FBG is given by [14] (7) where is the Bragg wavelength of the FBG, is the modulation depth of refractive index, is the dc index change along the grating length, is the effective refractive index of the fiber, and is the length of the grating. The weak grating regime is described by. Its null-to-null reflection bandwidth is given by, where is the number of periods, which means that the bandwidth is inversely proportional to the grating length for weak gratings. In the strong grating regime, where, the reflection bandwidth is proportional to the index variation given by the ac index change,, which means that the bandwidth for strong gratings is independent of grating length. As shown in Fig. 3, for the gratings with identical lengths, the stronger grating has a relative wider bandwidth; for the gratings with identical reflectivities, the shorter one has a relative wider bandwidth. Therefore, we conclude that the use of FBGs with different reflectivities and bandwidths will not only increase the complexity of the filter synthesis but also introduce many implementation errors during the grating writing processes. Based on the discussion above, to reduce both the implementation errors and the filter synthesis complexity, we propose to use FBGs with identical reflectivities and bandwidths, that is Then the design problem is now reduced to find the central wavelengths of the FBG array according to the power (8) Fig. 3. Simulated reflection spectra of two FBGs. (a) Identical length 1 mm, but different peak reflectivities. Solid line: =12 10 ; dotted line: =52 10. (b) Same peak reflectivities, but different lengths. Solid line: 10 mm; dotted line: 4 mm. spectrum of the modulated light. The effective filter coefficients can be further simplified if the reflection spectra of the FBGs are considered ideal with very narrow bandwidths So only one synthesis parameter is required for a given incident power spectrum. Compared with the other approaches, the filter design process is significantly simplified. More importantly, writing gratings with identical reflectivities and bandwidths will be much easier with better accuracy. III. EXPERIMENTS In this paper, a low-pass filter with four taps {0.40, 0.80, 0.80, 0.40} would be experimentally implemented to meet the desired frequency response specifications, which include the 1) normalized passband frequency, 2) normalized stopband edge frequency, and 3) stopband (9)

1414 JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 23, NO. 3, MARCH 2005 Fig. 4. Experimental setup of the FBG-based all-optical microwave filter. magnitude response at least 15 db lower than the passband magnitude response, where is the tapping interval and its reciprocal is the FSR of the filter. Based on the theoretical analysis in Section II, the design of an FBG-based all-optical microwave filter can be carried out with the following steps. First, employ the modified digital filter design method to obtain the desired filter coefficients; second, measure the power spectrum of the light expected to incident into the FBG array; third, since all the FBGs are supposed to have identical reflectivities, lengths, and narrow bandwidths, only the Bragg wavelengths for the gratings are required to be determined, which can be calculated by substituting the measured power spectrum of the light source and the designed coefficients into (9). Based on this procedure, the synthesis parameters for the gratings in the array are determined. Then the FBG array is fabricated and the experimental implementation of the all-optical microwave filter employing the FBG array is carried out. A. Experimental Setup The experimental setup is shown in Fig. 4. An EDFA is used as the broadband source, which is intensity modulated by an EOM and fed to an array of four FBGs via an optical circulator. A second EDFA is employed before the circulator to compensate for the power loss in the system. An optical spectrum analyzer (OSA) is used to monitor the spectrum of the light to be detected by the photodiode. The measured output spectrum after the second EDFA is shown in Fig. 5. The Bragg wavelengths of the FBGs are determined by letting and be 3-dB larger than and to satisfy the design requirements. However, due to the lack of available phase masks, four FBGs would have the Bragg wavelengths which are slightly different from those in the design. According to the phase masks used in this approach, the tap weights would be {0.42, 0.71, 0.80, 0.45}, which will be used to calculate the theoretical frequency response. The four gratings are fabricated by employing UV photoimprinting technique, and each grating is also apodized to reduce the sidelobe levels by employing a Gaussian-profile amplitude mask before the phase mask. The measured Fig. 5. Power spectrum of the incident light. TABLE I CHARACTERISTICS OF FBG ARRAY characteristics are summarized in Table I. From Table I, we can see that the variations of the peak reflectivities and 3-dB bandwidths for the four gratings are small, which verifies that high uniformity is realized when gratings are fabricated with identical reflectivities and bandwidths. B. Experimental Results The grating spacing between two adjacent FBGs is 10 cm, which corresponds to a time delay of 0.966 ns and an FSR of 1.036 GHz. The reflected spectrum of this FBG array is shown in Fig. 6. By using the four reflected peaks in Fig. 6, the filter coefficients are calculated to be {0.46, 0.78, 0.80, 0.45} and denoted as peak reflectivity based coefficients. Compared with the theoretical coefficients {0.42, 0.71, 0.80, 0.45} in Section III-A, there are small variations due to the implementation errors of the peak reflectivities and central wavelengths of the fabricated FBGs. The reflected time-delayed signals are summed at the photodetector. An Agilent vector network analyzer (E8364A) is used to generate the modulating radio-frequency signal and to observe the corresponding output of the photodetector. By sweeping the electrical signal from 45 MHz to 5 GHz while keeping the amplitude constant, the filter frequency response is obtained. Fig. 7(a) shows the frequency responses: 1) the measured response (solid), 2) the peak-reflectivity-based response (dashed), and 3) the theoretical response (dotted). It can be seen that the peak reflectivity based filter response agrees well with the theoretical response. The variation between

ZENG AND YAO: ALL-OPTICAL MICROWAVE FILTERS USING UNIFORM FIBER BRAGG GRATINGS 1415 Fig. 6. Reflection power spectrum of the FBG array. these two curves is very small and negligible. Since this variation is caused by the small nonuniformity of the gratings peak reflectivities and the small errors in the central wavelengths, we can conclude that the implementation errors on the grating peak reflectivities and central wavelengths can be neglected. Another variation in the stopband between the peak reflectivity based and the measured response is also observed in Fig. 7(a), which is due to the bandwidth errors caused during the FBG fabrication. Based on the analysis in Section II and the measured 3-dB, 10-dB grating reflection bandwidths shown in Table I, tap weights of 0.35, 0.67, 0.8, 0.35 and 0.34, 0.69, 0.8, 0.36 are calculated by use of the total reflected power within the 3- and 10-dB reflection bandwidths, which are denoted as 3-dB bandwidth based and 10-dB bandwidth based in the text. Fig. 7(b) shows four frequency responses: 1) the measured response (solid), 2) the peak reflectivity based response (dotted), 3) the 3-dB bandwidth based response (dash-dotted), and 4) the 10-dB bandwidth based response (dashed). An excellent agreement between the measured and the 10-dB-bandwidth-based response is observed. The result shows that in order to further reduce the error of the implemented filter frequency response, the FBG bandwidth needs to be considered. IV. FURTHER IMPROVEMENT In the above example, only four taps are used. In real filter design, more taps are required to improve the filter performance, such as reduced sidelobe levels as well as the passband ripples. Basically, different weighting windows can be applied to reduce the sidelobe levels. By observing the power spectrum of the incident light shown in Fig. 5, only one up and down is available, which cannot provide enough wavelength choices if three or more FBGs are required to be located at or very close to one wavelength. This problem is solved by incorporating an optical comb filter to control the power spectrum of the broadband light source to obtain more ups and downs. Consequently, a large number of taps, even some of them required to be identical or very close, can be realized by choosing proper Bragg Fig. 7. Measured and theoretical frequency responses. (a) Measured (solid), peak reflectivity based (solid), and theoretical (dotted). (b) Measured (solid), peak-reflectivity-based (dotted), 3-dB bandwidth based (dash-dotted), 10-dB bandwidth based (dashed). wavelengths based on the optical comb filter modified broadband spectrum. Furthermore, by tuning the comb filter, different window functions can be applied to the proposed microwave filter. A. Lyot Sagnac Fiber Loop Fig. 8 shows the configuration of the optical comb filter, which has a Lyot Sagnac loop structure composed of a 3-dB coupler, a polarization controller, and two sections of high-birefringence (HiBi) fiber spliced with an angle between the two principal axes. When the light is incident into one port, i.e., port 1 of the coupler, it will be split into two counterpropagating beams. Each of the two counterpropagating beams is then disassembled into 2 beams after traveling through the two HiBi fiber sections. These counterpropagating light beams interfere with each other when they propagate in the loop.

1416 JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 23, NO. 3, MARCH 2005 Fig. 8. Configuration of the Lyot Sagnac fiber loop. The transfer function of this birefringence filter can be derived by use of the Jones formalism [15], [16]. If the input polarization state at port 1 is assumed to be (10) and the Jones matrices of the fiber along the clockwise and the counterclockwise direction inside the loop are (11) with for a lossless reciprocal system. According to the reciprocity of the Jones matrix, the output field at port 2 is given by [15] (12) where represents the imaginary part of, which is the off-diagonal term of the Jones matrix of or. Then the intensity transfer function of this optical filter can be obtained by (13) Fig. 9. Intensity transfer functions of the Lyot Sagnac fiber loop. Solid line: before tuning; dotted line: after tuning to another specific value. A maximum optical power change of 13 db can be observed at =1542:95 nm. which introduces high-order birefringence; this implies that the FSR of this Lyot Sagnac loop may be tuned by changing some of the parameters in the loop. In this paper, we are more interested in the term ; it can be seen that when we change the orientation angle of the polarizer, a continuous wavelength tuning is possible. Fig. 9 shows the output optical power spectra of the Lyot Sagnac loop when the polarizer is tuned. B. Sidelobe Suppression The Lyot Sagnac fiber loop is incorporated into the microwave filter after the second EDFA to modify the spectrum of the broadband source. Four frequency responses of the microwave filter with four different window functions are shown in Fig. 10. The corresponding optical reflection spectra of the FBG array are also shown in Fig. 10. It can be seen that the shape of the proposed microwave filter frequency response is changed by tuning the orientation angle of the polarizer in the Lyot Sagnac loop. The sidelobe level is reduced significantly when proper tap weights are selected. Although in the experiment, the filter has only four taps, the effectiveness of this approach can be extended to microwave filters with more taps. By carefully selecting FBGs central wavelengths, and tuning the comb filter, proper window function can be achieved to reduce the mainlobe-to-sidelobe ratio (MSR). As shown in Fig. 10(d), one low-pass filter with an MSR of 22 db and a notch filter with a 37 db notch rejection level are demonstrated. where in our setup; are the lengths of the two HiBi fiber sections; is the length difference between these two sections; is the difference between the two effective refractive indexes of the HiBi fiber that supports two linearly orthogonal fundamental modes and is the polarization angle difference between the incident light and the polarizer. is a tedious trigonometric function V. CONCLUSION In this paper, we proposed an approach to implementing alloptical microwave filters using FBGs with identical reflectivities. The filters were designed using the Bragg wavelengths of the FBGs and the spectrum profile of the broadband source as the synthesized parameters to achieve required tap weights. A

ZENG AND YAO: ALL-OPTICAL MICROWAVE FILTERS USING UNIFORM FIBER BRAGG GRATINGS 1417 Fig. 10. Experimental results. (a) Optical power spectra of the reflected light from the FBG-array for two different polarization states in the Lyot Sagnac loop. (b) Filter frequency responses corresponding to (a). (c) Optical power spectra of the reflected light from the FBG-array for another two different polarization states in the Lyot Sagnac loop. (d) Filter frequency responses corresponding to (c). microwave filter using four FBGs was experimentally implemented. By properly choosing the Bragg wavelengths of the FBGs and tuning the optical comb filter, filter frequency responses with different sidelobe levels were demonstrated. Since the FBGs were fabricated with identical reflectivities, the fabrication process was greatly simplified and the fabrication errors were much reduced. REFERENCES [1] K. P. Jackson, S. A. Newton, B. Moslehi, M. Tur, C. C. Cutler, J. W. Goodman, and H. J. Shaw, Optical fiber delay line signal processing, IEEE Trans. Microw. Theory Tech., vol. MTT-33, pp. 193 209, Mar. 1985. [2] K. Wilner and A. P. Van den Heuvel, Fiber-optic delay lines for microwave signal processing, in Proc. IEEE, vol. 64, 1976, pp. 805 807. [3] S. Sales, J. Capmany, J. Martin, and D. Pastor, Experimental demonstration of fiber-optic delay filters with negative coefficients, Electron. Lett., vol. 31, no. 13, pp. 1095 1096, June 1995. [4] T. A. Cusick, S. Iezekiel, R. E. Miles, S. Sales, and J. Capmany, Synthesis of all-optical microwave filters using Mach-Zehnder lattices, IEEE Trans. Microw. Theory Tech., vol. 45, pp. 1458 1462, Aug. 1997. [5] D. Norton, S. Johns, C. Keefer, and R. Soref, Tunable microwave filtering using high dispersion fiber time delays, IEEE Photon. Technol. Lett., vol. 6, pp. 831 832, Jul. 1994. [6] V. Polo, B. Vidal, J. L. Corral, and J. Marti, Novel tunable photonics microwave filter based on laser arrays and N 2 N AWG-based delay lines, IEEE Photon. Technol. Lett., vol. 15, pp. 584 586, Apr. 2003. [7] D. B. Hunter and R. Minasian, Reflectivity tapped fiber optic transversal filter using in-fiber Bragg gratings, Electron. Lett., vol. 31, pp. 1010 1012, Jun. 1995. [8] J. Marti, F. Ramos, and R. I. Laming, Photonic microwave filter employing multimode optical sources and wideband chirped fiber gratings, Electron. Lett., vol. 34, no. 18, pp. 1760 1761, Sep. 1998. [9] G. Yu, W. Zhang, and J. A. R. Williams, High-performance microwave transversal filter using fiber Bragg grating arrays, IEEE Photon. Technol. Lett., vol. 12, pp. 1183 1185, Sep. 2000. [10] J. Capmany, D. Pastor, and B. Ortega, Efficient sidelobe suppression by source power apodization in fiber optic microwave filters composed of linearly chirped fiber grating by laser array, Electron. Lett., vol. 35, no. 8, pp. 640 642, Apr. 1999.

1418 JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 23, NO. 3, MARCH 2005 [11] D. B. Hunter and R. A. Minasian, Microwave optical filters using in-fiber Bragg grating arrays, IEEE Microw. Guided Wave Lett., vol. 6, pp. 103 105, Feb. 1996. [12] D. Pastor, J. Capmany, and B. Ortega, Broad-band tunable microwave transversal notch filter based on tunable uniform fiber Bragg gratings as slicing filters, IEEE Photon. Technol. Lett., vol. 13, pp. 726 728, Jul. 2001. [13] F. Zeng, J. P. Yao, and S. Mihailov, Genetic algorithm for fiber Bragg grating based all-optical microwave filter synthesis, Opt. Eng., vol. 42, no. 8, pp. 2250 2256, Aug. 2003. [14] T. Erdogan, Fiber grating spectra, J. Lightw. Technol., vol. 15, pp. 1277 1294, Aug. 1997. [15] X. Fang and R. O. Claus, Polarization-independent all-fiber wavelength-division multiplexer based on a Sagnac interferometer, Opt. Lett., vol. 20, no. 20, pp. 2146 2148, Oct. 1995. [16] X. Fang, H. Ji, C. T. Allen, K. Demarest, and L. Pelz, A compound high-order polarization-independent Birefringence filter using Sagnac interferometers, IEEE Photon. Technol. Lett., vol. 9, pp. 458 460, Apr. 1997. Fei Zeng (S 04) received the B.Eng. degree in optoelectronic engineering from Huazhong University of Science and Technology, Wuhan, China, in 1993 and the M.A.Sc. degree in electrical engineering from the University of Ottawa, Ottawa, ON, Canada, in 2003, where he is currently pursuing the Ph.D. degree in the School of Information Technology and Engineering. From 1993 to 2001, he was with NEC Fiber Optical Communications Ltd., Wuhan, working on synchronous digital hierarchy and dense wavelength-division multiplexing system verification. His current research interests include microwave photonics, all-optical microwave filters, all-optical packet switching, and radio-over-fiber systems. Mr. Zeng is a Student Member of the Optical Society of America (OSA). Jianping Yao (M 99 SM 01) received the Ph.D. degree in electrical engineering from the University of Toulon, France, in 1997. He is an Associate Professor in the School of Information Technology and Engineering, University of Ottawa. From January 1998 to July 1999 he was a Research Fellow, and from July 1999 to December 2001 he was an Assistant Professor, both with the School of Electrical and Electronic Engineering, Nanyang Technological University, Singapore. He has published more than 70 papers in refereed journals and conference proceedings. His current research interests include optical signal processing, optically controlled phased array antennas, photonic generation of microwave signals, radio-over-fiber systems, fiber lasers and amplifiers, broadband infrared wireless home networking, and fiber-optic sensors. Dr. Yao is a Member of The International Society for Optical Engineers (SPIE) and the Optical Society of America (OSA).