- Email: [email protected]
Employing optical code division multiple access technology in the all fiber loop vibration sensor system
Optical Fiber Technology 19 (2013) 627–637
Contents lists available at ScienceDirect
Optical Fiber Technology www.elsevier.com/locate/yofte
Employing optical code division multiple access technology in the all fiber loop vibration sensor system Shin-Pin Tseng a, Chih-Ta Yen b, Rong-Shun Syu c, Hsu-Chih Cheng c,⇑ a
Department of Electronic Engineering, National United University, Miaoli, Taiwan Department of Electrical Engineering, National Formosa University, Yunlin, Taiwan c Department of Electro-Optical Engineering, National Formosa University, Yunlin, Taiwan b
a r t i c l e
i n f o
Article history: Received 2 April 2013 Revised 19 August 2013 Available online 22 October 2013 Keywords: Optical code division multiple access (OCDMA) Optical fiber bending Optical sensor Spectral amplitude coding (SAC) Vibration frequency measurement
a b s t r a c t This study proposes a spectral amplitude coding-optical code division multiple access (SAC-OCDMA) framework to access the vibration frequency of a test object on the all fiber loop vibration sensor (AFLVS). Each user possesses an individual SAC, and fiber Bragg grating (FBG) encoders/decoders using multiple FBG arrays were adopted, providing excellent orthogonal properties in the frequency domain. The system also mitigates multiple access interference (MAI) among users. When an optical fiber is bent to a point exceeding the critical radius, the fiber loop sensor becomes sensitive to external physical parameters (e.g., temperature, strain, and vibration). The AFLVS involves placing a fiber loop with a specific radius on a designed vibration platform. A 1 K coupler was adopted for the sensor system to divide a broadband light source into K light sources, which were then transmitted to various FBG encoders. The AFLVS was placed between the optical circulators of the various FBG encoders and multiple FBG arrays, and the stepping motor was directly placed on the fiber loops of the various AFLVSs. A signal generator was then used to input different frequencies into the stepping motors of the various sensors. After the light intensity for the reflectance spectrum, which was outputted by the FBG encoder, was modulated by the AFLVS, the modulated reflected signals were outputted to the K K star coupler through the optical circulator and transmitted to the FBG decoders for the users. A balanced photodetector (BPD) was employed in this study to convert the light output of the FBG decoder into an electrical signal, and a digitizing oscilloscope was employed to conduct a Fourier transform on the BPD electrical signal output, thereby acquiring the vibration frequency of the test object. The results of the experiment are compared to a piezoelectric accelerometer. The comparison results indicate that the piezoelectric accelerometer is less sensitive when the frequency is lower than 90 Hz, whereas the AFLVS exhibits excellent measurement results at a low frequency ranging between 50 and 200 Hz. Ó 2013 Elsevier Inc. All rights reserved.
1. Introduction Optical fiber sensors (OFSs) are widely applied in measuring physical parameters, including rotation, acceleration, electromagnetic fields, temperature, pressure, sound, vibration, and humidity. OFSs are also commonly used in civil engineering, environmental and biochemical testing, and clinical biomedical fields. In addition, OFSs are advantageous because of their small size, light weight, low power, high sensitivity, high bandwidth, and ability to resist electromagnetic interference (EMI). Furthermore, OFSs possess the characteristic of light signal transmission, which enables multiplex measurements from multiple points by combining various multiplex network frameworks [e.g., wavelength division multiplexing (WDM), time division
⇑ Corresponding author. E-mail address: [email protected] (H.-C. Cheng). 1068-5200/$ - see front matter Ó 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.yofte.2013.10.001
multiplexing (TDM), frequency division multiplexing (FDM), and optical code division multiple access (OCDMA)]. Traditional electric sensors (e.g., piezoelectric accelerometers), when used for vibration sensing, possess a broad range of working temperatures and can sense or detect higher frequencies. However, a relatively low signal-to-noise-ratio (SNR) is detected when traditional electric sensors are applied for low-frequency access. Moreover, these sensors are easily subjected to EMI; thus, they can only be applied to single-point simplex measurements. The sensitivity of the fiber loop sensor for external physical parameters is increased when the optical fiber is bent to a point exceeding the critical radius, and the sensor can be used to measure temperature, the refractive index, and vibration [1,2]. Theoretical and experimental results indicate that attributes regarding the significant influence of optical fiber diameter on bending loss can be used to develop vibration [3] and shift [4] sensors.
628
S.-P. Tseng et al. / Optical Fiber Technology 19 (2013) 627–637
OFSs can be integrated with multiplex techniques to achieve distributed sensing and decrease the cost and complexity of the multiple-point measurement system. OFSs are commonly applied to optical remote sensing (e.g., strain, temperature, and pressure sensing). Among various multiplex techniques, WDM is most commonly employed, and is a simple technique used to identify multiple OFSs. The number of OFSs applicable for this multiplexing system is limited by the light source bandwidth and fiber Bragg grating (FBG) reflected wavelength intervals [5]. TDM is another common technique, which uses a pulsed laser light source to emit pulses to FBG sensors at various distances. The reflected wavelengths of various FBG sensors differ, and the minimum distance for different FBG sensors is determined based on the time slot width of the light source. The pulse bandwidth must be equivalent to or less than the round-trip time for two random FBG sensors. The reflected wavelengths of various FBG sensors are separated according to the delay time of the optical fiber, and a rapid timedomain signal analysis is then employed to identify various sensor signals. However, significant power is required so that short pulses from light sources can increase SNR values for photodetector output signals [6]. To improve deficiencies in the TDM, the OCDMA framework can be used to implement additional FBG sensors by transmitting or emitting a modulated broadband light source to the FBG sensor arrays through an optical circulator. Based on the FBG encoding pattern, this light pulse possesses a unique sequence comprising zeros and ones. The reflected pulse sequence propagates through the detection system comprising the edge filter, balanced photodetectors (BPDs), and electrical signal processing units. The optical fiber delay line performs spatial modulations of the pulse sequence according to different sensor locations. Codes with various delay times can be used to calculate the autocorrelation function of the reflected signal while the wavelength messages of all sensors are demodulated. The edge filter is used as a passive wavelength demodulation device by converting wavelength variations into amplitude variations. Two photodetectors are employed to calculate the correlation function ratio of the wavelength shift in the FBG sensor. The analog-to-digital converter is then applied for sampling received signals [7]. In 2011, the research team for the present study proposed a fiber vibration sensor system combining an OCDMA framework and two optical collimators. This sensor system was capable of accurately measuring the vibration frequency of an object, and was successfully applied to complete a code division access experiment [8]. However, the framework employed in [8] was primarily applied to short-distance transmissions. Consequently, this study also adopted a balanced incomplete block design (BIBD) [9] in which each user group had a specific spectral amplitude coding (SAC). FBGs were used as a primary component for the encoder/decoder for each user, thereby decreasing the number of FBGs in these devices. In this way the encoder/ decoder was simplified and relevant costs reduced. SAC encoding can be used by a great number of active users because of its characteristic of eliminating interference and preventing phase-induced intensity noise during the photodetection process. The excellent characteristics of this system framework can be adopted for multiple-point multiplex measurements by using a distributed vibration sensing system combined with a sensor and encoding/ decoding device. For the vibration sensor, this study proposed using an all fiber loop vibration sensor (AFLVS) to replace the fiber vibration sensor consisting of two optical collimators. This improvement transforms the entire sensor framework into an economical, simple, and flexible all fiber optical sensor design, facilitating the applicability of the optical sensor system for various situations in which vibration measurements are required.
2. Theories and principles Optical fiber bending introduces considerable problems in fiberoptic communication, such as optical power loss and signal interference. The bending loss theory for the fundamental mode of fibers has been extensively researched for decades. The effects of microbending and macrobending are applied to OFSs, including effects for shift, pressure, and temperature sensors. Recent research has focused on relevant theories regarding macrobending loss characteristics for mono-mode fibers with a bend radius between 8.5 and 12 mm, which are used for edge filter wavelength measurements. Furthermore, when the bend radius of the optical fiber is less than 1 cm, the fiber can be applied to sensing. Fiber loop sensors are sensitive to external physical parameters when the optical fiber is bent to a point smaller than the critical radius; they can be used to measure temperature, strain, and vibration. Because of their unique characteristics, fiber loop sensors can measure the vibration frequency of mechanical equipment, and since these sensors require minimal optical power, potential fire risk is low. Fiber loop sensors can also be applied to measure high- and low-frequency vibrations as well as to detect shift by the millimeter. Numerous existing physical models have been employed to predict optical waveguide bending loss; the most common method adopted in this process is the bending loss equation [1,2]:
! 1=2 1 p j2 2c3 Re 2aB ¼ exp 2 2 c3 Re 3b20 V 2 k1 ðcaÞ 1=2
c ¼ ðb0 k2 n2cl Þ
j ¼ ðk2 n2co b20 Þ
1=2
V ¼ akðn2co n2cl Þ k ¼ 2p=k
1=2
ð1Þ
ð2Þ ð3Þ ð4Þ ð5Þ
2aB represents the bending loss coefficient per unit length, where c, k, and V denote the propagation constant; b0 indicates the propagation constant for the leaky mode of the straight optical fiber; a symbolizes the core radius; Re expresses the effective bend radius; nco and ncl respectively refer to the refractive indices of the core and cladding; and K1(ca) is equivalent to the deformed first order-Bessel function of the second kind. Figs. 1 and 2 respectively show the AFLVS structure and the practical AFLVS. A plastic card or tube is used to bend the optical fiber and construct a fiber loop sensor. Two wooden blocks are placed at an appropriate distance on a rubber sheet, and sheet metal is placed atop the two wooden blocks. The loop sensor is subsequently attached to the sheet metal, and the sensing platform for the AFLVS is complete. Fig. 3 shows the power loss of two fiber loop sensors that possess distinct fiber loop diameters. The power loss curve shows numerous intermediate peaks, which are caused when light is reflected from the cladding-coating (jacket) and coating (jacket)-air junction and combines with leaked core light, generating the resonance peaks in the whispering gallery mode [1–3]. Although the two fiber loop sensors are composed of the same single-mode fiber, certain differences exist between their power loss curves (Fig. 3). Whispering gallery mode has been used to develop several sensor applications because of its substantial sensitivity. Fig. 3 shows the maximal slopes of Fiber Loop Sensor 1 from diameters of 1.55 cm to 1.60 cm and Fiber Loop Sensor 2 from diameters of 1.57–1.62 cm, indicating that power within this range is easily affected, and, thus, possesses the highest sensitivity level. Conse-
629
S.-P. Tseng et al. / Optical Fiber Technology 19 (2013) 627–637
Fig. 1. The structure of all fiber loop vibration sensor.
encoding/decoding, as each user’s code array. This consequently increased the amount of access for users of the OCDMA system. Table 1 shows that a total of j users existed in Group 1; thus, the string of the WDM/M-sequence code can be presented as follows [8,9]:
Ak;l ¼ jak;l ðmÞjX k El ¼ ½el;0 X k ; el;1 X k ; . . . ; el;N1 X k
Fig. 2. The practical all fiber loop vibration sensor.
Fig. 3. The power loss curves of the fiber loops for vibration sensors 1 (black line) and 2 (red line). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
quently, the fiber loop diameters for Sensors 1 and 2 were fixed at 1.57 cm and 1.59 cm, respectively. Regarding the OCDMA, the SAC design involves assigning all users into N groups (N represents a positive integer), and each user is referred to as #(k, l). The code string for each user is Ak,l[k = (0, 1, 2, . . ., m 1)], which denotes the number of the Kth user. l = (0, 1, 2, . . ., N 1) expresses the Nth group, and the encoding/ decoding for each user constitutes a binary message bit of a zeros or ones. Multiple FBG arrays are developed into an original encoding/decoding device, and the message information of each user is connected to a star coupler. When all of the user signals are transmitted to the star coupler, the accumulated signal is transmitted through the output end, and then enters the decoding device of each user, subsequently demodulating unique encoded message information. A previous study proposed the use of WDM/BIBD encoding in optics communication systems [9]. However, this study employed an M-sequence code, which occupies less system capacity compared to the traditional unipolar code (i.e., code arrays comprising zeros and ones) and reduces the number of multiple FBG arrays in
ð6Þ
where m = (0, 1, . . ., MN 1), and ak,l(m) represents the mth element used by user Ak,l and the corresponding wavelength kmþ1 . Furthermore, X k ¼ ½xk;0 ; xk;1 ; . . . ; xk;m1 ¼ T k ; X0 expresses the SAC of the jth user (T represents the shift of the code spectrum one periodic vector to the right); El ¼ ðel;0 ; el;1 ; . . . ; el;N1 Þ indicates the vector unit transmitted by the N 1th group (l is a binary message bit composed of zeros or ones); symbolizes the Kronecker product; and X0k sðk ¼ 0; . . . ; M 1Þ represents the SAC code with a length of M (code length). WDM/M-sequence encoding enables the application of a single code array for numerous different group blocks, thereby increasing the number of users without using tedious code arrays and providing greater coding flexibility. In this context, the SAC’s greatest advantage is its excellent cyclical characteristics. When the code base for only one group of users is required, the code arrays for all users can be acquired. This method of code design is easily produced and simply structured. In addition, compared to that for the BIBD, the SAC’s greater number of code bases increases user and system capacity when employed for the OCDMA system. This study proposed adopting FBGs in the OCDMA system, and placing an AFLVS between the FBG encoder and optical circulator. As shown in Fig. 4, this framework comprises a superluminescent diode (SLD), a K K star coupler, a 1 K coupler, FBG encoders, and FBG decoders. A 1 K coupler is used in this system to divide a broadband light source into K light sources as output to the FBG encoders of each transmitting end. Different grating wavelengths are used for encoding and an AFLVS is employed to modulate the reflected spectral intensity. The message bit reflected by each transmitting end enters the K K star coupler through an optical circulator. The star coupler then accumulates and transmits all message bits
Table 1 Code string for WDM/M-sequence. For SAC, N = 2 and M = 7. l
j
Codeword A(k,l)
0 0 0 0 0 0 0 1 1 1 1 1 1 1
0 1 2 3 4 5 6 0 1 2 3 4 5 6
1 0 1 0 0 1 1 0 0 0 0 0 0 0
1 1 0 1 0 0 1 0 0 0 0 0 0 0
1 1 1 0 1 0 0 0 0 0 0 0 0 0
0 1 1 1 0 1 0 0 0 0 0 0 0 0
0 0 1 1 1 0 1 0 0 0 0 0 0 0
1 0 0 1 1 1 0 0 0 0 0 0 0 0
0 1 0 0 1 1 1 0 0 0 0 0 0 0
0 0 0 0 0 0 0 1 0 1 0 0 1 1
0 0 0 0 0 0 0 1 1 0 1 0 0 1
0 0 0 0 0 0 0 1 1 1 0 1 0 0
0 0 0 0 0 0 0 0 1 1 1 0 1 0
0 0 0 0 0 0 0 0 0 1 1 1 0 1
0 0 0 0 0 0 0 1 0 0 1 1 1 0
0 0 0 0 0 0 0 0 1 0 0 1 1 1
630
S.-P. Tseng et al. / Optical Fiber Technology 19 (2013) 627–637
Vibration 2
1
FBG Encoder#1 FBG Decoder#1 Oscilloscope Vibration SLD
FFT
1
1×3 Coupler
3×3 Star coupler
FBG Encoder#2 FBG Decoder#2 Vibration 2
1
FBG Encoder#3 FBG Decoder#3 Fig. 4. The all fiber loop vibration sensor system based on OCDMA technique.
outputted from the transmitting ends to each receiving end. The receiving ends then use FBG decoders to access the message bits of the correct encoder. The spectrum uses the fiber vibration sensor to facilitate light intensity variations in the reflected wavelengths, which are then converted into photoelectrical signals through a BPD. A digitizing oscilloscope (DOS) is subsequently used to present the frequency of each FBG encoder. Based on the encoders (Fig. 4) an AFLVS was placed between the optical circulator and the multiple FBG arrays, which were designed according to the system SAC codeword. When receiving stimulation from external physical parameters, the AFVLS conducts light intensity modulations on the reflected spectra of the multiple FBG arrays. These modulations are reflected to the optical circulator, and sensor code signals with vibration messages are outputted to the K K star coupler, which accumulates and then evenly transmits all signals outputted by the FBG encoder to the FBG decoder. However, the axial-strain controls of FBG encoders and decoders are used in each codec to compensate for temperature sensitivity [10]. The FBG wavelength of the FBG encoder is encoded using 0- or 1-bits, where 1 represents the existence of an FBG reflected wavelength and 0 indicates no FBG reflected wavelength. Therefore, this study adopted a 0- and 1-bit encoding technique for the FBG encoder’s reflected wavelength to design the FBG encoder. The SAC codeword for user #(k, l) is Ak,l, and Ik,l(t) is set as the light intensity variation after the Ak,l spectrum is modulated through the AFLVS by the vibration source; thus, the following equation can be inferred:
Hk;l ðtÞ ¼ Ak;l Ik;l ðtÞ ¼ ½X k El Ik;l ðtÞ
ð7Þ
where Hk,l(t) represents the SAC codeword for user #(k, l) under the interference of the vibration source, that is, the spectrum light intensity change caused by the influence of the vibration signal on the reflected spectrum. In the OCDMA technical framework proposed in this study, a WDM/SAC M-sequence codeword with N = 2 groups and an M = 3
code length was designed (Table 2). A total of three users were in each group. If the SAC codeword A0,0 for user #(0, 0) was [1, 1, 0, 0, 0, 0], the 0- and 1-bits respectively represented the absence and presence of a reflected spectrum after the light source propagated through the multiple FBG arrays. The central wavelengths of each FBG in the multiple FBG arrays differ; therefore, FBG encoders and decoders can be designed for numerous users according to different wavelengths. Fig. 5 shows the corresponding FBG reflected spectrum of the FBG encoder for each user of the WDM/SAC system; these correspond with the SAC codewords shown in Table 2. The FBG decoder at the receiving end is composed of two optical circulators, two sets of multiple FBG arrays, and a BPD, as shown in the decoders of Fig. 4. The FBGs of the FBG decoder were designed in line with the corresponding FBG encoder, accessing messages from this encoder. It should be noted that the FBG decoders are reversed compared with their matching encoders, and the delay lines are placed in the lower arms of the all decoders. This simultaneously eliminates the round-trip time of the encoders and enables obtaining the correct vibration signal. BPDs were used for correlation detection. The multiple access interference (MAI) from other users is eliminated by subtracting any two given SAC codewords, thereby decreasing the bit error rate of the access signal. The BPDs then convert the decoded signals of the FBG decoder into electrical signals, which are connected to the DOS to detect the expected
Table 2 WDM/SAC codeword; N = 2 and M = 3. l
k
Codeword Ak,l
0 0 0 1 1 1
0 1 2 0 1 2
110000 011000 101000 000110 000011 000101
631
S.-P. Tseng et al. / Optical Fiber Technology 19 (2013) 627–637 ð1Þ
A0,0
A0,1
A1,0
A1,1
and Hk;l ðtÞ respectively represent the spectral intensity distributions for the upper and lower arms of the BPD. Assuming that the signal of user #(k, l) is the expected signal, whereas that for user #(r, s) is not expected, the correlation funcðdÞ ðdÞ tion RAA for Ak;l and Ar,s is defined as follows:
A2,0
A2,1
RAA ðk; l; r; sÞ ¼
ðdÞ
M 1X N1 X
ðdÞ
ak;l ðj M þ iÞar;s ðj M þ iÞ; ðd ¼ 0; 1Þ
ð12Þ
i¼0 j¼0
Fig. 5. The spectrograms of the WDM/SAC codewords in Table 2.
frequency signal. The objective of this process is to use a multiplexing signal for distinguishing the encoded signals of users in the different groups. First, the FBG decoder receives the incident light of the K K star coupler, which includes all output signals of the FBG encoder. When the incident light of the K K star coupler is transmitted to ð0Þ the complementary multiple FBG array address Ak;l through the first optical circulator, it is reflected to the lower arm of the BPD. The remaining spectrum is transmitted to the multiple FBG array ð1Þ address Ak;l through the second optical circulator, and is then reflected to the upper arm of the BPD. Two diodes of the BPD conduct correlation detections after receiving these two complementary light signals. When the sensor encoding signal of FBG Encoder #2 propagates through FBG Decoder #1, the mismatch between the FBG reflected wavelengths causes FBG Decoder #1 to not receive the signal from FBG Encoder #2. Therefore, of the two multiple FBG arrays for the ð0Þ FBG decoder, the complementary multiple FBG array address Ak;l is used to reflect FBG wavelength interference from other users. The ð1Þ multiple FBG array address Ak;l is identical to the multiple FBG arrays of the encoder; thus, it can reflect its own FBG wavelengths, and light signal subtractions or deductions can be conducted using BPDs. This mechanism eliminates MAI caused by other users of the system, thereby increasing the SNR. MAI is generally directly proportional to the number of system users. Furthermore, BPDs are typically used in the laboratory for increasing the SNR of a detected signal and for signal amplification. The operating principle of a BPD involves subtracting the photoelectric currents received by two diodes to eliminate laser or common mode noise. During the decoding process, the signals for two given FBG decoders are complementary. This mechanism enables subtractions during the process of photodetection to obtain the desired signal. For example, regarding user #(k, 1), the decoded vectors for two complementary multiple FBG arrays can be presented as follows: ð0Þ Ak;l
¼ X k El
ð8Þ
Ak;l ¼ X k El
ð9Þ
ð1Þ
ð0Þ Ak;l
is set as the decoded signal detected by the lower arm of the ð1Þ BPD and Ak;l is set as the decoded signal detected by the upper arm of the BPD, where the SAC codewords for X k and Xk are complementary. These two light signals then undergo decoding through the BPD to acquire the expected signal. ð0Þ ð1Þ According to Ak;l and Ak;l of (8) and (9), respectively, the spectral intensity distributions for the FBG decoder before the signal enters the upper and lower arms of the BPD are inferred as follows: ð0Þ
ð0Þ
ð10Þ
ð1Þ
ð1Þ
ð11Þ
Hk;l ðtÞ ¼ Ak;l Ik;l ðtÞ ¼ ½X k El Ik;l ðtÞ Hk;l ðtÞ ¼ Ak;l Ik;l ðtÞ ¼ ½X k El Ik;l ðtÞ
where Ik,1(t) refers to the light intensity variations of the sensor for ð0Þ user #(k, l), which are influenced by the vibration source, and Hk;l ðtÞ
where d = 0, 1 represents the correlation functions for the user ðdÞ entering the upper or lower arm of the BPD and ak;l ðj M þ iÞ and ðdÞ ar,s(j ⁄ M + i) respectively indicate the ðj M þ iÞth element for ak;l and ar;s : i ¼ 0 . . . ; M 1. denotes the SAC codeword with a code length of M; j = 0, . . ., N – 1 expresses the groups. Based on this equation, the properties of the correlation functions for the upper and lower arms of the BPD are as follows:
8 k ¼ r; l ¼ s > <0 ðdÞ RAA ðk; l; r; sÞ ¼ ðM þ 1Þ=4 k – r; l ¼ s > : 0 l–s
ð13Þ
where (M + 1)/4 represents the correlation function value of the lower arm of the BPD when k – r (different users) and l = s (same group), and when k = r, l = s, or l – s, the correlation function value is zero.
8 > < ðM þ 1Þ=2 k ¼ r; l ¼ s ð1Þ RAA ðk; l; r; sÞ ¼ ðM þ 1Þ=4 k – r; l ¼ s > : 0 l–s
ð14Þ
where (M + 1)/2 represents the correlation function value of the upper arm of the BPD when k = r and l = s; (M + 1)/4 expresses the correlation function value of the lower arm of the BPD when k – r and l = s; and when l – s, the correlation function value is zero. When these two light signals enter the BPD for subtraction, the decoded signal can be expressed as follows: ð1Þ
ð0Þ
RAA ðk; l; r; sÞ RAA ðk; l; r; sÞ ¼
ðM þ 1Þ=2 k ¼ r; l ¼ s 0
otherwise
ð15Þ
where (M + 1)/2 represents the energy decoded through the BPD, and 0 represents no decoded energy. In experiment for this study, when N = 2 and M = 3, the correlaðdÞ tion function RAA ðk; l; r; sÞ for the two WDM/SAC decoded signals before entering the BPD is as follows:
8 > < 0 k ¼ r; l ¼ s ð0Þ RAA ðk; l; r; sÞ ¼ 1 k – r; l ¼ s > : 0 l–s
ð16Þ
where 1 represents the correlation function value entering the lower arm of the BPD when k – r and l = s, and l – s indicates that the correlation function value is zero.
8 > < 2 k ¼ r; l ¼ s ð1Þ RAA ðk; l; r; sÞ ¼ 1 k – r; l ¼ s > : 0 l–s
ð17Þ
where 2 represents the correlation function value of the upper arm of the BPD when k = r and l = s; 1 denotes the correlation function value entering the lower arm of the BPD when k – r and l = s; and l – s indicates that the correlation function value is zero. When these two signals enter the BPD, the decoded signal output can be written as follows: ð1Þ
ð0Þ
RAA ðk; l; r; sÞ RAA ðk; l; r; sÞ ¼
2 k ¼ r; l ¼ s 0
otherwise
ð18Þ
632
S.-P. Tseng et al. / Optical Fiber Technology 19 (2013) 627–637
Fig. 6. (a) The reflected spectra for FBG Encoder #1 and (b) the reflected spectra for FBG Encoder #2.
where 2 represents the decoded energy of the BPD for the matching signal output when k = r and l = s; 0 indicates no decoded energy output of the BPD for mismatching signals. When (17) and (18) are modulated by the optical power Ik,l(t) of the AFLVS, the decoded energy of the upper and lower arms of the BPD can be inferred as follows: ð1Þ
ð1Þ
Pk;l ðtÞ ¼ RAA ðk; l; r; sÞ Ik;l ðtÞ ¼
M1 N1 XX
ð1Þ
ak;l ðj M þ iÞar;s ðj M þ iÞ Ik;l ðtÞ
ð19Þ
i¼0 j¼0
ð0Þ
ð0Þ
Pk;l ðtÞ ¼ RAA ðk; l; r; sÞ Ik;l ðtÞ ¼
M1 N1 XX
ð0Þ
ak;l ðj M þ iÞar;s ðj M þ iÞ Ik;l ðtÞ
ð20Þ
i¼0 j¼0 ð1Þ
ð0Þ
When optical powers P k;l ðtÞ and P k;l ðtÞ, which are reflected by the decoder, enter the BPD, the BPD conducts a correlation operað1Þ ð0Þ tion Pk;l ðtÞ Pk;l ðtÞ during the decoding process. When the decoder address and the encoded signal match, a light intensity variation 2Ik,1 for the code is created through modulation, and when the decoder address and encoded signal mismatch, the BPD yields no light intensity output.
The WDM/SAC codeword for Group 1 in Table 2 was adopted in the experiment for this study to prove that the operating principles of the BPD can effectively eliminate MAI and increase the SNR for signal accessing. When the transmitting ends of users #(0, 0) and #(1, 0) simultaneously transmit message bits to the star coupler, and user #(2, 0) does not transmit message bits, the star coupler accumulates the modulated light intensity of all users, which is written as follows: (I0,0(t), I0,0(t) + I1,0(t), I1,0(t), 0, 0, 0). The receiving end of user #(0, 0) reflects two complementary and modulated ð1Þ ð0Þ light signals (i.e., P 0;0 ðtÞ and P 0;0 ðtÞ) to the BPD when receiving light intensity signals from the star coupler. The corresponding spectral light intensities are (I0,0(t), I0,0(t) + I2,0(t), 0, 0, 0, 0) and (0, 0, I2,0(t), 0, 0, 0), respectively. Therefore, the light intensity for these two light signals upon ð1Þ ð0Þ entering the BPD is P0;0 ðtÞ P0;0 ðtÞ ¼ 2I0;0 ðtÞ, indicating that the correct message bits can be accessed and inputted into the DOS to present a fast Fourier transform (FFT) spectrum, thereby enabling observations of the modulated frequency. This study examined whether message bits from other users could be eliminated at the receiving end for user #(2, 0) without light signal output, under the premise that no message bits were transmitted by user #(2, 0). Similarly, when the receiving end of user #(2, 0) receives light signals from the star coupler, two comð1Þ ð0Þ plementary and modulated light signals (i.e., P 2;0 ðtÞ and P2;0 ðtÞ)
Fig. 7. Output spectrum from star coupler after summing all FBG encoders. (a) The optical spectra of output port 1 in star coupler and (b) the optical spectra of output port 2 in star coupler.
S.-P. Tseng et al. / Optical Fiber Technology 19 (2013) 627–637
633
Fig. 8. (a) Reflected spectrum for the lower arm of the FBG Decoder #1 and (b) reflected spectra for the upper arm of the FBG Decoder #1.
are reflected and subsequently entered into the BPD. The corresponding spectral light intensities for these two light signals are (I0,0(t), 0, I2,0(t), 0, 0, 0) and (0, I0,0(t) + I2,0(t), 0, 0, 0, 0), respectively. Thus, the light intensity of these two light signals when entering ð1Þ ð0Þ the BPD is P2;0 ðtÞ P2;0 ðtÞ ¼ 0, indicating that the message bits of other users can be eliminated. Because the FBG decoder accesses no modulated light signals, the BPD does not demodulate any signals. When the AFLVS is placed between the optical circulator of the FBG encoder and the multiple FBG arrays, the light source is inputted into the FBG encoder through the optical circulator; after entering the multiple FBG arrays through the AFLVS, the FBG encoded wavelength is reflected to the AFLVS. Because of external effects, the AFLVS modulates the reflected wavelength of the multiple FBG arrays, which is then entered into the star coupler through the output end of the optical circulator and subsequently inputted into the FBG decoder. The light signal output of the FBG decoder is converted into an electrical signal through the BPD, and the vibration frequency of the test object is acquired through an FFT conducted by the DOS. 3. Experimental framework and results The objective of this experiment was to use the spectral amplitude coding-optical code division multiple access (SAC-OCDMA) system to access the vibration frequency of the stepping motor
(sensed by the AFLVS). Furthermore, this study endeavored to prove that the correct vibration frequency for the two sets of encoders could be accessed by their corresponding FBG decoders without receiving mutual interference when frequency was simultaneously inputted into the stepping motor. For the SLD model, NXTAR SLD-2000 was adopted in this study to provide a broadband light source with stable power for the system. Moreover, dual window wideband couplers (Fiber Optic Communications, Inc., Hsinchu, Taiwan) were employed as the 1 2 and 2 2 couplers of the study experiment. The Anritsu MS9710C optical spectrum analyzer was used to monitor the accuracy of the encoder/decoder spectral output. The BPD model employed for the decoding end was Model-1817 (New Focus Inc., California, USA), which conducted light signal subtractions in the optical domain and subsequently converted the results into electrical signals. The ±15 V current-limited power supply (New Focus Inc., California, USA) was then used to provide power to the BPD. In addition, the Tektronix TDS2102B digital oscilloscope was employed to conduct a Fourier transform on the electrical signal output of the BPD. Finally, the Twintex TFG 3510 signal generator simultaneously provided the vibration frequency for the two stepping motors. The FBG resonance wavelengths used for the FBG encoder/decoder in this study were 1548, 1551, and 1554 nm, representing k1, k2, and k3, respectively. These wavelengths corresponded to binary sequences of zeros and ones. To verify that the OCDMA system used in this study could eliminate the crosstalk from other users
Fig. 9. (a) Reflected spectrum for the lower arm of the FBG Decoder #2 and (b) reflected spectra for the upper arm of the FBG Decoder #2.
634
S.-P. Tseng et al. / Optical Fiber Technology 19 (2013) 627–637
and further access accurate vibration frequencies, the researchers hypothesized the following: (1) the SAC codewords for three users in the same group are A0,0, A1,0, and A2,0; (2) the corresponding binary sequence A0,0 = (1, 1, 0, 0, 0, 0) when the FBG central wavelength of FBG Encoder #1 for User #(0, 0) is ðk1 ; k2 Þ; (3), the corresponding binary sequence A1,0 = (0, 1, 1, 0, 0, 0) when the FBG central wavelength of FBG Encoder #2 for User #(1, 0) is (k2, k3); and (4) the corresponding binary sequence A0,3 = (1, 0, 1, 0, 0, 0) when the FBG central wavelength of FBG Encoder #3 for User #(2, 0) is ðk1 ; k3 Þ. For the study experiment, only the SAC codewords for users #(0, 0) and #(1, 0) were referenced to design the FBG encoder and decoder. A stepping motor was used as the vibration source for this experiment (Fig. 4) and was directly placed on the AFLVS of each FBG encoder, and various frequencies were inputted to observe whether this framework could eliminate crosstalk from other users and further demodulate the correct vibration frequency. Fig. 6(a) shows the measured reflected spectra (k1, k2) of FBG Encoder #1. The represented SAC codeword was (1, 1, 0); the corresponding central wavelengths were 1547.656 and 1550.648 nm, respectively; and the light intensities were 34.77 and 33.96 dB m, respectively. Fig. 6(b) shows the measured reflected spectra ðk2 ; k3 Þ of FBG Encoder #1. The represented SAC codeword was (0, 1, 1); the corresponding central wavelengths were 1550.664 and 1553.704 nm, respectively; and the light intensities were 32.96 and 33.78 dB m, respectively. Fig. 7 indicates the output reflected spectra for FBG Encoders #1 and #2. After propagating through the star coupler, the reflected spectra for all FBG encoders were accumulated and transmitted to FBG Decoders #1 and #2. Fig. 7(a) illustrates the reflected spectra ðk1 ; k2 ; k3 Þ that entered FBG Decoder #1, with a represented SAC codeword of (1, 2, 1); corresponding central wavelengths of
Decoder 1
(a)
Decoder 2
(d)
(b)
1547.656, 1550.664, and 1553.704 nm, respectively; and light intensities of 37.78, 33.76, and 37.07 dB m, respectively. Fig. 7(b) illustrates the reflected spectra ðk1 ; k2 ; k3 Þ that entered FBG Decoder #2, with a represented SAC codeword of (1, 2, 1); corresponding central wavelengths of 1547.656, 1550.664, and 1553.704 nm, respectively; and light intensities of 37.78, 33.76, and 37.07 dB m, respectively. The degree of outputted light intensity from the star coupler shows an approximately 3dB loss for the reflected spectra. Because of the star coupler, the wavelength 1550.664 nm created a superposition effect for the two spectra. Therefore, the wavelength for k2 was approximately 4 dB greater that for k1 and k3 ; that is, the light intensity was double the other wavelengths. When the light signals, which were accumulated by the star coupler, entered FBG Decoder #1, the reflected spectrum (k3 ) entered the lower arm of the BPD [Fig. 8(a)], with a represented codeð0Þ word A0;0 of (0, 0, 1), a corresponding central wavelength of 1553.688 nm, and a light intensity of 39.67 dB m. Fig. 8(b) shows the reflected spectra ðk1 ; k2 Þ of FBG Decoder #1 entering the upper ð1Þ arm of BPD, with a represented SAC codeword A0;0 of (1, 1, 0); corresponding central wavelengths of 1547.64 and 1553.688 nm, respectively; and light intensities of 41.29 and 37.04 dB m, respectively. When the light signals, which were accumulated by the star coupler, entered FBG Decoder #2, the reflected spectrum ðk1 Þ entered the lower arm of the BPD [Fig. 9(a)], with a represented codeð0Þ word A0;0 of (1, 0, 0), a corresponding central wavelength of 1547.656 nm, and a light intensity of 39.76 dB m. Fig. 9(b) shows the reflected spectra ðk1 ; k2 Þ of FBG Decoder #2 entering the upper ð1Þ arm of BPD, with a represented SAC codeword A0;0 of (0, 1, 1); corresponding central wavelengths of 1550.632 and 1553.688 nm, respectively; and light intensities of 36.48 and 40.36 dB m,
Decoder 1
(c)
Decoder 2
Decoder 2
(e)
(f)
Fig. 10. Decoding results for inputting a vibration signal of 0, 50, and 110 Hz into Encoder #1 and a fixed 90 Hz vibration signal for Encoder #2. (a) Decoded signal for Decoder #1 under a 0 Hz vibration signal input for Encoder #1 and a 90 Hz input for Encoder #2. (b) Decoded signal for Decoder #1 under a 50 Hz input for Encoder #1 and a 90 Hz input for Encoder #2. (c) Decoded signal for Decoder #1 under a 110 Hz input for Encoder #1 and a 90 Hz input for Encoder #2. (d) Decoded signal for Decoder #2 under a 0 Hz vibration signal input for Encoder #1 and a 90 Hz input for Encoder #2. (e) Decoded signal for Decoder #2 under a 50 Hz input for Encoder #1 and a 90 Hz input for Encoder #2. (f) Decoded signal for Decoder #2 under a 110 Hz input for Encoder #1 and a 90 Hz input for Encoder #2.
635
S.-P. Tseng et al. / Optical Fiber Technology 19 (2013) 627–637
Decoder 1
Decoder 1
(a)
(b)
Decoder 2
(d)
Decoder 2
(e)
(c)
Decoder 2
(f)
Fig. 11. Decoding results for inputting a vibration signal of 0, 50, and 110 Hz into Encoder #1 and a fixed 190 Hz vibration signal for Encoder #2. (a) Decoded signal for Decoder #1 under a 0 Hz vibration signal input for Encoder #1 and a 190 Hz input for Encoder #2. (b) Decoded signal for Decoder #1 under a 50 Hz input for Encoder #1 and a 190 Hz input for Encoder #2. (c) Decoded signal for Decoder #1 under a 110 Hz input for Encoder #1 and a 190 Hz input for Encoder #2. (d) Decoded signal for Decoder #2 under a 0 Hz vibration signal input for Encoder #1 and a 190 Hz input for Encoder #2. (e) Decoded signal for Decoder #2 under a 50 Hz input for Encoder #1 and a 190 Hz input for Encoder #2. (f) Decoded signal for Decoder #2 under a 110 Hz input for Encoder #1 and a 190 Hz input for Encoder #2.
Decoder 1
(a)
Decoder 1
(b)
Decoder 2
(d)
(e)
Decoder 1
(c)
Decoder 2
(f)
Fig. 12. Decoding results for inputting a fixed 90 Hz vibration signal for Encoder #1 and a vibration signal of 0, 50, and 110 Hz into Encoder #2. (a) Decoded signal for Decoder #1 under a 0 Hz vibration signal input for Encoder #2 and a 90 Hz input for Encoder #1. (b) Decoded signal for Decoder #1 under a 50 Hz input for Encoder #2 and a 90 Hz input for Encoder #1. (c) Decoded signal for Decoder #1 under a 110 Hz input for Encoder #2 and a 90 Hz input for Encoder #1. (d) Decoded signal for Decoder #2 under a 0 Hz vibration signal input for Encoder #2 and a 90 Hz input for Encoder #1. (e) Decoded signal for Decoder #2 under a 50 Hz input for Encoder #2 and a 90 Hz input for Encoder #1. (f) Decoded signal for Decoder #2 under a 110 Hz input for Encoder #2 and a 90 Hz input for Encoder #1.
636
S.-P. Tseng et al. / Optical Fiber Technology 19 (2013) 627–637
Decoder 1
(a)
Decoder 1
(b) Decoder 2
(d)
(e)
Decoder 1
(c)
Decoder 2
(f)
Fig. 13. Decoding results for inputting a fixed 190 Hz vibration signal for Encoder #1 and a vibration signal of 0, 50, and 110 Hz into Encoder #2. (a) Decoded signal for Decoder #1 under a 0 Hz vibration signal input for Encoder #2 and a 190 Hz input for Encoder #1. (b) Decoded signal for Decoder #1 under a 50 Hz vibration signal input for Encoder #2 and a 190 Hz input for Encoder #1. (c) Decoded signal for Decoder #1 under a 110 Hz vibration signal input for Encoder #2 and a 190 Hz input for Encoder #1. (d) Decoded signal for Decoder #2 under a 0 Hz vibration signal input for Encoder #2 and a 190 Hz input for Encoder #1. (e) Decoded signal for Decoder #2 under a 50 Hz vibration signal input for Encoder #2 and a 190 Hz input for Encoder #1. (f) Decoded signal for Decoder #2 under a 110 Hz vibration signal input for Encoder #2 and a 190 Hz input for Encoder #1.
Piezoelectric accelerometer 50Hz
(a)
Piezoelectric accelerometer 170Hz
(d)
Piezoelectric accelerometer 90Hz
(b)
Piezoelectric accelerometer 110Hz
(c)
Piezoelectric accelerometer 190Hz
(e)
Fig. 14. Results for using the piezoelectric accelerometer to measure the stepping motor. Measurement results under a vibration frequency for the stepping motor (a) 50 Hz, (b) 90 Hz, (c) 110 Hz, (d) 170 Hz and (e) 190 Hz.
S.-P. Tseng et al. / Optical Fiber Technology 19 (2013) 627–637
respectively. Finally, the BPD was used to convert the reflected spectra into light signals, and the digital oscilloscope or DOS was employed to access the vibration frequencies for FBG Encoders #1 and #2. In Experiment 1, vibration frequencies of 0, 50, and 110 Hz were inputted into FBG Encoder #1, and a fixed 90 Hz vibration frequency was used for FBG Encoder #2. Fig. 10(a–d) shows the FFT spectrograms for using FBG Decoder #1 to access the vibration frequency of the stepping motor, and Fig. 10(e and f) shows the FFT spectrograms for using FBG Decoder #2 to access the vibration frequency of the stepping motor. Subsequently, in Experiment 2, vibration frequencies of 0, 50, and 110 Hz were inputted into FBG Encoder #1 and a fixed 190 Hz was used for FBG Encoder #2. Fig. 11(a–d) shows the FFT spectrograms for using FBG Decoder #1 to access the vibration frequency of the stepping motor, and Fig. 11(e and f) shows the FFT spectrograms for using FBG Decoder #2 to access the vibration frequency of the stepping motor. Experiment 3 involved using a fixed vibration frequency for Encoder #1 and sequentially changing the vibration frequency input for Encoder #2. A fixed vibration frequency of 90 Hz and vibration frequencies of 0, 50, and 110 Hz were inputted into FBG Encoders #1 and #2, respectively. Fig. 12(a–d) and (e and f) respectively shows the FFT spectrograms of the stepping motor vibration frequency accessed using FBG Decoders #1 and #2. Finally, Experiment 4 involved inputting a fixed vibration frequency of 190 Hz and vibration frequencies of 0, 50, and 110 Hz into FBG Encoders #1 and #2, respectively. Fig. 13(a–d) and (e and f) respectively shows the FFT spectrograms of the stepping motor vibration frequency accessed using FBG Decoders #1 and #2. The experimental results indicate that when the digital oscilloscope or DOS was used to measure and obtain the FFT spectrograms, a corresponding frequency peak value could be observed in the FFT spectrogram when the signal generator inputted the frequency into the stepping motor. Although crosstalk from other encoders was conspicuous in the FFT spectrograms, the existence of an input frequency from other FBG encoders that possessed a level approximate to that for noise was verified. This does not affect the vibration frequency required for access. Thus, the experiment results prove that the experimental settings and arrangement can effectively eliminate MAI caused by other FBG encoders. Regarding the piezoelectric accelerometer, the FFT spectrograms for the stepping motor under a frequency of 50, 90, 110, 170, and 190 Hz were measured, and almost no frequency was detected when the frequency was lower than 90 Hz (Fig. 14), indicating that the piezoelectric accelerometer exhibited measurement difficulties at a low frequency (lower than 90 Hz). 4. Conclusion The AFLVS adopted in this study was constructed using optical fibers; thus, it possessed attributes such as small size, light weight,
637
and resistance to EMI. Consequently, this sensor is an ideal replacement for traditional electric or mechanical sensors. In this study, a combination of the SAC-OCDMA system and an AFLVS was used to conduct multiplex measurements at multiple points. Because this technique combines SAC and WDM, which exhibit excellent orthogonality, and uses FBG as the encoder/decoder, users are able to exclude MAI from other users, thereby increasing the SNR. In addition, the number of users increases with code length; therefore, more users are supported and designs are more flexible. The experimental results indicate that the various FBG decoders are capable of correctly accessing the vibration frequency of the corresponding FBG encoder sensor without experiencing MAI from other FBG encoders. In other words, the occurrence of crosstalk generated by other FBG encoders is avoided in the FFT spectrogram, thereby preventing vibration frequency reading errors. The piezoelectric accelerometer measurement results for the vibration frequency of the stepping motor indicate that because of size limitations, the detected signals at a frequency lower than 90 Hz were less than ideal. Consequently, large-scale network establishments such as that for OFSs cannot be implemented, and signals must be transmitted through cables; hence, applications involving large-scale vibration sensors are more costly. Acknowledgments The current authors gratefully acknowledge the financial support provided to this study by the National Science Council, Taiwan under the grant NSC 102-2221-E-150-002 and NSC 102-2221-E239-004. References [1] N.Q. Nguyen, N. Gupta, Power modulation based fiber-optic loop-sensor having a dual measurement range, J. Appl. Phys. 106 (2009) 033502. [2] H. Renner, Bending losses of coated single-mode fibers: a simple approach, IEEE J. Lightw. Technol. 10 (1992) 544–551. [3] P. Wang, Q. Wu, A. Sun, Y. Semenova, G. Farrell, A macrobending fiber based vibration sensor using whispering gallery mode’’, in: Proc. SPIE 7726, Optical Sensing and Detection, 2010. [4] P. Wang, Y. Semenova, Q. Wu, G. Farrell, A macrobending fiber based microdisplacement sensor utilizing whispering-gallery modes, in: Proc. SPIE 7503, 20th International Conference on Optical Fibre Sensors, 2009. [5] L.T. Blair, S.A. Cassidy, Wavelength division multiplexed sensor network using Bragg fibre reflection gratings, Electron. Lett. 28 (1992) 1734–1735. [6] R.S. Weis, A.D. Kersey, T.A. Berkoff, A four-element fiber grating sensor array with phase-sensitive detection, IEEE Photonics Technol. Lett. 6 (1994) 1469– 1472. [7] Y.H. Huang, C. Lum, P.K.A. Wai, H.Y. Tam, Large-scale FBG sensor utilizing code division multiplexing, in: Laser and Electro-Optics, 2008 and 2008 Conference on Quantum Electronics and Laser Science. Conference on CLEO/QELS 2008, 2008, pp. 1–2. [8] H.C. Cheng, C.H. Wu, C.C. Yang, Y.T. Chang, Wavelength division multiplexing/ spectral amplitude coding applications in fiber vibration sensor systems, IEEE Sens. J. 11 (10) (2011) 2518–2526. [9] C.C. Yang, Hybrid wavelength-division-multiplexing/spectral-amplitudecoding optical CDMA system, IEEE Photonics Technol. Lett. 17 (2005) 1343– 1345. [10] K.O. Hill, G. Meltz, Fiber Bragg grating technology fundamentals and overview, J. Lightw. Technol. 15 (8) (1997) 1263–1276.
Contents lists available at ScienceDirect
Optical Fiber Technology www.elsevier.com/locate/yofte
Employing optical code division multiple access technology in the all fiber loop vibration sensor system Shin-Pin Tseng a, Chih-Ta Yen b, Rong-Shun Syu c, Hsu-Chih Cheng c,⇑ a
Department of Electronic Engineering, National United University, Miaoli, Taiwan Department of Electrical Engineering, National Formosa University, Yunlin, Taiwan c Department of Electro-Optical Engineering, National Formosa University, Yunlin, Taiwan b
a r t i c l e
i n f o
Article history: Received 2 April 2013 Revised 19 August 2013 Available online 22 October 2013 Keywords: Optical code division multiple access (OCDMA) Optical fiber bending Optical sensor Spectral amplitude coding (SAC) Vibration frequency measurement
a b s t r a c t This study proposes a spectral amplitude coding-optical code division multiple access (SAC-OCDMA) framework to access the vibration frequency of a test object on the all fiber loop vibration sensor (AFLVS). Each user possesses an individual SAC, and fiber Bragg grating (FBG) encoders/decoders using multiple FBG arrays were adopted, providing excellent orthogonal properties in the frequency domain. The system also mitigates multiple access interference (MAI) among users. When an optical fiber is bent to a point exceeding the critical radius, the fiber loop sensor becomes sensitive to external physical parameters (e.g., temperature, strain, and vibration). The AFLVS involves placing a fiber loop with a specific radius on a designed vibration platform. A 1 K coupler was adopted for the sensor system to divide a broadband light source into K light sources, which were then transmitted to various FBG encoders. The AFLVS was placed between the optical circulators of the various FBG encoders and multiple FBG arrays, and the stepping motor was directly placed on the fiber loops of the various AFLVSs. A signal generator was then used to input different frequencies into the stepping motors of the various sensors. After the light intensity for the reflectance spectrum, which was outputted by the FBG encoder, was modulated by the AFLVS, the modulated reflected signals were outputted to the K K star coupler through the optical circulator and transmitted to the FBG decoders for the users. A balanced photodetector (BPD) was employed in this study to convert the light output of the FBG decoder into an electrical signal, and a digitizing oscilloscope was employed to conduct a Fourier transform on the BPD electrical signal output, thereby acquiring the vibration frequency of the test object. The results of the experiment are compared to a piezoelectric accelerometer. The comparison results indicate that the piezoelectric accelerometer is less sensitive when the frequency is lower than 90 Hz, whereas the AFLVS exhibits excellent measurement results at a low frequency ranging between 50 and 200 Hz. Ó 2013 Elsevier Inc. All rights reserved.
1. Introduction Optical fiber sensors (OFSs) are widely applied in measuring physical parameters, including rotation, acceleration, electromagnetic fields, temperature, pressure, sound, vibration, and humidity. OFSs are also commonly used in civil engineering, environmental and biochemical testing, and clinical biomedical fields. In addition, OFSs are advantageous because of their small size, light weight, low power, high sensitivity, high bandwidth, and ability to resist electromagnetic interference (EMI). Furthermore, OFSs possess the characteristic of light signal transmission, which enables multiplex measurements from multiple points by combining various multiplex network frameworks [e.g., wavelength division multiplexing (WDM), time division
⇑ Corresponding author. E-mail address: [email protected] (H.-C. Cheng). 1068-5200/$ - see front matter Ó 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.yofte.2013.10.001
multiplexing (TDM), frequency division multiplexing (FDM), and optical code division multiple access (OCDMA)]. Traditional electric sensors (e.g., piezoelectric accelerometers), when used for vibration sensing, possess a broad range of working temperatures and can sense or detect higher frequencies. However, a relatively low signal-to-noise-ratio (SNR) is detected when traditional electric sensors are applied for low-frequency access. Moreover, these sensors are easily subjected to EMI; thus, they can only be applied to single-point simplex measurements. The sensitivity of the fiber loop sensor for external physical parameters is increased when the optical fiber is bent to a point exceeding the critical radius, and the sensor can be used to measure temperature, the refractive index, and vibration [1,2]. Theoretical and experimental results indicate that attributes regarding the significant influence of optical fiber diameter on bending loss can be used to develop vibration [3] and shift [4] sensors.
628
S.-P. Tseng et al. / Optical Fiber Technology 19 (2013) 627–637
OFSs can be integrated with multiplex techniques to achieve distributed sensing and decrease the cost and complexity of the multiple-point measurement system. OFSs are commonly applied to optical remote sensing (e.g., strain, temperature, and pressure sensing). Among various multiplex techniques, WDM is most commonly employed, and is a simple technique used to identify multiple OFSs. The number of OFSs applicable for this multiplexing system is limited by the light source bandwidth and fiber Bragg grating (FBG) reflected wavelength intervals [5]. TDM is another common technique, which uses a pulsed laser light source to emit pulses to FBG sensors at various distances. The reflected wavelengths of various FBG sensors differ, and the minimum distance for different FBG sensors is determined based on the time slot width of the light source. The pulse bandwidth must be equivalent to or less than the round-trip time for two random FBG sensors. The reflected wavelengths of various FBG sensors are separated according to the delay time of the optical fiber, and a rapid timedomain signal analysis is then employed to identify various sensor signals. However, significant power is required so that short pulses from light sources can increase SNR values for photodetector output signals [6]. To improve deficiencies in the TDM, the OCDMA framework can be used to implement additional FBG sensors by transmitting or emitting a modulated broadband light source to the FBG sensor arrays through an optical circulator. Based on the FBG encoding pattern, this light pulse possesses a unique sequence comprising zeros and ones. The reflected pulse sequence propagates through the detection system comprising the edge filter, balanced photodetectors (BPDs), and electrical signal processing units. The optical fiber delay line performs spatial modulations of the pulse sequence according to different sensor locations. Codes with various delay times can be used to calculate the autocorrelation function of the reflected signal while the wavelength messages of all sensors are demodulated. The edge filter is used as a passive wavelength demodulation device by converting wavelength variations into amplitude variations. Two photodetectors are employed to calculate the correlation function ratio of the wavelength shift in the FBG sensor. The analog-to-digital converter is then applied for sampling received signals [7]. In 2011, the research team for the present study proposed a fiber vibration sensor system combining an OCDMA framework and two optical collimators. This sensor system was capable of accurately measuring the vibration frequency of an object, and was successfully applied to complete a code division access experiment [8]. However, the framework employed in [8] was primarily applied to short-distance transmissions. Consequently, this study also adopted a balanced incomplete block design (BIBD) [9] in which each user group had a specific spectral amplitude coding (SAC). FBGs were used as a primary component for the encoder/decoder for each user, thereby decreasing the number of FBGs in these devices. In this way the encoder/ decoder was simplified and relevant costs reduced. SAC encoding can be used by a great number of active users because of its characteristic of eliminating interference and preventing phase-induced intensity noise during the photodetection process. The excellent characteristics of this system framework can be adopted for multiple-point multiplex measurements by using a distributed vibration sensing system combined with a sensor and encoding/ decoding device. For the vibration sensor, this study proposed using an all fiber loop vibration sensor (AFLVS) to replace the fiber vibration sensor consisting of two optical collimators. This improvement transforms the entire sensor framework into an economical, simple, and flexible all fiber optical sensor design, facilitating the applicability of the optical sensor system for various situations in which vibration measurements are required.
2. Theories and principles Optical fiber bending introduces considerable problems in fiberoptic communication, such as optical power loss and signal interference. The bending loss theory for the fundamental mode of fibers has been extensively researched for decades. The effects of microbending and macrobending are applied to OFSs, including effects for shift, pressure, and temperature sensors. Recent research has focused on relevant theories regarding macrobending loss characteristics for mono-mode fibers with a bend radius between 8.5 and 12 mm, which are used for edge filter wavelength measurements. Furthermore, when the bend radius of the optical fiber is less than 1 cm, the fiber can be applied to sensing. Fiber loop sensors are sensitive to external physical parameters when the optical fiber is bent to a point smaller than the critical radius; they can be used to measure temperature, strain, and vibration. Because of their unique characteristics, fiber loop sensors can measure the vibration frequency of mechanical equipment, and since these sensors require minimal optical power, potential fire risk is low. Fiber loop sensors can also be applied to measure high- and low-frequency vibrations as well as to detect shift by the millimeter. Numerous existing physical models have been employed to predict optical waveguide bending loss; the most common method adopted in this process is the bending loss equation [1,2]:
! 1=2 1 p j2 2c3 Re 2aB ¼ exp 2 2 c3 Re 3b20 V 2 k1 ðcaÞ 1=2
c ¼ ðb0 k2 n2cl Þ
j ¼ ðk2 n2co b20 Þ
1=2
V ¼ akðn2co n2cl Þ k ¼ 2p=k
1=2
ð1Þ
ð2Þ ð3Þ ð4Þ ð5Þ
2aB represents the bending loss coefficient per unit length, where c, k, and V denote the propagation constant; b0 indicates the propagation constant for the leaky mode of the straight optical fiber; a symbolizes the core radius; Re expresses the effective bend radius; nco and ncl respectively refer to the refractive indices of the core and cladding; and K1(ca) is equivalent to the deformed first order-Bessel function of the second kind. Figs. 1 and 2 respectively show the AFLVS structure and the practical AFLVS. A plastic card or tube is used to bend the optical fiber and construct a fiber loop sensor. Two wooden blocks are placed at an appropriate distance on a rubber sheet, and sheet metal is placed atop the two wooden blocks. The loop sensor is subsequently attached to the sheet metal, and the sensing platform for the AFLVS is complete. Fig. 3 shows the power loss of two fiber loop sensors that possess distinct fiber loop diameters. The power loss curve shows numerous intermediate peaks, which are caused when light is reflected from the cladding-coating (jacket) and coating (jacket)-air junction and combines with leaked core light, generating the resonance peaks in the whispering gallery mode [1–3]. Although the two fiber loop sensors are composed of the same single-mode fiber, certain differences exist between their power loss curves (Fig. 3). Whispering gallery mode has been used to develop several sensor applications because of its substantial sensitivity. Fig. 3 shows the maximal slopes of Fiber Loop Sensor 1 from diameters of 1.55 cm to 1.60 cm and Fiber Loop Sensor 2 from diameters of 1.57–1.62 cm, indicating that power within this range is easily affected, and, thus, possesses the highest sensitivity level. Conse-
629
S.-P. Tseng et al. / Optical Fiber Technology 19 (2013) 627–637
Fig. 1. The structure of all fiber loop vibration sensor.
encoding/decoding, as each user’s code array. This consequently increased the amount of access for users of the OCDMA system. Table 1 shows that a total of j users existed in Group 1; thus, the string of the WDM/M-sequence code can be presented as follows [8,9]:
Ak;l ¼ jak;l ðmÞjX k El ¼ ½el;0 X k ; el;1 X k ; . . . ; el;N1 X k
Fig. 2. The practical all fiber loop vibration sensor.
Fig. 3. The power loss curves of the fiber loops for vibration sensors 1 (black line) and 2 (red line). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
quently, the fiber loop diameters for Sensors 1 and 2 were fixed at 1.57 cm and 1.59 cm, respectively. Regarding the OCDMA, the SAC design involves assigning all users into N groups (N represents a positive integer), and each user is referred to as #(k, l). The code string for each user is Ak,l[k = (0, 1, 2, . . ., m 1)], which denotes the number of the Kth user. l = (0, 1, 2, . . ., N 1) expresses the Nth group, and the encoding/ decoding for each user constitutes a binary message bit of a zeros or ones. Multiple FBG arrays are developed into an original encoding/decoding device, and the message information of each user is connected to a star coupler. When all of the user signals are transmitted to the star coupler, the accumulated signal is transmitted through the output end, and then enters the decoding device of each user, subsequently demodulating unique encoded message information. A previous study proposed the use of WDM/BIBD encoding in optics communication systems [9]. However, this study employed an M-sequence code, which occupies less system capacity compared to the traditional unipolar code (i.e., code arrays comprising zeros and ones) and reduces the number of multiple FBG arrays in
ð6Þ
where m = (0, 1, . . ., MN 1), and ak,l(m) represents the mth element used by user Ak,l and the corresponding wavelength kmþ1 . Furthermore, X k ¼ ½xk;0 ; xk;1 ; . . . ; xk;m1 ¼ T k ; X0 expresses the SAC of the jth user (T represents the shift of the code spectrum one periodic vector to the right); El ¼ ðel;0 ; el;1 ; . . . ; el;N1 Þ indicates the vector unit transmitted by the N 1th group (l is a binary message bit composed of zeros or ones); symbolizes the Kronecker product; and X0k sðk ¼ 0; . . . ; M 1Þ represents the SAC code with a length of M (code length). WDM/M-sequence encoding enables the application of a single code array for numerous different group blocks, thereby increasing the number of users without using tedious code arrays and providing greater coding flexibility. In this context, the SAC’s greatest advantage is its excellent cyclical characteristics. When the code base for only one group of users is required, the code arrays for all users can be acquired. This method of code design is easily produced and simply structured. In addition, compared to that for the BIBD, the SAC’s greater number of code bases increases user and system capacity when employed for the OCDMA system. This study proposed adopting FBGs in the OCDMA system, and placing an AFLVS between the FBG encoder and optical circulator. As shown in Fig. 4, this framework comprises a superluminescent diode (SLD), a K K star coupler, a 1 K coupler, FBG encoders, and FBG decoders. A 1 K coupler is used in this system to divide a broadband light source into K light sources as output to the FBG encoders of each transmitting end. Different grating wavelengths are used for encoding and an AFLVS is employed to modulate the reflected spectral intensity. The message bit reflected by each transmitting end enters the K K star coupler through an optical circulator. The star coupler then accumulates and transmits all message bits
Table 1 Code string for WDM/M-sequence. For SAC, N = 2 and M = 7. l
j
Codeword A(k,l)
0 0 0 0 0 0 0 1 1 1 1 1 1 1
0 1 2 3 4 5 6 0 1 2 3 4 5 6
1 0 1 0 0 1 1 0 0 0 0 0 0 0
1 1 0 1 0 0 1 0 0 0 0 0 0 0
1 1 1 0 1 0 0 0 0 0 0 0 0 0
0 1 1 1 0 1 0 0 0 0 0 0 0 0
0 0 1 1 1 0 1 0 0 0 0 0 0 0
1 0 0 1 1 1 0 0 0 0 0 0 0 0
0 1 0 0 1 1 1 0 0 0 0 0 0 0
0 0 0 0 0 0 0 1 0 1 0 0 1 1
0 0 0 0 0 0 0 1 1 0 1 0 0 1
0 0 0 0 0 0 0 1 1 1 0 1 0 0
0 0 0 0 0 0 0 0 1 1 1 0 1 0
0 0 0 0 0 0 0 0 0 1 1 1 0 1
0 0 0 0 0 0 0 1 0 0 1 1 1 0
0 0 0 0 0 0 0 0 1 0 0 1 1 1
630
S.-P. Tseng et al. / Optical Fiber Technology 19 (2013) 627–637
Vibration 2
1
FBG Encoder#1 FBG Decoder#1 Oscilloscope Vibration SLD
FFT
1
1×3 Coupler
3×3 Star coupler
FBG Encoder#2 FBG Decoder#2 Vibration 2
1
FBG Encoder#3 FBG Decoder#3 Fig. 4. The all fiber loop vibration sensor system based on OCDMA technique.
outputted from the transmitting ends to each receiving end. The receiving ends then use FBG decoders to access the message bits of the correct encoder. The spectrum uses the fiber vibration sensor to facilitate light intensity variations in the reflected wavelengths, which are then converted into photoelectrical signals through a BPD. A digitizing oscilloscope (DOS) is subsequently used to present the frequency of each FBG encoder. Based on the encoders (Fig. 4) an AFLVS was placed between the optical circulator and the multiple FBG arrays, which were designed according to the system SAC codeword. When receiving stimulation from external physical parameters, the AFVLS conducts light intensity modulations on the reflected spectra of the multiple FBG arrays. These modulations are reflected to the optical circulator, and sensor code signals with vibration messages are outputted to the K K star coupler, which accumulates and then evenly transmits all signals outputted by the FBG encoder to the FBG decoder. However, the axial-strain controls of FBG encoders and decoders are used in each codec to compensate for temperature sensitivity [10]. The FBG wavelength of the FBG encoder is encoded using 0- or 1-bits, where 1 represents the existence of an FBG reflected wavelength and 0 indicates no FBG reflected wavelength. Therefore, this study adopted a 0- and 1-bit encoding technique for the FBG encoder’s reflected wavelength to design the FBG encoder. The SAC codeword for user #(k, l) is Ak,l, and Ik,l(t) is set as the light intensity variation after the Ak,l spectrum is modulated through the AFLVS by the vibration source; thus, the following equation can be inferred:
Hk;l ðtÞ ¼ Ak;l Ik;l ðtÞ ¼ ½X k El Ik;l ðtÞ
ð7Þ
where Hk,l(t) represents the SAC codeword for user #(k, l) under the interference of the vibration source, that is, the spectrum light intensity change caused by the influence of the vibration signal on the reflected spectrum. In the OCDMA technical framework proposed in this study, a WDM/SAC M-sequence codeword with N = 2 groups and an M = 3
code length was designed (Table 2). A total of three users were in each group. If the SAC codeword A0,0 for user #(0, 0) was [1, 1, 0, 0, 0, 0], the 0- and 1-bits respectively represented the absence and presence of a reflected spectrum after the light source propagated through the multiple FBG arrays. The central wavelengths of each FBG in the multiple FBG arrays differ; therefore, FBG encoders and decoders can be designed for numerous users according to different wavelengths. Fig. 5 shows the corresponding FBG reflected spectrum of the FBG encoder for each user of the WDM/SAC system; these correspond with the SAC codewords shown in Table 2. The FBG decoder at the receiving end is composed of two optical circulators, two sets of multiple FBG arrays, and a BPD, as shown in the decoders of Fig. 4. The FBGs of the FBG decoder were designed in line with the corresponding FBG encoder, accessing messages from this encoder. It should be noted that the FBG decoders are reversed compared with their matching encoders, and the delay lines are placed in the lower arms of the all decoders. This simultaneously eliminates the round-trip time of the encoders and enables obtaining the correct vibration signal. BPDs were used for correlation detection. The multiple access interference (MAI) from other users is eliminated by subtracting any two given SAC codewords, thereby decreasing the bit error rate of the access signal. The BPDs then convert the decoded signals of the FBG decoder into electrical signals, which are connected to the DOS to detect the expected
Table 2 WDM/SAC codeword; N = 2 and M = 3. l
k
Codeword Ak,l
0 0 0 1 1 1
0 1 2 0 1 2
110000 011000 101000 000110 000011 000101
631
S.-P. Tseng et al. / Optical Fiber Technology 19 (2013) 627–637 ð1Þ
A0,0
A0,1
A1,0
A1,1
and Hk;l ðtÞ respectively represent the spectral intensity distributions for the upper and lower arms of the BPD. Assuming that the signal of user #(k, l) is the expected signal, whereas that for user #(r, s) is not expected, the correlation funcðdÞ ðdÞ tion RAA for Ak;l and Ar,s is defined as follows:
A2,0
A2,1
RAA ðk; l; r; sÞ ¼
ðdÞ
M 1X N1 X
ðdÞ
ak;l ðj M þ iÞar;s ðj M þ iÞ; ðd ¼ 0; 1Þ
ð12Þ
i¼0 j¼0
Fig. 5. The spectrograms of the WDM/SAC codewords in Table 2.
frequency signal. The objective of this process is to use a multiplexing signal for distinguishing the encoded signals of users in the different groups. First, the FBG decoder receives the incident light of the K K star coupler, which includes all output signals of the FBG encoder. When the incident light of the K K star coupler is transmitted to ð0Þ the complementary multiple FBG array address Ak;l through the first optical circulator, it is reflected to the lower arm of the BPD. The remaining spectrum is transmitted to the multiple FBG array ð1Þ address Ak;l through the second optical circulator, and is then reflected to the upper arm of the BPD. Two diodes of the BPD conduct correlation detections after receiving these two complementary light signals. When the sensor encoding signal of FBG Encoder #2 propagates through FBG Decoder #1, the mismatch between the FBG reflected wavelengths causes FBG Decoder #1 to not receive the signal from FBG Encoder #2. Therefore, of the two multiple FBG arrays for the ð0Þ FBG decoder, the complementary multiple FBG array address Ak;l is used to reflect FBG wavelength interference from other users. The ð1Þ multiple FBG array address Ak;l is identical to the multiple FBG arrays of the encoder; thus, it can reflect its own FBG wavelengths, and light signal subtractions or deductions can be conducted using BPDs. This mechanism eliminates MAI caused by other users of the system, thereby increasing the SNR. MAI is generally directly proportional to the number of system users. Furthermore, BPDs are typically used in the laboratory for increasing the SNR of a detected signal and for signal amplification. The operating principle of a BPD involves subtracting the photoelectric currents received by two diodes to eliminate laser or common mode noise. During the decoding process, the signals for two given FBG decoders are complementary. This mechanism enables subtractions during the process of photodetection to obtain the desired signal. For example, regarding user #(k, 1), the decoded vectors for two complementary multiple FBG arrays can be presented as follows: ð0Þ Ak;l
¼ X k El
ð8Þ
Ak;l ¼ X k El
ð9Þ
ð1Þ
ð0Þ Ak;l
is set as the decoded signal detected by the lower arm of the ð1Þ BPD and Ak;l is set as the decoded signal detected by the upper arm of the BPD, where the SAC codewords for X k and Xk are complementary. These two light signals then undergo decoding through the BPD to acquire the expected signal. ð0Þ ð1Þ According to Ak;l and Ak;l of (8) and (9), respectively, the spectral intensity distributions for the FBG decoder before the signal enters the upper and lower arms of the BPD are inferred as follows: ð0Þ
ð0Þ
ð10Þ
ð1Þ
ð1Þ
ð11Þ
Hk;l ðtÞ ¼ Ak;l Ik;l ðtÞ ¼ ½X k El Ik;l ðtÞ Hk;l ðtÞ ¼ Ak;l Ik;l ðtÞ ¼ ½X k El Ik;l ðtÞ
where Ik,1(t) refers to the light intensity variations of the sensor for ð0Þ user #(k, l), which are influenced by the vibration source, and Hk;l ðtÞ
where d = 0, 1 represents the correlation functions for the user ðdÞ entering the upper or lower arm of the BPD and ak;l ðj M þ iÞ and ðdÞ ar,s(j ⁄ M + i) respectively indicate the ðj M þ iÞth element for ak;l and ar;s : i ¼ 0 . . . ; M 1. denotes the SAC codeword with a code length of M; j = 0, . . ., N – 1 expresses the groups. Based on this equation, the properties of the correlation functions for the upper and lower arms of the BPD are as follows:
8 k ¼ r; l ¼ s > <0 ðdÞ RAA ðk; l; r; sÞ ¼ ðM þ 1Þ=4 k – r; l ¼ s > : 0 l–s
ð13Þ
where (M + 1)/4 represents the correlation function value of the lower arm of the BPD when k – r (different users) and l = s (same group), and when k = r, l = s, or l – s, the correlation function value is zero.
8 > < ðM þ 1Þ=2 k ¼ r; l ¼ s ð1Þ RAA ðk; l; r; sÞ ¼ ðM þ 1Þ=4 k – r; l ¼ s > : 0 l–s
ð14Þ
where (M + 1)/2 represents the correlation function value of the upper arm of the BPD when k = r and l = s; (M + 1)/4 expresses the correlation function value of the lower arm of the BPD when k – r and l = s; and when l – s, the correlation function value is zero. When these two light signals enter the BPD for subtraction, the decoded signal can be expressed as follows: ð1Þ
ð0Þ
RAA ðk; l; r; sÞ RAA ðk; l; r; sÞ ¼
ðM þ 1Þ=2 k ¼ r; l ¼ s 0
otherwise
ð15Þ
where (M + 1)/2 represents the energy decoded through the BPD, and 0 represents no decoded energy. In experiment for this study, when N = 2 and M = 3, the correlaðdÞ tion function RAA ðk; l; r; sÞ for the two WDM/SAC decoded signals before entering the BPD is as follows:
8 > < 0 k ¼ r; l ¼ s ð0Þ RAA ðk; l; r; sÞ ¼ 1 k – r; l ¼ s > : 0 l–s
ð16Þ
where 1 represents the correlation function value entering the lower arm of the BPD when k – r and l = s, and l – s indicates that the correlation function value is zero.
8 > < 2 k ¼ r; l ¼ s ð1Þ RAA ðk; l; r; sÞ ¼ 1 k – r; l ¼ s > : 0 l–s
ð17Þ
where 2 represents the correlation function value of the upper arm of the BPD when k = r and l = s; 1 denotes the correlation function value entering the lower arm of the BPD when k – r and l = s; and l – s indicates that the correlation function value is zero. When these two signals enter the BPD, the decoded signal output can be written as follows: ð1Þ
ð0Þ
RAA ðk; l; r; sÞ RAA ðk; l; r; sÞ ¼
2 k ¼ r; l ¼ s 0
otherwise
ð18Þ
632
S.-P. Tseng et al. / Optical Fiber Technology 19 (2013) 627–637
Fig. 6. (a) The reflected spectra for FBG Encoder #1 and (b) the reflected spectra for FBG Encoder #2.
where 2 represents the decoded energy of the BPD for the matching signal output when k = r and l = s; 0 indicates no decoded energy output of the BPD for mismatching signals. When (17) and (18) are modulated by the optical power Ik,l(t) of the AFLVS, the decoded energy of the upper and lower arms of the BPD can be inferred as follows: ð1Þ
ð1Þ
Pk;l ðtÞ ¼ RAA ðk; l; r; sÞ Ik;l ðtÞ ¼
M1 N1 XX
ð1Þ
ak;l ðj M þ iÞar;s ðj M þ iÞ Ik;l ðtÞ
ð19Þ
i¼0 j¼0
ð0Þ
ð0Þ
Pk;l ðtÞ ¼ RAA ðk; l; r; sÞ Ik;l ðtÞ ¼
M1 N1 XX
ð0Þ
ak;l ðj M þ iÞar;s ðj M þ iÞ Ik;l ðtÞ
ð20Þ
i¼0 j¼0 ð1Þ
ð0Þ
When optical powers P k;l ðtÞ and P k;l ðtÞ, which are reflected by the decoder, enter the BPD, the BPD conducts a correlation operað1Þ ð0Þ tion Pk;l ðtÞ Pk;l ðtÞ during the decoding process. When the decoder address and the encoded signal match, a light intensity variation 2Ik,1 for the code is created through modulation, and when the decoder address and encoded signal mismatch, the BPD yields no light intensity output.
The WDM/SAC codeword for Group 1 in Table 2 was adopted in the experiment for this study to prove that the operating principles of the BPD can effectively eliminate MAI and increase the SNR for signal accessing. When the transmitting ends of users #(0, 0) and #(1, 0) simultaneously transmit message bits to the star coupler, and user #(2, 0) does not transmit message bits, the star coupler accumulates the modulated light intensity of all users, which is written as follows: (I0,0(t), I0,0(t) + I1,0(t), I1,0(t), 0, 0, 0). The receiving end of user #(0, 0) reflects two complementary and modulated ð1Þ ð0Þ light signals (i.e., P 0;0 ðtÞ and P 0;0 ðtÞ) to the BPD when receiving light intensity signals from the star coupler. The corresponding spectral light intensities are (I0,0(t), I0,0(t) + I2,0(t), 0, 0, 0, 0) and (0, 0, I2,0(t), 0, 0, 0), respectively. Therefore, the light intensity for these two light signals upon ð1Þ ð0Þ entering the BPD is P0;0 ðtÞ P0;0 ðtÞ ¼ 2I0;0 ðtÞ, indicating that the correct message bits can be accessed and inputted into the DOS to present a fast Fourier transform (FFT) spectrum, thereby enabling observations of the modulated frequency. This study examined whether message bits from other users could be eliminated at the receiving end for user #(2, 0) without light signal output, under the premise that no message bits were transmitted by user #(2, 0). Similarly, when the receiving end of user #(2, 0) receives light signals from the star coupler, two comð1Þ ð0Þ plementary and modulated light signals (i.e., P 2;0 ðtÞ and P2;0 ðtÞ)
Fig. 7. Output spectrum from star coupler after summing all FBG encoders. (a) The optical spectra of output port 1 in star coupler and (b) the optical spectra of output port 2 in star coupler.
S.-P. Tseng et al. / Optical Fiber Technology 19 (2013) 627–637
633
Fig. 8. (a) Reflected spectrum for the lower arm of the FBG Decoder #1 and (b) reflected spectra for the upper arm of the FBG Decoder #1.
are reflected and subsequently entered into the BPD. The corresponding spectral light intensities for these two light signals are (I0,0(t), 0, I2,0(t), 0, 0, 0) and (0, I0,0(t) + I2,0(t), 0, 0, 0, 0), respectively. Thus, the light intensity of these two light signals when entering ð1Þ ð0Þ the BPD is P2;0 ðtÞ P2;0 ðtÞ ¼ 0, indicating that the message bits of other users can be eliminated. Because the FBG decoder accesses no modulated light signals, the BPD does not demodulate any signals. When the AFLVS is placed between the optical circulator of the FBG encoder and the multiple FBG arrays, the light source is inputted into the FBG encoder through the optical circulator; after entering the multiple FBG arrays through the AFLVS, the FBG encoded wavelength is reflected to the AFLVS. Because of external effects, the AFLVS modulates the reflected wavelength of the multiple FBG arrays, which is then entered into the star coupler through the output end of the optical circulator and subsequently inputted into the FBG decoder. The light signal output of the FBG decoder is converted into an electrical signal through the BPD, and the vibration frequency of the test object is acquired through an FFT conducted by the DOS. 3. Experimental framework and results The objective of this experiment was to use the spectral amplitude coding-optical code division multiple access (SAC-OCDMA) system to access the vibration frequency of the stepping motor
(sensed by the AFLVS). Furthermore, this study endeavored to prove that the correct vibration frequency for the two sets of encoders could be accessed by their corresponding FBG decoders without receiving mutual interference when frequency was simultaneously inputted into the stepping motor. For the SLD model, NXTAR SLD-2000 was adopted in this study to provide a broadband light source with stable power for the system. Moreover, dual window wideband couplers (Fiber Optic Communications, Inc., Hsinchu, Taiwan) were employed as the 1 2 and 2 2 couplers of the study experiment. The Anritsu MS9710C optical spectrum analyzer was used to monitor the accuracy of the encoder/decoder spectral output. The BPD model employed for the decoding end was Model-1817 (New Focus Inc., California, USA), which conducted light signal subtractions in the optical domain and subsequently converted the results into electrical signals. The ±15 V current-limited power supply (New Focus Inc., California, USA) was then used to provide power to the BPD. In addition, the Tektronix TDS2102B digital oscilloscope was employed to conduct a Fourier transform on the electrical signal output of the BPD. Finally, the Twintex TFG 3510 signal generator simultaneously provided the vibration frequency for the two stepping motors. The FBG resonance wavelengths used for the FBG encoder/decoder in this study were 1548, 1551, and 1554 nm, representing k1, k2, and k3, respectively. These wavelengths corresponded to binary sequences of zeros and ones. To verify that the OCDMA system used in this study could eliminate the crosstalk from other users
Fig. 9. (a) Reflected spectrum for the lower arm of the FBG Decoder #2 and (b) reflected spectra for the upper arm of the FBG Decoder #2.
634
S.-P. Tseng et al. / Optical Fiber Technology 19 (2013) 627–637
and further access accurate vibration frequencies, the researchers hypothesized the following: (1) the SAC codewords for three users in the same group are A0,0, A1,0, and A2,0; (2) the corresponding binary sequence A0,0 = (1, 1, 0, 0, 0, 0) when the FBG central wavelength of FBG Encoder #1 for User #(0, 0) is ðk1 ; k2 Þ; (3), the corresponding binary sequence A1,0 = (0, 1, 1, 0, 0, 0) when the FBG central wavelength of FBG Encoder #2 for User #(1, 0) is (k2, k3); and (4) the corresponding binary sequence A0,3 = (1, 0, 1, 0, 0, 0) when the FBG central wavelength of FBG Encoder #3 for User #(2, 0) is ðk1 ; k3 Þ. For the study experiment, only the SAC codewords for users #(0, 0) and #(1, 0) were referenced to design the FBG encoder and decoder. A stepping motor was used as the vibration source for this experiment (Fig. 4) and was directly placed on the AFLVS of each FBG encoder, and various frequencies were inputted to observe whether this framework could eliminate crosstalk from other users and further demodulate the correct vibration frequency. Fig. 6(a) shows the measured reflected spectra (k1, k2) of FBG Encoder #1. The represented SAC codeword was (1, 1, 0); the corresponding central wavelengths were 1547.656 and 1550.648 nm, respectively; and the light intensities were 34.77 and 33.96 dB m, respectively. Fig. 6(b) shows the measured reflected spectra ðk2 ; k3 Þ of FBG Encoder #1. The represented SAC codeword was (0, 1, 1); the corresponding central wavelengths were 1550.664 and 1553.704 nm, respectively; and the light intensities were 32.96 and 33.78 dB m, respectively. Fig. 7 indicates the output reflected spectra for FBG Encoders #1 and #2. After propagating through the star coupler, the reflected spectra for all FBG encoders were accumulated and transmitted to FBG Decoders #1 and #2. Fig. 7(a) illustrates the reflected spectra ðk1 ; k2 ; k3 Þ that entered FBG Decoder #1, with a represented SAC codeword of (1, 2, 1); corresponding central wavelengths of
Decoder 1
(a)
Decoder 2
(d)
(b)
1547.656, 1550.664, and 1553.704 nm, respectively; and light intensities of 37.78, 33.76, and 37.07 dB m, respectively. Fig. 7(b) illustrates the reflected spectra ðk1 ; k2 ; k3 Þ that entered FBG Decoder #2, with a represented SAC codeword of (1, 2, 1); corresponding central wavelengths of 1547.656, 1550.664, and 1553.704 nm, respectively; and light intensities of 37.78, 33.76, and 37.07 dB m, respectively. The degree of outputted light intensity from the star coupler shows an approximately 3dB loss for the reflected spectra. Because of the star coupler, the wavelength 1550.664 nm created a superposition effect for the two spectra. Therefore, the wavelength for k2 was approximately 4 dB greater that for k1 and k3 ; that is, the light intensity was double the other wavelengths. When the light signals, which were accumulated by the star coupler, entered FBG Decoder #1, the reflected spectrum (k3 ) entered the lower arm of the BPD [Fig. 8(a)], with a represented codeð0Þ word A0;0 of (0, 0, 1), a corresponding central wavelength of 1553.688 nm, and a light intensity of 39.67 dB m. Fig. 8(b) shows the reflected spectra ðk1 ; k2 Þ of FBG Decoder #1 entering the upper ð1Þ arm of BPD, with a represented SAC codeword A0;0 of (1, 1, 0); corresponding central wavelengths of 1547.64 and 1553.688 nm, respectively; and light intensities of 41.29 and 37.04 dB m, respectively. When the light signals, which were accumulated by the star coupler, entered FBG Decoder #2, the reflected spectrum ðk1 Þ entered the lower arm of the BPD [Fig. 9(a)], with a represented codeð0Þ word A0;0 of (1, 0, 0), a corresponding central wavelength of 1547.656 nm, and a light intensity of 39.76 dB m. Fig. 9(b) shows the reflected spectra ðk1 ; k2 Þ of FBG Decoder #2 entering the upper ð1Þ arm of BPD, with a represented SAC codeword A0;0 of (0, 1, 1); corresponding central wavelengths of 1550.632 and 1553.688 nm, respectively; and light intensities of 36.48 and 40.36 dB m,
Decoder 1
(c)
Decoder 2
Decoder 2
(e)
(f)
Fig. 10. Decoding results for inputting a vibration signal of 0, 50, and 110 Hz into Encoder #1 and a fixed 90 Hz vibration signal for Encoder #2. (a) Decoded signal for Decoder #1 under a 0 Hz vibration signal input for Encoder #1 and a 90 Hz input for Encoder #2. (b) Decoded signal for Decoder #1 under a 50 Hz input for Encoder #1 and a 90 Hz input for Encoder #2. (c) Decoded signal for Decoder #1 under a 110 Hz input for Encoder #1 and a 90 Hz input for Encoder #2. (d) Decoded signal for Decoder #2 under a 0 Hz vibration signal input for Encoder #1 and a 90 Hz input for Encoder #2. (e) Decoded signal for Decoder #2 under a 50 Hz input for Encoder #1 and a 90 Hz input for Encoder #2. (f) Decoded signal for Decoder #2 under a 110 Hz input for Encoder #1 and a 90 Hz input for Encoder #2.
635
S.-P. Tseng et al. / Optical Fiber Technology 19 (2013) 627–637
Decoder 1
Decoder 1
(a)
(b)
Decoder 2
(d)
Decoder 2
(e)
(c)
Decoder 2
(f)
Fig. 11. Decoding results for inputting a vibration signal of 0, 50, and 110 Hz into Encoder #1 and a fixed 190 Hz vibration signal for Encoder #2. (a) Decoded signal for Decoder #1 under a 0 Hz vibration signal input for Encoder #1 and a 190 Hz input for Encoder #2. (b) Decoded signal for Decoder #1 under a 50 Hz input for Encoder #1 and a 190 Hz input for Encoder #2. (c) Decoded signal for Decoder #1 under a 110 Hz input for Encoder #1 and a 190 Hz input for Encoder #2. (d) Decoded signal for Decoder #2 under a 0 Hz vibration signal input for Encoder #1 and a 190 Hz input for Encoder #2. (e) Decoded signal for Decoder #2 under a 50 Hz input for Encoder #1 and a 190 Hz input for Encoder #2. (f) Decoded signal for Decoder #2 under a 110 Hz input for Encoder #1 and a 190 Hz input for Encoder #2.
Decoder 1
(a)
Decoder 1
(b)
Decoder 2
(d)
(e)
Decoder 1
(c)
Decoder 2
(f)
Fig. 12. Decoding results for inputting a fixed 90 Hz vibration signal for Encoder #1 and a vibration signal of 0, 50, and 110 Hz into Encoder #2. (a) Decoded signal for Decoder #1 under a 0 Hz vibration signal input for Encoder #2 and a 90 Hz input for Encoder #1. (b) Decoded signal for Decoder #1 under a 50 Hz input for Encoder #2 and a 90 Hz input for Encoder #1. (c) Decoded signal for Decoder #1 under a 110 Hz input for Encoder #2 and a 90 Hz input for Encoder #1. (d) Decoded signal for Decoder #2 under a 0 Hz vibration signal input for Encoder #2 and a 90 Hz input for Encoder #1. (e) Decoded signal for Decoder #2 under a 50 Hz input for Encoder #2 and a 90 Hz input for Encoder #1. (f) Decoded signal for Decoder #2 under a 110 Hz input for Encoder #2 and a 90 Hz input for Encoder #1.
636
S.-P. Tseng et al. / Optical Fiber Technology 19 (2013) 627–637
Decoder 1
(a)
Decoder 1
(b) Decoder 2
(d)
(e)
Decoder 1
(c)
Decoder 2
(f)
Fig. 13. Decoding results for inputting a fixed 190 Hz vibration signal for Encoder #1 and a vibration signal of 0, 50, and 110 Hz into Encoder #2. (a) Decoded signal for Decoder #1 under a 0 Hz vibration signal input for Encoder #2 and a 190 Hz input for Encoder #1. (b) Decoded signal for Decoder #1 under a 50 Hz vibration signal input for Encoder #2 and a 190 Hz input for Encoder #1. (c) Decoded signal for Decoder #1 under a 110 Hz vibration signal input for Encoder #2 and a 190 Hz input for Encoder #1. (d) Decoded signal for Decoder #2 under a 0 Hz vibration signal input for Encoder #2 and a 190 Hz input for Encoder #1. (e) Decoded signal for Decoder #2 under a 50 Hz vibration signal input for Encoder #2 and a 190 Hz input for Encoder #1. (f) Decoded signal for Decoder #2 under a 110 Hz vibration signal input for Encoder #2 and a 190 Hz input for Encoder #1.
Piezoelectric accelerometer 50Hz
(a)
Piezoelectric accelerometer 170Hz
(d)
Piezoelectric accelerometer 90Hz
(b)
Piezoelectric accelerometer 110Hz
(c)
Piezoelectric accelerometer 190Hz
(e)
Fig. 14. Results for using the piezoelectric accelerometer to measure the stepping motor. Measurement results under a vibration frequency for the stepping motor (a) 50 Hz, (b) 90 Hz, (c) 110 Hz, (d) 170 Hz and (e) 190 Hz.
S.-P. Tseng et al. / Optical Fiber Technology 19 (2013) 627–637
respectively. Finally, the BPD was used to convert the reflected spectra into light signals, and the digital oscilloscope or DOS was employed to access the vibration frequencies for FBG Encoders #1 and #2. In Experiment 1, vibration frequencies of 0, 50, and 110 Hz were inputted into FBG Encoder #1, and a fixed 90 Hz vibration frequency was used for FBG Encoder #2. Fig. 10(a–d) shows the FFT spectrograms for using FBG Decoder #1 to access the vibration frequency of the stepping motor, and Fig. 10(e and f) shows the FFT spectrograms for using FBG Decoder #2 to access the vibration frequency of the stepping motor. Subsequently, in Experiment 2, vibration frequencies of 0, 50, and 110 Hz were inputted into FBG Encoder #1 and a fixed 190 Hz was used for FBG Encoder #2. Fig. 11(a–d) shows the FFT spectrograms for using FBG Decoder #1 to access the vibration frequency of the stepping motor, and Fig. 11(e and f) shows the FFT spectrograms for using FBG Decoder #2 to access the vibration frequency of the stepping motor. Experiment 3 involved using a fixed vibration frequency for Encoder #1 and sequentially changing the vibration frequency input for Encoder #2. A fixed vibration frequency of 90 Hz and vibration frequencies of 0, 50, and 110 Hz were inputted into FBG Encoders #1 and #2, respectively. Fig. 12(a–d) and (e and f) respectively shows the FFT spectrograms of the stepping motor vibration frequency accessed using FBG Decoders #1 and #2. Finally, Experiment 4 involved inputting a fixed vibration frequency of 190 Hz and vibration frequencies of 0, 50, and 110 Hz into FBG Encoders #1 and #2, respectively. Fig. 13(a–d) and (e and f) respectively shows the FFT spectrograms of the stepping motor vibration frequency accessed using FBG Decoders #1 and #2. The experimental results indicate that when the digital oscilloscope or DOS was used to measure and obtain the FFT spectrograms, a corresponding frequency peak value could be observed in the FFT spectrogram when the signal generator inputted the frequency into the stepping motor. Although crosstalk from other encoders was conspicuous in the FFT spectrograms, the existence of an input frequency from other FBG encoders that possessed a level approximate to that for noise was verified. This does not affect the vibration frequency required for access. Thus, the experiment results prove that the experimental settings and arrangement can effectively eliminate MAI caused by other FBG encoders. Regarding the piezoelectric accelerometer, the FFT spectrograms for the stepping motor under a frequency of 50, 90, 110, 170, and 190 Hz were measured, and almost no frequency was detected when the frequency was lower than 90 Hz (Fig. 14), indicating that the piezoelectric accelerometer exhibited measurement difficulties at a low frequency (lower than 90 Hz). 4. Conclusion The AFLVS adopted in this study was constructed using optical fibers; thus, it possessed attributes such as small size, light weight,
637
and resistance to EMI. Consequently, this sensor is an ideal replacement for traditional electric or mechanical sensors. In this study, a combination of the SAC-OCDMA system and an AFLVS was used to conduct multiplex measurements at multiple points. Because this technique combines SAC and WDM, which exhibit excellent orthogonality, and uses FBG as the encoder/decoder, users are able to exclude MAI from other users, thereby increasing the SNR. In addition, the number of users increases with code length; therefore, more users are supported and designs are more flexible. The experimental results indicate that the various FBG decoders are capable of correctly accessing the vibration frequency of the corresponding FBG encoder sensor without experiencing MAI from other FBG encoders. In other words, the occurrence of crosstalk generated by other FBG encoders is avoided in the FFT spectrogram, thereby preventing vibration frequency reading errors. The piezoelectric accelerometer measurement results for the vibration frequency of the stepping motor indicate that because of size limitations, the detected signals at a frequency lower than 90 Hz were less than ideal. Consequently, large-scale network establishments such as that for OFSs cannot be implemented, and signals must be transmitted through cables; hence, applications involving large-scale vibration sensors are more costly. Acknowledgments The current authors gratefully acknowledge the financial support provided to this study by the National Science Council, Taiwan under the grant NSC 102-2221-E-150-002 and NSC 102-2221-E239-004. References [1] N.Q. Nguyen, N. Gupta, Power modulation based fiber-optic loop-sensor having a dual measurement range, J. Appl. Phys. 106 (2009) 033502. [2] H. Renner, Bending losses of coated single-mode fibers: a simple approach, IEEE J. Lightw. Technol. 10 (1992) 544–551. [3] P. Wang, Q. Wu, A. Sun, Y. Semenova, G. Farrell, A macrobending fiber based vibration sensor using whispering gallery mode’’, in: Proc. SPIE 7726, Optical Sensing and Detection, 2010. [4] P. Wang, Y. Semenova, Q. Wu, G. Farrell, A macrobending fiber based microdisplacement sensor utilizing whispering-gallery modes, in: Proc. SPIE 7503, 20th International Conference on Optical Fibre Sensors, 2009. [5] L.T. Blair, S.A. Cassidy, Wavelength division multiplexed sensor network using Bragg fibre reflection gratings, Electron. Lett. 28 (1992) 1734–1735. [6] R.S. Weis, A.D. Kersey, T.A. Berkoff, A four-element fiber grating sensor array with phase-sensitive detection, IEEE Photonics Technol. Lett. 6 (1994) 1469– 1472. [7] Y.H. Huang, C. Lum, P.K.A. Wai, H.Y. Tam, Large-scale FBG sensor utilizing code division multiplexing, in: Laser and Electro-Optics, 2008 and 2008 Conference on Quantum Electronics and Laser Science. Conference on CLEO/QELS 2008, 2008, pp. 1–2. [8] H.C. Cheng, C.H. Wu, C.C. Yang, Y.T. Chang, Wavelength division multiplexing/ spectral amplitude coding applications in fiber vibration sensor systems, IEEE Sens. J. 11 (10) (2011) 2518–2526. [9] C.C. Yang, Hybrid wavelength-division-multiplexing/spectral-amplitudecoding optical CDMA system, IEEE Photonics Technol. Lett. 17 (2005) 1343– 1345. [10] K.O. Hill, G. Meltz, Fiber Bragg grating technology fundamentals and overview, J. Lightw. Technol. 15 (8) (1997) 1263–1276.