Hybrid WDM and Optical Spectral Amplitude Coding ...
Hybrid WDM and Optical Spectral Amplitude Coding with Array Waveguide Gratings over Fiber-to-the-Home Network Yao-Tang Chang The Institute of Computer and Communication Engineering, Department of Electrical Engineering, National Cheng Kung University and National Communications Commission, Taiwan
Abstract -The hybrid wavelength-division multiplexing and optical code-division multiple-access scheme for EPON (called WDM/OCDMA-EPON) is presented over fiber-to-the-home (FTTH) network. In this study, each ONU (optical network unit) group is assigned the maximal-length sequence (M-sequence) code which is popularly used for spectral amplitude coding in OCDMA network. Hence, the proposed scheme is suitable for various distances between ONU and OLT (optical line terminal). By employing the inherent cyclic and free spectral range (FSR) periodic properties of arrayedwaveguide grating (AWG) router, the AWG coder/decoder (codec) is shared by various ONUs group and then the number of AWGs is significantly decreased. Moreover, the variable degree of polarization (DOP) is investigated to evaluate the influence of the ratio of signal to phase-noise (SNRPIIN). The system performance in terms of simultaneous active ONUs is improved approximating 20% compared to the average DOP of 0.5 when we set up a depolarizer in front of balanced photo-detector. Keywords: Hybrid wavelength-division multiplexing and optical code-division multiple-access (WDM/OCDMA), Array-Waveguide-Grating (AWG), Fiber-to-the-home (FTTH), The degree of polarization (DOP).
1
Introduction
The spectral amplitude coding for optical codedivision multiple-access (i.e., called SAC/OCDMA) techniques are suggested to be a more flexible solution over optical local area networks (LANs) because the multiple access interference (MAI) is significantly cancelled [1-3]. Hence, the SAC/OCDMA scheme is suitable for burst and asynchronously environments. Moreover, the entire coded wavelength of each user is mixed via the optical coupler and then is capable of confidential enhancement. Therefore, the SAC-OCDMA scheme can provide more promise solution on fiber-to-thehome (FTTH) access network than wavelength-division multiplexing for Ethernet passive optical network (i.e., called WDM-EPON) [4].
Jen-Fa Huang and Che-Chih Hsu The Institute of Computer and Communication Engineering Department of Electrical Engineering National Cheng Kung University, Taiwan [email protected] The combined WDM and SAC/OCDMA scheme with fiber Bragg grating (FBG) based encoder/decoders (FBGbased codecs) over LAN was proposed by Yang [5]. By employing balanced incomplete block design code as the SAC code of each user group, the ability for MAI cancellation is preserved. Since the FTTH network is characterized by various distance between optical network unit (ONU) and optical line terminal (OLT), this scheme [5] is not directly applied to implement a flexible and confidential access network. In our previous done work by Huang et al. [6], the array waveguide grating based (AWG-based) codecs, which combined the inherent cyclic properties and maximal length sequence code (M-sequence code), was configured over optical local area networks (LANs). In current proposed WDM/OCDMA scheme, combining the inherent cyclic and free spectral range (FSR) periodic properties of AWG router, the AWG-based codec is shared by the ONUs which distributed and separated in various group (different area). Also, the M-sequence code, which is popular used for SAC/OCDMA scheme, is assigned as individual signature address code. Hence, under the consideration of costing, flexibility and scalability over FTTH network, the AWG-based codecs of proposed WDM/OCDMA not only is shared more ONUs (subscribers) but also set up flexible to different ONUs group. Compared to previous scheme [6], the proposed scheme applies the same NxN AWG router and achieves the more costing-efficiency performance over FTTH network. For SAC/OCDMA scheme, a crucial phase-induced intensity-noise (PIIN) is generated when mixed incoherent light fields are incident upon a photo-detector [3]. However, the PIIN noise has never investigated on the influence of the degree of polarization (DOP). In practical, the degree of polarization, P is not only dependent on the light source but also vary via the long haul network transmission. In order to achieve higher signal-to-PIIN noise ratio, in this study, we can set up a depolarizer in front of photodetector to ensure that the polarization-dependent properties of the photo-detector can be eliminated [7]. The remainder of this paper is organized as follows. In the section 2, AWG-based WDM/OCDMA network is
depicted. Especially, the signature codeword assignment with M-sequence is demonstrated under the cyclic and free spectral range (FSR) periodic properties of AWG router. Section 3 describes the AWG-based codecs of proposed WDM/OCDMA scheme and the illustrative example is demonstrated for multiple access interference (MAI) cancellation. The Section 4 evaluates the system performance in terms of the BER (bit error rate) and the maximum number of permissible simultaneous active ONUs. Especially, the PIIN noise affected the degree of polarization is investigated. Finally, Section 5 provides some concluding remarks.
2 2.1
The WDM/OCDMA Scheme The system configuration
In the current study, the flexible WDM/OCDM technique on EPON (i.e., the so-called WDM/OCDMAEPON) is set up in access and premise network shown as Fig. 1. Considering variable distributed area for users (custom), all the ONUs can be divided into specified groups and each group accommodate the same number of ONUs.
For the down stream transmitted from OLT, the data bit coming from each ONU is encoded by NxN AWG router and summed together in optical combiner. At the receiver end (ONU end), a pair of NxN AWG router, which is matched OLT AWG router and pre-written with Msequence as signature address codes, is used to implement optical correlation processes. The desired data bit of individual ONU is then recovered via balanced photodetector. Similarly, for the down stream, the symmetry and reciprocal architecture is set up implement bi-direction network. First of all, the proposed WDM/OCDMA scheme preserves the ability of multiple-access interference (MAI) cancellation. As seen in Fig. 1, since the FTTH network is characterized by various and unequal distance between ONUs and OLT, the proposed coder/decoder (codec) can be constructed with AWG router located on centralized OLT. Meanwhile, the corresponding codecs located on ONUs are constructed with AWG or fiber Bragg grating (FBG). Note that the FBG-based codecs are more suitable for a separated ONUs than a centralized ONUs. That is, the AWG-based ONU is selected when the position of supported ONU is approaching each other. Conversely, the FBG-based is selected when the position of supported ONU is isolated. Since the FBG-based codecs with Msequence have demonstrated by previous authors’ own work [8], we only focus on the AWG-based codecs in current study.
2.2 The Codeword Assignment
Fig. 1 The schematic diagram of proposed scheme configured by AWG and FBG based codecs for down stream example. AWG: array waveguide grating, FBG: fiber Bragg grating. The previous AWG-based codecs was configured over local area network by Huang et al. [6]. Here, by exploiting the inherent cyclic and FSR periodic properties for AWG routers, M-sequence code family is used as signature address code. Thus, for the down stream example shown as Fig. 1, a shared AWG router is located on centralized OLT. At the ONUs end, the matched AWGbased codecs is composed of a pair of corresponding AWG router, a pair of photo-detector and a decision unit. Hence, the AWG-based codecs with M-sequence scheme become simple and are easily implemented over FTTH network.
In the proposed WDM/OCDMA scheme, the ONU capacity is divided into G groups, where G is positive integer. and depend on the spectrum of light source. Every ONU is indexed as #(g, h) and assigned a code word Cg, h. The subscript of g denotes the g-th group which is divided by FSR interval range. The subscript of h denotes the h-th code sequence which is assigned as M-sequence code in the g-th group. Here, the g and h set belong to {1, 2, 3,…, G} and {1, 2, 3,…, N}, respectively. Let Mh(i)= (Mh(1), Mh(2), …, Mh(N)) be (0, 1) sequences of length N and the h-th sequence is followed by M-sequence code matrix, where h = 1, ..., N and i is the i-th chip (element) of Mh sequence. Thus, the j-th sequence of Mj(i) (i.e., j = 1, ..., N) can be obtained by cyclic shifting the original sequence of M1(i) (i.e., Mj(i) = T j-1 M1(i)), where T is the operator shifts vectors cyclically to the right by j-1 place). Let Fg(k)= (Fg(1), Fg(2), …, Fg(G)) is a vector whose the g-th element is one and others are zero, where k = 1, ...,G and the adjacent element of Fg(k) is equal to the FSR wavelength interval range of AWG router (i.e., so called Δλf). For example, the difference wavelength of Fg(1) and Fg(2) is one FSR (i.e., Δλf). Similarly, the difference of Fg(1) and Fg(3) is double FSR (i.e., 2*Δλf). Hence, the proposed WDM/OCDMA codeword of the h-th ONU in the g-th group is expressed as
Cg, h =[ Cg, h (m)] = [Mh(i)* Fg(k), …, Mh(i)*Fg(k)] (1) where m =1, 2, 3, …, G*N and Cg, h (m) is the m-th element of Cg, h and the symbol of * denote multiple operator. Based on Eq. (1) for M-sequence code length of N=7, the total coded spectrum of proposed WDM/OCDMA scheme is illustrated as Fig. 2. As seen in Fig. 2, the wavelength spectrum of light source is divided into G groups following the inherent FSR interval range of AWG router. Each group consists of seven wavelengths based on M-sequence of N=7. Hence, the codeword of Cg, h and Cg+1, h is located on the adjacent group. Also, the i-th and its corresponding wavelength (i.e., chip element) of adjacent group is far away one Δλf (i.e., equal to one FSR interval range). For simple example, there are seven ONU capacity from C g,1 to C g,7 in the g-th group. As a result, the Cg, 1 and Cg+1, 1 are characterized by the same M-sequence coding pattern of (1, 1, 1, 0, 0, 1, 0) in the adjacent group (i.e., in the group g and g+1) but different wavelength range. the Cg, 2 and Cg+1, 2 are characterized by the same M-sequence coding pattern of (0, 1, 1, 1, 0, 0, 1) but different wavelength range in the adjacent group of g and g+1. Note that the λ g , h denotes
signify information bit “0”. The unpolarized and broadband incoherent light source (BLS), whose spectrum is filtered within one free spectral range (FSR) of the AWG router, is employed. By applying the inherent cyclic and FSR periodic properties of AWG routers, the resulting optical signals of each ONU, which is assigned as M-sequence code, are directed to the corresponding input ports of 7x7 AWG-bsed encoder.
the h-th wavelength in group #g. Fig. 3 The proposed WDM/OCDMA techniques over FTTH for M-sequence codeword length of N=7 The each input port of AWG-based encoder accommodates the specified ONU which is assigned as the same M-sequence codeword but different wavelength group #g. At the AWG-based decoder, its corresponding output ports of 7x7 AWG-based decoder (original and complementary part) recover the data bit of desired ONU. That is, the AWG-based codecs is shared by all the ONU number of G*7 (i.e., 7 is codeword length of the assigned M-sequence) and illustrated as Table 1. TABLE 1.The shared AWG-based for M-sequence codeword length of N=7.
Fig. 2 The coded spectrum of proposed WDM/OCDMA scheme for M-sequence codeword length of N=7.
2.3
The AWG-based Encoder/Decoder
For simple explanation for WDM/OCDMA techniques over FTTH network, the 7x7 AWG routers is employed and shown as Fig. 3. As a result, there are G*7 ONU capacity share 7x7 AWG router. Here, G is positive integer and depends on the spectrum of light source. As seen in Fig. 3, in order to fulfill the E-O modulation, the data bits of each ONU #(g, h) is On–Off shift keying (OOK), the energy of which will be transmitted for information bit “1” and an absence of optical energy will
p
As seen in Table 1, the subscript of λ g , h denote the h-th wavelength in group #g relative to the spectrum of light source. Also, the superscript denotes the input port of 7x7 AWG-based encoder. The different group from ONU #(1, 1) to #(G, 1) signal combine together and enter into the same input port #1 of the 7x7 AWG-based encoder. Similarly, the different group from ONU #(1, 2) to #(G, 2) signal combine together and enter into the same input port #2 of the 7x7 AWG-based encoder. Here, each group is occupied seven wavelengths such that λ1 to λ7 is assigned in the group #1 and λG, 1 to λG, 7 is assigned in the group #G etc. For down stream example, the coded spectrum created on OLT is summed together via combiner and then transmit down to splitter located on ONUs. Since the summed optical signal is mixed up for simultaneous active ONUs, the proposed WDM/OCDMA is capable of confidentiality compared to conventional WDM-EPON and WDM-EPON even code division multiplexing EPON (CDM-EPON) scheme [6]. At the ONUs end, the wavelength-selective filter is used to separate the down stream into #g groups. Here, the range of wavelength selective filter is applied to filter out flexibly the desired wavelength from λg, 1 to λg, 7 (λg, i corresponding to λ(g-1)*7 + i ). Since the system performance of SAC scheme only depends on the correlation property of the underlined coding pattern (i.e., M-sequence code). The wavelength-selective filter can be selected in relaxed requirement and decrease the system costing. Subsequently, at the desired receiver (ONU group #g) end, a pair of AWG-based decoder pre-written with Msequence signature address codes is used to implement optical correlation processes. The desired data bit of “1” or ‘0” coming from ONU #(g, h) is then obtained via balanced photo-detector. Note that the scenario of present (on) or absent (off), which is wired on the input ports of AWG decoder, is employed with M-sequence code pattern for logic one or zero. Also, the input ports of the original and its complementary AWG-based decoder are matched the wired scenario on the output ports of AWG-based encoder. Compared to conventional EPON scheme over FTTH network, the qualitative comparisons is described as Table 2. TABLE 2. The comparisons of popular EPON scheme on several aspect.
3
MAI cancellation
The light sources used in this system are incoherent ideally un-polarized sources. Here, we use the unipolar sequences to encode the amplitudes of light source spectrum. Let Cg, h= (Cg, h(1), Cg, h(2),.., Cg, h(N)) be (0, 1) sequences of length N assigned as the codeword of ONU #(g, h), where the maximal of g*h is equal to G*N. Note that G, N denote the number of group and the number of M-sequence codeset (i.e., equal to code length, N), respectively. Based on the common rule for previous SAC OCDMA scheme with M-sequence code [8], the periodic correlation between Cg, h and Cr, s is defined as Rcc ( g , h , r , s ) and Rcc ( g , h , r , s ) , respectively. The proposed scheme satisfies the following relationship. Rcc ( g , h , r , s ) =
⎧( N + 1) / 2, g = r , h = s ⎪ ∑ C g, h ( i ) Cr , s ( i ) = ⎨( N + 1) / 4, g = r , h ≠ s ⎪ i =1 0, otherwise ⎩
(2-a)
⎧⎪ ( N + 1) / 4, g = h , r ≠ s otherwise ⎩ 0,
(2-b)
N
Rcc ( g , h , r , s ) =
N
∑ C g, h ( i ) Cr , s ( i ) = ⎨⎪
i =1
where the subscript g, h, r and s of C denote that the codeword C is assigned to ONU #h and #s in the g-th and r-th group, respectively. The symbol i denote the i-th chip for codeword C. Note that Ch is a unipolar M-sequence in the g-th group arbitrarily. The code length is equal to N and the codeword of the s-th ONU’s sequence from Ch can be written as Cs = TsCh by cyclic shifting the original sequence by s place, where T is the operator shifts vectors cyclically to the right by one place. That is, C2=TC1 = (C1(N), C1(1), …, C1(N-2), C1(N-1)). Following the Eq.(2), the proposed WDM/SACOCDMA retains the quasi-orthogonal property via the additional wavelength-selective filter which is characterized by relax requirement (i.e., less. costing) for achieving better flexibility and scalability. Thus, a balanced detector for desired ONU #(r, s) will implement the correlation subtractions expressed as follows: Z = Rcc ( g , h , r , s ) − Rcc ( g , h , r , s )
N +1 N +1 N +1 ⎧ , g = r, h = s ⎪⎪ 2 × 2 − 2 = 2 =⎨ ⎪ 2× N +1 − N +1 = 0 , otherwise ⎪⎩ 4 2
(3)
Eq. (3) is shown that MAI coming from other ONUs will be completely cancelled. With this excellent MAI-
∞
cancellation ability and the cyclic manner of M-sequence signature sequences, a compact WDM/OCDMA scheme with AWG-based codecs will operate without MAI theoretically.
4
The System Performance evaluation
In evaluating the performance of the proposed WDM/OCDMA scheme, the present study adopts a similar analysis model that applied by the conventional SAC scheme. The symbols which appear in the following discussions are defined in [3] and [9]. The present evaluation assumes that each light source is unpolarized and that the optical source is an ideal flat spectrum with a magnitude of Psr /Δν, where Psr is the effective power from a single source at the receiver and Δν is the optical source bandwidth in hertz.
4.1
Signal power evaluation
Following the result of Eq. (2), the PSD of the received optical signal can be written as N P G* N s ( v ) = sr ∑ b ( g,h ) ∑ C ( g,h ) ( i ) { rect ( i )} Δ v ( g,h ) =1 i =1
(4)
(5)
where N denote the codeword length of M-sequence and u is unit step function. The symbol of Δν is optical bandwidth. Assuming bit synchronism for now, the power spectrum density (PSD) at PD1 (photo-detector #1) of the desired ONU #(r, s) receiver during one bit period can be written as N P G* N G1 ( v ) = sr ∑ b( g,h ) ∑ C( g,h ) ( i ) C( r,s ) ( i ) { rect ( i ) } Δv ( g,h )=1 i =1
=
R Psr ⎛ N + 1 ⎞ b( r, s ) N ⎜⎝ 2 ⎟⎠
(7)
where R is the photodiode responsivity and b(r, s) ∈ {0,1}
4.2
PIIN power evaluation
This study evaluates only the PIIN noise which results from several thermal sources having a phase variation at each frequency. The PIIN noise results are compared with those obtained from a conventional SAC scheme for evaluation purposes. Referring to [3, Eq. (19)], the thermal noise is independent of the effective power of a broad-band source at the receiver (i.e., Psr) and the shot noise is only proportional to Psr. However, the phase induced intensity noise (PIIN) is proportional to Psr squared. Hence, it is shown that the effective power Psr from each ONU is large (e.g., Psr more than -10dBm), both the shot and thermal noises are negligibly small compared with PIIN noise, which becomes the main limitation factor of the system performance of signal to noise ratio (SNR). Hence, the variance of photo-detector current is expressed as ≈ I2 (1+P2)τc B
where (g, h) is the number of active ONU #(g, h) less than or equal G*N. Also, b(g, h) = “1” or “0” represents the data bit of ONU #(g, h). The rect(i) function in (4) is given by Δv Δv ⎡ ⎤ ⎡ ⎤ rect ( i) = u⎢v− v0− ( − N+ 2i− 2) ⎥ − u⎢v− v0− ( − N+ 2i) ⎥ 2N 2N ⎣ ⎦ ⎣ ⎦
∞
I = I1 − I 2 = R ∫ G1 ( v ) d v − R ∫ G2 ( v ) d v 0 0
(6-a)
(8)
where I,τc , B and P denotes the average photocurrent; the noise-equivalent electrical bandwidth of the receiver; the coherence time of the source and the degree of polarization (DOP), respectively. We temporally assume that the degree of polarization P is ideal on Eq. (8) such that ≈ I2 τc B. Hence, by substituting G12 ( ν ) and G22 ( ν ) into Eq. (7) and assuming all the ONUs are transmitting bit “1”, The result is given by: ∞ ⎡ ∞ ⎤ 2 i 2 = I PIIN = B R 2 ⎢ ∫ G12 ( v ) d v + ∫ G22 ( v ) d v ⎥ 0 0 ⎣ ⎦ ⎤ B R2 Psr2 N ⎧⎪ ⎡ N ⎨ ⎢ ∑ b ( g, h ) C ( g, h )( i ) ⎥ ∑ N Δ v i=1⎪ ⎢h=1 ⎥⎦ ⎩⎣
⎡N ⎤ ⎫⎪ ⎢ ∑ b ( r, s ) C ( r, s )( i ) ⎥ ⎬ ⎢⎣ s=1 ⎥⎦ ⎭⎪
⎫ P 2 B R 2 ⎧ ⎛ N +1 ⎞ = sr K ( K +1) ⎬ ⎨⎜ ⎟ N Δ v ⎩⎝ 4 ⎠ ⎭
(9)
Similarly, the PSD at PD2 (photo-detector #2) of the desired ONU #(r, s) receiver during one bit period can be written as
4.3 The Bit Error Rate (BER) evaluation
N P G* N G2 ( v ) = sr ∑ b(g , h) ∑ C(g ,h) ( i ) C(r , s ) ( i ) { rect ( i ) } Δv (g , h) =1 i =1
Dividing Eq. (7) by Eq. (9), the signal-to-noise ratio (SNR) and bit-error-rate, under the assumption that Gaussian approximation, is written as
(6-b)
Because of C(r , s) ( i ) = 1 − C(r , s) ( i ) and using Eq.(2), the signal from the desired ONU is given by the difference of the photodiode current outputs
SNR =
Ib =1 − Ib = 0 2 I PIIN
2 =
( N +1)Δν N B K ( K +1)
(10)
where Δν is optical band-width and K is number of active ONUs. As seen the Eq. (8), the degree of polarization, P, is defined as:
P
2
(s =
1
2
+ s2
2
+ s3
2
2
)
(11)
s0 In Eq. (11), s0, s1, s2, and s3 are referred to as the stoke parameters and are expressions of the state of polarization (SOP). The bracket < > denotes the average value of these parameters over wavelength, time or space. It is well known that the degree of polarization, P is not only dependent on the light source. Also, the degree of polarization, P is varied via the long haul network transmission. In order to inspect the DOP value in general light source, we select a common amplified spontaneous emission (ASE) as light source randomly and make simple experiment on back to back configuration. We connect ASE source to polarization analyzer (HP 8509C). The spectrum of unpolarized ASE light source is shown as Fig. 4(a) and the measured DOP approximates 0.035 at wavelength 1550 nm shown as Fig. 4(b). Note that the power of light source is properly attenuated.
Subsequently, by setting (i.e., selecting switch) internal linear polarizer of polarization analyzer (HP 8509C), the spectrum of polarized thermal source is shown as Fig. 5(a) and the detected DOP value is 0.989 approximately shown as Fig. 5(b). It reveals that the long haul network transmission become main factor to affect the DOP variation. In order to achieve higher signal-to-PIIN noise ratio, we can set up a depolarizer in front of photo-detector to ensure the polarization-dependent properties of the photo-detector can be eliminated. Hence, the average value of s1, s2, and s3 from Eq. (11) is approaching to zero and the degree of polarization (DOP) can be significantly decreased. That is, the depolarizer may remove the polarization sensitivity of the photo-detector for proposed WDM/OCDMA scheme in theory. For previous work done by Dale R. Lutz [7], a depolarizer was configured in front of balanced photodetector to achieve the DOP of 0.03 (i.e., P = 0.03).
(a)
(a)
(b) Fig. 5 The (a) spectrum and (b) measured degree of polarization (DOP) of polarized amplified spontaneous emission (ASE) light source.
(b) Fig. 4 The (a) spectrum and (b) measured degree of polarization (DOP) of un-polarized amplified spontaneous emission (ASE) light source.
In order to analyze the BER performance with and without depolarizer in convenience, we take the average DOP of 0.5 as a general scenario without depolarizer compensation after a long haul distance transmission. Following Eqs. (8) and (11), the SNR (signal power divided by PIIN power) affected by degree of polarization is evaluated and then established the corresponding BER (BER = (1/2)*erfc[(SNRPIIN / 8)1/2]) shown as Fig. 6.
As seen in Fig. 6, the BER performance of the proposed WDM/OCDMA scheme is characterized by an upper bound of P=1 for the worst case and a lower bound of P=0 for the ideal case. In particular, by placing a depolarizer in front of the photo-detector to achieve P=0.03, as configured by Dale R. Lutz [7], it can be seen that the curve of P = 0.03 virtually overlaps that of the proposed WDM/OCDMA scheme with M-sequence and HadmardWalsh code for the ideal case (P=0). Compared to the average DOP of 0.5 (i.e., the DOP value varies from 1 to 0 randomly), the active ONU is improved approximating 20%. Furthermore, it indicates that the BER performance with M-sequence is approximating with Hadamard-Walsh code when the active users is increased (i.e., SNR = Δν / B K ( K +1) with Hadamard-Walsh code). In particular, the number of AWGs with M-sequence code is significantly decreased due to the cyclic and FSR periodic properties.
Fig. 6 BER performance under various degree of polarization setting of P in proposed WDM/OCDMA scheme while Δν= 6.25THz, B=80MHz and codeword length of N=127.
5
Conclusions
Under the consideration of costing, flexibility and confidentiality, the WDM/OCDMA-EPON technique is demonstrated over FTTH network. Here, the AWG-based codecs is configured on optical network unit and optical line terminal, respectively. The ONU capacity is divided into groups and the M-sequence code is assigned as signature address code. By employing the inherent cyclic and free spectral range periodic properties of AWG router, the number of AWG router is significantly decreased because that the AWG-based codecs can be shared by the number of group. Furthermore, the various degree of polarization (DOP) is investigated to evaluate the signal to phase induced intensity noise ratio (SNRPIIN). By setting the depolarizer in front of balanced photo-detector, the system performance of simultaneous active ONUs is
approximately improved 20% compared to the average DOP of 0.5. In this study, the proposed WDM/OCDMA is more flexible, confidential and scalable than conventional TDMEPON and WDM-EPON scheme even code division multiplexing EPON (CDM-EPON). However, the additional wavelength-selective filter and depolarizer is used so that the system complexity is increased.
6
References
[1] Kavehrad, M. and D. Zaccarin, “Optical CodeDivision-Multiplexed Systems Based on Spectral Encoding of Noncoherent Sources,” IEEE J. Lightwave Technol., vol. 13, no. 3, pp. 534-545, March 1995. [2] X. Zhou, H. M. H. Shalaby, C. Lu, and T. Cheng, “Code for spectral amplitude coding optical CDMA systems,” Electron. Lett., vol. 36, pp. 728–729, Apr. 13, 2000. [3] Z. Wei; Shalaby, H.M.H.; Ghafouri-Shiraz, H., “Modified quadratic congruence codes for fiber Bragggrating-based spectral-amplitude-coding optical CDMA systems,” IEEE J. Lightwave Technol., vol. 19, no. 9, pp. 1274-1281, Sept. 2001. [4] S.-J. Park, S. Kim, K.-H. Song, and J.-R. Lee, “DWDM-based FTTC access network,” J. Lightwave Technol., vol. 9, pp. 1851-1855, Dec. 2001. [5] C. C. Yang, “Hybrid Wavelength-DivisionMultiplexing/Spectral-Amplitude-Coding Optical CDMA system,” IEEE Photonics Technology Letters, vol. 17, no. 6, pp. 1343-1345, June 2005. [6] C. C. Yang, J. F. Huang, and H. P. Tseng, “Optical CDMA Network Codecs Structured with M-Sequence Codes over Waveguide-Grating Routers,” IEEE Photonics Technology Letters, vol. 16, no. 2, pp. 641643, February 2004. [7] Dale R. Lutz, “A Passive Fiber-Optic Depolarizer,” IEEE Photonics Technology Letters, vol. 4, no. 4, April 1993. [8] J. F. Huang, and D. Z. Hsu, “Fiber-grating-based optical CDMA spectral coding with nearly orthogonal M-sequence codes,” IEEE Photon. Technol. Letter, vol. 12, pp. 1252-1254, Sept. 2000. [9] E. D. J. Smith, R.J. Baikie, and D.P. Taylor, “Performance enhancement of spectral- amplitudecoding optical CDMA using pulse-position modulation,” IEEE Trans. on Commun, vol. 46, no. 9, pp. 1176-1185, September1998.
Abstract -The hybrid wavelength-division multiplexing and optical code-division multiple-access scheme for EPON (called WDM/OCDMA-EPON) is presented over fiber-to-the-home (FTTH) network. In this study, each ONU (optical network unit) group is assigned the maximal-length sequence (M-sequence) code which is popularly used for spectral amplitude coding in OCDMA network. Hence, the proposed scheme is suitable for various distances between ONU and OLT (optical line terminal). By employing the inherent cyclic and free spectral range (FSR) periodic properties of arrayedwaveguide grating (AWG) router, the AWG coder/decoder (codec) is shared by various ONUs group and then the number of AWGs is significantly decreased. Moreover, the variable degree of polarization (DOP) is investigated to evaluate the influence of the ratio of signal to phase-noise (SNRPIIN). The system performance in terms of simultaneous active ONUs is improved approximating 20% compared to the average DOP of 0.5 when we set up a depolarizer in front of balanced photo-detector. Keywords: Hybrid wavelength-division multiplexing and optical code-division multiple-access (WDM/OCDMA), Array-Waveguide-Grating (AWG), Fiber-to-the-home (FTTH), The degree of polarization (DOP).
1
Introduction
The spectral amplitude coding for optical codedivision multiple-access (i.e., called SAC/OCDMA) techniques are suggested to be a more flexible solution over optical local area networks (LANs) because the multiple access interference (MAI) is significantly cancelled [1-3]. Hence, the SAC/OCDMA scheme is suitable for burst and asynchronously environments. Moreover, the entire coded wavelength of each user is mixed via the optical coupler and then is capable of confidential enhancement. Therefore, the SAC-OCDMA scheme can provide more promise solution on fiber-to-thehome (FTTH) access network than wavelength-division multiplexing for Ethernet passive optical network (i.e., called WDM-EPON) [4].
Jen-Fa Huang and Che-Chih Hsu The Institute of Computer and Communication Engineering Department of Electrical Engineering National Cheng Kung University, Taiwan [email protected] The combined WDM and SAC/OCDMA scheme with fiber Bragg grating (FBG) based encoder/decoders (FBGbased codecs) over LAN was proposed by Yang [5]. By employing balanced incomplete block design code as the SAC code of each user group, the ability for MAI cancellation is preserved. Since the FTTH network is characterized by various distance between optical network unit (ONU) and optical line terminal (OLT), this scheme [5] is not directly applied to implement a flexible and confidential access network. In our previous done work by Huang et al. [6], the array waveguide grating based (AWG-based) codecs, which combined the inherent cyclic properties and maximal length sequence code (M-sequence code), was configured over optical local area networks (LANs). In current proposed WDM/OCDMA scheme, combining the inherent cyclic and free spectral range (FSR) periodic properties of AWG router, the AWG-based codec is shared by the ONUs which distributed and separated in various group (different area). Also, the M-sequence code, which is popular used for SAC/OCDMA scheme, is assigned as individual signature address code. Hence, under the consideration of costing, flexibility and scalability over FTTH network, the AWG-based codecs of proposed WDM/OCDMA not only is shared more ONUs (subscribers) but also set up flexible to different ONUs group. Compared to previous scheme [6], the proposed scheme applies the same NxN AWG router and achieves the more costing-efficiency performance over FTTH network. For SAC/OCDMA scheme, a crucial phase-induced intensity-noise (PIIN) is generated when mixed incoherent light fields are incident upon a photo-detector [3]. However, the PIIN noise has never investigated on the influence of the degree of polarization (DOP). In practical, the degree of polarization, P is not only dependent on the light source but also vary via the long haul network transmission. In order to achieve higher signal-to-PIIN noise ratio, in this study, we can set up a depolarizer in front of photodetector to ensure that the polarization-dependent properties of the photo-detector can be eliminated [7]. The remainder of this paper is organized as follows. In the section 2, AWG-based WDM/OCDMA network is
depicted. Especially, the signature codeword assignment with M-sequence is demonstrated under the cyclic and free spectral range (FSR) periodic properties of AWG router. Section 3 describes the AWG-based codecs of proposed WDM/OCDMA scheme and the illustrative example is demonstrated for multiple access interference (MAI) cancellation. The Section 4 evaluates the system performance in terms of the BER (bit error rate) and the maximum number of permissible simultaneous active ONUs. Especially, the PIIN noise affected the degree of polarization is investigated. Finally, Section 5 provides some concluding remarks.
2 2.1
The WDM/OCDMA Scheme The system configuration
In the current study, the flexible WDM/OCDM technique on EPON (i.e., the so-called WDM/OCDMAEPON) is set up in access and premise network shown as Fig. 1. Considering variable distributed area for users (custom), all the ONUs can be divided into specified groups and each group accommodate the same number of ONUs.
For the down stream transmitted from OLT, the data bit coming from each ONU is encoded by NxN AWG router and summed together in optical combiner. At the receiver end (ONU end), a pair of NxN AWG router, which is matched OLT AWG router and pre-written with Msequence as signature address codes, is used to implement optical correlation processes. The desired data bit of individual ONU is then recovered via balanced photodetector. Similarly, for the down stream, the symmetry and reciprocal architecture is set up implement bi-direction network. First of all, the proposed WDM/OCDMA scheme preserves the ability of multiple-access interference (MAI) cancellation. As seen in Fig. 1, since the FTTH network is characterized by various and unequal distance between ONUs and OLT, the proposed coder/decoder (codec) can be constructed with AWG router located on centralized OLT. Meanwhile, the corresponding codecs located on ONUs are constructed with AWG or fiber Bragg grating (FBG). Note that the FBG-based codecs are more suitable for a separated ONUs than a centralized ONUs. That is, the AWG-based ONU is selected when the position of supported ONU is approaching each other. Conversely, the FBG-based is selected when the position of supported ONU is isolated. Since the FBG-based codecs with Msequence have demonstrated by previous authors’ own work [8], we only focus on the AWG-based codecs in current study.
2.2 The Codeword Assignment
Fig. 1 The schematic diagram of proposed scheme configured by AWG and FBG based codecs for down stream example. AWG: array waveguide grating, FBG: fiber Bragg grating. The previous AWG-based codecs was configured over local area network by Huang et al. [6]. Here, by exploiting the inherent cyclic and FSR periodic properties for AWG routers, M-sequence code family is used as signature address code. Thus, for the down stream example shown as Fig. 1, a shared AWG router is located on centralized OLT. At the ONUs end, the matched AWGbased codecs is composed of a pair of corresponding AWG router, a pair of photo-detector and a decision unit. Hence, the AWG-based codecs with M-sequence scheme become simple and are easily implemented over FTTH network.
In the proposed WDM/OCDMA scheme, the ONU capacity is divided into G groups, where G is positive integer. and depend on the spectrum of light source. Every ONU is indexed as #(g, h) and assigned a code word Cg, h. The subscript of g denotes the g-th group which is divided by FSR interval range. The subscript of h denotes the h-th code sequence which is assigned as M-sequence code in the g-th group. Here, the g and h set belong to {1, 2, 3,…, G} and {1, 2, 3,…, N}, respectively. Let Mh(i)= (Mh(1), Mh(2), …, Mh(N)) be (0, 1) sequences of length N and the h-th sequence is followed by M-sequence code matrix, where h = 1, ..., N and i is the i-th chip (element) of Mh sequence. Thus, the j-th sequence of Mj(i) (i.e., j = 1, ..., N) can be obtained by cyclic shifting the original sequence of M1(i) (i.e., Mj(i) = T j-1 M1(i)), where T is the operator shifts vectors cyclically to the right by j-1 place). Let Fg(k)= (Fg(1), Fg(2), …, Fg(G)) is a vector whose the g-th element is one and others are zero, where k = 1, ...,G and the adjacent element of Fg(k) is equal to the FSR wavelength interval range of AWG router (i.e., so called Δλf). For example, the difference wavelength of Fg(1) and Fg(2) is one FSR (i.e., Δλf). Similarly, the difference of Fg(1) and Fg(3) is double FSR (i.e., 2*Δλf). Hence, the proposed WDM/OCDMA codeword of the h-th ONU in the g-th group is expressed as
Cg, h =[ Cg, h (m)] = [Mh(i)* Fg(k), …, Mh(i)*Fg(k)] (1) where m =1, 2, 3, …, G*N and Cg, h (m) is the m-th element of Cg, h and the symbol of * denote multiple operator. Based on Eq. (1) for M-sequence code length of N=7, the total coded spectrum of proposed WDM/OCDMA scheme is illustrated as Fig. 2. As seen in Fig. 2, the wavelength spectrum of light source is divided into G groups following the inherent FSR interval range of AWG router. Each group consists of seven wavelengths based on M-sequence of N=7. Hence, the codeword of Cg, h and Cg+1, h is located on the adjacent group. Also, the i-th and its corresponding wavelength (i.e., chip element) of adjacent group is far away one Δλf (i.e., equal to one FSR interval range). For simple example, there are seven ONU capacity from C g,1 to C g,7 in the g-th group. As a result, the Cg, 1 and Cg+1, 1 are characterized by the same M-sequence coding pattern of (1, 1, 1, 0, 0, 1, 0) in the adjacent group (i.e., in the group g and g+1) but different wavelength range. the Cg, 2 and Cg+1, 2 are characterized by the same M-sequence coding pattern of (0, 1, 1, 1, 0, 0, 1) but different wavelength range in the adjacent group of g and g+1. Note that the λ g , h denotes
signify information bit “0”. The unpolarized and broadband incoherent light source (BLS), whose spectrum is filtered within one free spectral range (FSR) of the AWG router, is employed. By applying the inherent cyclic and FSR periodic properties of AWG routers, the resulting optical signals of each ONU, which is assigned as M-sequence code, are directed to the corresponding input ports of 7x7 AWG-bsed encoder.
the h-th wavelength in group #g. Fig. 3 The proposed WDM/OCDMA techniques over FTTH for M-sequence codeword length of N=7 The each input port of AWG-based encoder accommodates the specified ONU which is assigned as the same M-sequence codeword but different wavelength group #g. At the AWG-based decoder, its corresponding output ports of 7x7 AWG-based decoder (original and complementary part) recover the data bit of desired ONU. That is, the AWG-based codecs is shared by all the ONU number of G*7 (i.e., 7 is codeword length of the assigned M-sequence) and illustrated as Table 1. TABLE 1.The shared AWG-based for M-sequence codeword length of N=7.
Fig. 2 The coded spectrum of proposed WDM/OCDMA scheme for M-sequence codeword length of N=7.
2.3
The AWG-based Encoder/Decoder
For simple explanation for WDM/OCDMA techniques over FTTH network, the 7x7 AWG routers is employed and shown as Fig. 3. As a result, there are G*7 ONU capacity share 7x7 AWG router. Here, G is positive integer and depends on the spectrum of light source. As seen in Fig. 3, in order to fulfill the E-O modulation, the data bits of each ONU #(g, h) is On–Off shift keying (OOK), the energy of which will be transmitted for information bit “1” and an absence of optical energy will
p
As seen in Table 1, the subscript of λ g , h denote the h-th wavelength in group #g relative to the spectrum of light source. Also, the superscript denotes the input port of 7x7 AWG-based encoder. The different group from ONU #(1, 1) to #(G, 1) signal combine together and enter into the same input port #1 of the 7x7 AWG-based encoder. Similarly, the different group from ONU #(1, 2) to #(G, 2) signal combine together and enter into the same input port #2 of the 7x7 AWG-based encoder. Here, each group is occupied seven wavelengths such that λ1 to λ7 is assigned in the group #1 and λG, 1 to λG, 7 is assigned in the group #G etc. For down stream example, the coded spectrum created on OLT is summed together via combiner and then transmit down to splitter located on ONUs. Since the summed optical signal is mixed up for simultaneous active ONUs, the proposed WDM/OCDMA is capable of confidentiality compared to conventional WDM-EPON and WDM-EPON even code division multiplexing EPON (CDM-EPON) scheme [6]. At the ONUs end, the wavelength-selective filter is used to separate the down stream into #g groups. Here, the range of wavelength selective filter is applied to filter out flexibly the desired wavelength from λg, 1 to λg, 7 (λg, i corresponding to λ(g-1)*7 + i ). Since the system performance of SAC scheme only depends on the correlation property of the underlined coding pattern (i.e., M-sequence code). The wavelength-selective filter can be selected in relaxed requirement and decrease the system costing. Subsequently, at the desired receiver (ONU group #g) end, a pair of AWG-based decoder pre-written with Msequence signature address codes is used to implement optical correlation processes. The desired data bit of “1” or ‘0” coming from ONU #(g, h) is then obtained via balanced photo-detector. Note that the scenario of present (on) or absent (off), which is wired on the input ports of AWG decoder, is employed with M-sequence code pattern for logic one or zero. Also, the input ports of the original and its complementary AWG-based decoder are matched the wired scenario on the output ports of AWG-based encoder. Compared to conventional EPON scheme over FTTH network, the qualitative comparisons is described as Table 2. TABLE 2. The comparisons of popular EPON scheme on several aspect.
3
MAI cancellation
The light sources used in this system are incoherent ideally un-polarized sources. Here, we use the unipolar sequences to encode the amplitudes of light source spectrum. Let Cg, h= (Cg, h(1), Cg, h(2),.., Cg, h(N)) be (0, 1) sequences of length N assigned as the codeword of ONU #(g, h), where the maximal of g*h is equal to G*N. Note that G, N denote the number of group and the number of M-sequence codeset (i.e., equal to code length, N), respectively. Based on the common rule for previous SAC OCDMA scheme with M-sequence code [8], the periodic correlation between Cg, h and Cr, s is defined as Rcc ( g , h , r , s ) and Rcc ( g , h , r , s ) , respectively. The proposed scheme satisfies the following relationship. Rcc ( g , h , r , s ) =
⎧( N + 1) / 2, g = r , h = s ⎪ ∑ C g, h ( i ) Cr , s ( i ) = ⎨( N + 1) / 4, g = r , h ≠ s ⎪ i =1 0, otherwise ⎩
(2-a)
⎧⎪ ( N + 1) / 4, g = h , r ≠ s otherwise ⎩ 0,
(2-b)
N
Rcc ( g , h , r , s ) =
N
∑ C g, h ( i ) Cr , s ( i ) = ⎨⎪
i =1
where the subscript g, h, r and s of C denote that the codeword C is assigned to ONU #h and #s in the g-th and r-th group, respectively. The symbol i denote the i-th chip for codeword C. Note that Ch is a unipolar M-sequence in the g-th group arbitrarily. The code length is equal to N and the codeword of the s-th ONU’s sequence from Ch can be written as Cs = TsCh by cyclic shifting the original sequence by s place, where T is the operator shifts vectors cyclically to the right by one place. That is, C2=TC1 = (C1(N), C1(1), …, C1(N-2), C1(N-1)). Following the Eq.(2), the proposed WDM/SACOCDMA retains the quasi-orthogonal property via the additional wavelength-selective filter which is characterized by relax requirement (i.e., less. costing) for achieving better flexibility and scalability. Thus, a balanced detector for desired ONU #(r, s) will implement the correlation subtractions expressed as follows: Z = Rcc ( g , h , r , s ) − Rcc ( g , h , r , s )
N +1 N +1 N +1 ⎧ , g = r, h = s ⎪⎪ 2 × 2 − 2 = 2 =⎨ ⎪ 2× N +1 − N +1 = 0 , otherwise ⎪⎩ 4 2
(3)
Eq. (3) is shown that MAI coming from other ONUs will be completely cancelled. With this excellent MAI-
∞
cancellation ability and the cyclic manner of M-sequence signature sequences, a compact WDM/OCDMA scheme with AWG-based codecs will operate without MAI theoretically.
4
The System Performance evaluation
In evaluating the performance of the proposed WDM/OCDMA scheme, the present study adopts a similar analysis model that applied by the conventional SAC scheme. The symbols which appear in the following discussions are defined in [3] and [9]. The present evaluation assumes that each light source is unpolarized and that the optical source is an ideal flat spectrum with a magnitude of Psr /Δν, where Psr is the effective power from a single source at the receiver and Δν is the optical source bandwidth in hertz.
4.1
Signal power evaluation
Following the result of Eq. (2), the PSD of the received optical signal can be written as N P G* N s ( v ) = sr ∑ b ( g,h ) ∑ C ( g,h ) ( i ) { rect ( i )} Δ v ( g,h ) =1 i =1
(4)
(5)
where N denote the codeword length of M-sequence and u is unit step function. The symbol of Δν is optical bandwidth. Assuming bit synchronism for now, the power spectrum density (PSD) at PD1 (photo-detector #1) of the desired ONU #(r, s) receiver during one bit period can be written as N P G* N G1 ( v ) = sr ∑ b( g,h ) ∑ C( g,h ) ( i ) C( r,s ) ( i ) { rect ( i ) } Δv ( g,h )=1 i =1
=
R Psr ⎛ N + 1 ⎞ b( r, s ) N ⎜⎝ 2 ⎟⎠
(7)
where R is the photodiode responsivity and b(r, s) ∈ {0,1}
4.2
PIIN power evaluation
This study evaluates only the PIIN noise which results from several thermal sources having a phase variation at each frequency. The PIIN noise results are compared with those obtained from a conventional SAC scheme for evaluation purposes. Referring to [3, Eq. (19)], the thermal noise is independent of the effective power of a broad-band source at the receiver (i.e., Psr) and the shot noise is only proportional to Psr. However, the phase induced intensity noise (PIIN) is proportional to Psr squared. Hence, it is shown that the effective power Psr from each ONU is large (e.g., Psr more than -10dBm), both the shot and thermal noises are negligibly small compared with PIIN noise, which becomes the main limitation factor of the system performance of signal to noise ratio (SNR). Hence, the variance of photo-detector current is expressed as ≈ I2 (1+P2)τc B
where (g, h) is the number of active ONU #(g, h) less than or equal G*N. Also, b(g, h) = “1” or “0” represents the data bit of ONU #(g, h). The rect(i) function in (4) is given by Δv Δv ⎡ ⎤ ⎡ ⎤ rect ( i) = u⎢v− v0− ( − N+ 2i− 2) ⎥ − u⎢v− v0− ( − N+ 2i) ⎥ 2N 2N ⎣ ⎦ ⎣ ⎦
∞
I = I1 − I 2 = R ∫ G1 ( v ) d v − R ∫ G2 ( v ) d v 0 0
(6-a)
(8)
where I,τc , B and P denotes the average photocurrent; the noise-equivalent electrical bandwidth of the receiver; the coherence time of the source and the degree of polarization (DOP), respectively. We temporally assume that the degree of polarization P is ideal on Eq. (8) such that ≈ I2 τc B. Hence, by substituting G12 ( ν ) and G22 ( ν ) into Eq. (7) and assuming all the ONUs are transmitting bit “1”, The result is given by: ∞ ⎡ ∞ ⎤ 2 i 2 = I PIIN = B R 2 ⎢ ∫ G12 ( v ) d v + ∫ G22 ( v ) d v ⎥ 0 0 ⎣ ⎦ ⎤ B R2 Psr2 N ⎧⎪ ⎡ N ⎨ ⎢ ∑ b ( g, h ) C ( g, h )( i ) ⎥ ∑ N Δ v i=1⎪ ⎢h=1 ⎥⎦ ⎩⎣
⎡N ⎤ ⎫⎪ ⎢ ∑ b ( r, s ) C ( r, s )( i ) ⎥ ⎬ ⎢⎣ s=1 ⎥⎦ ⎭⎪
⎫ P 2 B R 2 ⎧ ⎛ N +1 ⎞ = sr K ( K +1) ⎬ ⎨⎜ ⎟ N Δ v ⎩⎝ 4 ⎠ ⎭
(9)
Similarly, the PSD at PD2 (photo-detector #2) of the desired ONU #(r, s) receiver during one bit period can be written as
4.3 The Bit Error Rate (BER) evaluation
N P G* N G2 ( v ) = sr ∑ b(g , h) ∑ C(g ,h) ( i ) C(r , s ) ( i ) { rect ( i ) } Δv (g , h) =1 i =1
Dividing Eq. (7) by Eq. (9), the signal-to-noise ratio (SNR) and bit-error-rate, under the assumption that Gaussian approximation, is written as
(6-b)
Because of C(r , s) ( i ) = 1 − C(r , s) ( i ) and using Eq.(2), the signal from the desired ONU is given by the difference of the photodiode current outputs
SNR =
Ib =1 − Ib = 0 2 I PIIN
2 =
( N +1)Δν N B K ( K +1)
(10)
where Δν is optical band-width and K is number of active ONUs. As seen the Eq. (8), the degree of polarization, P, is defined as:
P
2
(s =
1
2
+ s2
2
+ s3
2
2
)
(11)
s0 In Eq. (11), s0, s1, s2, and s3 are referred to as the stoke parameters and are expressions of the state of polarization (SOP). The bracket < > denotes the average value of these parameters over wavelength, time or space. It is well known that the degree of polarization, P is not only dependent on the light source. Also, the degree of polarization, P is varied via the long haul network transmission. In order to inspect the DOP value in general light source, we select a common amplified spontaneous emission (ASE) as light source randomly and make simple experiment on back to back configuration. We connect ASE source to polarization analyzer (HP 8509C). The spectrum of unpolarized ASE light source is shown as Fig. 4(a) and the measured DOP approximates 0.035 at wavelength 1550 nm shown as Fig. 4(b). Note that the power of light source is properly attenuated.
Subsequently, by setting (i.e., selecting switch) internal linear polarizer of polarization analyzer (HP 8509C), the spectrum of polarized thermal source is shown as Fig. 5(a) and the detected DOP value is 0.989 approximately shown as Fig. 5(b). It reveals that the long haul network transmission become main factor to affect the DOP variation. In order to achieve higher signal-to-PIIN noise ratio, we can set up a depolarizer in front of photo-detector to ensure the polarization-dependent properties of the photo-detector can be eliminated. Hence, the average value of s1, s2, and s3 from Eq. (11) is approaching to zero and the degree of polarization (DOP) can be significantly decreased. That is, the depolarizer may remove the polarization sensitivity of the photo-detector for proposed WDM/OCDMA scheme in theory. For previous work done by Dale R. Lutz [7], a depolarizer was configured in front of balanced photodetector to achieve the DOP of 0.03 (i.e., P = 0.03).
(a)
(a)
(b) Fig. 5 The (a) spectrum and (b) measured degree of polarization (DOP) of polarized amplified spontaneous emission (ASE) light source.
(b) Fig. 4 The (a) spectrum and (b) measured degree of polarization (DOP) of un-polarized amplified spontaneous emission (ASE) light source.
In order to analyze the BER performance with and without depolarizer in convenience, we take the average DOP of 0.5 as a general scenario without depolarizer compensation after a long haul distance transmission. Following Eqs. (8) and (11), the SNR (signal power divided by PIIN power) affected by degree of polarization is evaluated and then established the corresponding BER (BER = (1/2)*erfc[(SNRPIIN / 8)1/2]) shown as Fig. 6.
As seen in Fig. 6, the BER performance of the proposed WDM/OCDMA scheme is characterized by an upper bound of P=1 for the worst case and a lower bound of P=0 for the ideal case. In particular, by placing a depolarizer in front of the photo-detector to achieve P=0.03, as configured by Dale R. Lutz [7], it can be seen that the curve of P = 0.03 virtually overlaps that of the proposed WDM/OCDMA scheme with M-sequence and HadmardWalsh code for the ideal case (P=0). Compared to the average DOP of 0.5 (i.e., the DOP value varies from 1 to 0 randomly), the active ONU is improved approximating 20%. Furthermore, it indicates that the BER performance with M-sequence is approximating with Hadamard-Walsh code when the active users is increased (i.e., SNR = Δν / B K ( K +1) with Hadamard-Walsh code). In particular, the number of AWGs with M-sequence code is significantly decreased due to the cyclic and FSR periodic properties.
Fig. 6 BER performance under various degree of polarization setting of P in proposed WDM/OCDMA scheme while Δν= 6.25THz, B=80MHz and codeword length of N=127.
5
Conclusions
Under the consideration of costing, flexibility and confidentiality, the WDM/OCDMA-EPON technique is demonstrated over FTTH network. Here, the AWG-based codecs is configured on optical network unit and optical line terminal, respectively. The ONU capacity is divided into groups and the M-sequence code is assigned as signature address code. By employing the inherent cyclic and free spectral range periodic properties of AWG router, the number of AWG router is significantly decreased because that the AWG-based codecs can be shared by the number of group. Furthermore, the various degree of polarization (DOP) is investigated to evaluate the signal to phase induced intensity noise ratio (SNRPIIN). By setting the depolarizer in front of balanced photo-detector, the system performance of simultaneous active ONUs is
approximately improved 20% compared to the average DOP of 0.5. In this study, the proposed WDM/OCDMA is more flexible, confidential and scalable than conventional TDMEPON and WDM-EPON scheme even code division multiplexing EPON (CDM-EPON). However, the additional wavelength-selective filter and depolarizer is used so that the system complexity is increased.
6
References
[1] Kavehrad, M. and D. Zaccarin, “Optical CodeDivision-Multiplexed Systems Based on Spectral Encoding of Noncoherent Sources,” IEEE J. Lightwave Technol., vol. 13, no. 3, pp. 534-545, March 1995. [2] X. Zhou, H. M. H. Shalaby, C. Lu, and T. Cheng, “Code for spectral amplitude coding optical CDMA systems,” Electron. Lett., vol. 36, pp. 728–729, Apr. 13, 2000. [3] Z. Wei; Shalaby, H.M.H.; Ghafouri-Shiraz, H., “Modified quadratic congruence codes for fiber Bragggrating-based spectral-amplitude-coding optical CDMA systems,” IEEE J. Lightwave Technol., vol. 19, no. 9, pp. 1274-1281, Sept. 2001. [4] S.-J. Park, S. Kim, K.-H. Song, and J.-R. Lee, “DWDM-based FTTC access network,” J. Lightwave Technol., vol. 9, pp. 1851-1855, Dec. 2001. [5] C. C. Yang, “Hybrid Wavelength-DivisionMultiplexing/Spectral-Amplitude-Coding Optical CDMA system,” IEEE Photonics Technology Letters, vol. 17, no. 6, pp. 1343-1345, June 2005. [6] C. C. Yang, J. F. Huang, and H. P. Tseng, “Optical CDMA Network Codecs Structured with M-Sequence Codes over Waveguide-Grating Routers,” IEEE Photonics Technology Letters, vol. 16, no. 2, pp. 641643, February 2004. [7] Dale R. Lutz, “A Passive Fiber-Optic Depolarizer,” IEEE Photonics Technology Letters, vol. 4, no. 4, April 1993. [8] J. F. Huang, and D. Z. Hsu, “Fiber-grating-based optical CDMA spectral coding with nearly orthogonal M-sequence codes,” IEEE Photon. Technol. Letter, vol. 12, pp. 1252-1254, Sept. 2000. [9] E. D. J. Smith, R.J. Baikie, and D.P. Taylor, “Performance enhancement of spectral- amplitudecoding optical CDMA using pulse-position modulation,” IEEE Trans. on Commun, vol. 46, no. 9, pp. 1176-1185, September1998.
