Performance enhancement of OCDMA systems for LAN consideration

IET Optoelectronics Research Article Performance enhancement of OCDMA systems for LAN consideration ISSN 1751-8768 Received on 13th April 2016 Revised 19th May 2016 Accepted on 5th June 2016 doi: 10.1049/iet-opt.2016.0044 www.ietdl.org Hichem Mrabet1 , Sofien Mthali2 1IT Department, College of Computation and Informatics Saudi Electronic University, Saudi Arabia Lab., Tunisia Polytechnic School, La Marsa, Tunis, Tunisia E-mail: [email protected] 2SERCOM Abstract: In this study, the performance of optical code-division-multiple-access (OCDMA) systems using 2D optical codes, taking into account an optical channel in local area network context is investigated. The authors demonstrate that one can enhance the 3-dB multi-mode fibre baseband bandwidth up to 4 GHz by exciting only two mode groups with the optimal axial launching. It has been demonstrated, that the OCDMA system performance can achieve a very low bit-error-rates in the range of −9 dBm. Furthermore, a large number of simultaneous users can be supported. In the authors’ treatment, the effect of both multiple access interference and modal dispersion are considered. Nonetheless, an OCDMA successive interference cancellation receiver is used to foster a better system performance. 1 Introduction In a matter of fact, it is known that optical code-division-multipleaccess (OCDMA) is an advantageous technique for sharing the total bandwidth of the optical channel. The common problems, in this case, are due to the effects of multiple access interference (MAI), optical beat noise and channel dispersion. Previous works have shown that optical beat noise can be minimised by choosing a longer code with minimal cross-correlation constraint [1]. OCDMA technique based on code attribution for each user using the system. There exist three families of OCDMA codes, 1D optical codes, 2D optical codes [2] and 3D optical codes [3]. The 2D optical codes are used to avoid problems associated with long codes and low weights of 1D optical codes. Several families of 2D optical codes, such as 2D OOC [4], prime hop system (PHS) and hybrid code (HC) [5] can be generated. In this paper, we try to investigate an analytic study with the help of the error probability of OCDMA system using 2D-OCDMA codes. The performance of OCDMA systems can be improved using a successive interference cancellation (SIC) based receivers compared with conventional correlation receiver (CCR) [1]. This improvement is translated into a very low error rates, and an enhancement of system performance. Recently, optical communication has shown to play an important role in multimedia transmission through the Internet using a fibre channel. In literature, there are two families of fibres, such as, multi-mode fibre (MMF) and single mode fibre (SMF). On one hand, MMFs are used in short distance, like local area network (LAN) and metropolitan area network [6, 7]. On the other hand, SMFs are used in long haul system, such as wide area network, and wavelength-division multiplexing system [8, 9]. According to IUT recommendation G651.1 [10] for MMF, one can support from 10 Mb/s to 10 Gb/s and hence their deployment as backbone architecture in LANs, storage area networks and wireless LANs [11]. Several applications using MMF in many fields are developed, such as, Ethernet [12], MIMO system using MGDM techniques [13] and fibre-channel redundant array of independent disks system [14]. Furthermore, there are two kinds of MMF related to the shape of index, such as, step-index used in short distance and graded-index to mitigate modal dispersion related to step-index and can reach a distance over then 100 km [15]. In LAN context, communication is characterised by a short distance and a small number of simultaneous (active) users, MMF is a good candidate used in the 850 and 1300 nm region [15]. In such networks, communications should support high levels of IET Optoelectron. © The Institution of Engineering and Technology 2016 security, low bit-error rates (BERs), and high data rates to ensure good quality of services (QoS). On one hand, the high-security levels can be assured through the use of 2D, time and wavelength domains, optical codes. On the other hand, low BERs in the range of −9 dBm can be achieved using MMF with a 3-dB baseband bandwidth that is more than 4 GHz (i.e. exciting a limited number of mode groups) [16, 17]. In addition, we have performed the modal dispersion of MMF in [16] with different kind of excitation techniques. Hence, this study is the logic continuity of our previous works [1, 16], and we try to investigate system performance of 2D-OCDMA codes using MMF channel in LAN context. We also try to employ the OCDMA SIC receiver to enhance the system performance. The rest of the paper is organised as follows. In Section 2, the system model is presented. In Section 3, we investigate a performance analysis of the OCDMA system, in terms of BER with considering an analytic model. Nonetheless, in Section 4, a power budget calculation of the proposed system is performed. Finally, our conclusions are deduced. 2 System model As shown in Fig. 1, the system model is composed of three main elements; an OCDMA transmitter, an optical star channel model and OCDMA receiver. After the channel model, an OCDMA receiver is used to recover the transmitted user data. 2.1 Transmitter model The OCDMA transmitter is composed of OCDMA encoder followed by a vertical-cavity-surface-emitting laser (VCSEL). In this work, a VCSEL is used to couple a small number of mode groups into the MMF core in order to improve the MMF baseband bandwidth and penalty signal in the LAN context. As depicted in Fig. 2, the transmitter architecture is composed of an OCDMA encoder, a multiplier and a VCSEL optical transmitter. The OCDMA encoder generates the 2D-OCDMA code ck(t), and a VCSEL transmitter converts the electrical signal to optical signal using an optical carrier. At the end of each user transmitter system we have the following output signal: �� � = �� � �� � exp �2��c� (1) 1 �(�, �) = �1(�) 1 − 2Δ(�) � � � 1/2 �1(�)[1 − 2Δ(�)]1/2 0≤�≤� (5) �≥� where r is an offset distance from the core centre, λ is the wavelength emitted by the laser, α is the index exponent, n1(λ) is the core peak index, Δ(λ) is the refractive-index contrast and a is the core radius of the fibre. The MMF is known to support a finite number of optical modes which can be analysed by solving Fig. 1 OCDMA system architecture Fig. 2 OCDMA transmitter architecture Fig. 3 Star channel model Where bk(t) is the kth user data, and ck(t) is the optical code of the kth user, exp() is the exponential function, and fc is the optical carrier of the VCSEL transmitter. It is known that for PHS and HC generated with a prime number (P), the OCDMA system can support P(P−1) and P(P + 1) users, respectively, and the cross correlation varies between 0 and 1 [1]. In our development, we employ both PHS and HC due to their attractive features such as the large multiplexing capacity and good correlation properties. 2.2 Channel model As shown in Fig. 3, the star channel model of the proposed system is composed of a coupler, a MMF channel and a number of spans block loop. The role of the number of the spans block loop is to provide the number of MMF to meet the length link user requirement. At the coupler output, the overall transmitted signal through the optical fibre s(t), can be expressed as: �(�) = � ∑ ��(�)��(�)���(�2����) (2) ℎmodal(�)����(� − �)exp(�2��c�) (3) �=1 where N is the total number of active users. The received OCDMA signal at the MMF output is given by: �(�) = � ∑ �=1 Where ϕ is the propagation delay caused by the MMF channel. The modal transfer function in terms of the inverse Fourier transform is given by: ℎmodal(�) = 1 2� ∫ +∞ −∞ �(�, �, �)e���d� . (4) In this study, we consider a graded-index optical MMF, characterised by a-class refractive index profile which can be expressed as: 2 Maxwell's equations. In this case, guided optical modes can be grouped into families of modes with propagation properties described by the same propagation constant. In an MMF, the total number of mode groups that can be guided is given by: �(�) = 2�� �1(�) �Δ(�) � �+2 1/2 . (6) The modal transfer function of the MMF is represented by [17]: �modal(�, �, �) = ∫ 1 1/� 2�ℜ(�, �, �, �)d� (7) where z is the fibre transmission length, ω represents the baseband angular frequency, ℜ(�, �, �, �) is the modal power distribution, and x is the normalised mode group number, defined as x = m/M(λ) with m being the principal mode number. Note that the mode number m is a discrete integer parameter which takes values ranging from unity to M(λ). The modal power distribution is described by the following power flow equation: ∂ℜ(�, �, �, �) = − ���(�, �) + �(�, �)ℜ(�, �, �, �) ∂� 1 ∂ ∂ℜ(�, �, �, �) + �d(�, �) � ∂� ∂� (8) where τ(x, λ) is the modal delay, γ(x, λ) is the modal attenuation, and d(x, λ) is the mode-coupling coefficient. Note that the power flow equation represents the modal power distribution as a function of frequency in Fourier domain. In fact, several techniques are used for exciting modes in MMF as the axial launching and the offset launching [18, 19]. In the case of axial excitation with Gaussian input beam, the excitation mode can be simulated by computing the launching efficiency as the overlap integral of the electrical field of each fibre mode with the electrical field of the incident light. A particular centre launching technique, named mode-field matched centre excites only the fundamental mode in MMF and avoids the inherent limitation caused by the differential mode delay [20]. In our study, we have considered the fibre axial excitation with a Gaussian input beam. The numerical aperture (N.A.) of silica IET Optoelectron. © The Institution of Engineering and Technology 2016 Fig. 4 Receiver architecture optical fibre is in the range of 0.16–0.65 whereas the numerical aperture of the beam spot is approximately given by the relation N.A. = λ/πw, with w being the spot beam radius measured from the core centre. We can reasonably assume that for a relatively large spot size, the N.A of the Gaussian beam is smaller than that of the optical fibre at w. In this case, the excited number of mode groups with the Gaussian beam is: � � �GB(�) = �(�) 2 + 2�2�21(�)Δ(�) −1 2 (� + 2)/(2�) � � . (9) An optimised Gaussian spot radius can lead to excitation of a small number of mode groups. Hence, the optimal spot radius is given by: 1 ���/2 �opt = 1/(� + 2) ��1(�) [�Δ(�)] 2/(� + 2) (10) . In [16], it was shown that an optimal axial launching technique can enhance the baseband bandwidth of the underlying MMF and also reduce the signal penalty. Furthermore, a spot radius equal to 10 μm which produces two excited groups of modes in silica optical fibre was shown to be optimal. 2.3 Receiver model As depicted in Fig. 4, the received architecture is composed of PIN receiver, and a CCR or SIC receiver. The PIN receiver convert the optical signal to electrical signal and generates the following signal: �1(�) = � ∑ �=1 ℎmodal(�)����(� − �) + �(�) (11) Where n(t) is the transceiver and the photo-detector noise. In this study, we consider the transceiver and photo-detector noise to be small and hence can be neglected. Knowing the desired code user ck(t) the CCR or SIC receiver give the estimated bit user ��(�). To reduce the effect of MAI, several multi-user detectors have been introduced in the literature. Among these detectors, the parallel interference cancellation and SIC receivers have shown to be promising [1]. In this paper, two types of OCDMA receivers are analysed; namely the OCDMA CCR and the OCDMA SIC based receiver. 3 BER analysis We consider a 50/125 μm MMF with short length equal to 5 Km operating at 1300 nm. We investigate the OCDMA system Table 1 System parameters Parameter undesired user threshold (ThN) desired user threshold (Th1) prime number (P) core/cladding radius MMF mode-coupling constant refractive index exponent intrinsic fibre attenuation total number of excited groups Value W W-2 17 50/125 µm 6.5 × 10–5 km–1 2 0.55 dB/km 25 IET Optoelectron. © The Institution of Engineering and Technology 2016 performance for both ideal and realistic MMF channels. Table 1 presents the system parameters used in our study. 3.1 Ideal channel For comparison purposes, we first consider the BER of OCDMA systems in ideal channels. Let the number of interfering users be modelled as a binomial distribution with parameter N−1 and PrI (the average probability of hits). Then, the probability of error can be expressed as [1]: Pe ≤ 1 �−1 � − 1 ( Pr )�(1 − Pr )� − 1 − � � � 2 � =∑�ℎ � (12) where Th is the receiver threshold value. The BER as a function of the minimum probability of error is then given by [21]: BER�� = 10log10 ��min (13) where Pemin is the minimum probability of error. Fig. 5 shows the BER of an OCDMA system using both the PHS and the HC as a function of the number of active users with the optimal threshold employed in the CCR and SIC with different interference cancellation stages and P = 17. It can be seen from Fig. 5 that the SIC receiver with PHS can support a larger number of simultaneous users than the CCR at a BER value equal to 10−9. The same observation is also clear for the HC case. We can say that when the number of users is increasing, then MAI is increasing also. In this case, the probability of error is increasing, but BER is decreasing. Hence, we can conclude that we have a degradation of OCDMA system performance, in terms of BER when the number of users is increasing. If we would increase the number of active users in OCDMA system, we should increase the parameters of 2D-OCDMA codes, called the prime number P. At a given prime number P, the OCDMA system can support P(P − 1) and P(P + 1) users, respectively, for the PHS and HC. As shown in Table 1, the prime number is taken equal to 17, hence the maximum theoretical number of users with considering MAI only is equal to 272 and 306, respectively, for PHS and PC as shown in Table 2. As depicted in Fig. 5, we consider only the number of active users for BER value equal to 10−9 to take the maximum advantage of the fibre at rate speed equal to 1 Gbit/s. If we want to increase the number of active users, so the BER value is decreased. As an example from Fig. 5, we can reach a number of users equal to 200 and 300, respectively, with a BER value equal to 10−3 and 10−4 for PHS and HC, respectively, when a CCR receiver is used. To make a comparison, in terms of system performance and to show the added value of our SIC receiver compared with CCR receiver, we add the Q-factor in dB related to the BER with the following expressing: � dB = 20log10( 2 erfc−1(2BER)) (14) According to Fig. 5, at 200 users for the PHS code, the BER value is equal to 10−3 and 10−6 for the CCR and the SIC receiver for 5 stages, respectively. Hence, the Q-factor when using (14) is equal to 12.59, and 14.98 dB, respectively. As a result, we can deduce that the SIC receiver can improve the system performance by about 14.98–12.59 = 2.39 dB for the given number of active users. As depicted in Fig. 5, at 300 users for the HC code, the BER value is equal to 10−4 and 10−7 for the CCR and the SIC receiver 3 Fig. 5 OCDMA system performances with an ideal channel (a) PHS, (b) HC Fig. 6 OCDMA system performances with MMF channel (a) PHS, (b) HC for 5 stages respectively. Hence, the Q-factor when using (14) is equal to 11.40, and 14.98 dB, respectively. As a result, we can conclude that the SIC receiver can improve the system performance by about 14.98–11.40 = 3.39 dB at 300 active users. 3.2 MMF channel When considering a realistic channel such as the MMF, both MAI and modal dispersion need to be examined. Here we examine the impact of modal dispersion caused by the propagation of several groups of modes in MMF standard core on system performance. The probability error of OCDMA system considering the MMF channel can be written as: Pe ≤ )d� 1 2 ∫ 1 1/� �−1 ∑ � = Th �−1 ( Pr� )�(1 − Pr� )� − 1 − � 2�ℜ(�, �, �, � � (15) Table 2 Users capacity with P = 17 and BER equal to –9 dBm Parameter PHS HC theoretical number ideal channel with CCR ideal channel with SIC MMF channel with SIC MMF channel with optimal axial launching 4 272 70 120 110 120 306 55 150 140 150 In Fig. 6, we present the BER of the OCDMA system using 2D optical codes and SIC receiver with three and five stages of interference cancellation and taking into account the effect of modal dispersion. The system performance is reported with the optimal axial launching technique for excitation in the case of P = 17. From Fig. 6, it can be observed that the number of simultaneous users is smaller than the ideal case for both the PHS and the HC due to the effects of modal dispersion. According to Fig. 6, at 200 users for the PHS code, the BER value is equal to 10−5 and 10−6 for the SIC receiver with three stages and the SIC receiver with five stages, respectively, with considering modal dispersion. Hence, the Q-factor when using (14) is equal to 11.4, and 12.59 dB, respectively. As a result, we can deduce that the SIC receiver with five stages can improve the system performance compared with SIC receiver with three stages by about 12.59–11.4 = 1.19 dB for the given number of active users. As depicted in Fig. 6, at 300 users for the HC code, the BER value is equal to 10−6 and 10−7 for the SIC receiver with three stages and the SIC receiver with five stages, respectively, with considering modal dispersion. Hence, the Q-factor when using (14) is equal to 12.59, and 13.54 dB, respectively. As a result, we can conclude that the SIC receiver with five stages can improve the system performance compared with SIC with three stages by about 13.54–12.59 = 0.95 dB at 300 active users. Fig. 7 shows the BER of the OCDMA system using 2D optical codes and SIC receiver with five stages of interference cancellation and different axial launching techniques (i.e. full mode excitation, optimal axial excitation, and mode field matched axial launching). From these results, one can see that the system can support a large IET Optoelectron. © The Institution of Engineering and Technology 2016 Fig. 7 OCDMA system performances as a function of various axial launching techniques (a) PHS, (b) HC Table 3 Bubget power calculation Parameter emitted average power (dBm) -connectors attenuation (dB) -star coupler attenuation (dB) -system margin (dB) -received average power (dBm) =available attenuation (a) � maximal fibre length ∝ LED VCSL −13 −2 −3 −6 −(−42) 18 5.14 Km 6.9 −2 −3 −6 −(−42) 37.5 10.82 Km number of active users for the PHS and HC when the optimal axial launching is used. However, the user capacity can be enhanced further by employing mode field matched axial launching with P = 17. According to Fig. 7, at 200 users for the PHS code, the BER value is equal to 9.4 10−7, 4.810−7 and 1.110−7 for the full mode excitation, optimal axial launching and mode field launching, respectively. Hence, the Q-factor when using (14) is equal to 12.62, 12.92 and 13.5 dB, respectively. As a result, we can deduce that the optimal axial launching can improve the system performance by about 12.92 –12.62 = 0.3 dB compared with full mode excitation for the given number of active users. As depicted in Fig. 7, at 300 users for the HC code, the BER value is equal to 1.310−7, 6.910−8 and 1.710−8 for the full mode excitation, optimal axial launching and mode field launching. Hence, the Q-factor when using (14) is equal to 13.44, 13.67 and 14.15 dB, respectively. As a result, we can conclude that the optimal axial launching can improve the system performance by about 13.67 –13.44 = 0.23 dB compared with full mode excitation at 300 active users. As shown in Table 2, we can reach a user capacity system equal to 120 and 150 for PHS and HC, respectively, when MMF channel is considered with optimal axial launching and a BER value equal to –9 dBm(10−9). 4 Power budget The cost performance of OCDMA system is calculated as optical power budget when using a LED and a VCSEL laser in the transmitter and a PIN for the receiver as shown in Table 3. As depicted in Table 3, the emitted average power of the transmitter is equal to 50 µW and 5 mW for the LED and the VCSEL respectively. In addition, The receiver average power of a PIN receiver is equal to –42 dBm for a BER equal to –9 dBm acceptable in optical context. According to Agrawal [15], the fibre cable loss(α) for MMF is equal to 3.5 dB/Km, the available attenuation for the proposed OCDMA system is equal to 18 and 37.5 for LED and VCSEL respectively. As a result, we retrieve that the maximal distance between the transmitter and the receiver for the proposed system is equal to 5.14 and 10.82 Km for LED, and VCSEL respectively. 5 We have analysed the performance of OCDMA systems with 2D optical codes taking into account the effect of both MAI and modal dispersion caused by the MMF channel in LAN context. We have demonstrated that optimal axial launching is upper bounded by the full mode excitation and lower bounded by the mode matched axial launching. It has shown also that using SIC-based receiver, the OCDMA system can deliver high QoS with BERs lesser than –9 dBm with large user capacity for both PHS and HC. A numerical result is given express that OCDMA system can reach a capacity user equal to 120 and 150 for PHS and HC, when taking into account MMF channel with optimal axial launching and MAI for a prime number equal to 17. 6 [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] IET Optoelectron. © The Institution of Engineering and Technology 2016 Conclusion References Mrabet, H., Dayoub, I., Attia, R., et al.: ‘Performance improving of OCDMA system using 2-D optical codes with optical SIC receiver’, J. Lightwave Technol., 2009, 27, (21), pp. 4744–4753 Wang, X., Wada, N., Hamanaka, T., et al.: ‘10-user, truly-asynchronous OCDMA experiment with 511-chip SSFBG En/decoder and SC-based optical thresholder’. Proc. of the Optical Fiber Communication Conf., Anaheim, CA, 2005 McGeehan, J.E., Nezam, S.M.R.M., Saghari, P., et al.: ‘Experimental demonstration of OCDMA transmission using a three dimensional(timewavelength-polarization) codeset’, J. Lightwave Technol., 2005, 23, (10), pp. 3282–3289 Saghari, P., Omrani, R., Willner, E., et al.: ‘Analytical interference model for two-dimensional (time-wavelength) asynchronous O-CDMA systems using various receiver structures’, J. Lightwave Technol., 2005, 23, (10), pp. 3260– 3269 Lin, S., Huang, J., Yang, C.: ‘Optical CDMA network codecs with merged-Mcoded wavelength-hopping and prime-coded time-spreading’, Opt. Fiber Technol., 2006, 13, (2), pp. 117–128. Chen, H., van den Boom, H.P.A., Tangdiongga, E., et al.: ‘30-Gb/s bidirectional transparent optical transmission with an MMF access and an indoor optical wireless link’, PTL., 2012, 24, (7), pp. 572–574 Mikroulis, S., Omomukuyo, O., Thakur, M.P., et al.: ‘Investigation of a SMFMMF Link for a remote heterodyne 60-GHz OFDM RoF based gigabit wireless access topology’, J. Lightwave Technol., 2014, 32, (20), pp. 3645– 3653 Xia, T.J., Wellbrock, G.A., Huang, M., et al.: ‘Transmission of 400 G PM-16QAM channels over long-haul distance with commercial all-distributed Raman amplification system and aged standard SMF in field’. Proc. of Optical Fiber Communications Conf. and Exhibition (OFC), 2014, pp. 1–3 Zhang, J., Yu, J., Dong, Z., et al.: ‘Transmission of 20 × 440-Gb/s superNyquist-filtered signals over 3600 km based on single-carrier 110-GBaud PDM QPSK with 100-GHz Grid’. Proc. of Optical Fiber Communications Conf. and Exhibition (OFC), 2014, pp. 1–3 ITU-T Recommendation G.651.1: ‘Characteristics of a 50/125 μm multimode graded index optical fiber cable for the optical access network’, 2007 Sauer, M., Kobyakov, A., George, J.: ‘Radio over fiber for picocellular network architectures’, J. Lightwave Technol., 2007, 25, (11), pp. 3301–3320 ITU-T manual: ‘Optical fiber, cables and systems’, 2009 5 [13] [14] [15] [16] [17] [18] 6 Baklouti, F., Dayoub, I., Haxha, S., et al.: ‘Novel method for improving the capacity of MIMO system using MGDM’, IEEE Photonics J., 2014, 6, (6) Tan, Z., Xie, C., Wan, J.: ‘Performance benchmark for target module in fiber channel RAID systems’. Proc. of IEEE Conf. on High-Performance Switching and Routing, 2008 Agrawal, G.P.: ‘Fiber-optic communication systems’ (Wiley, 2012, 4th edn.) Mrabet, H., Dayoub, I., Attia, R., et al.: ‘Wavelength and beam launching effects on silica optical fiber in local area networks’, J. Opt. Commun., 2010, 283, (21), pp. 4234–4241 Mrabet, H., Dayoub, I., Attia, R.: ‘Performance analysis of OCDMA systems with 2-D optical codes in MMF channels’. Proc. of TELECOM2013 and 8th JFMMA, Marrakech, Morocco, 2013 Yabre, G.: ‘Comprehensive theory of dispersion in graded-index optical fibers’, J. Lightwave Technol., 2000, 18, pp. 166–177 [19] [20] [21] Raddatz, L., White, I.H., Cunningham, D.G., et al.: ‘An experimental and theoretical study of the offset launch technique for the enhancement of the bandwidth of multimode fiber links’, J. Lightwave Technol., 1998, 16, (3), pp. 324–331 Sim, D.H., Takushima, Y., Chung, Y.C.: ‘High-speed multimode fiber transmission by using mode-field matched center-launching technique’, J. Lightwave Technol., 2009, 27, pp. 1018–1026 Salehi, J., Brackett, C.: ‘Code division multiple-access techniques in optical fiber networks—Part III: optical AND logic gate receiver structure with generalized optical orthogonal codes’, IEEE Trans. Commun., 2006, 54, (8), pp. 1457–1842 IET Optoelectron. © The Institution of Engineering and Technology 2016