Open Access
Add to BibSonomyAbstract
We describe experimental demonstration of spectrum-sliced elastic optical path network (SLICE) architecture. We employ optical orthogonal frequency-division multiplexing (OFDM) modulation format and bandwidth-variable optical cross-connects (OXC) to generate, transmit and receive optical paths with bandwidths of up to 1 Tb/s. We experimentally demonstrate elastic optical path setup and spectrally-efficient transmission of multiple channels with bit rates ranging from 40 to 140 Gb/s between six nodes of a mesh network. We show dynamic bandwidth scalability for optical paths with bit rates of 40 to 440 Gb/s. Moreover, we demonstrate multihop transmission of a 1 Tb/s optical path over 400 km of standard single-mode fiber (SMF). Finally, we investigate the filtering properties and the required guard band width for spectrally-efficient allocation of optical paths in SLICE.
©2010 Optical Society of America
1. Introduction
Recent developments in optical modulation and transmission technologies have opened new horizons for transparent transmission in optical networks. These developments, in conjunction with the growing capacity of optical cross-connects (OXCs), enable the end-to-end provisioning of optical network resources [1]. When deployed as a transparent mesh network, the optical end-to-end connections can be reconfigured upon request, thereby forming a dynamic wavelength pool to support client’s bandwidth demands [2].
From the network client’s side the growing demand for the capacity and agility of interconnects stems from the existing as well as emerging network services. New, high-end applications, such as grid computing [3], storage area networks [4], or distribution of ultra-high definition TV (UHDTV) and digital cinematic productions [5] will require bandwidths in excess of the currently existing interfaces. Moreover, the envisaged connection time scale is expected to be shorter than that of the optical circuits, which nowadays are configured manually and require significant lead time to set up. In order to support the coming client demands, new standards such as the higher-rate Ethernet (100 GbE) are being developed [6]. A similar scenario can be applied to future 400 Gb/s or 1 Tb/s optical transport networks which will accommodate traffic of e.g. Terabit local area networks [7]. Interfaces adhering to those standards will provide large bandwidth transmission pipes. However, in the case when the service does not occupy the entire allocated bandwidth, the remaining part will become a wasted resource. This results in stranding of bandwidth which may become an issue in highly spectrally-efficient networks.
In order to introduce some flexibility into the otherwise fixed, large bandwidth optical transmission system, several approaches have been presented in the literature. Among others is the optical packet switching (OPS) providing granularity in the time domain [8]. The OPS approaches the problem of providing the bandwidth in a flexible manner but its development has not reached the stage of deployment. The dominant problem is the lack of viable optical buffer technology. Another approach is the mixed-rate transmission [9]. It envisages a link capacity upgrade and simplification of node control, but does not provide the necessary flexibility for efficient accommodation of dynamically changing traffic flows.
In order to address the issue of stranded bandwidth in optical transmission systems, we proposed a network architecture concept based on optical spectrum slicing [10]. The architecture, called spectrum-sliced elastic optical path network (SLICE) is presented in Fig. 1 .
Fig. 1 The concept of spectrum-sliced elastic optical path network (SLICE). Flexible-rate transceivers allocate end-to-end optical paths with right-sized bandwidth adjusted to traffic demand. The optical paths are transmitted through bandwidth-variable optical cross-connects (BV OXC) which enable allocation of optical spectrum exactly matching the capacity of optical paths.
SLICE uses the spectrum domain (as opposed to time domain) to provide the flexibility to the optical channel. The bandwidth is allocated to the service according to the demand by ‘slicing off’ the spectral resources. This lies in contrast to the fixed-bandwidth allocation scheme, in which the same, maximum bandwidth (spectrum) is allocated regardless of the current user needs. In the scenario presented in Fig. 1, the traffic is established between the nodes arranged in mesh network. The routers are connected to SLICE network edge nodes with optical interfaces (IF) supporting bit rates of up to 100 Gb/s (e.g. 100 GbE). The peak volume of traffic on each of the interfaces ranges from 30 Gb/s to 100 Gb/s. In order to efficiently transmit the varied traffic through the optical core, the SLICE network allocates end-to-end optical paths in OXCs with bandwidth matching each of the traffics while the SLICE transmitters generate the optical paths with the corresponding spectral width. Each of the optical paths can be switched independently, and when the demand for traffic changes, the bandwidths can be adjusted, accordingly. By using only the necessary part of spectral resources and by flexibly adjusting to the traffic volume, SLICE offers several advantages when compared to the current architectures. It provides a spectrally-efficient platform for transparent optical data transmission. SLICE also enables direct allocation of traffic flows onto optical paths without constraining the bit rate of the channel. Therefore, it enables the accommodation of terabit class flows without employing ultra-high frequency components or optical time-division multiplexing. Finally, SLICE can potentially increase the network reliability through path restoration with bandwidth squeezing [11].
In this paper we report the experimental demonstration of the concept and functionality of SLICE. In Section 2, we present the key technologies enabling the realization of SLICE architecture. In Section 3, we explain the experimental setup and present the results of generation, transmission and detection of elastic optical paths with bit rates of up to 1 Tb/s. In Section 4, we discuss the filtering properties of OXC providing the elastic optical paths and the spectral allocation. Finally, in Section 5 we summarize and conclude the results.
2. Slice key technologies
The concept of SLICE network is based upon the efficient allocation of traffic to elastic optical paths and the establishment of end-to-end paths in OXCs covering the corresponding optical spectrum. This concept is realized using two key technologies – the bandwidth-variable (BV) OXC and BV transponder. The BV OXC enables setting up the elastic optical path end-to-end with bandwidth exactly accommodating the spectrum of the path. It is schematically depicted in Fig. 2(a) . The OXC is a broadcast-and-select switch, employing optical splitters at the input and wavelength-selective switches (WSSs) at the output ports.
Fig. 2 Key technologies for SLICE network – BV OXC: (a) OXC with continuously BV wavelength-selective switches (WSS); comparison of spectra of: (b) optical OFDM path; (c) conventional WDM channels; (d) ultra-high speed single channel.
The WSS is a 1xN switch or filter providing continuously-tunable and variable seamless transmission spectrum. Such functionality can be realized using liquid crystal (LC) or micro-electro mechanical systems (MEMS) technology [12–14]. The architecture of the BV OXC enables forwarding of channels with arbitrary spectral widths to arbitrary output ports, add and drop as well as broadcasting functionality. This functionality is illustrated with the spectra of optical channels at the inputs, outputs, and the add/drop ports of the OXC in Fig. 2(a).
The second key technology of SLICE is the BV transponder. The transponder has to provide variable granularity in the spectral domain and enable dynamic adjustment of capacity according to the service demand. In SLICE, optical orthogonal frequency-division multiplexed (OFDM) modulation [15] is employed to enable flexible bandwidth adjustment of the signal while maintaining high SE of transmission. The OFDM signal consists of multiple subcarrier signals, each modulated at a rate lower than the aggregate bit rate of the channel. The subcarriers are optically multiplexed in the frequency domain. This approach is an alternative to the OFDM signals generated by inverse Fourier transform in the electrical domain [16,17] and it alleviates the limitation of the optical path capacity imposed by the electrical bandwidth of transmitter’s digital-to-analog converter.
The spectrum of OFDM-modulated signal is schematically presented in Fig. 2(b). The subcarriers of the OFDM signal overlap spectrally, which is in contrast to the dense wavelength division multiplexed (DWDM) transmission shown in Fig. 2(c). Maintaining the integer ratio between the subcarrier spacing and the symbol rate ensures that the OFDM carriers remain orthogonal (mutually independent), and can be recovered error-free at the receiver. At the same carrier modulation speed, the OFDM signal provides higher overall SE than DWDM as no spectral guard band is required between the subcarriers. Also, when compared to the single-channel ultra-high speed modulation [Fig. 2(d)], OFDM signal provides higher flexibility in adjustment of bandwidth.
The diagram of the BV transmitter is shown in Fig. 3 . The client data entering the SLICE network IF is split into low bit rate data streams which are fed into individual modulators. The modulators imprint the data streams onto the CW subcarriers generated by the multicarrier source. The carriers are subsequently multiplexed to form optical OFDM signal. Carrier spacing and modulation bit rate are frequency-locked in order to satisfy the orthogonality condition of OFDM [18]. The optical no-guard interval optical OFDM [19] modulation employed in SLICE allows maintaining the bit rate of individual subcarriers low enough to mitigate the influence of transmission impairments stemming from accumulated chromatic dispersion (CD) and polarization-mode dispersion (PMD). Since, it is does require pilot tone or guard interval, it does away with the overhead and allows better utilization of the optical spectrum than the traditional OFDM.
If the client signal bandwidth exceeds the maximum rate of the transponder, it may be distributed over a number of lower rate BV transmitters. The outputs of the transmitters would be aggregated into a spectrally continuous optical path, such as presented in [20]. This allows treating the transmitters as a resource pool shared among multiple optical paths. With transponders deployed according to the amount of traffic added and dropped in a given node, such an approach enables a small start while ensuring scalability to multi-hundred Gb/s optical paths.
3. Experimental demonstration of SLICE
3.1 Experimental setup
In the experimental demonstration of the SLICE concept, we generated, transmitted and received OFDM elastic optical paths with varying bit rates. The paths were generated using the setup shown in Fig. 4(a) . A continuous wave (CW) source, emitting at 1576 nm, was launched into a multicarrier generator consisting of a discrete Mach-Zehnder modulator (MZM) and discrete phase modulator producing a comb of evenly spaced carriers. The modulators were driven by a 10 GHz clock synchronized with the pulse pattern generator (PPG). The multicarrier spectrum was equalized to obtain 45 equal power optical carriers spaced by 10 GHz [Fig. 4(b)]. The CW carriers were split into even and odd channels by an interleave filter (ILF). A MZDI with free spectral range (FSR) of 20 GHz and extinction ratio of 35 dB was used as the ILF. The odd and even channels were differential phase-shift keying (DPSK) modulated with two different 10 Gb/s pseudo-random bit sequence (PRBS) signals. The PPG driving the modulators was synchronized with the clock driving the multicarrier generator, thereby assuring the orthogonality between the subcarriers. The spectrum of modulated carriers in the even branch of the modulator is shown in Fig. 4(c). Subsequently, the signals in both branches were bit-aligned and coupled to form the optical OFDM signal.
Fig. 4 Experimental setup: (a) optical OFDM DPSK transmitter; optical spectra of: (b) equalized multicarrier comb, (c) modulated subcarriers in even channels, (d) modulated even and odd channels merged into optical OFDM signal. CW – continuous wave, EQL – equalizer, PPG – pulse pattern generator.
The coupled subcarriers were co-polarized. The spectrum of the OFDM signal carrying 45 DPSK OFDM subcarriers is shown in Fig. 4(d).
In the experiment, we simultaneously transmitted multiple optical paths. However, because only one OFDM signal source was available, we divided the generated spectrum into multiple optical paths with bit rates ranging from 40 to 140 Gb/s. In order to split and distribute the signal to the add ports of the nodes, we used a commercially available WSS. The LC-based WSS was a 1x9 device with minimum addressable bandwidth of 50 GHz. The WSS filter roll-off characteristics approximated the Gaussian shape of the order 2.5 at the 3 dB bandwidth of 50 GHz. The employed WSS enabled concatenation of neighboring channels, forming a continuous transmission spectrum. The transmission properties of the WSS resulted in one carrier being suppressed, forming a guard band (GB) between the neighboring paths. The subcarriers in the optical paths were not aligned to the ITU-T frequency grid. Instead, we employ the term frequency slot, as described in [21]. The filtering properties and necessary GB are discussed in more detail in Section 4.
At the receiver side, shown in Fig. 5(a) , non-coherent detection of individual subcarriers was employed [22]. The incoming DPSK OFDM optical path [Fig. 5(b)] entering the receiver was split into even and odd subcarriers [Figs. 5(c) and 5(d)] by a MZDI with 1/2 bit delay (FSR of 20 GHz). An MZM following an automatic polarization adjuster and driven by a 10 GHz clock was used as an optical gate. An arrayed waveguide grating (AWG) was used to select a single subcarrier [Fig. 5(e)]. The AWG provided a Gaussian transmission spectrum with 3-dB bandwidth of 6.5 GHz and channel spacing of 25 GHz. The isolated subcarrier was received using the standard DPSK receiver consisting of a 1 bit delay MZDI, balanced photodiode (BPD), clock and data recovery (CDR) and bit error rate tester (BERT). The Q-factor performance was calculated from the obtained BER. Since the AWG channel spacing did not match the OFDM subcarrier spacing, only one channel at a time could be received. The desired channel could be selected by controlling the temperature of the AWG and choosing the appropriate port. A demultiplexer matched to the carrier spacing would enable the simultaneous reception of all odd subcarriers provided that each output port were equipped with a DPSK receiver. Receiving the even channels would require placing a similar set of polarization adjuster, optical gate and AWG (shifted by 10 GHz with respect to the odd branch) followed by DPSK receivers at the second output of the T/2 MZDI. It should be noted that in the above case, the simultaneous reception of all carriers would require compensation of CD in order to remove the skew between the subcarriers of the path.
Fig. 5 (a) Experimental DPSK OFDM receiver setup; (b) optical spectra corresponding to the input of the receiver; (c) even and odd (d) subcarriers demodulated by 1/2 bit delay Mach-Zehnder delay interferometer (MZDI); (e) subcarriers after the arrayed-waveguide grating (AWG) filter. The received signal waveforms are shown without and with the optical gate (f and g, respectively). BPD – balanced photodiode, CDR – clock and data recovery, BERT – bit-error rate tester.
The T/2 MZDI, optical gate and the narrowband filter used in the receiver serve as an optical equivalent of the discrete Fourier transform (DFT) performed in the DSP to separate the individual subcarriers [18,23,24]. The waveforms of the received signal without and with the optical gate are shown in the Fig. 5(f) and Fig. 5(g), respectively.
In order to ensure the error-free performance of the elastic optical path employing DPSK OFDM modulation a number of parameters needs to be optimized. The OFDM signal reception employing the 1/2 bit delay interferometer and the optical gate requires that the signals in the neighboring channels be bit-aligned. The misalignment between the channels leads to eye closure and penalty, as plotted in Fig. 6(a) . In order to limit the penalty to less than 1 dB, it is necessary to ensure that the bit misalignment does not exceed ± 0.2 bit period, which for a 10 Gb/s signal corresponds to ± 20 ps. Accordingly, the skew between the neighboring subcarriers due to accumulated chromatic dispersion must be less than ± 20 ps. For subcarriers spaced by 10 GHz this corresponds to compensating the residual CD to within ± 250 ps/nm.
Fig. 6 (a) Receiver OSNR penalty as a function of bit position misalignment between even and odd channels of OFDM signal; (b) OSNR penalty as a function of coherent cross-talk; (c) OSNR penalty as a function fiber input power level.
Another source of penalty may be the coherent cross-talk within the optical channel. In the presented OFDM transmitter, the multicarrier signal is split into even and odd branches for individual modulation. If the FSR or spectral alignment of the ILF does not match that of the optical carrier frequencies, the suppression of odd (even) carriers in the even (odd) branch may be incomplete, leading to coherent cross-talk after carrier coupling. The cross-talk results in a penalty at the receiver. We deliberately detuned the ILF and analyzed the optical signal-to-noise ratio (OSNR) penalty due to the coherent cross-talk between the OFDM subcarriers. The results are plotted in Fig. 6(b). In order to limit the penalty to less than 1 dB, it is necessary to suppress the offending carriers to −13 dB. This is approximately 3 dB higher cross-talk power for 1 dB penalty than reported for DPSK signals [25]. The better cross-talk performance may be explained by the phase locking between the clock generating the subcarriers and the modulation speed.
The third investigated parameter is the optical input power launched to the fiber. The transmission of the narrowly spaced OFDM subcarriers over the low-CD fibers, such as dispersion-shifted fiber (DSF) (ITU-T G.651), is likely to cause four-wave mixing (FWM) interaction between the subcarriers and lead to a penalty. We investigate the level of nonlinear penalty by transmitting the 440 Gb/s elastic optical path over a 50-km long span of DSF at power levels ranging from −8 to 0 dBm per subcarrier and comparing the Q-factor performance to the back-to-back case. The results are plotted in Fig. 6(c). For an input power of up to −8 dBm per subcarrier the nonlinear penalty is not observed. Increasing the power gives rise to penalty, reaching 0.2 dB for −4 dBm and 1.7 dB for 0 dBm input power. It should be noted that only two data patterns were used for modulating the odd and even subcarriers. This is likely to increase the level of nonlinear interactions such as FWM between the carriers. The exact level of nonlinear penalty depends on the number and spacing of the subcarriers as well as the number of data patterns and will be investigated elsewhere. Basing on the obtained results, in the transmission experiment we maintained the fiber input power below −8 dBm.
The elastic optical paths were transmitted between nodes of a transparent mesh network. The network is schematically presented in Fig. 7 . Six network nodes (A~F) are the broadcast-and-select type OXCs, with the architecture shown in Fig. 2(a). The nodes are interconnected with pairs of 50-km long spans of DSF (one for each direction) indicated by lowercase letters a~e. The end nodes of paths α~λ are shown in the table in Fig. 7. The paths with bandwidths determined by the distributor of OFDM transmitter are launched into the network through the add ports of the respective nodes. The ports of the distributor shown in Fig. 4 are devoted to supply the optical paths to the respective add ports of nodes A, C, E, and F. The transmitted paths are added, dropped and multicast within the OXCs and the Q-factor performance of the subcarriers is analyzed in node F.
Fig. 7 Experimental network setup – six nodes A~F arranged in mesh topology are connected by links a~i. The table represents connections established in the network experiment
3.2 Experimental results
The spectra and the assignment of the elastic optical paths transmitted through the network are shown in Fig. 8 . The spectra of paths in links a~e correspond to transmission in the direction from West to East in Fig. 7, while the spectrum of link b’ corresponds to the transmission in the opposite direction. The highest SE considering the amount of allocated traffic with respect the occupied bandwidth was 0.85 b/s/Hz, obtained in link c.
Fig. 8 Optical spectra of paths traversing the network shown in Fig. 7 captured in respective links a~e. The lower plot shows the Q-factor performance of path δ at node F.
The lower part of Fig. 8 illustrates the Q-factor performance of path δ observed in node F. The bit rate of path δ is 140 Gb/s. It is added in node A, forwarded in node C and multicast to nodes B and F in node D. The total transmission distance is 150 km. The Q-factor performance of all 14 subcarriers is above the limit set by the forward error-correction coding (FEC) of 9.1 dB. The subcarriers indicated by triangles in the figure were measured to have Q-factor performance better than 16.5 dB.
In addition to the demonstration of multiple optical path transmission, we also show the bandwidth scaling of the elastic optical path. The optical path was added in node A, transmitted and multicast before being received in node F (see Fig. 7). The bit rate of the path was scaled from 40 to 440 Gb/s and the corresponding spectra are shown in Fig. 9(a) . The path bit rate was scaled from 40 Gb/s in steps of 50 Gb/s due to the minimum channel bandwidth of the WSS (50 GHz) and the presence of GB of 1 subcarrier. The Q-factor performance of the 440 Gb/s optical path is analyzed in node F after transmission over 150 km of DSF. The results are shown in Fig. 9(b) where the Q-factor values are plotted for 44 subcarriers. All subcarriers perform above the FEC Q limit and the Q-factor value of subcarriers indicated by triangles is higher than 16.5 dB. A deterioration of performance of the optical path edge carriers can be observed in the results shown in Fig. 8 and in Fig. 9. The cause of the penalty is the filtering in the WSS. This phenomenon is further investigated in Section 4.
Fig. 9 Optical spectra of bandwidth-scalable path carrying 40~440 Gb/s traffic. The lower plot shows the Q-factor performance of the 440 Gb/s optical path dropped at node F.
The scaling of optical path bandwidth was also demonstrated employing multiple optical paths. The spectra of 5 optical paths carried on link b’ (network in Fig. 7) are shown at consecutive instants in Fig. 10(a) through Fig. 10(d). As a dynamic case is considered, the allocation of paths differs from that of table shown in Fig. 7. In Fig. 10(a), three paths are allocated: α (40 Gb/s), λ (90 Gb/s), and ι (140 Gb/s). In the instant shown in Fig. 10(b), path ι is spectrally shifted in order to allocate spectrum to paths β and κ, each carrying traffic of 40 Gb/s. The instant of all paths being allocated is shown in Fig. 10(c). Eventually, in the instant presented in Fig. 10(d), path κ is dropped and the freed spectrum is allocated to path ι whose bandwidth demand increased from 140 Gb/s to 190 Gb/s. The bandwidth of optical paths, their spectral locations as well as the setup of OXCs is controlled by a routing and spectrum-assignment (RSA) algorithm, managing the spectral resources of SLICE. The RSA algorithm is a target of future study.
Finally, in order to the show the ability of SLICE to carry ultra-high capacity services, we demonstrate the generation, transmission, and reception of a 1 Tb/s elastic optical path. A 1 Tb/s optical path may be generated through optical aggregation, or by using a single 1 Tb/s transmitter. In the experiment we employ the second approach. The setup is shown in Fig. 11(a) . The 1 Tb/s optical path is formed by 50 OFDM subcarriers, each carrying 21.4 Gb/s differential quadrature phase-shift keying (DQPSK) modulation. The transmitter setup is similar to that shown in Fig. 4(a). In this case the optical modulators were driven to generate the multilevel signal. The spectrum of generated signal covering the bandwidth of 4.4 nm is shown in Fig. 11(b). The path was transmitted in a recirculating loop. The loop consisted of 20 km of standard single-mode fiber (SMF), a coupler and a WSS forming the OXC, an optical spectrum equalizer (EQL) and optical amplifiers. The filter in the WSS was open to pass the entire spectrum of the 1 Tb/s optical path. After exiting the loop, the signal was transmitted through a tunable optical dispersion compensator (TODC) to compensate the residual CD. Finally, the path was received using the setup shown in Fig. 5(a). We analyzed the OSNR performance of the edge subcarrier at BER of 1x10−3. The results are plotted in Fig. 11(c). The OSNR penalty is plotted as a function of the number of transmitted loops. The
Fig. 11 1 Tb/s transmission recirculating loop experiment: (a) experimental setup; (b) 50 subcarrier optical signal spectrum; (c) OSNR penalty as a function of the number of loops.
subcarriers of the 1 Tb/s path experience a 1 dB OSNR penalty after approximately 13 hops, corresponding to the transmission distance of 260 km. In this experiment we analyzed the performance of the subcarrier for up to 20 hops (400 km) at which point the OSNR penalty reached 2 dB.
In the described experiments, we demonstrated the transmission of multiple elastic optical paths in a mesh network, scaling of optical paths from 40 to 440 Gb/s as well as transmission of a 1 Tb/s optical path over 400 km. The results prove the feasibility and the flexibility of the SLICE platform. In the obtained results, a variation of Q factor performance can be observed between the individual carriers of the OFDM signal. This is especially clear in the case of the edge subcarriers. In order to address this problem, in the next section, we investigate the filtering properties of the elastic optical paths.
4. Filtering properties
In the plots representing the Q-factor values in Fig. 8 and Fig. 9(b), it can be observed that the edge subcarriers suffer an excess Q penalty. In here, we consider two sources of penalty, as presented in Fig. 12 . In Fig. 12(a), the spectra of 3 optical paths allocated to a single link are shown. Figure 12(b) focuses on the optical spectrum corresponding to the overlap point between two directly neighboring paths. The paths are separated by a GB of idle spectrum equal to the bandwidth of a single subcarrier. The modulation sideband of the edge subcarriers of both paths overlap. After several add/drop stages in the network nodes the incoherent cross-talk between the neighboring channels will reduce the Q-factor performance of the edge subcarrier. This penalty may be mitigated by applying a narrower filter. This, however, affects the spectrum of the edge subcarrier leading to the filtering penalty, as illustrated in Fig. 12(c). The penalty experienced in the experiment by the edge subcarrier of the optical path when transmitted through consecutive nodes is plotted in Fig. 12(d). The nodes consist of WSSs with the filtering properties described in Section 3. The penalty reaches 1 dB after only two filtering stages, which significantly limits the scale of the network. The filtering performance could be improved by employing partial DPSK modulation [26,27]. However, in here we focus on the characteristics of the network platform without evaluating specific modulation formats.
Fig. 12 Filtering and cross-talk penalty: (a) elastic optical path spectrum, (b) spectrum of two neighboring paths; (c) spectrum of the edge subcarrier; (d) edge subcarrier penalty as a function of filtering stages; (e) model for investigation of guard band width and filter bandwidth.
The Q-factor performance of the edge subcarrier can be improved by increasing the width of the GB and optimizing the shape and the bandwidth of the optical filter of the WSS. We performed a numerical simulation in order to find the optimum GB width and filter shape for the conditions replicating that of the experimental setup. In the simulation, two optical paths, carrying 4 DPSK OFDM subcarriers each, are separated by a GB of 1, 2, or 3 subcarriers, as shown in Fig. 12(e). The carrier modulation bit rate is 10 Gb/s and, correspondingly, the subcarrier spacing is 10 GHz. At the 0th stage, the paths are assumed to be transmitted from a frequency locked source, such as the one shown in Fig. 4. The optical path carrying the analyzed subcarrier is transmitted over multiple WSS filtering stages, while the neighboring path is dropped and added at every node. The filters defining the paths have identical, Gaussian shapes described by the 3 dB bandwidth (BWfilt) and Gaussian order.
We investigate the optimum shape of the filter by calculating the power penalty as a function of the number of filtering stages for transmission through a core network with the longest path transmitted over 20 nodes. We consider the penalty caused by signal filtering and crosstalk. Transmission impairments, such as the accumulated CD, PMD, and reduction of OSNR are not taken into account. The results for GB of 1, 2, and 3 subcarriers are plotted in Figs. 13(a) , 13(b), and 13(c), respectively. In the case of narrow guard band equal to 1 subcarrier, only the filter with bandwidth of 50 GHz enabled transmission over multiple filtering stages. Even though such a narrow filter causes a filtering penalty, in this case the cross-talk penalty dominates. Consequently, employing a filter with a steeper response (higher order) allows transmission over a larger number of WXCs. However, even employing a 6th order filter causes a filtering penalty higher than 1 dB after fewer than 10 stages. Therefore, it is necessary to use a wider GB.
Fig. 13 Guard band width and filter shape simulation result. Edge subcarrier power penalty in function of filtering stages for guard band of: (a) 1 subcarrier, (b) 2 subcarriers, and (c) 3 subcarriers.
The case of GB equal to 2 subcarriers is presented in Fig. 13(b). With the wider GB it is possible to use the filter of 60 GHz, which reduces the filtering penalty, while reducing the influence of cross-talk. In this case a filter with 4th order Gaussian shape enables transmission of the OFDM signal over more than 20 filtering stages, while keeping the penalty below 1 dB. Employing a filter of 6th order would further reduce the penalty due to cross-talk. For comparison, a 4th order 70 GHz filter could also be used; however it would again lead to the cross-talk dominating the penalty as the neighboring path would not be suppressed sufficiently.
Finally, we investigate the case of GB equal to 3 subcarriers, when the cross-talk component becomes insignificant. Employing a 60 GHz filter produces results similar to the previous case. Extending the filter bandwidth to 70 GHz enables transmission of the OFDM signal over more than 20 filtering stages with penalty of less than 1 dB. We consider the case of GB width equal to 3 subcarriers to provide satisfactory performance while maintaining high SE of the link in systems employing elastic optical paths. In this case, the bandwidth of the optical filter in WSS for a 4-subcarrier path is 70 GHz. This result can be interpreted as a requirement to provide filters with 3 dB roll-off spaced by 20 GHz from the center frequency of the edge subcarrier.
It should be stressed that the shape of the actual filter can only be approximated by the Gaussian shape. Moreover, depending on the bandwidth of the filter, the edge steepness for a Gaussian filter with given order changes. Therefore, we calculated the orders of 50 and 70 GHz Gaussian filters which most closely match the shape of the filter used in the experiment. The corresponding filter orders for 50 GHz case and 70 GHz case are 2.5 and 4, respectively. We verified the simulation results in the experiment by transmitting the optical path in a recirculating loop consisting a WSS, 20 km of SMF and optical amplifiers. We varied the bandwidth of the filter in WSS to correspond to the case of 50 GHz and 70 GHz filter. The results, considering the impairment due to the loop transmission are plotted in open symbols in Fig. 13(c) and are compared to the simulation results for GB of 3 subcarriers. The simulation and experimental results show a good agreement, which proves the correct approximation of the model.
5. Conclusion
In this contribution we described the experimental demonstration of the spectrum-sliced elastic optical path network architecture. We showed the generation and reception of bandwidth-variable OFDM optical paths with bandwidth ranging from 40 to 440 Gb/s as well as transmission of such paths between six nodes of mesh network connected by DSF. We also demonstrated the multi-hop transmission of a 1 Tb/s optical path over 400 km of SMF. The deployment of bandwidth-variable optical paths in a network realized the highly-spectrally efficient transmission and flexible accommodation of client signal supporting 100 GbE or faster traffic. We believe that the new functionality brought about by the SLICE platform will open a new role for optical transport networks intelligently accommodating both the aggregated existing traffic flows, and future high-capacity services.
Acknowledgments
The authors thank Yutaka Miyamoto, Takayuki Kobayashi, and Yoshiaki Sone of NTT Network Innovation Laboratories for their fruitful discussions.
References and links
1. M. Jinno, “Emerging advanced optical transmission technologies: their impact on optical transport networks and optical packet switching,” Proc. Photonics in Switching 2008, S-06–3 (2008).
2. M. Jinno, Y. Miyamoto, and Y. Hibino, “Optical transport networks in 2015,” Nat. Photonics 1(3), 157–159 (2007). [CrossRef]
3. L. Smarr, “Riding the light towards new science,” Nat. Photonics 1(3), 133–135 (2007). [CrossRef]
4. A. Gumaste and S. Q. Zheng, “Next-generation optical storage area networks: the light-trails approach,” IEEE Commun. Mag. 43(3), 72–79 (2005). [CrossRef]
5. D. Shirai, T. Kawano, T. Fujii, K. Kaneko, N. Ohta, S. Ono, S. Arai, and T. Ogoshi, “Real time switching and streaming transmission of uncompressed 4K motion pictures,” Future Gener. Comput. Syst. 25(2), 192–197 (2009). [CrossRef]
6. J. McDonough, “Moving standards to 100 Gbe and beyond,” IEEE Commun. Mag. 45(11), 6–9 (2007). [CrossRef]
7. M. Tomizawa, J. Yamawaku, Y. Takigawa, M. Koga, Y. Miyamoto, T. Morioka, and K. Hagimoto, “Terabit LAN with optical virtual concatenation for Grid applications with super-computers,” Proc. OFC/NFOEC 2005, OThG6 (2005).
8. T. S. El-Bawab and Jong-Dug Shin, “Optical packet switching in core networks: between vision and reality,” IEEE Commun. Mag. 40(9), 60–65 (2002). [CrossRef]
9. S. Chandrasekhar, and X. Liu, “Mixed 10/40/100-Gb/s transmission through bandwidth-managed ROADMs”, Proc. OECC 2008, WeQ-1 (2008).
10. M. Jinno, H. Takara, B. Kozicki, Y. Tsukishima, T. Yoshimatsu, T. Kobayashi, Y. Miyamoto, K. Yonenaga, A. Takada, O. Ishida, and S. Matsuoka, “Demonstration of novel spectrum-efficient elastic optical path network with per-channel variable capacity of 40 Gb/s to over 400 Gb/s,” Proc. ECOC 2008, Th.3.F.6 (2008).
11. Y. Sone, A. Watanabe, W. Imajuku, Y. Tsukishima, B. Kozicki, H. Takara, and M. Jinno, “Highly survivable restoration scheme employing optical bandwidth squeezing in spectrum-sliced elastic optical path (SLICE) network,” Proc. OFC/NFOEC 2009, OThO2 (2009).
12. J.-K. Rhee, F. Garcia, A. Ellis, B. Hallock, T. Kennedy, T. Lackey, R. G. Lindquist, J. P. Kondis, B. A. Scott, J. M. Harris, D. Wolf, and M. Dugan, “Variable passband optical add-drop multiplexer using wavelength selective switch,” Proc. ECOC 2001, 550–551 (2001).
13. R. Ryf, Y. Su, L. Moller, S. Chandrasekhar, D. T. Xiang Liu, Neilson, and C. R. Giles, “Wavelength blocking filter with flexible data rates and channel spacing,” J. Lightwave Technol. 23(1), 54–61 (2005). [CrossRef]
14. G. Baxter, S. Frisken, D. Abakoumov, H. Zhou, I. Clarke, A. Bartos, and S. Poole, “Highly programmable wavelength selective switch based on liquid crystal on silicon switching elements,” Proc. OFC/NFOEC2006, OTuF2 (2006).
15. Y. Miyamoto, A. Sano, H. Masuda, E. Yoshida, and S. Aisawa, “Enhancing the capacity beyond terabit per second for transparent optical transport network,” Proc. ECOC 2007, 10.5.1 (2007).
16. W. Shieh and C. Athaudage, “Coherent optical orthogonal frequency division multiplexing,” Electron. Lett. 42(10), 587–589 (2006). [CrossRef]
17. S. L. Jansen, I. Morita, and H. Tanaka, “16x52.5-Gb/s, 50-GHz spaced, POLMUX- CO-OFDM transmission over 4,160 km of SSMF enabled by MIMO processing,” Proc. ECOC 2007, PD1.3 (2007).
18. H. Sanjoh, E. Yamada, and Y. Yoshikuni, “Optical orthogonal frequency division multiplexing using frequency/time domain filtering for high spectral efficiency up to 1 bit/s/Hz,” Proc. OFC 2002, ThD1 (2002).
19. A. Sano, H. Masuda, E. Yoshida, T. Kobayashi, E. Yamada, Y. Miyamoto, F. Inuzuka, Y. Hibino, Y. Takatori, and K. Hagimoto, “30 x 100-Gb/s all-optical OFDM transmission over 1300 km SMF with 10 ROADM nodes,” ECOC2007, PD1.7 (2007).
20. B. Kozicki, H. Takara, Y. Tsukishima, T. Yoshimatsu, T. Kobayashi, K. Yonenaga, and M. Jinno, “Optical path aggregation for 1 Tb/s transmission in spectrum-sliced elastic optical path network,” IEEE Photon. Technol. Lett. 22(17), 1315–1317 (2010). [CrossRef]
21. M. Jinno, H. Takara, B. Kozicki, Y. Tsukishima, Y. Sone, and S. Matsuoka, “Spectrum-efficient and scalable elastic optical path network: architecture, benefits, and enabling technologies,” IEEE Commun. Mag. 47(11), 66–73 (2009). [CrossRef]
22. T. Kobayashi, A. Sano, E. Yamada, Y. Miyamoto, H. Takara, and A. Takada, “Electro-optically multiplexed 110 Gbit/s optical OFDM signal transmission over 80 km SMF without dispersion compensation,” Electron. Lett. 44(3), 225–226 (2008). [CrossRef]
23. K. Takiguchi, M. Oguma, T. Shibata, and H. Takahashi, “Optical OFDM demultiplexer using silica PLC based optical FFT circuit,” in Proc. OFC/NFOEC 2009, OWO3 (2009).
24. M. E. Marhic, “Discrete Fourier transforms by single-mode star networks,” Opt. Lett. 12(1), 63–65 (1987). [CrossRef] [PubMed]
25. X. Liu, Y.-H. Kao, M. Movassaghi, and R. C. Giles, “Tolerance to in-band coherent crosstalk of differential phase-shift-keyed signal with balanced detection and FEC,” IEEE Photon. Technol. Lett. 16(4), 1209–1211 (2004). [CrossRef]
26. B. Mikkelsen, C. Rasmussen, P. Mamyshev, and F. Liu, “Partial DPSK with excellent filter tolerance and OSNR sensitivity,” Electron. Lett. 42(23), 1363–1364 (2006). [CrossRef]
27. S. Chandrasekhar, X. Liu, and A. R. Chraplyvy, “Improving the filtering tolerance of 42.7-Gb/s partial-DPSK by optimized power imbalance,” Proc. OFC 2009, OMM1 (2009).
