WO 2008/040053 PCT/AU2007/001465 "Single Sideband Orthogonal Frequency Division Multiplexed Optical Fibre Transmission" Cross-Reference to Related Applications 5 The present application claims priority from Australian Provisional Patent Application No 2006905457 filed on 3 October 2006, the content of which is incorporated herein by reference. Technical Field 10 The present invention relates to data transmission over an optical fibre or optical network, and in particular relates to transmission of data using orthogonal frequency division multiplexing (OFDM) and single sideband (SSB) transmission. Background of the Invention 15 Optical fibre chromatic dispersion (FCD) arises due to the variation in refractive index of an optical fibre medium for differing wavelengths of optical transmissions, such that a signal component having a first wavelength propagates at a velocity which is different to the velocity of a signal component having a second wavelength. FCD leads to pulse spreading, which constrains the bandwidth distance product (BPD) for a given optical 20 medium, and is thus a limiting factor upon the data rate and/or optical link distance. Similarly, polarisation mode dispersion (PMD) causes optical signals polarised in differing planes to propagate at differing velocities. PMD also constrains link performance. FCD and PMD present a major problem in multi-gigabit fibre transmission. 25 To combat fibre chromatic dispersion, techniques have been developed in the optical domain including the use of dispersion compensated fibre (DCF), the addition of a second fibre with opposite dispersion (which is costly), and/or the use of optically created single sideband (O-SSB) transmissions or optically created vestigial sideband 30 (O-VSB) transmissions. In the electrical domain, proposed solutions include the use of optical filters or the phasing of optical modulators, provision of electrical dispersion WO 2008/040053 PCT/AU2007/001465 2 compensation (EDC), use of a linear feed forward equaliser (FFE), use of nonlinear decision feed back (DFE) and maximum likelihood sequence equalisation (MLSE). Orthogonal frequency division multiplexed optical transmissions offer another prospect 5 to improve the BPD, and involve the use of many RF sub-carriers with low bit-rates to transmit a high bit rate signal, in a similar manner to wireless OFDM systems. However such solutions tend to be complex as they require up-conversion to an orthogonal microwave carrier for each OFDM sub-carrier before optical transmission, and corresponding down-conversion at the receiver before FFT processing. 10 Any discussion of documents, acts, materials, devices, articles or the like which has been included in the present specification is solely for the purpose of providing a context for the present invention. It is not to be taken as an admission that any or all of these matters form part of the prior art base or were common general knowledge in the 15 field relevant to the present invention as it existed before the priority date of each claim of this application. Throughout this specification the word "comprise", or variations such as "comprises" or "comprising", will be understood to imply the inclusion of a stated element, integer or 20 step, or group of elements, integers or steps, but not the exclusion of any other element, integer or step, or group of elements, integers or steps. Summary of the Invention According to a first aspect the present invention provides a method of data transmission 25 over an optical medium, the method comprising: performing signal processing upon data to be transmitted, to generate two baseband broadband orthogonal frequency division multiplexed 90 degree feed signals; and driving a quadrature dual optical modulator with the two 90 degree feed signals, 30 wherein the signal processing is performed such that the two generated 90 degree feed signals cause an output of the modulator to be single sideband.
WO 2008/040053 PCT/AU2007/001465 3 According to a second aspect the present invention provides a transmitter for data transmission over an optical medium, the transmitter comprising: a signal processor for performing signal processing upon data to be transmitted, 5 to generate two baseband broadband orthogonal frequency division multiplexed 90 degree feed signals; and a quadrature dual optical modulator configured to be driven by the two 90 degree feed signals, wherein the signal processor is configured to generate the two 90 degree feed 10 signals in a manner which causes an output of the modulator to be single sideband. By transmitting a SSB optical OFDM signal, the present invention provides a solution in which FCD will affect only the phase of the OFDM sub-carriers, which may thus be corrected by a training sequence. 15 Further, by providing baseband feed signals to the optical modulator, the present invention positions the single sideband immediately adjacent to the optical carrier frequency, in contrast to proposals utilising a RF sub-carrier which position the sidebands in a range away from the optical carrier frequency, for example in the range 20 between 5 GHz and 10 GHz away from the optical carrier frequency. By positioning the single sideband immediately adjacent to the optical carrier frequency, the present invention permits more dense spectral spacing of a plurality of such optical carriers. In preferred embodiments, the signal processing comprises QAM modulation of the 25 data to be transmitted, to N QAM symbols. In further preferred embodiments of the invention, the signal processing comprises parallel processing along two processing arms to generate each feed signal. In particularly preferred embodiments, the 90 degree phase shift between the drive signals is imposed upon the N QAM symbols of one of the processing arms, before OFDM processing in that arm. 30 WO 2008/040053 PCT/AU2007/001465 4 The parallel processing in each processing arm preferably comprises reverse complex conjugate addition of the N QAM symbols, to 2N symbols. The 2N symbols are preferably processed by an inverse fast Fourier transform (IFFT). Such embodiments recognise that the necessary phasing of OFDM sidebands will arise during IFFT 5 processing of the 90 degree shifted N QAM symbols. Such embodiments are particularly advantageous in avoiding the need for a Hilbert transformer, or similar block such as a FIR filter or other DSP filter to effect the 90 degree phase difference in the drive signals. To implement baseband OFDM with low frequency sidebands would require a very large number of filter taps, and thus the present embodiment provides for 10 particularly efficient implementations which may permit application in very high data rate environments. Real processing of the IFFT output is preferably utilised. Cyclic prefixing is preferably applied to each 2N block, although inter-block guard bands may additionally or 15 alternatively be utilised. The feed signal from each processing arm is preferably produced in analogue form by way of up sampling, interpolation and digital to analogue (D/A) conversion. Such interpolation preferably comprises post-IFFT packing. 20 The modulator is preferably a dual electrode Mach Zehnder modulator biased for SSB generation. According to a third aspect the present invention provides a receiver for receiving SSB OFDM transmissions, the receiver comprising: 25 a channel phase correctional device for correcting a phase of the channel as a whole under the control of a channel phase shift bias value; a sub-carrier group divider for dividing the channel into a plurality of sub-carrier groups; and for each sub-carrier group, a group phase correctional device for correcting the 30 phase of the sub-carrier group under the control of a group phase shift bias value.
WO 2008/040053 PCT/AU2007/001465 5 The receiver may utilise a training sequence to correct phase drift in each group. Brief Description of the Drawings An example of the invention will now be described with reference to the 5 accompanying drawings, in which: Figure la is a schematic of an optical transmission system for transmitting baseband SSB OFDM optical data signals, and Figure lb is a more detailed schematic of the transmitter-side of the schematic of Figure la, illustrating the generation of the two 90 degree baseband drive signals; 10 Figure 2a is a generalised schematic, and Figure 2b is a detailed schematic, of a system used in a computer simulation of a first embodiment of the present invention for 10 Gb/s rate and having a zero length of optical fibre between transmitter and receiver, Figure 2c illustrates the transmitted SSB spectrum of the system of Figure 2b, and Figure 2d illustrates the received QAM constellation in the system of Figure 2b; 15 Figures 3a and 3b illustrate bit sequence insertion and extraction points and a constellation rotation input of the schematic of Figures 2a and 2b, respectively, Figure 3c illustrates a transmitted pseudo-random bit sequence, Figure 3d illustrates the received pseudo-random bit sequence, and Figure 3e illustrates the post-rotation QAM constellation; 20 Figure 4a is a detail of the schematic of Figures 2a and 3a, and Figure 4b is a detail of the schematic of Figures 2b and 3b, each illustrating a channel constellation rotation input, sub-carrier constellation rotation inputs, and signal outputs, Figure 4c illustrates the received channel QAM constellation for all 128 sub-carriers after transmission over 40 km of fibre, Figure 4d illustrates the received QAM constellation 25 for sub-carriers 1 to 11 of the channel, Figure 4e illustrates the received QAM constellation for sub-carriers 101 to 111 of the channel, and Figure 4f illustrates the post- sub-carrier rotation QAM constellation of sub-carriers 101-111 of the channel; Figure 5a illustrates the received channel QAM constellation for all 128 sub carriers after transmission over 400 km of fibre, Figure 5b illustrates the received QAM 30 constellation for sub-carriers 1 to 11 of the channel, and Figure 5c illustrates the received QAM constellation for sub-carriers 101 to 111 of the channel; and WO 2008/040053 PCT/AU2007/001465 6 Figure 6a illustrates the received channel QAM constellation for all 128 sub carriers after transmission over 1000 km of fibre, Figure 6b illustrates the received QAM constellation for sub-carriers 1 to 11 of the channel, and Figure 6c illustrates the received QAM constellation for sub-carriers 101 to 111 of the channel. 5 Description of the Preferred Embodiments Figure 1 is a schematic of an optical transmission system 100 for transmitting baseband SSB OFDM optical data signals. System 100 takes as an input a pseudo-random bit sequence 110 which is passed to a QAM modulator 120 for conversion to blocks of N 10 QAM symbols. The N symbols are in turn passed to a baseband OFDM processor 130, adapted to generate two feed signals 132 and 134. An optical source in the form of laser 140 is modulated by a dual Mach Zehnder modulator 142, the modulator being biased such that the I and Q feed signals used as 15 inputs produce a SSB output of the modulator. The baseband feed signals 132 and 134 generated by the processor 130 are used to drive the dual Mach Zehnder modulator 142. The modulated optical signal is then transmitted over a length of fibre to a receiver, where it is detected by a PIN receiver 150, A/D converted by A/D converter 160, and processed by FFT 170. Processor 180 performs 2N to N processing, and 20 demodulator 190 performs QAM demodulation to recover data bits. OFDM processor 130 generates feed signals 132 and 134 in such a manner that the transmitted optical signal is a SSB OFDM optical signal. To achieve this end processor 130 uses reverse complex conjugate addition, IFFT, cyclic prefix, dual 90 degree 25 processing, and interpolation and D/A conversion, which in a sense can be considered as a Hilbert transform. Importantly, in the present embodiment the digital processing 130 makes it possible to achieve constant 90 degree phase difference between the two driving arms 132, 134 of 30 the dual MZ modulator 142. Baseband real processing is used in the Inverse Fast Fourier transform (IFFT) as opposed to complex signals. This is illustrated in more WO 2008/040053 PCT/AU2007/001465 7 detail in Figure Ib, which shows that the N QAM symbols output by modulator 120 are fed into two parallel processing arms. The first processing arm in Figure lb includes the following processing steps: reverse 5 complex conjugate addition 1312; IFFT 1314; cyclic prefixing 1316 and interpolation and D/A conversion 1318. The second arm includes an initial digital phase shifter 1320 which performs R to P conversion to effect 90 degree rotation of the QAM symbols in the second arm. The second arm then includes the same processing steps as the first arm, namely: reverse complex conjugate addition 1322; IFFT 1324; cyclic prefixing 10 1326 and interpolation and D/A conversion 1328. Notably, all processing within CMOS 130 occurs at baseband, and the present embodiment thus does not require any intermediate RF frequency stage. Importantly, by providing a simple 90 degree rotation of the input QAM constellation 15 at 1320, the present embodiment exploits the recognition that the necessary phasing of the OFDM sidebands will then arise through the subsequent IFFT processing. Such a 90 degree rotation is a simple CMOS processing operation, and in real time parallel processing this technique will be substantially faster than the alternative of including an FIR or IIR Hilbert transform stage. Moreover, the broadband requirements of SSB 20 OFDM demand a high number of taps in such filters, further hampering processing speeds. It is further noted that the present embodiment is particularly suitable for OFDM over optical SSB with direct detection, as in such applications it is desirable to transmit the optical carrier for direct detection in a simple low cost optical system. 25 Figure 2a is a generalised schematic, and Figure 2b is a detailed schematic, of a system used in a computer simulation of a first embodiment of the present invention for 10 Gb/s data rate and having a zero length of optical fibre between transmitter and receiver. Parallel processing paths 200a and 200b are utilised in order to generate the two feed signals for Mach Zehnder modulator 210. A 90 degree phase shift is applied 30 at 202b at the start of path 200b to provide the 90 degree processing. Figure 2c illustrates the SSB spectrum of the signal transmitted at 220 in the system of Figure 2b.
WO 2008/040053 PCT/AU2007/001465 8 As can be seen the signal spectrum is SSB, and notably the SSB spectrum is achieved by appropriate signal processing without need for optical filtering to convert DSB to SSB. Further, the signal spectrum is immediately adjacent to the carrier frequency which allows close spacing of multiple such channels and thus permits high spectral 5 efficiency. Figure 2d illustrates the QAM constellation received at 230 in the system of Figure 2b. As can be seen there exists tight grouping of the points of the 4-QAM constellation, albeit with 45 degrees of rotation from expected constellation point positions. 10 Figure 3a is another representation of the schematic of Figure 2a, and Figure 3b is another representation of the schematic of Figure 2b, illustrating bit sequence insertion point 310 and bit sequence extraction point 330, and a constellation rotation input 320. As for Figures 2a to 2d, Figures 3a to 3e relate to an embodiment in which the 15 transmitter and receiver are back to back with zero length of optical fibre between them. Figure 3c illustrates a transmitted pseudo-random bit sequence inserted at 310, and Figure 3d illustrates the received pseudo-random bit sequence extracted from 330, showing proper bit sequence recovery. 20 Constellation rotation input 320 allows a phase shift bias value to be applied to the receiver in order to rotate the received QAM constellation from the received position, such as that shown in Figure 2d, to the expected 4-QAM point positions. Figure 3e illustrates the post-rotation QAM constellation, illustrating the phase shift of 135 degrees achieved by a phase shift bias value of (1.57/2 + 1.57) applied at 320. The 25 phase shift bias point 320 thus provides a very simple means for phase adjustment of the received channel for rotation of the constellation back to the expected QAM positions. Figure 4a is a detail of the schematic of Figures 2a and 3a, while Figure 4b is a detail of 30 the schematic of Figures 2b and 3b. Figures 4a and 4b illustrate in the receiver a channel constellation rotation input 410, sub-carrier group constellation rotation inputs WO 2008/040053 PCT/AU2007/001465 9 420, and signal outputs 430. Channel constellation rotation input 410 rotates the entire received constellation in the same manner as noted in the preceding with respect to Figures 3e and 2d. 5 Figure 4c illustrates the received channel QAM constellation for all 128 sub-carriers output at 430a, after transmission over a 40 km length of fibre, and after correctional channel rotation provided by phase shift bias 410. As can be seen, each point in the QAM constellation has undergone spreading over the 40 km optical link, for example due to FCD and/or PMD. Notably, transmission of a SSB signal in accordance with the 10 present invention ensures that such spreading manifests only as rotation in the QAM plane. In considering in more detail the manner of spreading of the sub-carriers making up the channel, it can be seen in Figure 4d that the received QAM constellation points at 430b 15 for sub-carriers 1 to 11 of the channel are closely grouped and undergo little or no constellation rotation. However, Figure 4e illustrates that the received QAM constellation points at 430c for sub-carriers 101 to 111 of the channel suffer a greater amount of spreading than sub-carriers 1-11, and further that, when phase shift bias 420b is inactive, the QAM constellation points for sub-carriers 101-111 undergo a 20 notable amount of constellation rotation. Thus, the present invention recognises that sub-carriers in such a baseband SSB OFDM transmission can be dealt with in groups. The present embodiment thus provides sub carrier group constellation rotation inputs 420, each of which can be individually 25 adjusted to rotate that group's QAM constellation points by a desired angle. Figure 4f illustrates the effect of activation of phase shift bias 420b upon the output at 430c, showing that the sub-carrier QAM constellation of sub-carriers 101-111 of the channel can be rotated as a group to be located substantially at the ideal 4-QAM points. Applying suitable phase shift biases to each sub-carrier group thus allows the spreading 30 evident in Figure 4c to be accounted for and for the data to be accurately recovered by QAM demodulation.
WO 2008/040053 PCT/AU2007/001465 10 The use of baseband modulation further causes constellation rotation to be roughly progressive or incremental from one sub-carrier to the next, and thus from one sub carrier group to the next, such that a deterministic derivation of appropriate phase shift 5 biases 420 may be obtained, rather than requiring feedback phase shift bias control for all groups. To investigate the severity of spreading which can be dealt with by the present embodiment, further simulations were carried out using the simulation schematic of 10 Figures 2b, 3b and 4b, for fibre lengths of 400 kin, and 1000 km, respectively. Figure 5a illustrates the received channel QAM constellation for all 128 sub-carriers output at 430b after transmission over 400 km of fibre. Visual inspection of Fig 5a reveals substantial spreading of the QAM points of the overall channel. However, 15 Figure 5b illustrates the received QAM constellation at 430b for sub-carriers 1 to 11 of the channel. Clearly, it would be possible to QAM demodulate sub-carriers 1 to 11 from such compact QAM point clusters. Further, Figure 5c illustrates the received QAM constellation at 430c for sub-carriers 101 to 111 of the channel, with phase shift bias 420b inactive. While the QAM points for carriers 101 to 111 have undergone 20 substantial rotation in the I/Q plane this can be corrected by appropriate control of phase shift bias 420b. Further, sub-carriers 101-111 remain adequately grouped together that QAM demodulation can be performed to accurately recover the original data. The present simulations thus suggest that the present invention has promise in applications having optical links of 400 km. 25 Figure 6a illustrates the received channel QAM constellation for all 128 sub-carriers output at 430b after transmission over 1000 km of fibre. Visual inspection of Fig 6a reveals substantial spreading of the QAM points of the overall channel. However, Figure 6b illustrates the received QAM constellation at 430b for sub-carriers 1 to 11 of 30 the channel. Clearly, it would be possible to QAM demodulate sub-carriers 1 to 11 from such compact QAM point clusters. Further, Figure 6c illustrates the received WO 2008/040053 PCT/AU2007/001465 11 QAM constellation at 430c for sub-carriers 101 to 111 of the channel, with phase shift bias 420b inactive. While the QAM points for carriers 101 to 111 have undergone substantial rotation in the I/Q plane this can be corrected by appropriate control of phase shift bias 420b. Further, even after transmission over a 1000 km fibre link, sub 5 carriers 101-111 can still be seen to be grouped together so that QAM demodulation can be performed to accurately recover the original data. The present simulations thus suggest that the present invention has promise in applications having optical links of 1000 km. 10 The retention of adequate sub-carrier QAM point grouping, and the resistance of the QAM points to spreading, can be attributed to the use of baseband modulation of the optical source, laser 140. Further, it is noted that while the QAM points for sub carriers 101-111 have spread to an extent that QAM demodulation may cause errors, it is to be noted that a reduction of group size (group having been chosen somewhat 15 arbitrarily in the present simulations) will counteract such spreading. For example by inspection of Figure 6c it might be deduced that a sub-carrier grouping of only five or six sub-carriers would provide for more clearly distinguishable QAM point groups in the IQ plane, thus improving the likelihood of error-free QAM demodulation, even in such an extreme application as a 1000 km 10 GB / s fibre link. Moreover, while these 20 simulations assume the presence of erbium doped fibre amplifiers (EDFAs) along the transmission link, promising results have also been obtained when instead considering the presence of Raman amplifiers and pump sources. The present embodiment of the invention thus illustrates that it is possible to modify 25 the baseband silicon processor generating the baseband OFDM sub-carriers, and in effect realise a broadband Hilbert transform for directly driving a dual Mach-Zehnder (MZ) optical modulator. The present embodiment directly generates, electrically, a SSB multi-gigabit optical OFDM signal without the complication of microwave up conversion, without the need for broadband filtering with a high number of filter taps, 30 and without the complication of optically filtering a DSB OFDM signal to remove one sideband.
WO 2008/040053 PCT/AU2007/001465 12 Furthermore, the close spacing of OFDM sub-carriers to the optical carrier in this invention, illustrated in Figure 2c, means that even for transmission distances greater than 1000 km low order OFDM sidebands may need no electrical dispersion 5 compensation at the receiver and higher order OFDM sidebands may be electrically compensated simply using block channel phase compensation techniques (and not necessarily individual channel compensation). The present invention recognises that a key objective of system vendors and network 10 providers is transmission of 10 Gbit/s and 40 GBit/s optical signals over standard single mode (SSM) optical fibre, without dispersion compensated fibre (DCF), and in the presence of polarisation mode dispersion (PMD). The present embodiment provides baseband OFDM over optical SSB and may present a valuable tool in achieving this objective. Further, the present embodiment may be appropriate for use in optically 15 switched networks where either small electrical phase compensation is required or rapid electrical adaptive compensation is necessary as signals are switched dynamically. Sub-carrier groupings can be chosen to include a number of channels appropriate to the 20 application in question. For example a large number of sub-carriers may be grouped together for correctional rotation by phase shift bias 420 in applications where the constellation rotation is sufficiently similar for sub-carriers throughout that grouping. Alternatively, sub-carriers may be phase shift biased in small groups, or even on a one by-one basis by a larger number of phase shift biases 420, in applications such as long 25 haul where constellation rotation causes excessive spreading of the constellation of adjacent sub-carriers. While the present embodiments have been described with reference to 4-QAM modulation it is to be appreciated that alternative embodiments of the present invention 30 may be applied to M-QAM or alternative I/Q modulation schemes.
WO 2008/040053 PCT/AU2007/001465 13 It will be appreciated by persons skilled in the art that numerous variations and/or modifications may be made to the invention as shown in the specific embodiments without departing from the spirit or scope of the invention as broadly described. The present embodiments are, therefore, to be considered in all respects as illustrative and 5 not restrictive.