METHOD FOR PROCESSING DATA IN AN OPTICAL NETWORK ELEMENT AND OPTICAL NETWORK ELEMENT

Abstract

A method for processing data in an optical network element is provided, wherein a multicarrier signal is linear pre-coded, and wherein the linear pre-coded signal is modulated. An according optical network element is also suggested.

Description

The invention relates to a method and to a device for processing data in an optical network element and to an according optical network element.

With broadband Internet connections and mobile data transfer becoming ubiquitous technologies, requirements regarding bitrates over WDM optical channels are increasing. In this context, a spectral efficiency of a modulation format used is of significant relevance.

The spectral efficiency of an optical signal can be increased, e.g., by multilevel modulation, polarization multiplexing, orthogonal frequency multiplexing or a combination thereof. However, the complexity of the system noticeably increases with the modulation.

When multilevel modulation format and polarization multiplexing were chosen to increase the spectral efficiency (SE) of a transmission system, orthogonal frequency division modulation (OFDM) is a subsequent step that doubles the SE by overlapping the spectra of a multitude of subcarriers.

For example, a binary on-off-keying (OOK) signal with a data rate of 100 Gbps uses 200 GHz of optical bandwidth (BW). If OFDM is used with quadrature-phase-shift-keying (QPSK) modulated subcarriers and polarization multiplexing (PolMux) a 100 Gbps line rate signal would roughly use 25 GHz optical bandwidth.

However, such an advanced modulation format would require the use of digital coherent detection, and therefore, greatly increase the complexity of the implementation, because of the need for advance digital signal processing (DSP) and an optical local oscillator.

A reduced complexity of the implementation could be achieved by using a direct detectable OFDM signal, but a reference carrier has to be sent along the data signal situated at a distance in spectrum equal to the BW of the OFDM signal, which reduces the SE. Another possibility is the use of a Compatible Single Sideband OFDM modulation (CompSSB-OFDM). However, a very high power carrier is required in this case, also compromising an overall performance of the system.

In addition, all such cases of enhanced SE require complex DSP algorithms that need to be implemented in the transmitter and the receiver.

For high data rates (>10 Gbps), the use of DSP is a challenging problem not only concerning the developments of high-speed electronics, but also with regard to an energy consumption of next generation high-speed systems.

The problem to be solved is to overcome the disadvantages mentioned above and in particular to provide a high degree of spectral efficiency without a high complexity of digital signal processing in an optical system.

This problem is solved according to the features of the independent claims. Further embodiments result from the depending claims.

In order to overcome this problem, a method is suggested for processing data in an optical network element,

• • wherein a multicarrier signal is linear pre-coded,
  • wherein the linear pre-coded signal is modulated.

The proposed method in particular doubles the spectral efficiency of an optical signal while maintaining a low complexity of the system. This allows for example, the use of high-speed 100 Gbps signals in dense WDM systems without the need of polarization multiplexing or complex digital signal processing at the receiver, therefore allowing for a cost efficient approach.

The approach further increases the spectral efficiency of signals and provides compatibility with direct detection receivers. Hence, in particular no local oscillator is required at the receiver.

In an embodiment, the pre-coded signal is modulated by a differential phase modulation or by an amplitude modulation.

It is noted that direct detection modulation formats can be utilized, in particular OOK, DPSK, DQPSK, D8PSK, Star-D8QAM, Star-D16QAM, PAM, etc.

In another embodiment, the multicarrier signal is linear pre-coded, wherein each subcarrier is a linear combination of all other subcarriers.

Hence, after direct detection at a receiver, at each k-th sampling point, the electrical signal has a value proportional to the k-th subcarrier.

In a further embodiment, the multicarrier signal is linear pre-coded by a matrix T, wherein the coefficients of a linear combination of a k-th subcarrier correspond to the k-th row of an DFT matrix.

In a next embodiment, a channel transfer function is pre-compensated by an equalization.

This equalization can be conducted at the optical network element after the linear pre-coding. Advantageously, the signal can be sent via a dispersion-unmanaged link.

It is also an embodiment that said equalization is a one tap equalization.

Pursuant to another embodiment, a subcarrier is discarded by a receiver in case the subcarrier experiences deforming (e.g., due to distortions) and/or attenuating effects above a given threshold.

In such case, the receiver may (temporarily) discard at least one subcarrier. The receiver may inform the transmitter of the discarded subcarrier and the transmitter may no longer use this subcarrier towards this receiver.

According to an embodiment, a dummy subcarrier is used to provide a guard band.

Advantageously, such guard band is useful to reduce (inter-symbol) interference from a previously pre-coded block.

According to another embodiment, a zero padding is conducted prior to an inverse discrete Fourier transform.

Advantageously, this zero padding simplifies filtering at the receiver. Zero padding can be adjusted to the filtering capability of the respective transmitter.

In yet another embodiment, a feedback channel is provided to convey information from a receiver to the optical network element.

Such feedback channel can be utilized for various purposes. The optical network element may in particular adjust its signals to increase the efficiency based on the information obtained via the feedback channel.

According to a next embodiment, the optical network element is an optical transmitter, e.g., an optical line terminal or an optical network unit.

The problem stated above is also solved by an optical network element comprising

• • a linear pre-coder processing a multicarrier signal,
  • a modulator processing the linear pre-coded signal.

Pursuant to yet an embodiment, the optical network element further comprises a zero padding unit prior to an inverse discrete Fourier transform unit prior to said modulator.

It is noted that said modulator may advantageously comprise at least one digital operation.

It is also an option that the modulator is a differential modulator or an amplitude modulator

According to an embodiment, the optical network element comprises a control unit that is arranged such that the method as described herein can be executed.

The problem stated supra is further solved by an optical communication system comprising the optical network element as described herein.

Embodiments of the invention are shown and illustrated in the following figures:

FIG. 1 shows a transmission scheme comprising a block diagram of a transmitter, a channel and a receiver;

FIG. 2 shows another representation of the transmitter;

FIG. 3 shows a spectrum of an optical output signal of the transmitter according to FIG. 2.

The solution provided in particular suggests a discrete inverse Fourier transform in the transmitter that increases the spectral efficiency of an optical signal by shaping its spectrum in an OFDM-like fashion. In addition, it makes the signal compatible with direct detection, without the need for digital signal processing and without having to send an optical carrier along the signal. Hence, without such need for complex DSP, the complexity is feasible even for high data rates (e.g., >10 Gbps) and a high spectral efficiency is achieved as no bandwidth is to be reserved for an optical carrier.

At the transmitter, an optical OFDM-like signal is generated, wherein each optical subcarrier is a linear combination of all other subcarriers, so that after direct detection, at each k-th sampling point, the electrical signal has a value proportional to the k-th subcarrier.

The transmitter maybe implemented using a moderate amount of DSP, digital to analog converters (DACs) and one optical IQ modulator. No DSP is needed at the receiver (therefore also no ADCs) to demodulate the OFDM-like signal. The receiver maintains the complexity of its single-carrier counterpart.

It is also an option that after the electrical I- and Q-signals are generated, they can be up-mixed to an intermediate frequency by an electrical IQ-modulator. The output thereof can be used to modulate a laser diode via an optical amplitude modulator, e.g., a mach-zehnder modulator. Then, an optical filter can be used to filter one sideband of the optical signal.

As increased SE is desired, one attractive application would be to use this technique with DQPSK mapped subcarriers to form a 100 Gbps optical signal without PolMux and without DSP at the receiver. Instead, a simple DQPSK demodulator can be used at the receiver. In this case, the signal shows a 50 GHz bandwidth and therefore would be compatible with DWDM systems.

It is also an option to use PolMux and higher level modulation, therefore increasing the SE even further but retaining the complexity of a receiver equal to the case where direct detected single carrier is used.

Another advantage is that, as each subcarrier is available at the transmitter separately; simple one tap equalization can be performed for pre-compensating the channel transfer function. This allows, e.g., to send such a signal through a dispersion-unmanaged link (i.e., without dispersion compensation modules).

It is a further option of this approach to allow the receiver to perform simple subcarrier selection. Some detrimental effects on the link (i.e., narrowband filtering) may affect some subcarriers in a stronger way than others. Such bad subcarriers can be discarded easily by the receiver by just ignoring certain sample instants. Similar, dummy subcarriers can be used (containing no useful information or no information at all, i.e. always zero amplitude) to form guard bands that help maintaining the performance.

It is noted that the approach presented can be used with modulation formats that are in particular compatible with direct detection, e.g., OOK, DPSK, DQPSK, DQPSK, Star-D8QAM, Star-D16QAM, PAM, etc.

Here, DQPSK is an example for such modulation format. Other modulation formats may be applicable as well.

FIG. 1 shows a transmission scheme comprising a block diagram of a transmitter 101, a channel 102 and a receiver 103.

Transmitter:

The transmitter 101 according to FIG. 1 will be described in detail.

• • (1) A logical binary data sequence is input to an M*N serial to parallel module 104. M depends on the modulation format: For OOK or BPSK, M equals 1, for QPSK, M amounts to 2, and so on (M-ary modulation).
  • (2) After parallelization each of the N sequences are mapped independently according to the desired modulation format (see block 105).
    • The symbol duration (Ts) is equal to the inverse of the total data rate (Br) multiplied by (M*N).
    • A vector x is provided as an output of the block 105.
  • (3) At each Ts instant the vector x composed by the N mapped DQPSK symbols is multiplied by a transform matrix T in a block 106 providing a vector Tx.
  • (4) Then in order to invert the channel response, one tap equalization (see block 107) is performed on each element of the vector Tx.
  • (5) Before performing the inverse DFT at the transmitter, as an option, zero padding (ZP) is used to ease the filtering of the images of the digital signal that causes aliasing (see block 108).
  • (6) After this block 108, the resulting vector is multiplied by the inverse DFT matrix in a block 109.
  • (7) The output vector of the block 109 is serialized in a block 110 and the real part and imaginary part of the sequence are fed to an optical IQ modulator 111. As an alternative, an electrical IQ-modulator can be used as indicated above.
  • (8) The output signal of the optical IQ modulator 111 is fed over the channel 102 towards the receiver 103.

The Transform Matrix

The output of the transmitter 101 is an OFDM-like signal, each subcarrier is a combination of the data vector x. The matrix T is chosen so that the coefficients of the linear combination of each k-th subcarrier correspond to the k-th row of the DFT matrix.

The purpose of this linear combination is that at the k-th sampling instant the value of the optical signal is proportional to the k-th element of the vector x. Hence, squaring of the direct detector will only affect the k-th element.

The transform matrix T is dependent on the amount of zero padding (ZP). If the amount of ZP is equal or greater than N, then T is equal to the DFT matrix. If the zero padding is smaller than N then T correspond to a permutation of the rows of the DFT matrix. The permutation is dependent on the amount of ZP.

The Receiver

Advantageously, a legacy direct detection receiver 103 may be used depending on the modulation format of the subcarriers. In the case of DQPSK, two delay interferometers with balance detection may be used to separate and demodulate the I- and the Q-components of the optical signal.

Preferably, the delay time is matched to the symbol duration Ts (that is the duration of the “OFDM-like block”).

Sampling is done at a rate equal to the data rate divided by the modulation order (i.e. Br/M).

Details Regarding the Processing of Data

The following provides further details of the solution provided and explains its impact on the receiving side.

x is regarded as a vector comprising N elements, i.e.

$\begin{array}{cc}x=\left[\begin{array}{c}{x}_0\\ x_1\\ ⋮\\ x_{N-1}\end{array}\right].& \left(1\right)\end{array}$

A transform matrix of N×N dimension is defined as

$\begin{array}{cc}T=\left[\begin{array}{cccc}{t}_{00}& t_{01}& \dots & t_{0\left(N-1\right)}\\ t_{10}& t_{11}& \dots & t_{1\left(N-1\right)}\\ ⋮& ⋮& \ddots & ⋮\\ t_{\left(N-1\right)0}& t_{\left(N-1\right)1}& \dots & t_{\left(N-1\right)\left(N-1\right)}\end{array}\right].& \left(2\right)\end{array}$

Further, t′_n is a row of the matrix T, i.e.

t′ _n =[t _n0 t _n1 . . . t _n(N−1)].   (3)

The transmitter in particular generates a signal as follows:

$\begin{array}{cc}s\left(t\right)=\sum _{n=0}^{N-1}t_n^{\prime }·x·{}^{j2\pi f_n·t}.& \left(4\right)\end{array}$

As a first example, the vector x is an intensity modulated signal (ON-OFF-keying), i.e.

x_n ε [0; 1].   (5)

A photodiode may be used at the receiver side to detect and demodulate the signal transmitted. The received sampled signal ŝ is sampled at sampling points

$\begin{array}{cc}t=\frac{k·\Delta t}{N}\mathrm{with}k=0,\dots ,N-1.& \left(6\right)\end{array}$

By utilizing said photodiode the square of the magnitude of the optical signal is received, which (prior to the sampling phase) amounts to

$\begin{array}{cc}{s\left(t\right)}^2={\sum _{n=0}^{N-1}t_n^{\prime }·x·{}^{j2\pi f_n·t}}^2.& \left(7\right)\end{array}$

After sampling, i.e.

$t=\frac{k·\Delta t}{N}\mathrm{and}$ $f_n=\frac{n}{\Delta t}$

the signal amounts to

$\begin{array}{cc}{s\left(\frac{k·\Delta t}{N}\right)}^2={\sum _{n=0}^{N-1}t_n^{\prime }·x·{}^{j2\pi \frac{n}{\Delta t}·\frac{k·\Delta t}{N}}}^2,& \left(8\right)\end{array}$

which corresponds to the inverse Fourier transform of T·x. Equation (8) can be rewritten in matrix form as follows:

$\begin{array}{cc}{\hat{s}}_k={s\left(\frac{k·\Delta t}{N}\right)}^2=\left(e_k·W^{-1}·T·x\right)·\left(e_k·W^{-1}·T·x\right)*,& \left(9\right)\end{array}$

wherein ( . . . )* denotes a transposed conjugated matrix and e_k identifies a k-th row of the identity matrix, e.g.,

e₂=[0 1 0 0 . . . 0].

In addition, W⁻¹ refers to an inverse discrete Fourier transform (DFT) matrix.

Equation (9) can be rewritten as follows:

ŝ _k =e _k ·W ⁻¹ ·T·x·x*·T*·W ⁻¹ *·e* _k.   (10)

The transform matrix T may be equal to the DFT matrix W (transforming with DFT). Thus, equation (10) results in

ŝ _k =e _k ·x·x*·e* _k   (11)

and thus in

ŝ _k =|x _k|².   (12)

Therefore, at the suitable sampling point

$t=\frac{k·\Delta t}{N}$

if the transform matrix T equals the DFT matrix W, the k-th sampling value equals the k-th element of the vector x (magnitude square according to equation (12)).

If x_k was intensity-modulated (i.e. x_k ε [0; 1]), the signal |x_k|² will also amount to either 0 or 1.

It is noted that the sampling points may be determined based on, e.g., a bit error rate. A suitable sampling point may thus correspond to an optimized or suitable bit error rate. Hence, the sample points maybe chosen that allow for such an acceptable (or optimal) bit error rate.

It is further noted that the sampling points may be iteratively or dynamically adjusted by tracing the bit error rate. The receiver may hence at a given time check whether an adjustment of the sampling points result in an improved bit error rate and thus adjust the timing accordingly.

In addition to the OOK example described above, DQPSK modulation is another example that could be utilized, which in further detail is described hereinafter.

As shown with regard to OOK, the transmitter generates a signal s(t) pursuant to equation (4). Now, the vector x is DQPSK modulated, i.e. information is encoded in the phase difference of subsequent signals x_k(t) and x_k(t−Δt) for k==0, . . . , N−1.

A delay-interferometer plus balance detection can be used to obtain signals I(t) and Q(t) from the signal s(t). The signals I(t) and Q(t) received are sampled at sampling points

$\begin{array}{cc}t=\frac{k·\Delta t}{N}\mathrm{with}k=0,\dots ,N-1& \left(13\right)\end{array}$

to obtain signal vectors Î and {circumflex over (Q)}.

Prior to the sampling phase at the receiver, according to the transfer function (in time) of the demodulator, the signal received can be denoted as

I(t)={s(t)·s*(t−Δt)·e^−jπ/1}  (14).

Hence, the signal I(t) is derived from the real part { . . . }, wherein the signal Q(t) is accordingly derived from the imaginary part ℑ{ . . . }.

After the sampling phase, equation (14) results in

$\begin{array}{cc}I\left(\frac{k·\Delta t}{N}\right)=\Re \left\{s\left(\frac{k·\Delta t}{N}\right)·s^{*}\left(\frac{\left(k-1\right)·\Delta t}{N}\right)·{}^{-j\frac{\pi }{4}}\right\}& \left(15\right)\\ \mathrm{or}& \\ I\left(\frac{k·\Delta t}{N}\right)=\Re \left\{\begin{array}{c}\sum _{n=0}^{N-1}t_n^{\prime *}·x^{*}\left(k-1\right)·{}^{-j2\pi n\left(k-1\right)-j\frac{\pi }{4}}··\\ \sum _{n=0}^{N-1}t_n^{\prime *}·x^{*}\left(k-1\right)·{}^{-j2\pi n\left(k-1\right)-j\frac{\pi }{4}}\end{array}\right\}.& \left(16\right)\end{array}$

With

$\varphi =\frac{\pi }{4},$

equation (14) can be rewritten in matrix form

Î _k= {(e_k·W⁻¹·T·x(k))(e_k·W⁻¹·T·x_φ(k−1))*}  (17)

and further to

Î _k = {e _k ·W ⁻¹ ·T·x(k)·x_φ*(k−1)·T*·W⁻¹*·e*_k}.   (18)

The transform matrix T may be equal to the DFT matrix W (transforming with DFT). Thus, equation (18) results in

Î _k = {e _k ·x(k)·x_φ*(k−1)·e*_k}  (19)

and thus in

Î _k = {x _k(k)·x*_kφ(k−1)},   (20)

which can be denoted as

Î _k=cos(Φ_k(k)−Φ_k(k−1)−φ),   (21)

which further equals the conventional DQPSK with

$\Phi =\left[\begin{array}{c}{\Phi }_0\\ {\Phi }_1\\ ⋮\\ {\Phi }_{N-1}\end{array}\right]$

and

Φ_n being the phase of {x_n}.

It is in particular noted that the transform matrix may consider channel characteristics that could be determined in advance to or during data processing. In this case, the transform matrix allows precoding of the data to be conveyed across such a channel in a way that the channel's distortions are at least partially compensated. Hence, noise and/or interference imposed on the channel, e.g., near end and/or far end cross talk, can (at least partially) be compensated. In addition, dispersion of an optical fiber could be compensated by the transform matrix.

In order to consider channel characteristics, the transform matrix may be

T=H _D ⁻¹ ·W,

wherein H_D⁻¹ denotes a diagonal matrix comprising the channel's characteristics.

Common channel estimation techniques could be utilized to determine the characteristics of the channel. One example is a receiver that conveys information regarding the channel quality back to the transmitter (e.g., via a physical or logical feedback channel). In addition, loops can be used at the transmitter to determine crosstalk from adjacent fibers (channels).

However, such predistortion based on channel properties is an option and not necessarily required for the approach presented herein. Insofar, predistortion in the context of this document also comprises a mere transformation utilized by the transform matrix as described and does not require consideration of particular channel characteristics.

FIG. 2 shows another representation of the transmitter. A vector x 201 is generated by a modulator MOD with guard bands (GB) being inserted (see blocks 216, 217 and 218), wherein the vector x 201 pursuant to equation (1) is fed to a processing unit 206, where it is transformed with the matrix T to {circumflex over (x)}=T·x. Furthermore, zero padding is conducted at a processing unit 207 with a vector z of dimension M×1. An output 202 amounts to

$\begin{array}{cc}\tilde{x}=\left[\begin{array}{c}\hat{x}\\ z\end{array}\right]=\left[\begin{array}{c}{\tilde{x}}_0\\ {\tilde{x}}_1\\ ⋮\\ {\tilde{x}}_{N+M-1}\end{array}\right].& \left(22\right)\end{array}$

This vector {tilde over (x)} as output 202 is input to an iDFT 208 as indicated by the square matrix W⁻¹ of dimension (N+M)×(N+M). An output 203 of the block 208 can be denoted as

$\begin{array}{cc}y=W^{-1}·\tilde{x}& \left(23\right)\\ \mathrm{with}y=\left[\begin{array}{c}{y}_0\\ y_1\\ ⋮\\ y_{N+M-1}\end{array}\right].& \left(24\right)\end{array}$

Each element of the vector y can be written as

$\begin{array}{cc}{y}_k=\sum _{n=0}^{N+M-1}{\tilde{x}}_n·{}^{j2\pi \frac{\mathrm{nk}}{N+M}},& \left(25\right)\end{array}$

which stems from the fact that the matrix W⁻¹ comprises elements

$\begin{array}{cc}{w}_{\mathrm{ij}}={}^{j2\pi \frac{j}{N+M}}.& \left(26\right)\end{array}$

The elements of the vector y can be converted into serial signals by a block 209 and further the imaginary part and the real part can each be serially input to a DAC 210, 211 at time intervals

$\begin{array}{cc}t=\frac{k·\Delta t}{N+M}& \left(27\right)\end{array}$

• • with 0<k<N+M−1.

With

$f_n=\frac{n}{\Delta t},$

the output signal 204 of the DAC 211 after a low pass filter 213 amounts to

$\begin{array}{cc}{y}_{}\left(t\right)=\left\{\sum _{n=0}^{N+M-1}{\tilde{x}}_n·{}^{j2\pi f_nt}\right\}.& \left(28\right)\end{array}$

Thus, signal 204 comprises the imaginary part of the complex baseband (BB) representation of the signal that is going to be upconverted at an IQ modulator 214 to the optical carrier frequency provided by a laser diode LD 215.

The baseband (BB) signal can be rewritten as follows:

$\begin{array}{cc}{y}_{\mathrm{BB}}\left(t\right)=\sum _{n=0}^{N-1}{\tilde{x}}_n·{}^{j2\pi f_nt}+\sum _{n=M}^{N+M-1}{\tilde{x}}_n·{}^{j2\pi f_nt}.& \left(29\right)\end{array}$

If

z _n=0 ∀n ε*|0<n<M−1,   (30)

hence

$z=\left[\begin{array}{c}0\\ 0\\ ⋮\\ 0\end{array}\right]$

is a vector comprising zeros. Accordingly,

{tilde over (x)} _n=0 for N<n<N+M.   (31)

Therefore,

$\begin{array}{cc}\sum _{n=N}^{N+M-1}{\tilde{x}}_n·{}^{j2\pi f_nt}=0& \left(32\right)\end{array}$

and equation (29) results in

$\begin{array}{cc}{y}_{\mathrm{BB}}\left(t\right)=\sum _{n=0}^{N-1}{\tilde{x}}_n·{}^{j2\pi f_nt},& \left(33\right)\end{array}$

and based on

{tilde over (x)} _n ={tilde over (x)} _n for 0<n<N−1   (34)

equation (33) can be denoted as

$\begin{array}{cc}{y}_{\mathrm{BB}}\left(t\right)=\sum _{n=0}^{N-1}{\hat{x}}_n·{}^{j2\pi f_nt},& \left(35\right)\end{array}$

which equals

$\begin{array}{cc}{y}_{\mathrm{BB}}\left(t\right)=\sum _{n=0}^{N-1}T_n·x·{}^{j2\pi f_nt},& \left(36\right)\end{array}$

being the signal to be upconverted by the modulator 214.

After being processed by the modulator 214 with an optical carrier of a frequency f_c, an optical output signal 205 amounts to

$\begin{array}{cc}y\left(t\right)=\sum _{n=0}^{N-1}T_n·x·{}^{j2\pi \left(f_c+f_n\right)t}.& \left(37\right)\end{array}$

The spectrum of this optical output signal 205 is visualized by FIG. 3.

Further Advantages

The approach provided in particular bears the following advantages:

• • (a) The spectral efficiency can be significantly increased utilizing an OFDM-like approach to be compatible with direct detection.
  • (b) No power is required for sending an optical carrier signal.
  • (c) A one tap pre-compensation is possible, permitting the transmission over dispersion uncompensated links.
  • (d) A simple subcarrier selection is possible in the receiver, e.g., by discarding samples, therefore allowing a selection of samples that suffered less from detrimental effects (i.e. use “good subcarriers”).
  • (e) Side information can be sent multiplexed in some subcarriers (pilot tone, training symbols, coding data, protocol data, etc.).
  • (f) There is no need for complex digital signal processing to be supplied with the receiver.
  • (g) The approach allows for inexpensive highly spectral efficient systems, e.g., 100 Gbps systems with direct detection, without polarization multiplexing and without DSP for DWDM systems.
  • (h) There is no need for a local oscillator in the receiver and no need to waste energy transmitting a pilot signal.

LIST OF ABBREVIATIONS

ADC Analog-to-Digital Converter

BER Bit Error Rate

Br Bit Rate

BW Bandwidth

CompSSB-OFDM Compatible Single Sideband OFDM modulation

DAC Digital-to-Analog Converter

DFT Discrete Fourier Transform

DOSM Digital Orthogonal Subcarrier Multiplexing

DPSK Differential Phase Shift Keying

DQPSK Differential Quaternary Phase Shift Keying

DSP Digital Signal Processing

DWDM Dense WDM

IDFT Inverse DFT

OFDM Orthogonal Frequency-Division Multiplexing

OOK On Off Keying

PolMux Polarization Multiplexing

QPSK Quaternary Phase Shift Keying

SE Spectral Efficiency

WDM Wavelength Division Multiplexing

ZP Zero Padding

Claims

1-15. (canceled)

16. A method for processing data in an optical network element, the method comprising: receiving a vector of symbols and processing a multi carrier signal;linear pre-coding the multicarrier signal to form a linear pre-coded signal, wherein the amplitude of each subcarrier is a linear combination of said symbols, and wherein the coefficients of this linear combination are taken from a matrix T.

17. The method according to claim 16 wherein the value of the optical signal at the k-th sampling point is proportional to the k-th element of the vector x.

18. The method according to claim 16, wherein a channel transfer function is pre-compensated by an equalization.

19. The method according to claim 18, wherein said equalization is a one tap equalization.

20. The method according to claim 16, wherein a subcarrier is discarded by a receiver in case the subcarrier experiences deforming and/or attenuating effects above a given threshold.

21. The method according to claim 16, wherein a dummy subcarrier is used to provide a guard band.

22. The method according claim 16, wherein a zero padding is conducted prior to an inverse discrete Fourier transform.

23. The method according to claim 16, wherein a feedback channel is provided to convey information from a receiver to the optical network element.

24. The method according to claim 16, wherein the optical network element is an optical transmitter.

25. An optical network element comprising a linear pre-coder receiving a vector of symbols x and processing a multicarrier signal, wherein the linear pre-coder pre-codes the multicarrier signal to form a linear pre-coded signal, wherein the amplitude of each subcarrier is a linear combination of said symbols, and wherein the coefficients of this linear combination are taken from a matrix T.

26. The optical network element according to claim 25 comprising a zero padding unit prior to an inverse discrete Fourier transform unit prior to said modulator.

27. The optical network element according to claim 25 wherein the value of the optical signal at the k-th sampling point is proportional to the k-th element of the vector x.

28. The optical network element comprising a control unit that is arranged such that the method according to any of claim 16 can be executed.