method for encoding digital data for transmission over a plastic optical fiber, method for receiving

专利摘要:
METHOD FOR ENCODING DIGITAL DATA FOR TRANSMISSION THROUGH AN OPTICAL FIBER OF PLASTIC, METHOD FOR RECEIVING AND DECODING A DIGITAL SIGNAL CODED BY AN ENCODING METHOD, APPLIANCE FOR RECEIVING AND DECODING AN ENCRYPTED CODED DIGITAL SIGNAL. The present invention relates to an efficient coding and modulation system for transmitting digital data over plastic optical fibers. In particular, the digital signal is encoded by means of a three-level side class encoding. The spectral efficiency of the system is configurable by selecting the number of bits to be processed at each of the levels. The first level applies a binary BCH encoding to digital data and performs side class partitioning through constellation mapping and lattice transformations. Similarly, the second level applies another binary BCH coding, which can be performed selectively according to the desired configuration by two BCH codes with substantially the same coding rate, operating on code words of different sizes. The third level is uncoded. Both the second and third levels undergo mapping and lattice transformation. After adding the levels, one (...).
公开号:BR102012005220B1
申请号:R102012005220-2
申请日:2012-03-08
公开日:2020-12-22
发明作者:Carlos Pardo Vidal；Rúben Pérez de Aranda Alonso
申请人:Knowledge Development For Pof, S.L；
IPC主号:

专利说明:

The present invention relates to a transmission of data by a plastic optical fiber. In particular, the present invention relates to a method and apparatus for transmitting and receiving data over a plastic optical fiber using an adaptive error corrective code and modulation scheme. HISTORY OF THE INVENTION
Today's communications systems use several types of cable and radio interfaces. The most reliable are glass optical fibers that also allow very high transmission rates. On the other hand, copper cables still form part of the telephone lines that are also used for data transmission. Especially in the last few decades, wireless communications have developed rapidly. All of these means of data transport have their own characteristics and are suitable for use in different scenarios and architectures.
Glass optical fibers (GOF) are used today especially for communications that require very high bandwidth and very low attenuation. Since glass optical fibers have very small diameters and low numerical apertures (NA - numerical apertures), their installation requires tools of 0.5. However, there is also an SI-POF with a low numerical aperture of 0.3 that allows for higher data rates as well as PMMA GI-POF with a product with a bandwidth length close to 1 GHz x 100 meter. PMMA has several attenuation windows that allow the POF to be used with different sources of visible light from blue to red Emitting Diodes (LEDs) or red Lasers Diodes (LD - Lasers Diodes).
In comparison to GOF, plastic optical fibers have an advantage of very easy installation. They can be employed by professional and non-professional installation workers using basic tools such as scissors or cutters and inexpensive plastic connectors. It is resilient to misalignment and strong vibrations so that it can be installed in industrial or automotive environments without loss of communication capacity. POF connections also have much greater tolerance for residual dust on the end faces than GOF connections, due to the larger core diameter.
Since POF transmission is optical, plastic optical fibers are completely immune to electrical noise. In this way, the existing copper wiring will not interfere with the data passing through plastic optical fibers, so it can be installed right next to electrical wiring. Plastic fiber optic and optoelectronic connectors for POF are primarily low-cost consumer parts that allow installation workers to save on cable costs and installation, testing, and maintenance time. Plastic optical fibers have been widely used, in particular, for information and entertainment networks in cars and can now be seen as a global standard for high-speed networks on cars such as Media Oriented Systems Transport (MOST - Media Oriented Systems Transport).
Figure 1 illustrates an example of a system for transmitting and receiving data via POF. The transmission by plastic optical fibers is based on a modulation of the light intensity with direct detection. The signal to be transmitted is generated from a digital circuit 110 to encode and modulate the user bitstream information and passed to an analog front end (AFE - analogue front end) 120 of transmitter (Tx) for data conversion digital signals in an electrical signal to control the light emitting element 13 0. After this conversion of the electrical signal to an optical signal, the latter is then inserted into the optical fiber 150. Electrical optical converters used for plastic optical fibers are typically emission diodes of light (LED) characterized by properties such as a peak wavelength, a wavelength width or modal distribution of launch.
During signal transmission via plastic optical fibers 150, light is affected by severe attenuation as well as distortion mainly due to modal dispersion. Modal dispersion is caused by different light modes propagating in the fiber in different ways and with different speeds and attenuations, resulting in different arrival times at the receiver. The optical signal is also affected by a so-called mode coupling where the energy from higher order modes is transferred to lower order modes and vice versa. As a consequence, an optical pulse is amplified, which leads to lowering the signal bandwidth.
In a receiver, the optical signal from the plastic optical fiber 150 is converted to electrical intensity by means of an opto-electrical converter 170 as a photodiode. The electrical signal is then processed by the analog front end (AFE) 180. In particular, it is amplified, among other things, by a trans-impedance amplifier (TIA - trans-impedance amplifier) and connected to a digital receiver 190. 0 TIA it is typically the most important source of noise that limits the final sensitivity of the communication system.
With respect to data transmission technology, GOF has successfully used non-return-to-zero (NRZ) modulation. In particular, the current fiberglass communication systems mainly use NRZ 8b / 10b or NRZI 4b / 5b line coding which requires a transmission rate of 1.25 GHz and 125 MHz for 1 Gbps and 100 Mbps solutions, respectively . Current solutions with plastic optical fiber in this way have also adopted NRZ modulation for data communications. However, plastic optical fibers have a different frequency and time response than glass fibers and also have considerably greater attenuation. As a means of communication, plastic optical fibers show very high modal dispersion due to their important differential delay and differential attenuation. Large area photodiodes required for coupling with a fiber typically have limited bandwidth. In view of a plastic fiber optic frequency response, solutions supporting 100 or 150 Mbps are possible up to about 50 meters; however, 1 Gbps does not appear to be achievable without more advanced technology.
Figure 2A shows a variation of the POF optical bandwidth (y-axis, in MHz) as a function of the fiber length (x-axis, in meters). Figure 2B shows the variation of the product bandwidth-length (y-axis, in MHz-100m) as a function of the fiber length. Here, the fiber is a SI-POF with a numerical aperture NA of 0.5 (in particular, Mitsubishi Eska-GH4001 model), and the light source and an RCLED with launch condition FWHN NA of 0.31, peak of 658 nanometer wavelength and 21 nanometer FWHN wavelength. As can be seen from Figure 1, a suitable flat response for a desired 1.25 GHz transmission rate is only possible in the first few meters of the plastic optical fiber. For a laser light source, the optical bandwidth as a function of length is very similar. Therefore, the bandwidth bottleneck is produced by plastic optical fibers regardless of how fast the light source is because the limiting factor is, in particular, the modal dispersion by coupling modes in the fiber. SUMMARY OF THE INVENTION
In view of the aforementioned limitations of plastic optical fiber, the objective of the present invention is to provide an efficient and adaptive transmission system based on plastic optical fibers.
This is achieved by the resources of the independent claims.
Additional advantageous achievements are addressed in the dependent claims.
Employing plastic optical fiber provides many advantages. In particular, with respect to wireless and electrical transmission media, POF are resilient against electromagnetic interference. Compared to optical fiber glass, POF allows easier installation, is less expensive and provides greater robustness with respect to connections. The present invention explores the advantages of POF and provides an adaptive system that allows communication with a high data rate per POF. It is the particular approach of the present invention to transmit POF data encoded by a three-level side class code, where the first level includes BCH encoding, the second level includes BCH encoding with a higher code rate than the first level, and all three levels include respective mapping for a constellation and lattice transformation of the mapped symbols. The levels are then added and the resulting encoded symbols are mapped in a time domain modulation. The second level provides two selectable BCH codes with substantially the same code rate and different code word length.
In accordance with an aspect of the present invention a method is provided for encoding digital data for transmission over a plastic optical fiber. The method comprises the steps of: encoding digital input data by a three-level side class encoding and modulating the symbols encoded with the three-level side class encoding using a time domain modulation. Furthermore, the three-level side class encoding includes the steps of separating from the digital input data a first portion, a second portion and a third portion of data with the respective bit quantities; encode the first portion of data with a first BCH code at a first level and encode the second portion with either a Second or a third BCH code at a second level, where the third BCH code has a codeword length less than the code word length of the first and second BCH code and the third BCH code has substantially the same code rate as the second BCH code. At the first level, a step of mapping the first portion encoded in symbols of a first predefined constellation is performed, as well as a lattice transformation of the mapped symbols to achieve lateral class partitioning. At the second level, a step of mapping the second portion encoded in symbols of a second predefined constellation is performed depending on whether a second or third BCH code was used as well as a lattice transformation of the mapped symbols to achieve side class partitioning. At the third level, a step of mapping the third portion into symbols of a third predefined constellation is performed as well as a lattice transformation of the mapped symbols to achieve side class partitioning. Then, a step of adding the transformed symbols from the first, second and third level is applied.
According to another aspect of the present invention, a method is provided for decoding a digital signal encoded with a three-level side class encoder and received via a plastic optical fiber, the method comprising the steps of: demodulating the encoded digital signal with a time domain modulation to obtain symbol code words and transform them with an inverse lattice transformation, decode the demodulated and transformed symbols with a three-stage decoder including the steps of: extracting a first portion of a code applying reverse lattice transformation and a module operation to a symbol obtained by subtracting the first decoded portion from the demodulated signal; first decode the first portion with a first BCH decoder and based on the first decoded portion by selecting a first side class; extracting a second portion by applying reverse lattice transformation and a modulus operation to a symbol obtained by subtracting the first decoded portion from the demodulated symbol; decode in a second stage the second portion with a second or a third BCH decoder and based on the second decoded portion select a second side class; obtaining a third portion by subtracting the first and second decoded side classes from the demodulated symbol and applying an inverse lattice transformation and a module operation; and multiplexing the first, second and third decoded portion, in which in the second stage the second BCH code and the third BCH code are provided, in which the third BCH code has a code word length less than the length of codeword of the first and second BCH code, and the third BCH code has substantially the same code rate as the second BCH code.
The plastic optical fiber here is any commercially available optical fiber made from plastics. The present invention relates to digital processing to be carried out on the transmitter before the signal is converted to analog values to control a light emitting element to generate the optical signal injected into POF and / or the receiver after the optical signal has been detected by a photoelectric element.
The three-level side-class encoding is a multi-level side-class encoding that provides a multi-dimensional constellation in which the bits that select between side classes of the first level are better encoded than the bits selected between side classes of the second level and / or constellation points. The bits that select a side class are better protected than the bits that specify the constellation point. The bits that select the side class at a lower level are better protected than the bits that select a side class at a higher level of the multi-level side class code.
Adaptability to possible changes in channel conditions is provided by the possibility of defining / selecting the number of bits in the three data positions, for example, among valid pre-defined configurations supported by coding, mapping, lattice transformation, and number of states modulation of time domain employed. In addition, the selectability of either the second or third BCH code at the second level provides an even better chance of achieving the desired spectral efficiency, especially lower spectral efficiencies.
Advantageously, the method of the present invention further comprises a step of selecting the second or third BCH code at the second level based on, for example, channel conditions. The selection of processing to be performed at the second level must be conducted synchronously at the transmitter and receiver. This can be achieved, for example, by defining a suitable protocol for a message exchange between the transmitter and the receiver and / or rules / conditions for switching. In the same way, the selection of the number of bits inserted in the respective levels can be configured. The number of bits in the second and / or third levels can be zero for some configurations.
In general, an adaptive algorithm for spectral efficiency of the three-level side class encoder can be established based on the signal-to-noise ratio (SNR) or reception quality measured at the receiver (such as error rate bit rate, block error rate, or any other quality measure). The receiver can then feed this information to the transmitter. This is advantageously accomplished using physical layer headers (or portions of such headers) defined for this purpose and transmitted through a feedback channel. However, the present invention is not limited to these and any switching mechanism can be employed. For example, the receiver can measure a quality indicator as exemplified above, and based on this estimate an appropriate encoder configuration, which is then signaled to the transmitter. The feedback channel can be provided within packet headers of the physical layer packets, or, alternatively, in packets dedicated to such a feedback channel. The quality indication can be sent regularly at predefined time periods / at predefined time instances, or it can be sent irregularly as soon as a change in encoder settings is required, for example, due to a change in the quality of channel.
According to this information (quality indication) the transmitter can then switch the MLCC configuration, and signal the switching to the receiver by encoding a corresponding indication within the physical layer header. The configuration change is thus synchronized through its signaling within a header to ensure that the receiver can always properly decode the data. On the other hand, the header preferably uses a fixed MLCC coding scheme configured with the lowest spectral efficiency, so that the header can be robustly decoded under any noise and distortion conditions, for which the system has been designed. However, the switching mechanism described above is only an example, and the present invention can also work with other switching mechanisms. Preferably, the second and third binary BCH codes have the same primitive polynomial. This allows you to use the same second-level implementation for both the second and third code.
According to an advantageous embodiment of the present invention, the first BCH encoder generates code words with 2044 bits based on 1637 bits of input information, and / or the second BCH encoder generates code words with 2044 bits based on 2022 bits of input information, and the third BCH encoder generates 1022-bit code words based on 1000 bits of input information. This configuration is particularly suitable for the desired transmission rate of around IGbps. However, the present invention is not limited to this. The length of the primitive polynomial as well as the particular code should preferably be selected according to the requirements of the system.
Advantageously, the mapping is one of a QPSK Gray, BPSK, Z2 or RZ2 mapping, the lattice transformation includes translation, scaling and / or rotation of a symbol, and / or the time domain modulation is M-PAM. In particular, the first level mapper can be a QPSK Gray mapper, the second level mapper can be either a QPSK Gray mapper or a BPSK mapper depending on the selected BCH code, and the third level mapper is a Z2 or RZ2 mapper (as much as possible Gray). Gray mapping has the advantage that a detector error instead of a constellation point another nearby constellation point results in fewer bit errors, for example, in a single bit error. It is noted that the present invention is not limited to IM-PAM as a modulation. In general, the present invention can also work well with difference modulations, such as an M-PAM difference based on a quantization device. Other time domain modulations such as phase shift modulations can also be considered, however, these would likely lead to a less efficient system.
Preferably, the method additionally comprises a Tomlinson-Harashima pre-coding step applied to the modulated symbols. However, other equalization approaches are also possible for the present invention.
For example, instead of pre-coding, an Forward Feed Equalizer can be applied to the receiver. This may be more suitable for systems, where a feedback channel from the receiver to the transmitter is difficult to implement. Note that these were just examples and the present invention can also work with any other equalization techniques.
In accordance with yet another aspect of the present invention, an apparatus is provided for encoding digital data for transmission over a plastic optical fiber. The apparatus comprises a multi-level side class encoder for encoding digital input data by a three-level side class encoding, wherein the multi-level side class encoder additionally includes a demultiplexer to separate from the digital input data a first portion, a second portion and a third portion of data, each with a predetermined number of bits; a first BCH encoder that encodes the first portion of data with a first BCH code at a first level; a second BCH encoder at a second level, wherein the second BCH encoder is adapted to encode the second portion with either a second or a third BCH code provided both in the second BCH encoder, wherein the third BCH code has a code word length shorter than the code word length of the first and second BCH code, and the third BCH code has substantially the same code rate as the second BCH code; a first mapper on the first level to map the first portion encoded in symbols of a first predefined constellation and perform a lattice transformation of the mapped symbols to achieve side class partitioning; a second mapper on the second level to map the second portion encoded in symbols from a second predefined constellation depending on whether the second or third BCH code was used and perform a lattice transformation of the mapped symbols to achieve side class partitioning; a third mapper at the third level to map the third portion into symbols from a third predefined constellation and perform a lattice transformation of the mapped symbols; and an adder to add the transformed symbols of the first, second and third level. The device additionally comprises a modulator to modulate the symbols encoded with the three-level side class encoding using a time domain modulation.
In accordance with yet another aspect of the present invention, an apparatus is provided for decoding a digital signal encoded with a side class encoder of three levels and received via a deplastic optical fiber, the apparatus comprising: a demodulator for modulating the encoded digital signal with a domain modulation. time to obtain symbol code words, a transformation unit to transform demodulated symbols with an inverse lattice transformation; a multi-stage decoder to decode the demodulated and transformed symbols, the multi-stage decoder having three stages and additionally comprising: a first extractor to extract a first portion of a code word by applying a reverse lattice transformation and a module operation for a demodulated symbol; a first BCH decoder to first decode the first portion and based on the first decoded portion select a first side class; a second extractor for extracting a second portion by applying an inverse lattice transformation and a modulus operation for a symbol obtained by subtracting the first decoded portion from the demodulated symbol; a second and a third BCH decoder in a second stage applying a respective second BCH code and third BCH code, in j, i that the third BCH code has a shorter codeword length than the code of the first and second BCH code, and the third BCH code has substantially the same code rate as the second BCH code; a decoder for a second stage decoding the second portion with the second or third BCH decoder and based on the second decoded portion selecting a second side class; an extractor third to obtain a third portion by subtracting the first and second decoded side classes from the demodulated symbol and applying an inverse lattice transformation and a modulus operation; and a multiplexer for multiplexing the first, second and third portion w decoded.
According to a preferred embodiment of the present invention, an integrated circuit is provided, implementing ‘any of the methods described above.
Advantageously, a system is provided to transmit digital data via plastic optical fiber. The system comprises a transmitter including a coding apparatus as described above, an electro-optical converter to convert the encoded signal into an optical signal and to inject the optical signal into the POF, an opto-electrical detection element to transform an optical signal received from the POF into an electrical signal, and a decoder as described above to decode the received signal.
The above and other objects and resources of the present invention will become more apparent by following the description and preferred embodiment given in conjunction with the accompanying drawings in which:
Figure 12 is a block diagram illustrating an example of transforming the lattice to a third level;
Figure 13 is a block diagram illustrating an example of the implementation of a lattice adder (vector sum);
Figure 14 is a block diagram illustrating an example of a second stage lattice transformation;
Figure 15 is a block diagram illustrating an example of implementing a module operation;
Figure 16 is a block diagram illustrating M-PAM modulation output;
Figure 17 is a table including valid configurations of the MLCC code according to an advantageous embodiment of the present invention;
Figure 18 includes constellation diagrams for the first and second level, for constellations after mapping and lattice transformation, and for the final constellation, assuming the second configuration in Figure 17;
Figure 19 includes constellation diagrams for the first and second level, for constellations after mapping and lattice transformation, and for the final constellation, assuming the third configuration in Figure 17;
Figure 20 includes constellation diagrams for the first, second and third level, for constellations after mapping and lattice transformation, and for the final constellation, assuming the fourth configuration in Figure 17;
Figure 21 includes constellation diagrams for the first, second and third level, for constellations after lattice mapping and transformation, and for the final constellation, assuming the sixth configuration in Figure 17;
Figure 22 is a graph representing bit error rate curves for different BCH code rates at the first level;
Figure 23 is a graph representing bit error rate curves for the first level and the second level as well as for the entire scheme without BCH code on the second level;
Figure 24 is a graph representing bit error rate curves for the first level and the second level as well as for the entire scheme with a BCH code capable of correcting a single bit error at the second level, and where 1 bit per dimension it is coded at the second level;
Figure 25 is a graph representing bit error rate curves for the first level and the second level as well as for the entire scheme with a BCH code capable of correcting two bit errors at the second level, and where 1 bit per dimension is coded at the second level;
Figure 26 is a graph representing bit error rate curves for the first level and the second level as well as for the entire scheme with a BCH code capable of correcting two bit errors at the second level, and where 0.5 bits per dimension is coded at the second level;
Figure 27 is a graph representing bit error rate curves for the first level, the second level, and the third level as well as for the entire scheme;
Figure 28 is a graph illustrating power balance at the link for predefined conditions and three different bands;
Figure 29 is a flow diagram illustrating a coding method according to an embodiment of the present invention;
Figure 30 is a block diagram showing a decoder with a three-level side class decoding and demodulation in accordance with an embodiment of the present invention;
Figure 31A is a block diagram illustrating Tomlinson-Harashima precoding adapted in accordance with an embodiment of the present invention;
Figure 31B is a block diagram showing Forward Feed Equalizer, which can be used with the present invention;
Figure 31C is a block diagram showing Feedback Feedback Equalizer, which can also be used with the present invention;
Figure 32 is a block diagram illustrating a differential modulator according to an embodiment of the present invention;
Figure 33 is a block diagram illustrating a differential demodulator according to an embodiment of the present invention; and
Figure 34 is a flow diagram illustrating a method of decoding according to an embodiment of the present invention. DETAILED DESCRIPTION
The problem underlying the present invention is based on an observation that techniques typically used for optical glass fiber are not sufficient to achieve efficient data transmission through a plastic optical fiber. Due to a difference between characteristics of plastic optical fiber channels compared to glass, wireless or copper optical fiber channels, the techniques developed and employed for such channels are also not directly applicable to plastic optical fibers. It is an object of the present invention to allow data communications with high spectral efficiency by POF.
One of the general criteria for designing a communications system is to maximize the capacity of the channel. The channel capacity limit can be calculated according to the theory. Key digital techniques that impact the capacity limit approach are modulation, inter-symbol interference compensation and coding. These techniques have to be designed with regard to the characteristics of the communication channel and possibly in consideration of each other.
The crest factor (also called peak-to-average ratio) is a ratio of the peak amplitude of the waveform divided by the mean square of the root of the waveform. For optical systems, a modulation is appropriate that minimizes the crest factor and maximizes the variance of the optical signal for a given optical modulation amplitude (OMA) injected into the POF. Modulation techniques that allow this are M-aria pulse amplitude modulation (M-PAM) and the M-PAM difference. Assuming a zero mean constellation before the electro-optical conversion, the crest factor is minimized and the average energy of the symbol is minimal for the minimum distance of a given constellation, as several levels of the signal are uniformly distributed. The number of levels of pulse amplitude modulation can be defined as a function of the bandwidth, required bit rate, and / or encoding. In order to design the modulation properly, a power balance in the connection of the plastic optical fiber channel has to be analyzed. To maximize the power balance on the link, there is an optimal value for the number of levels and the signal bandwidth for a desired transmission rate as will be shown below. A high spectral efficiency communications system is required in order to maximize the power balance on the link. Based on this requirement, channel equalization and coding have to be designed with regard to modulation.
As a consequence of the signal amplification in the transmission medium, here POF, the neighboring data that carry symbols overlap when they are received, which makes it difficult to correctly detect and decode them. This effect is called inter-symbol interference. In order to recover such symbols, equalization techniques are typically employed. There are many equalization approaches on the receiver side available in the prior art including MMSE equalizer, zero-forcing, Forward Power Equalizer, feedback decision equalizer, etc.
In order to efficiently design a communication system, based on Volterra models that can be obtained for a particular channel by analyzing its measured characteristics, the linear and non-linear parts of the characteristics of the channel can be separated. For the linear part of the channel, the maximization of the power balance in the connection according to the information theory can be performed. Furthermore, equalization can be designed independently for the linear and non-linear part of the channel. On the transmitter and / or receiver side, a linearizer (a non-linear filter structure) can be employed to provide a linear channel far enough where the well-known equalization techniques can be used.
For example, an Feed Forward Forward Equalization (FFE) is an equalization technique employed at the receiver that corrects the received waveform based on information about the waveform itself, in particular the waveform current and waveform associated with previously received communication symbols. Equalization is performed on the waveform (voltage levels) before any decisions about the received bits are made. Another well-known technique is Decision Feedback Equalization (DFE). DFE calculates a correction value that adapts decision thresholds to detect multi-dimensional modulation symbols. In this way, DFE results in changing the threshold based on what new decisions are made (more details on DFE and equalization can be found in JG Proakis, Digital Communications, 4 th Edition, McGraw-Hill Book Co., New York, 2001 , incorporated herein by reference). A disadvantage of DFE is the propagation of error, resulting from decision errors in the output of the decision device that cause an incorrect estimate of the post-cursor Inter-Symbol Interference (1ST - Inter-Symbol Interference). Error propagation can be prevented by using transmitter pre-coding.
Pre-coding allows you to move the cancellation of the ISI post-cursor to the transmitter where data symbols are available. Furthermore, a feedback filter is employed to pre-code the signal using a current channel impulse response. The impulse response is typically estimated at the receiver using adaptive filtering techniques and feedback to the transmitter. There are several different variations of precoders (cf., for example, GD Forney and G. Ungerboeck "Modulation and coding for linear Gaussian channels", IEEE Trans, on Information Theory, vol. 44, no. 6, Oct. 1998, pp. 2384-2415, which is incorporated by reference). One of the precoding techniques, explicitly the Tomlinson-Harashima precoder (THP), is of particular interest. The Tomlinson-Harashima Pre-coding (for more details see, for example, RD Wessel, JM Cioffi, "Achievable rates for Tomlinson-Harashima Precoding", IEEE Trans, on Inf. Theory, vol. 44, no. 2, Mar 1998, pp. 824-831, which is also incorporated here by reference) is considered to be a prominent pre-coding scheme especially due to its ability to efficiently cancel known interference on the transmitter side. Therefore, the information rates achieved by THP are higher than those achieved by conventional linear pre-coding schemes.
Figure 3 illustrates a well-known use of THP with an M-PAM modulation. The Tomlinson-Harashima precoder moves the feedback filter 330 from a DFE frame to the transmitter and combines it with a 310 module operator to reduce the post-cursor ISI-compensated symbols for the pre-coding Voronoi region of the corresponding M-PAM constellation. The forward feed filter 340 remains in the receiver to compensate for the cursor and pre-cursor ISI and to clear the noise. Therefore, a module 320 operator analogous to the module 310 operator on the transmitter side is required to retrieve the transmitted symbols. THP is able to approach the realization of the ideal DFE without error propagation, for modulations with medium and high spectral efficiency.
However, THP equalization has four capacity losses inherent in loss of pre-coding, loss of crest factor, loss of modulus, and loss of molding, of which only the first two are relevant to the intended application for POF. These losses are mainly caused by the application of the module operator and depend on the number of modulation levels as shown below. The module operator with "the feedback filter on the transmitter converts a discrete uniform distribution of M-PAN symbols into a continuous uniform distribution spanning the entire Voronoi region of the original constellation (assuming the energy dispersion of the feedback filter) is large enough to completely fill the Voronoi region corresponding to the pre-coding. This results in an increase in the energy of the transmission signal, which needs to be compensated by the transmitter in order to enter the same average power for the POF. the increase in energy leads the receiver to a loss of the available SNR, which is called pre-coding loss. The pre-coding loss can be estimated as a function of the number of modulation levels M such as:

For example, for 2-level PAM (2-PAM), the loss of pre-coding is approximately 1.25 dB. For larger constellations, the loss of pre-coding decreases towards zero.
The translation of the discrete M-PAM constellation to the continuous Voronoi region performed by THP also results in an increase in the crest factor. The crest factor of an M-PAM modulation depends on M and varies between 0 dB for 2PAM and the asymptotic 4.77 dB for arbitrarily high number of modulation levels. A signal pre-encoded by THP has a constant crest factor of 4.77 dB, assuming that the entire Voronoi region is filled. The crest factor loss is a difference between the crest factor at the input and the output and is defined as:

Since the POF is a channel limited by peak power, the loss of the crest factor actually represents decreased performance.
Figure 4 shows the loss of performance (in dB) of a THP transmission considering both the loss of pre-coding and the loss of crest factor as a function of the number of modulation levels M = 2. The 420 curve represents the loss due to the crest factor of the M-PAM modulation that would be completely equalized by the receiver (Decision Feedback Equalizer or forward feed equalizer). For 2-PAM (k = l) there is no loss, since the crest factor of 2-PAM is 0 dB. Curve 430 shows the transmission loss for THP (the loss of pre-coding plus the loss of crest factor) which becomes asymptotically the same as the loss of crest factor for the high numbers of modulation levels. Finally, curve 410 illustrates the advantage of M-PAM with respect to THP as a function of M. Since the crest factor for THP is constant and equal for all values of M, explicitly 4.77 dB, it can be seen as an extra loss due to pre-coding in the very small range from M to 4. The performance loss is negligible for M equal to or greater than 4 (corresponding to k 2). When M is high enough, the pre-coded symbols are independent and uniformly distributed random variables. This implies that the statistics of the precoded symbols are very similar to the statistics of the original data symbols and the spectrum of the precoded symbols is white. Furthermore, since pre-coding is employed on the transmitter side, there is no problem in applying a more complicated modulation coding such as lattice-coded modulation or side-class coding, which requires delaying decisions and therefore cannot be combined well with a DFE on the receiver.
However, the THP used in the transmitter requires feedback from the receiver in order to get the channel back to the current. Despite this small implementation disadvantage, THP still remains suitable for the predominant part of the intended POF applications. For example, THP is suitable for any of a star topology, daisy chain topology or tree topology. In the star topology, each node is connected to the network through a packet switch node via a duplex POF having two fibers in the respective two directions. In the daisy chain topology, some nodes have packet switching capabilities and more than one duplex interface. A node is connected to the network and, at the same time, acts as a bridge between the different network domains with which it is interconnected. The tree topology is an evolution of the daisy chain topology, in which some nodes have more than two POF duplex interfaces. These three topologies are generally suitable for any type of sensor applications or distribution of media based on video, especially for applications in home networks, applications in industrial or automotive plants, in particular, cameras and interconnected screens.
However, current POF-based automotive applications also use a physical ring topology through a simplex POF. Correspondingly, several nodes are connected in series or are connected to a central unit. Such a topology is not necessarily optimal for sensor applications. Furthermore, the implementation of a feedback channel for each pair of nodes along a common ring is difficult to implement, especially for a higher number of nodes involved. For such topologies, therefore, equalization techniques other than THP may be more convenient. For example, an Forward Power Equalization (FFE), which does not require feedback from the receiver to the transmitter. When physical ring topology is required, FFE may perform better than DFE due to a high spectral efficiency M-PAM, despite the loss of performance due to increased noise. Explicitly, DFE can suffer a considerable error propagation in such a system. In order to achieve an efficient use of modulation, coding and pre-coding, it is important that these techniques are designed in consideration of each other. In particular, by employing multi-level side-class coding, additional losses can be avoided by designing module operators to separate each coding level in the multi-stage decoder structure and be congruent with the Voronoi region of the THP. The multi-stage decoder structure on the receiver thus performs both the separation of partition channels (encoding levels) and the reduction of THP in a single step so that the multi-level side class code decoder based on MSD can be connected directly to the power supply filter output as described below with reference to Figure 31A.
In view of the POF characteristics discussed above, the high spectral efficiency objectified by the present invention is only achievable when advanced coding and modulation schemes are employed, such as lattice coded modulation, interlaced binary code modulation, side class coding, or others coding-modulation schemes. For example, Interlaced Binary Code Modulation has the disadvantage of extensive latency caused by the interleaver. Furthermore, it has a lower performance when used for modulations of medium and high spectral efficiency, and a non-uniform encoding gain for adaptive bit rate.
A multi-level side class encoding is a spherical boundary encoding technique. The theoretical description and design of MLCC can be found in G. D. Forney et al., "Sphere-bound-achieving coset codes and multilevel coset codes", IEEE Trans, on Information Theory, vol. 46, no. 3, May 2000, pp. 820-850, In particular, in Sections V.E, V.F and VII. B as well as in U. Wachsmann et al. "Multilevel Codes: Theoretical Concepts and Practical Design Rules", IEEE Trans, on Information Theory, vol. 45, no. 5, July 1999, pp. 1361-1391, which are both incorporated herein by reference. Theoretical rules are formulated in terms of the code rate of the component codes, partition channel capacities and knurled noise per module present in each decoding level, assuming decoding by Multi-Stage Decoder (MSD - Multi-Stage Decoder). However, mathematical theory does not deal with the particular characteristics of binary component codes that are also suitable for implementation in a "real world", meaning, for example, hardware or software implementation. In the literature mentioned above, Low Density Parity Check Codes (LDPC) were studied as possible component codes for MLCC. A combination of LDPC at the first level and BCH at the second level has been suggested to support a spectral efficiency adaptability of up to 0.25 bits / s / Hz / dim.
However, LDPC codes require very high computational complexity to decode which, on the other hand, requires more area in the hardware implementation and causes greater power consumption. In terms of power balance in the optical connection, the improvement caused by using LDPC next to BCH seems to be negligible. Furthermore, with the LCPC code there is a potential error floor, a compensation for which could require the use of an additional external algebraic code. The binary codes of Bose, Chaudhuri, Hocquenghem (BCH) are almost perfect algebraic codes in terms of the minimum Hamming distance between the code words. BCH codes do not have an error floor when difficult decision coding is applied. BCH codes also provide the advantage of a simple implementation that can be easily embedded, for example, in an integrated circuit. For high code rates, BCH codes provide high coding gain, which, on the other hand, decreases for medium or lower code rates. As such, BCH codes are not particularly suitable for adaptability in terms of their configurable code rate.
In spite of this, according to the present invention, BCH codes are in fact employed in a system with spectral efficiency adaptation. However, the code rate of BCH codes employed is fixed and adaptation is carried out by means of crosshairs. In particular, the code rate of the BCH codes selectable at one level is substantially the same in order to avoid degradation of coding gain when adapting the spectral efficiency. It is beneficial to have selectable BCH codes at the same code rate. However, the code rate of these BCH codes can also vary as the code rate of the second and third BCH codes are selected according to the rate of the first BCH code, in order to avoid loss of performance of the entire scheme of MLCC. The two dimensional lattices used in one of the embodiments of the present invention allow adaptation with a step of 0.5 bits / s / Hz / dim. However, a finer step can be achieved, for example, through side-class partitioning implemented in 4-dimensional lattices. The 0.5-bit / s / Hz / dim step allows bit rate adaptation in 3 dB variations of the channel SNR. This variation of SNR only represents a variation of 1.5 dB of the optical power that enters the optical-to-electrical converter (photodiode) included in the optical communication system because a variation of N dB of the received optical power produces a variation of 2 times N dB of the electrical current amplitude of the photodiode, modulated by a time domain modulation such as an M-aria pulse amplitude modulation (M-PAM), which is considered here to be part of the MLCCr code 510, and the PAM symbols are additionally pre-encoded by a pre-encoder 530. The decoder 500b includes a time domain demodulator 580 to demodulate the receiver signal and a multi-stage decoder 590 to decode the demodulated symbols. If THP was not applied, FFE 570 can be used as an equalizer (cf. Figure 31B and the description below). If THP has been applied (cf. Figure 31A), FFE can equalize the cursor and post-cursor of the impulse response of the channel, as well as lighten the noise. In such a case, 570 is the forward feed filter 340 of the THP structure.
Figure 6 illustrates an MLCC 600 encoder that can be used in place of the MLCC 510 encoder shown in Figure 5A according to the present invention. The encoder 600 receives a bit sequence x of the ctMLCC length information belonging to an MLCC code word to be transmitted. The number of aMLCc bits can be selected according to the desired spectral efficiency with regard to channel quality. The information bits x to be encoded as an MLCC code word are first separated in an MLCC 610 demultiplexer at three MLCC levels. In particular, a piece of information with aMLcc bits is separated into portions having respectively β (V), Z (2), β (3) bits, each of which is inserted at a corresponding MLCC level, where aMLCC = β (V) + / (2) + β (3). The amount of information bits entered per level is configurable as an adaptive spectral efficiency function as will be shown below. The bit order must be established to minimize latency at the 32 MSD receiver. Assuming an input vector x = L ^ or "> ^ αMtc & .J for separator 610, after demultiplexing, the insertion on the first level is y = [x0, ..., XW) _J. The second and third level flows are included as a function of the spectral efficiency setting (by inserting a number of input bits other than zero). In particular, the second level can receive the vector y2 = [x (9 (1), ..., X / (1) + / (2) _l ^ (when a number of bits other than zero must be present) and the third level can receive the vector y3 = [^ (i> yj (2) v, ^ βwtcc -i] (when a non-zero number of bits must be present).
The first two levels of the MLCC encoder include early error correction coding 620a and 620b. The third level is uncoded. The 620a error correction encoder encodes the> 0 (1) bits in rtc (1) encoded bits, while the 620b error correction encoder encodes the / (2) bits in nc (2) encoded bits. At the third non-codified level, similarly nc (3) = β (T). In particular, like the MLCC early error correction component codes, the BCH codes were selected as discussed above. The selection of binary BCH codes in particular for each level is critical to ensure performance and avoid error floor of the MLCC scheme as will be shown below.
BCH encoding at the first level 620a is always performed. The resulting code words with nc (l) bits are then mapped using a QPSK 630a Gray mapper, which maps to the number of bits encoded by dimension r Ó (l) = l bit / dim. According to an advantageous embodiment of the present invention, 'c (l) t = 2044 is the length of the BCH code word, while the number of bits of information per code word A: c (l) = β (l) = 1637 bits is the length of the information word before encoding. The second level can employ two different BCH codes in order to provide scalability of the spectral efficiency through mapping with nb (2) of either 0.5 or 1 bit / dim. Alternatively, nb (2) can be zero, in which case also β (2) = o = / (3). Thus, in such a case, no coding is performed at the second level. The primitive polynomial of both BCH codes at the second level is the same as both nh (2) configurations and the shortening of the BCH code is performed to accommodate bits of information belonging to the second level to the configured spectral efficiency. In this way, the code rate of the BCH code remains substantially the same, while its code word length changes. In particular, the code rate rc (l) for the 1st. level is defined as a ratio between the number of bits of information per codeword and the length of the codeword: rc (1) = kc (1) / nc (1). Spectral efficiency 77 is 3 then defined as rj = ∑nb <J) rc (f) • For «A (2) = 0.5 bit / dim, an i = í binary phase shift modulation (BPSK) mapper 630b is used, while for «A (2) = l bit / dim, a quadrature phase shift modulation (QPSK) mapper is applied. According to the advantageous embodiment of the present invention, the BCH code for nb (2) = 1 bit / dim is a code (2044, 2022) and the BCH code for nò (2) = 0.5 bit / dim is a code (1022, 1000). Here, the first number refers to a number of bits of the codeword inserted from the respective BCH encoder and the second number refers to the number of bits of the information word at the input of the are inserted into the crosshair transformer 640a. For the second level, two different cases are distinguished. For wé (2) = l bit / dim, the rotation is not performed. For HÍ (2) = 0.5 bit / dim, rotation is required because the corresponding 2D constellation maps 1 bit by two dimensions (odd number). The reticle transformation architecture of the second level is illustrated in Figure 11. As can be seen from the figure, the value of nb (2) controls the multiplexer to perform the rotation or not.
The lattice transformation for the third level is shown in Figure 12. The rotation is implemented for nb (3) values of 0.5, 1.5, 2.5, 3.5, ... etc. bit / dim. For rtfc (3) = 1, 2, 3, 4, ... etc., rotation is disabled.
After performing the lattice transformations 640a, 640b, and 640c, the symbols transformed in the lattice of each of the three levels are added 650 in this way, performing the lateral class partitioning over the Z2 lattice and the final partitioning. In particular, the input phase and quadrature components of the three levels are added separately to guarantee a respective new component of the input phase S, and quadrature component SQ as illustrated in Figure 13. The symbols with component output of the phase input S, and quadrature component SQ of reticle adder 650 are then further transformed in order to obtain the final two-dimensional square constellation of zero mean over Z2 or RZ2. The lattice transformation of the second step 660 includes the following three steps: 1) -45 degree rotation for £ = 1.5, 2.5, 3.5, ... etc. bits per dimension (where £ =, 2) module operation that restricts the symbols in Figure 32 (modulator) and Figure 33 (demodulator). In particular, the differential modulation scheme illustrated in these figures is designed to prevent loss of module during multi-stage decoding. The 3200 differential modulator transforms M-PAN 3201 symbols of the alphabet {-M + 1, -M + 3, ..., M-3, Ml], where M = 2k, into M-PAN 3202 differential symbols of the alphabet {-M + 1, -M + 3, M-3, Ml}. Correspondingly, a dependency between a current differential symbol and a previously modulated symbol is created. Correspondingly, a demodulator 3300 transforms differential symbols of M-PAN 3301 of the alphabet {-M + l, -M + 3, ..., M-3, Ml} into symbols of M-PAN 3302 of the alphabet {-M + l , - M + 3, ..., M-3, Ml}. The block denoted as z'1 is a symbol delay element. The operation ">> 1" is an offset to the right by one bit and "<< 1" is an offset to the left by one bit.
Difference modulations are particularly suitable for designed receiver circuits coupled to AC (alternating current), and where the absolute value of the signal amplitude of the incoming optical signal is destroyed, for example, as a result of high pass filtering with a filter having lower cutoff frequency.
Figure 17 shows a table with all the configurations valid for the BCH codes discussed above, according to an advantageous embodiment of the present invention. The first column shows the number of M-PAM symbols in 1D per output code word for the entire MLCC. The second column shows number of bits of input information per MLCC code word. The third to fifth columns respectively specify the number of bits per level by which the input information code word is divided. The sixth column specifies the spectral efficiency corresponding to a respective configuration, while the seventh column shows the number of PAM states. In the eighth to tenth columns specify the number of bits encoded by dimension at each level.
Next, the effect of lattice mapping and transformation on symbol constellations is visualized. Assume £ = 1.5 bit / dim, «A (l) = 1 bit / dim, MÔ (2) = 0.5 bit / dim and nò (3) = 0 bit / dim (cf. second line of the Table in Figure 17). After the respective mappers 630, the first stage lattice transformations 640, the lattice adder 650, and the second stage lattice transformation 660, an 8-QAM constellation over RZ2 is obtained, which is additionally converted to 4-PAM .
Figure 18 shows this step-wisely. In particular, in (a), the constellation of the first level after the mapper 630a is shown and in (b), the constellation of the second level after the mapper 630b is shown. After the respective lattice transformation 640a and 640b, and after the vector sum 650, the lateral class partitioning is as shown in (c). Each of the eight possible points in the constellation is obtained by combining a first level point 1801 and a second level point 1802. In particular, by means of lattice transformation 640a, the points of the first level have been scaled to obtain a minimum distance 1 and translated to the first quadrant in 2D. By transforming the 640b lattice, the 1802 points on the second level were rotated by 45 degrees and translated to the first quadrant with a minimum distance of 2. The first level thus divides the space in 2D into four lateral classes, each corresponding to a point of the original QPSK constellation. The MSD at the receiver must first decide which of the side classes has been transmitted. When this side class is known, the next step is to decide between the two points 1802. The ratio between minimum constellation distance and noise standard deviation at the receiver is minimal for the first level (points 1801) and is increased twice for the second level. Thus, the first level requires a binary code with greater capacity to correct errors than the second level. The first-class side class is decided for all symbols belonging to an MLCC code word simultaneously, through difficult-decision and BCH block decoding. The remap of the decoded BCH codeword is performed to obtain the first level side class before the second level decoding in the MSD. After the transformation of second stage lattice 660, a zero mean square constellation over RZ is obtained as shown in (d) of Figure 18. Another example is illustrated in Figure 19 for £ = 2 bit / dim. Similar to the previous example, only the first and second levels are used, Hfc (l) = l bit / dim, «^ (2) =! bit / dim and nA (3) = 0 bit / dim (cf. third row of the Table in Figure 17). In particular, in (a), the constellation of the first level after the mapper 630a is shown and in (b), the constellation of the second level after the mapper 63 0b is shown (rotation is not applied at the second level). After the respective lattice transformation 640a and 640b, and after the vector sum 650, the lateral class partitioning is as shown in (c). After the transformation of the second stage lattice, the resulting zero mean constellation is shown in (d) of Figure 19, resulting in a 16-QAM constellation.
Figure 20 illustrates another example of configuring the MLCCr code as described above. Now £ £ 2.5 bit / dim, right (l) = l bit / dim, »is (2) = l bit / dim and« is (3) = 0.5 bit / dim (cf. fourth line of Table in Figure 17). This is the simplest configuration including all three levels. The constellation and bit mapping for the first, second and third levels are illustrated in (a), (b), and (c). In (d), the lateral class partitioning as performed by the first stage lattice transformations and the vector sum is shown. The first level separates the points in 2D into 4 side classes (points 2001) and each side class of the first level is further divided into another 4 side classes (point 2002) by the transformation of the second level. The MSD on the receiver decides the first level side classes corresponding to each symbol. This provides a 6dB increase in SNR in second level decoding with respect to the first level. Then the MSD decides between 4 side classes providing another increase of the SNR by 6dB. The third level performs decoding with a larger SNR of 12dB with respect to the first level since the minimum constellation distance has been increased by 4. The final constellation after transformation 660 with labeling is shown in (e) and corresponds to a 32-QAM rotated, the symbols of which must be converted into 8-PAM symbols by the 670 modulator.
Figure 21 provides a fourth example of configuring the MLCCr code. Now £ £ 3.5 bit / dim, nè (l) = l bit / dim, Hft (2) = l bit / dim and «h (3) = 1.5 bit / dim (cf. sixth row of the Table in Figure 17). The constellation and bit mapping for the first, second and third levels are illustrated in (a), (b), and (c). In (d), the lateral class partitioning as performed by the first stage lattice transformations and the vector sum is shown. The final constellation after reticle transformation of the second step 660 with labeling is shown in (e), its symbols must be converted into 16-PAM symbols by the 670 modulator.
According to a preferred embodiment of the present invention, for Plastic Optical Fiber the component codes BCH with binary primitive polynomials on Galois field GF (2m) with m = 11 are considered. The primitive polynomials of the first and second levels are advantageously 2047 bits long and have a minimum shortening of 3 bits, resulting in NMLCc = 2,044 symbols in 1D at the output. In the above example of a POF transmission described with reference to Figure 6, as an advantageous embodiment, the BCH (2044, 1637) on the first level and the BCH codes with (2044, 2022) and (1022, 1000) on the second level were applied.
The selection of binary BCH codes for each level is important to guarantee the performance and absence of the error floor of the presented MLCC scheme. In the following, the selection of BCH codes is discussed in detail in order to guarantee the operation of the entire MLCC-MSD system at very low error rates as well as in order to obtain a possibly high coding gain from the involved BCH codes. The performance of the BCH codes is estimated taking into account characteristics of a BCH difficult-decision decoder as well as the module operations performed by the MSD to separate the decoding of the three levels of MLCC.
In order to estimate the performance of the transmission system based on the MLCC according to the present invention, the Bit Error Rate (BER) as a Shannon gap function of the entire MLCC is analyzed under the following assumptions regarding the implementation of the Multi-Stage Decoder on the receiver: • Difficult decision is implemented for the detection of symbols at each decoding level. To separate the information received corresponding to each level, module operators are implemented as defined in G. D. Forney et al., "Sphere-bound-achieving coset codes and multilevel coset codes", IEEE Trans, on Information Theory, vol. 46, no. 3, May 2000, pp. 820-850. • Limited distance decoding like the Berlekamp-Massey Algorithm (BMA), is used for BCH decoding. BMA is widely used for decoding BCH in hardware implementations. Therefore, the estimates presented here are an exact estimate for performance achievable in real implementations. • The MLCC scheme is considered to be applied to an Additive White Gaussian Noise (AWGN) channel. Inter-Symbol Interference is considered to be eliminated by means of a well-designed equalizer. This is the case for Tomlinson-Harashima pre-coding, but not for Decision Feedback Equalizer (DFE) or linear Forward Feed Equalizer (FFE), because error propagation results in color noise in the detector. Nevertheless, the combination of MLCC presented with FFE is adequate, although with a certain loss of capacity produced by the detection noise autocorrelation. For THP, the side class partitioning presented above, was designed in such a way that no module operations are required before Multi-Stage Decoding. In this way, no additional loss is produced by a THP module. • The module operation is implemented at each level before decoding. When all three levels are enabled (because the configured spectral efficiency is high enough), the module operators in the first two levels are implemented to separate the information directed to each of these levels. The module operator at the third level has the same functionality that would be performed by the THP module after the Forward Power Filter, however, without any loss of capacity. The module operation is implemented over an ID grid in the precoder (transmitter side), but can be optimally decoded at the highest enabled decoding level of the MSD structure over a 2D grid.
The performance forecast takes into account the molding and module losses, but not the loss of pre-coding conferred by THP. The MLCC scheme causes a molding loss because the transmitted signal is evenly distributed (that is, all PAM symbols are equally likely) regardless of whether THP is applied or not. Since POF is a channel limited by peak power, constellation molding techniques lead to an increase in the crest factor resulting in a reduction in the channel capacity of the system. In fact, Shannon's real limit corresponds to the Sphere Limit (also called Mold Limit). On the other hand, the loss of module is considered in the evaluation of the probability of error of the symbol in each level of decoding.
Shannon showed that in an AWGN channel with a given SNR and a bandwidth B (Hz), the data rate R (bits / s) of a reliable transmission is superiorly limited by R <B-log2 (l + SNR) . Equally, Shannon's result shows that the spectral efficiency (bits / s / Hz) is superiorly limited by r <log2 (l + SNR), or, given a spectral efficiency r |, that the SNR needed for reliable transmission is limited inferiorly by SNR> 2q - 1. Let's define a normalized SNR parameter SNRn as SNRn = SNR / (2q - 1) . For any reliable coding scheme SNRn> 1, that is, the Shannon limit (lower limit) in SNRn is 1 (0 dB), regardless of the spectral efficiency r |. Furthermore, SNRn measures the gap for capacity, which is the difference in decibels (dB) between the SNR actually used by a coding scheme and the Shannon limit given in SNR for a given r].
The normalized probability of 2-dimensional symbol error is given by the following estimate of the boundary limit, as a function of the lattice A parameters and the noise variance, assuming AWGN and minimum distance detection:

Here, d2 (Δ) is the minimum distance squared between points in the lattice, K (A) is the lattice number (Kissing number) of the lattice, n is the number of dimensions and a2 is the variance of a n-dimensional noise. with Vufn volume, ^ 2) * (2neo2) a / 2, that is, a sphere of squared ray noise nc ^ Qf-) is the Gaussian probability of the error function. This equation means that the normalized average probability of the symbol error is a function of the ratio between the minimum square distance and the noise variance. This ratio, depending on the reticle and constellation used, is related to SNRn differently. The oscillation number is the average number of points at a minimum distance around a point in the constellation. Side class partitioning is based on 2D qZ2 lattices for QAM constellations, and qRZ2 for rotated QAM constellations, where q is the level-dependent scaling factor. For both reticles n = 2.
For 2k-QAM constellations, where k is even, d2 (qZ2J = q2, and the average constellation energy is
From this, the SNRn is given by:

Therefore, the average probability of the symbol error is estimated as follows:

The osculation number is 4 for a very large constellation within the reticle. However, for small lattices, the average number of points at the minimum distance is to be considered, as the border points typically have half the internal points. However, when MSD module operators are implemented before the difficult and noise detecting aliasing is produced, K (A) = 4 is considered accurate, especially when THP is applied. For rotated 2k-QAM constellations where k is odd, the following is an estimate for the symbol's probability of error. By series sum methods, the average energy of the constellation is calculated as
and thus, the SNRn is given by:
and the probability of error of the symbol is as follows:

The oscillation number will be considered to be 4 as for the k for QAM constellations.
From this, the average probability of error of the symbol for each level of decoding involved in MSD can now be calculated.
For the MLCC scheme configured with a total of f bits encoded by dimension and spectral efficiency of 77 bits / Hz / s per dimension, the normalized SNR for each level 1 is defined as a function of the complete MLCC scheme SNRn. When 2Ç is even, the SNRn, 1 is given by
assuming a current spectral efficiency of the system (27) being less than or equal to the number of bits encoded by dimension (Ç). Furthermore, it is assumed that the minimum constellation distance is increased at each level as a function of the sum of the number of bits encoded by dimension at previous levels. When 2Ç is odd, we get:
where the deviation due to the constellation 22 ^ -QAM rotated after A'J /) and A'2 was considered. Let us note the probability of error of the symbol for level 1 as a normalized SNR function as P ° l (SNRn) and the average probability of bit error as P ° bl (SNRn). The symbol's probability of error is then calculated as

In order to obtain the bit error probability, it must be evaluated how many links of minimum distance between points produce only a 1 bit change in the mapping and how many produce a change of 2 bits. For even QAM constellations, all links have a 1-bit change, as they are mapped by Gray. For odd QAM constellations, it is more difficult, but it can still be calculated analytically - in a closed way. When 2-n6 (!) Is even:
and when 2 - ^ (7) is odd, the average bit error probability is estimated by:
Furthermore, for the coded levels (first and possibly second level described with reference to Figure 6), the probability of BCH decoding error can be estimated from P ° bl (SNRn), because for constellations of both QPSK and type of the BPSK type applied at these levels, only one bit fails due to a failed symbol decision under a low error rate regime. For the 1 st level, a BCH code (rtc (/), £ c (/), re (/)) over Galois field GF (2m) is defined, where nc (l) is the number of bits per codeword nc (l) £ 2 ™ -1 and is even, kc (l) is the number of bits of information per codeword and tc (l) is the error correction capability in bits per codeword. BCH codes are designed separately for the first and second level, meeting different criteria. For sufficiently large BCH codes, the next equation estimates exactly the probability of decoding bit error P'eb t (Pe ° bz) as a function of the input bit error probability:
This equation applies to the first and second level BCH codes. The third level is uncoded and thus P ^ bJ = P BJ. An entire codeword is assumed to be erroneous when it cannot be corrected and the number of failed bits varies from tc (l) + l to nc (l). Using the estimates described above, it is now possible to estimate the bit error probability for the entire MSD decoded MLCC scheme at the receiver, as a function of P ^ bl at all enabled MLCC levels, Lmax denoting the index at level 1 higher than it is enabled. The bit error probability of the MLCC is estimated as
where rc (l) is the code-rate for each level, where rc (3) = l, as the 3rd. level is uncoded.
Now, according to an embodiment of the present invention, based on the estimated bit error probability derived above, the BCH codes are selected to ensure that the bit error rate (BER) of the entire MLCC-MSD system is less than a target BER denoted as BER0. The target BER is a system requirement depending on the target application. For example, for a lGbps Ethernet, BERQ = IO'10, and for 10 Gbps, BERo = 10'12. The method for selecting BCH codes includes the following steps: 1. Select m, NMLCc θ nc (l). For any level 1, β (l) = kc (l) is optimal. As nt> (l) = 1 bit / dim for any MLCC configuration according to Figure 6, NMLCC = nc (l) one-dimensional symbols per MLCC code word. On the other hand, for a BCH over GF (2m), choosing nc (l) as close as possible to 2m-l will maximize performance. The m is selected based on system requirements; the gain of BCH encoding increases with m. In addition, NMLCc must be even because the MLCC works in a two-dimensional lattice, and when rotated constellations are involved an odd number of bits will be mapped by 2 dimensions. Therefore: 1.1 m is selected as a function of the system requirements, taking into account that with a larger m, the coding gain increases as well as the complexity of decoding and latency. 1.2 nc (l) is an integer such that nc (l) <2m - 1. In order to maximize the coding gain, nc (l) must be selected as high as possible. 2. Selection of rc (l). The next step is the selection of the code rate for the BCH used in the first level. From a performance point of view, for a given m and nc (l), there is an optimal value rc (l) for which the Shannon gap is minimal. BCH codes perform very well for high code rates. When the code-rate is decreased from 1 to 0, the Shannon gap is decreased, so the coding gain increases up to a given value rc (l) Opt- From this value the gap increases monotonically. For the selection of rc (l), hardware implementation restrictions must also be considered. The computational complexity, hence the area of silicon, of the BCH BMA decoder is proportional to the factor m • t. 3. Calculation of kc (l) and the primitive polynomial. The BCH code (z C (1), (1), íc (l)) is a shortened version of a BCH code (2ra-l, kc0 (l), tc0 (l)), where kc0 (l ) = kc (l) + 2m - 1 - nc (l). From this first tenth value, we look for the valid primitive polynomial with k kc0 (l). From this, the BCH is strictly defined by nc (l) and kc (l) = k + nc (l) - 2a + 1 4. BCH selection for the second level: two different BCH codes are used at the second level, however they are a scaled-down version of a single primitive polynomial, thus, advantageously, only one BCH decoder needs to be implemented for this level. When nb (2) = 1 bit / dim, then nc (2) = nc (l). For nb (2) = 0.5 bits / dim, nc (2) = nc (l) / 2. If kc0 (2) is the number of information bits in the
As can be seen, MLCC BER is conditioned by the second level to BER <10 "7.
When the configuration is changed and tc (2) = 1 bit, the performance changes as shown in Figure 24. As can be seen, the performance of the entire MLCC is now closer to the Shannon limit. Figure 25 illustrates the performance for tc (2) = 2 bits per BCH code word. This setting is also valid for nb (2) = 0.5 bits / dim as shown in Figure 26. Figure 27 shows the performance for r] ~ 3.3 bits / s / Hz / dim and Ç = 3.5 bits / dim, with nb (l) = nb (2) = 1 bit / dim and nb (3) = 1.5 bits / dim. As already discussed above, the performance of the third level has no influence on performance when BCH codes are well selected for deployable BERo.
As already mentioned above, the power balance in the connection is an important measure of the quality of an optical communication system. It measures the maximum attenuation at which the communication system can still guarantee a given data rate with a bit error rate lower than a specified target (under defined conditions of noise, distortion, temperature, etc.). The objective is to select and develop the most appropriate telecommunications techniques that lead to the maximization of the power balance in the connection. This problem is related to the objective of addressing the capacity of the communication channel for which the system is designed. Therefore, criteria and limits of information theory can be used to estimate the transmission rate and number of optimal M-PAM levels that maximize the link balance as discussed above.
The following example considers that THP is implemented as an equalization technique with "the MLCC scheme of the present invention. The goal is to provide 1 Gbps using an LED with a 20 nm wavelength FWHM,
Numerical opening FWHM of 0.3 (with launch condition similar to END) and OMA of OdBm. Several -3dB analog LED electrical bandwidths are considered to show how the link balance depends on this. The link balance is for 50 meters of SI-POF (A4a.2) and the POF attenuation is included within the link balance. In other words, the POF response is considered normalized by direct current ("DC"), all modal dispersion and the consequent ISI being included in the evaluation of the power balance in the connection. A commercially available integrated large area optical receiver is considered. The responsiveness is 0.5 A / W at 650 nm wavelength and the trans-impedance is adjusted to optimize the noise bandwidth balance as a function of the transmission rate and the number of modulation levels.
Figure 28 shows the link balance (dBo) as a function of the signal transmission rate (Fs). As can be seen, the transmission rate that maximizes the link balance is around 300 MBaud, for LED bandwidths between 50 and 100 MHz, and something more than 350 MBaud for 350 MHz. signal band around 150 MHz and 16-PAM modulation, providing a spectral efficiency of -3.3 bits / symbol (6.6 bits / s / Hz) is the setting to maximize the link balance. When the power balance on the link is depleted because there is excess attenuation due to bad coupling, bad end, intermediate connectors or too many bends, there are two possibilities: not establishing the communication link or reducing the data rate (for example 8 00 Mbps from 1 Gbps) using automatic algorithms implemented in the system. From the end user's point of view, the adaptive bit rate solution can be the key to success, especially for amateur installations (easy-to-install requirement), which are likely to be prevalent in home network installations. On the other hand, keeping in mind cabling topologies such as daisy-chain or tree, which reduce the complexity of installation within the home, the adaptive bit rate capability can be advantageous in common sections of the network (sections that must handle the traffic from several other sections), where more than 1 Gbps may be required and provided due to possible excess margin. The present invention employs an MLCC scheme that supports programmable spectral efficiency for adaptive bit rate.
Figure 29 summarizes a method according to the present invention. Correspondingly, bits are inserted2810 into the three-level side class encoder and separated into its three levels. The bits in the first and second level are encoded 2820 by the BCH code. The encoded bits of the first two levels as well as bits on the third level are then mapped into respective predefined constellations 2830 and the resulting symbols are then transformed 2840 by means of a lattice transformation, thereby performing side-class partitioning. The symbols transformed in the lattice of each level are added and processed by a lattice transformation of the second step 2850 in order to obtain a zero-mean constellation with possibly uniform energy distribution. After three-level side-class coding, the symbols are modulated 2860 and either provided for transmission, or pre-coded 2870.
Figure 30 shows an example of a multi-stage decoder 3000 that can be used to decode encoded signal as described above. First, the encoded digital signal (M-PAM) having an NMLCc length is demodulated 3010 to obtain a two-dimensional symbol code word with an NMLCC / 2 length. The demodulated symbol is then transformed 3020 by an inverse lattice transformation (combined with the second encoder lattice transformation 660) and decoded with a three-stage decoder. The three-stage decoder applies in the first stage another reverse lattice transformation 3030a (combining lattice transformation 640a in the encoder) and difficult decision 3040a, then extracts a first portion of the transformed symbol code word by applying a 3050a module operation ( mod-Ai) to the demodulated symbol. Then, still in the first stage, the first extracted portion is demapped 3060a (combined with mapping 630a in the encoder) to obtain BCH code word with nc (l) bit length, which is decoded with a first BCH decoder 3070a (combined to the BCH 620a encoder in the encoder) to obtain β (l) bits of information. The code word decoded from the first portion with nc (l) bits selects a first side class. The decoded code word of the first portion is then mapped back in a similar way to 630a and transformed by a lattice transformation corresponding to 640a. The resulting first-level side class is then subtracted from the symbol's code word and the second portion is obtained in the second stage by applying module operation (mod-A2) to this after a reverse lattice transformation and difficult decision, which are similar to functional blocks described for the first stage. The second portion is then demapped (combined with mapping 630b in the encoder) to obtain BCH code word with nc (2) bit length, which is decoded with a second or third BCH decoder (corresponding to BCH encoder 620b in encoder) to obtain β (2) bits of information. The decoded code word of the second portion selects a second side class. A third í f> portion with a size of nc (3) = β (3) bits is obtained by subtracting the decoded code word from the second side class (mapped 630b and transformed reticle 640b again) and the side class from the first level from the demodulated symbol and by reverse lattice transformation (corresponding to the lattice transformation 640c in the encoder), difficult decision 3040c and module operation (mod-A3) 3050C. After a 3060c mapping (corresponding to the 630c mapping in the encoder), based on the third portion obtained, the code word is finally decoded by multiplexing 3080 (MLCC mux) the three decoded parts obtained in three decoder stages to obtain the decoding word of aMLcc- length
Figures 31A, 31B, and 31C show possible methods of equalization that can be used with the present invention. Advantageously, as described above, Tomlinson-Harashima Pre-coding is used, including 3110 module operation and 3130 feedback filtration, similar to the pre-coding known as shown in Figure 3 (cf. module 310 operation and 330 feedback filter) . However, on the decoder side, the module operator 320 of Figure 3 is advantageously excluded in order to reduce module loss, and in view of the module operation being carried out during multi-stage decoding. In this way, before entering the multi-stage decoder (for example the decoder in Figure 30), only forward feed filtration 3120 is performed.
Alternatively, when Tomlinson-Harashima precoding is not applied, Forward Feed Equalization can be performed as shown in Figure 31B. Alternatively, a Feedback Feedback Equalizer can be used as shown in Figure 31C.
Figure 34 summarizes steps of the method as performed by the higher bandwidths required by video-based sensors are achievable with POF and the technology described above, allowing the installation of high definition rear view cameras, 360 ° view with multi-cameras, parking assistance, side mirror replacement, rear seat monitoring, night vision, etc. at affordable cost.
Another embodiment of the invention relates to the implementation of the various embodiments described above using hardware and software. It is recognized that the various embodiments of the invention can be implemented or performed using computing devices (processors). A computing device or processor can, for example, be general purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), Field Programmable Port Array ( Field-programmable gate array (FPGA) or other programmable logic devices, etc. The various embodiments of the invention can also be carried out or carried out by a combination of these devices.
In addition, the various embodiments of the invention can also be implemented through software modules, which are executed by a processor or directly on hardware. A combination of software modules and hardware implementation may also be possible. The software modules can be stored on any type of computer-readable storage media, for example RAM, EPROM, EEPROM, flash memory, recorders, hard drives, CD-ROM, DVD, etc.
In summary, the present invention relates to an efficient coding and modulation system for transmitting digital data over plastic optical fibers. In particular, the digital signal is encoded by means of a three-level side class encoding. The spectral efficiency of the system is configurable by selecting the number of bits to be processed at each of the levels. The first level applies a binary BCH encoding to digital data and performs side class partitioning through constellation mapping and lattice transformations. Similarly, the second level applies another binary BCH coding, which can be performed selectively according to the desired configuration by two BCH codes with substantially the same coding rate, operating on code words of different sizes. The third level is uncoded. Both the second and third levels undergo mapping and lattice transformation. After adding the levels, a second stage lattice transformation is performed in order to obtain a zero mean constellation. The symbols coming out of such a three-level side class encoder are then further modulated.

权利要求:
Claims (15)
[0001]
1. METHOD FOR ENCODING DIGITAL DATA FOR TRANSMISSION THROUGH A PLASTIC OPTICAL FIBER, the method being characterized by understanding the steps of: selecting the number of bits in one second and a third portion of the digital input data as a function of spectral efficiency desired; encoding digital input data by a three-level side class encoding including: separating (2910) from the input digital data a first portion, the second portion and the third portion of data, each with a predetermined number of bits; encode (2920) the first piece of data with a first BCH code at a first level; provide on a second level a second BCH code and a third BCH code, where the third BCH code has a code word length less than the code word length of the first and second BCH code, and the third BCH code has the same error correction capability as the second BCH code; encode (2920) the second portion with either the second or third BCH code at the second level; on the first level, map (2930) the first portion encoded in symbols from a first GPSK constellation and perform a lattice transformation (2940) of the mapped symbols to achieve lateral class partitioning, the lattice transformation including translation and scale; on the second level, map (2930) the second portion encoded in symbols from a second predefined constellation using QPSK or BPSK mapping depending on whether the second or third BCH code was used and perform a lattice transformation (2940) of the mapped symbols to achieve lateral class partitioning, the transformation of the lattice including translation, 45 degree rotation and scale of the translated constellation when the number of bits per two dimensions of the second constellation is singular; on the third level, map (2930) the third portion into symbols of a third constellation on Z2 or RZ2 lattices, configured according to the desired spectral efficiency and perform a lattice transformation (2940) of the mapped symbols, the lattice transformation including translation , constellation rotation translated in 45 degrees when the number of bits per two dimensions of the configured constellation is singular, and scale; add (2950) the transformed symbols from the first, second and third level resulting in a combined constellation over Z2 or RZ2; perform (2950) a second stage lattice transformation of the combined constellation to achieve a zero mean constellation, the second stage of the lattice transformation including 45 degree rotation of the constellation when the number of bits per two dimensions of the combined constellation is singular and a module operation to restrict the symbols of the combined constellation to a square constellation over Z2 or RZ2; emodular (2960) symbols encoded with three-level side class encoding using a time domain modulation.
[0002]
2. METHOD according to claim 1, characterized in that the second and third binary BCH codes have the same primitive polynomial.
[0003]
3. METHOD, according to claim 1, characterized in that it additionally comprises a step of, based on the desired spectral efficiency, selecting the number of bits by two dimensions of each level and adapting the lattice transformations by applying the said constellation rotation when the number of bits per two dimensions of the combined constellation is singular and does not apply rotation, on the other hand.
[0004]
4. METHOD according to claim 1, characterized in that the first BCH encoder generates code words with 2044 bits based on 1637 bits of input information, and / or the second BCH encoder generates code words with 2044 bits based on 2022 bits of input information, and / or the third BCH encoder generates 1022-bit code words based on 1000 bits of input information.
[0005]
5. METHOD, according to claim 1, characterized in that the time domain modulation is M-PAM and the first stage of transformation of the A (l) lattice is given by:
[0006]
6. METHOD according to any of claims 1 to 5, characterized in that it additionally comprises a Tomlinson-Harashima pre-coding step applied to the modulated symbols.
[0007]
7. METHOD FOR RECEIVING AND DECODING A CODED DIGITAL SIGNAL, by an encoding method as defined in claim 1 and received by a plastic optical fiber, the method being characterized by comprising the steps of: demodulating (3420) the digital signal encoded with a modulation time domain to obtain symbol code words and transform them (3430) with an inverse lattice transformation, decode the demodulated and transformed symbols with a three-stage decoder including the steps of: extracting (3440) a first portion of a code word applying inverse lattice transformation, difficult QPSK decision, and then a module operation for a demodulated symbol; first decode (3440) the first portion with a first BCH decoder and based on the first decoded portion select a first side class; extract (3450) a second portion by applying reverse lattice transformation, difficult decision QPSK or BPSK and then a modulus operation for a symbol obtained by subtracting the first decoded portion of the demodulated symbol; decode (3450) in a second stage the second portion with a second or a third BCH decoder and based on the second decoded portion select a second side class; obtain (3460) a third portion by subtracting the first and second decoded side classes from the demodulated symbol and applying reverse lattice transformation, difficult decision Z2 or RZ2 and then a module operation; and multiplexing (3470) the first, second and third decoded portion, where in the second stage the second BCH code and the third BCH code are provided, where the third BCH code has a code word length less than the code word length of the first and second BCH code, and the third BCH code has the same error correction capability as the second BCH code.
[0008]
8. APPLIANCE FOR ENCODING DIGITAL DATA FOR TRANSMISSION THROUGH A PLASTIC OPTICAL FIBER, the apparatus being characterized by comprising: a multi-level side class encoder (510) for encoding digital input data by a three-class side encoding levels including: a demultiplexer (610) for separating from the digital input data a first portion, a second portion and a third portion of data, each with a predetermined number of bits selected as a function of the desired spectral efficiency; a first BCH encoder (620a) encoding the first portion of data with a first BCH code at a first level; a second BCH encoder (620b) at a second level, wherein the second BCH encoder is adapted to encode the second portion with either a second or a third BCH code provided both in the second BCH encoder, wherein the third code BCH code length is less than the code word length of the first and second BCH code, and the third BCH code has the same error correction capability as the second BCH code; a first mapper (630a) at the first level to map the first symbol-encoded portion of a first predefined QPSK constellation and perform a lattice transformation (640a) of the mapped symbols to achieve side class partitioning, the lattice transformation including translation and scale; a second mapper (630b) at the second level to map the second portion encoded in symbols of a second predefined constellation using QPSK or BPSK mapping depending on whether the second or third BCH code was used and perform a lattice transformation (640b) the symbols mapped to achieve side class partitioning, the transformation of the lattice including translation, 45 degree rotation and scale of the translated constellation when the number of bits per two dimensions of the second constellation is unique; a third mapper (630c) at the third level to map the third portion in symbols of a third predefined constellation over lattices Z2 or RZ2, configurable according to the desired spectral efficiency and perform a lattice transformation (640c) of the mapped symbols, the transformation of the lattice including translation, rotation of the constellation translated by 45 degrees when the number of bits per two dimensions of the configured constellation is singular, and scale; an adder (650) to add the transformed symbols from the first, second and third level resulting in a combined constellation over Z2 or RZ2, and a transformation unit (660) to perform a lattice transformation of the combined constellation in a way to achieve a zero mean constellation, the second stage of the reticle transformation including 45 degree rotation of the constellation when the number of bits per two dimensions of the combined constellation is singular and a module operation to restrict the symbols of the combined constellation to a square constellation over Z2 or RZ2; and a modulator (670) for modulating the symbols encoded with the three-level side class encoding using a time domain modulation.
[0009]
Apparatus according to claim 8, characterized in that the second and third binary BCH codes (620b) have the same primitive polynomial.
[0010]
10. APPARATUS, according to claim 8, characterized in that it additionally comprises a selector for, based on the desired spectral efficiency, selecting the number of bits by two dimensions of each step and adapting the transformations of the lattice through the application of said rotation of the constellation when the number of bits per two dimensions of the combined constellation is singular and does not apply rotation, otherwise.
[0011]
11. Apparatus according to claim 8, characterized in that the first BCH encoder generates code words with 2044 bits based on 1637 bits of input information, and / or the second BCH encoder generates code words with 2044 bits based on 2022 bits of input information, and the third BCH encoder generates 1022-bit code words based on 1000 bits of input information.
[0012]
Apparatus according to claim 8, characterized in that the first (630a), the second (630b), and / or the third (630c) mapper is configured to perform the first stage of the transformation of the At1 lattice (l) given by
[0013]
13. Apparatus according to any one of claims 8 to 12, characterized in that it additionally comprises a Tomlinson-Harashima precoder (530) for precoding the modulated symbols.
[0014]
14. APPARATUS FOR RECEIVING AND DECODING A CODED DIGITAL SIGNAL, with an apparatus as defined in claim 8 and received by a plastic optical fiber, the apparatus being characterized by comprising: a demodulator (3010) to demodulate the encoded digital signal with a modulation of time domain to obtain symbol code words, a transformation unit to transform (3020) demodulated symbols with an inverse lattice transformation; a multi-stage decoder to decode the demodulated and transformed symbols, the multi-stage decoder having three stages and additionally comprising: a first extractor to extract a first portion of a code word applying a difficult QPSK decision, and then a transformation of inverse lattice (3030a) and a module operation (3050a) for a demodulated symbol; a first BCH decoder (3070a) to first decode the first portion and based on the first decoded portion select a first side class; a second extractor to extract a second portion by applying an inverse lattice transformation (3030b), difficult decision QPSK or BPSK and then a modulus operation (3050b) to a symbol obtained by subtracting the first decoded portion from the demodulated symbol; a second and a third BCH decoder in a second stage applying a respective second BCH code and third BCH code, wherein the third BCH code has a shorter codeword length than the first codeword length and the second BCH code, and the third BCH code has the same error correction capability as the second BCH code; a decoder (3070b) for decoding in a second stage the second portion with the second or third BCH decoder and based on the second decoded portion selecting a second side class; a third extractor to obtain a third portion by subtracting the first and second decoded side classes from the demodulated symbol, applying an inverse lattice transformation (3030c), difficult decision Z2 or RZ2 and then a module operation (3050c); and a multiplexer (3080) to multiplex the first, second and third decoded portions.
[0015]
15. INTEGRATED CIRCUIT, characterized by incorporating the device as defined in claim 8.

类似技术:

---

公开号 | 公开日 | 专利标题

---

BR102012005220B1|2020-12-22|method for encoding digital data for transmission over a plastic optical fiber, method for receiving and decoding a digital signal encoded by an encoding method, apparatus for receiving and decoding a digital signal encoded with an apparatus and integrated circuit

---

US9094151B2|2015-07-28|Frame structure for adaptive data communications over a plastic optical fibre

---

KR101775466B1|2017-09-19|Two-level coset coding scheme for gigabit Ethernet over plastic optical fibre

---

US8199803B2|2012-06-12|Receiver structure and method for the demodulation of a quadrature-modulated signal

---

US7447278B2|2008-11-04|Apparatus for transmitting and receiving data

---

JP6127848B2|2017-05-17|Equalizer coefficient updating apparatus, method, receiver, and optical communication system

---

TWI311876B|2009-07-01|Prescribed response precoding for channels with intersymbol interfrrence

---

CN109328451B|2021-06-04|System and method for precoding faster-than-nyquist signaling

---

Prodaniuc et al.2016|Performance comparison between 4D trellis coded modulation and PAM-4 for low-cost 400 Gbps WDM optical networks

---

JP4927617B2|2012-05-09|Data transmission apparatus and transmission code generation method

---

US9143306B2|2015-09-22|Device and method for encoding bits to symbols for a communication system

---

EP2107702A1|2009-10-07|Device and method for transmitting optical data signals

---

JP2779973B2|1998-07-23|Trellis coding for modulation scheme

---

JP6259363B2|2018-01-10|Station-side optical line terminator and signal generation method

---

Zeolla et al.2018|Comparison of Different Modulation Formats for 1 Gigabit/s SI-POF Transmission Systems

---

WO2015192970A1|2015-12-23|Precoding method and device for partial response modulation

---

Fischer et al.1995|amics Limited PECQ

同族专利:

---

公开号 | 公开日

---

EP2498432B1|2014-06-04|

---

KR101898013B1|2018-09-21|

---

US8634450B2|2014-01-21|

---

JP2012217151A|2012-11-08|

---

JP5952598B2|2016-07-13|

---

CN102684830B|2017-03-01|

---

ES2489740T3|2014-09-02|

---

EP2498432A1|2012-09-12|

---

US20120250785A1|2012-10-04|

---

KR20120104129A|2012-09-20|

---

CN102684830A|2012-09-19|

---

BR102012005220A2|2016-10-04|

引用文献:

---

公开号 | 申请日 | 公开日 | 申请人 | 专利标题

---

---

US6523147B1|1999-11-11|2003-02-18|Ibiquity Digital Corporation|Method and apparatus for forward error correction coding for an AM in-band on-channel digital audio broadcasting system|

---

US7065147B2|2000-07-12|2006-06-20|Texas Instruments Incorporated|System and method of data communication using turbo trellis coded modulation combined with constellation shaping with or without precoding|

---

US7484165B2|2003-04-30|2009-01-27|Ericsson Ab|Forward error correction coding|

---

JP4622276B2|2004-03-18|2011-02-02|日本電気株式会社|Coded modulation apparatus and method|

---

RU2422992C2|2004-12-17|2011-06-27|Юроупиан Спейс Эйдженси|Method to develop set of extending codes of satellite navigation system|

---

ES2601556T3|2006-09-26|2017-02-15|Koninklijke Philips N.V.|PAPR reduction using a generalized constellation rotation|US9219550B2|2011-06-23|2015-12-22|Infinera Corporation|Forward carrier recovery using forward error correctionfeedback|

---

US9698939B2|2013-06-13|2017-07-04|Ciena Corporation|Variable spectral efficiency optical modulation schemes|

---

US9455799B2|2013-08-06|2016-09-27|OptCTS, Inc.|Dynamic control of quality of serviceusing derived QOS measures|

---

US10523490B2|2013-08-06|2019-12-31|Agilepq, Inc.|Authentication of a subscribed code table user utilizing optimized code table signaling|

---

US9444580B2|2013-08-06|2016-09-13|OptCTS, Inc.|Optimized data transfer utilizing optimized code table signaling|

---

US9374167B2|2013-09-20|2016-06-21|Alcatel Lucent|Level spacing for M-PAM optical systems with coherent detection|

---

US9246510B2|2013-10-01|2016-01-26|Texas Instruments Incorporated|Apparatus and method for multilevel coding in communication systems|

---

US9154156B2|2014-02-25|2015-10-06|Qualcomm Incorporated|Ternary line code design for controlled decision feedback equalizer error propagation|

---

EP3164942A1|2014-07-02|2017-05-10|Agilepq, Inc.|Data recovery utilizing optimized code table signaling|

---

US10693560B2|2014-10-10|2020-06-23|Nec Corporation|Optical transmitter, optical communication system, and optical communication method|

---

JP6763302B2|2014-10-10|2020-09-30|日本電気株式会社|Optical transmitters, optical communication systems, and optical communication methods|

---

US9912414B2|2014-11-14|2018-03-06|Zte Corporation|Iterative post-equalization for coherent optical receivers|

---

EP3046276B1|2015-01-14|2018-03-14|Knowledge Development for POF, S.L.|Two-level coset coding scheme for gigabit Ethernet over plastic optical fibre|

---

US20170052762A1|2015-08-20|2017-02-23|Futurewei Technologies, Inc.|System and method for representing complex numbers in fused floating point|

---

EP3151458B1|2015-10-02|2019-03-20|Mitsubishi Electric R&D Centre Europe B.V.|A method for determining features of an error correcting code system|

---

US10587399B2|2016-06-06|2020-03-10|Agilepq, Inc.|Data conversion systems and methods|

---

CN110521152B|2018-02-14|2021-05-18|华为技术有限公司|Encoding and decoding method, encoding and decoding device and encoding and decoding equipment|

---

CN110380999B|2018-04-12|2020-10-09|华为技术有限公司|Probability non-uniform modulation data transmission method and device|

---

US11196594B2|2019-02-28|2021-12-07|Nokia Solutions And Networks Oy|Probabilistic signal shaping using multiple codebooks|

---

CN110022278B|2019-03-06|2020-08-04|华中科技大学|Coding method and coder of , n-1) -PGC in communication modulation system|

法律状态:
2016-10-04| B03A| Publication of a patent application or of a certificate of addition of invention [chapter 3.1 patent gazette]|

---

2017-03-01| B25G| Requested change of headquarter approved|Owner name: KNOWLEDGE DEVELOPMENT FOR POF, S.L. (ES) |

---

2018-12-18| B06F| Objections, documents and/or translations needed after an examination request according [chapter 6.6 patent gazette]|

---

2019-10-15| B06U| Preliminary requirement: requests with searches performed by other patent offices: procedure suspended [chapter 6.21 patent gazette]|

---

2020-06-23| B06A| Patent application procedure suspended [chapter 6.1 patent gazette]|

---

2020-10-06| B09A| Decision: intention to grant [chapter 9.1 patent gazette]|

---

2020-12-15| B09X| Republication of the decision to grant [chapter 9.1.3 patent gazette]|

---

2020-12-22| B16A| Patent or certificate of addition of invention granted [chapter 16.1 patent gazette]|Free format text: PRAZO DE VALIDADE: 20 (VINTE) ANOS CONTADOS A PARTIR DE 08/03/2012, OBSERVADAS AS CONDICOES LEGAIS. |

优先权:

---

申请号 | 申请日 | 专利标题

---

EP11002046.8|2011-03-11|

---

EP11002046.8A|EP2498432B1|2011-03-11|2011-03-11|Adaptative error correcting code for data communications over a plastic optical fibre|

---