SE506643C2 - Multi-carrier wave transmission system e.g. for VDSL system - Google Patents

Abstract

The system includes a digital receiver unit and transmission unit. The receiver unit including a symbol detection unit. The system determines a parameter for each individual carrier wave. The parameter indicate a deviation in a received signal from a corresponding constellation point. The parameter is compared with an upper and a lower limit. When the parameter is outside the limits the used constellation is altered in order to modulate the carrier wave to a neighbouring constellation. The symbol detection unit is used to determine the parameter for a carrier wave. The parameter is a ratio d2/o2, where d is the shortest distance between adjoining constellations, o is a standard deviation, and o2 is the variance in the deviations in the input and output signal values in the symbol detection unit.

Description

506 645 data rates are transmitted in one direction only. Such solutions meet many of the requirements for high bandwidth services, such as video-on-demand, but in the long run symmetrical duplex systems will be required.

VDSL technology is very similar to ADSL, although ADSL has to provide much greater dynamic ranges and as a result is much more complex. VDSL is lower in cost and lower in energy (lower in power), and VDSL units in real estate (premises) need to implement an access control for media in the physical layer for multiplexing upstream data.

Four license codes have been proposed for VDSL: - CAP; AM / PM, a version of: AM with suppressed carrier, for passive NT- “Carrierless” configurations, CAP would use QPSK upstream and a type of TDMA for multiplexing (although CAP does not exclude a solution with FDM for upstream multiplexing); - DMT; Discrete Multi-Tone, a multi-carrier system that uses discrete Fourier Transforms to create and demodulate individual carriers, for passive NT configurations; DMT would use FDM for upstream multiplexing (although DMT does not exclude a TDMA multiplexing strategy); - DWXT; Discrete Wavelet multitone (Discrete Wavelet Multi-Tone), a multi-carrier system used for "wavelet transforms" to create and demodulate individual carriers; Un u: Un DWMT also uses FDM for upstream multiplexing, but also allows TDMA; and SLC; Simple Line Code, a four-level baseband signaling that filters the baseband and resets it at the receiver, for passive NT configurations; it is most likely that SLC will use TDMA for upstream multiplexing, although FDM is Possible.

Early versions of VDSL will use frequency division multiplexing to separate downstream from upstream channels, and both of these from POTS and ISDN. Echo quenching may be required for later generations of systems with symmetric data rates. A fairly large distance, in frequency, will be maintained between the lowest data channel and POTS to enable very simple and cost-effective POTS line splitters. Normal use would place the downstream duct above the upstream duct. However, the DAVIC specification reverses this scheme to enable the distribution of VDSL signals over coaxial cable systems in buildings.

Modern multi-carrier technology that uses orthogonal carriers with high-order QAM constellations for the transmission of a plurality of bits per carrier and symbol, uses some method to determine an appropriate constellation (bit charge) for each separate carrier. This is usually based on a value calculation of the value of the signal-to-noise ratio SNR (Signal-to-noise ratio) for each carrier and the maximum permissible symbol error rate. 50 30 ya LJ! An object of the present invention is to provide in a multi-carrier transmission system a simplified method of determining bit charge which utilizes the fact that all signals are normalized to the same scale when input and output in a symbol detector.

Another object of the present invention is to provide a multi-carrier transmission system in which a suitable constellation (bit charge) is determined for each carrier, the determination of the bit charge utilizing the fact that all signals are normalized to the same scale when input and output in a symbol detector.

V.L. According to the first aspect of the present invention: there is provided a multi-carrier transmission system using orthogonal carriers having: high order AM constellations for transmitting a plurality of bits per carrier and symbol, Q ~. wherein said system includes digital receiver and transmitter units, said receiver unit includes a symbol detection unit, characterized in that said system is arranged to determine a parameter for each individual carrier, said parameter indicating a deviation of a received signal from a corresponding constellation point; to compare the parameter with an upper and lower limit; and, if the parameter is outside said limits, changing the constellation used to modulate the carrier to an adjacent drilling constellation.

The symbol detection unit can be used to determine the parameter of a carrier.

Said parameter may be a ratio dz / G2, where d is the shortest distance between adjacent constellations, G is a standard deviation, and oz is the variance of the deviation of the input and output signal values in said symbol detection unit.

The symbol detection unit can be adapted to associate an input signal applied thereto, in an area around a specific constellation point, to a certain value for this point, where said distinct value is the value from said symbol detection unit, and to measure differences between said input and output signals. for the carrier, where said differences are squared and averaged to provide a value calculation of a value of Gz represented in the same units as the squared distance dz of the constellation used to modulate the carrier.

The differences between the input and output signal values in said symbol detection unit can vary in value, where the variance is a measure of the disturbance on the carrier and directly related to SNR. The variance may be proportional to SNR.

An upper limit of said variance for a specific constellation can be determined with a maximum allowable symbol error rate, and a lower limit for said variance can be equal to an upper limit for the second largest constellation.

A specific maximum symbol error rate can give a minimum ratio of d / 6.

According to a second aspect of the present invention, there is provided a multi-carrier transmission system using orthogonal carriers with high order QAM constellations for transmitting a plurality of bits per carrier and symbol, said system including digital receiver and transmitter units, said receiver unit including a symbol detection unit, a method of determining a constellation (bit charge) for each individual carrier, characterized by the steps of determining a parameter for each individual carrier, said parameter indicating a deviation of a received signal from a corresponding constellation point; comparison of the parameter with an upper and lower limit; and, if the parameter is outside said limits, changing the constellation of the carrier to a neighboring constellation.

In a preferred method, the determination of the parameter for each carrier is performed by said symbol detection unit, and the parameter is a ratio d2 / 62, where d is the shortest distance between adjacent constellations, 6 is a standard deviation, and the variance of the deviations of the input and output values in said symbol detection unit.

According to a preferred method, determining the parameter of a carrier path includes the steps of associating an input signal to said symbol detection unit, in an area around a specific constellation point, to a certain value for that point, said distinct value being the value of the output of said symbol detection unit. ; and measuring differences between said inputs and outputs of said symbol detection unit for the carrier, wherein said differences are squared and averaged to give a value calculation of the value of Gz represented in the same units as the squared distance dz of the constellation used to modulate the carrier.

The differences between the input and output signal values of said symbol detection unit may vary in value, where said variance is a measure of the disturbance on the carrier and directly related to SNR. The variance may be proportional to SNR.

According to a preferred method, an upper limit of said variance is determined for a specific constellation of a maximum allowable symbol error rate, and of that a lower limit of 10 kx) LI: 30 01 ~ 3 <3 \ said variance is equal to an upper limit of the the second largest constellation. A specific maximum symbol error rate can give a minimum ratio of d / G.

According to a third aspect of the present invention, there is provided a multi-carrier transmission system using orthogonal carriers with high order QAM constellations for transmitting a plurality of bits per carrier and symbol, said system including digital receiver and transmitter units, said receiver unit including a symball unit , characterized in that said system uses a method outlined in the preceding paragraph to determine a constellation for each individual carrier.

The multi-carrier transmission system can be a DMT system, for example a DMT-based VDSL system.

The foregoing and other features of the present invention will be better understood from the following description with reference to the accompanying figures, in which: Figure 1 shows, in schematic form, an asymmetric communication system.

Figure 2 shows, in schematic form, a DMT system.

Figure 3 shows, graphically, the channel separations used in an asymmetric DMT transmission system.

Figure 4 shows, in schematic form, the cornerstones of a multitone carrier system modem contemplated in the present invention.

Figure 5 shows, in schematic form, a partitioning of the multitone carrier system modem which 506 20 k) Un 30 645 is shown in Figure 4, and which is used to facilitate implementation.

Figure 6 shows, in graphical form, spectral allocation for copper pairs.

Figure 7 shows, in schematic form, the frame structure used in the multitone carrier system described here.

Figure 8 shows, in schematic form, the analog interface of the multitone carrier system modem shown in Figure 4.

Figure 9 shows, in graphical form, the dependence of the signal-to-noise ratio (SNR ratio) on frequency in the multitone carrier system described here.

Figure 16 shows, in schematic form, the FFT algorithm used in the multitone carrier system modem shown in Figure 4.

Figure 11 shows, in schematic form, the frame correlation principle used in the multitone carrier system modem shown in Figure 4.

Figure 12 shows, in schematic form, implementation of a correlator used in the multitone carrier system modem shown in Figure 4.

Figure 13 shows, in schematic form, the averager used in the correlator in Figure 12.

Figure 14 shows, in schematic form, a correlation position detector used for the multitone carrier system modem shown in Figure 4.

Figure 15 shows, in schematic form, an overview of the synchronization unit used in the multitone carrier system modem shown in Figure 4. u: (fl CD O \ (Y \ I> CN Figure 16 shows, in schematic form, an overview of the FFT / IFFT device used in the multitone carrier system modem shown in Figure 4.

Figure 17 shows, in schematic form, the use of a cyclic prefix.

Figure 18 shows, in schematic form, a “decision directed” channel value calculation and smoothing system for use in the multitone carrier system modem shown in Figure 4.

Figure 19 shows QAM coding for b = 6.

Figure 23 shows, in schematic form, the realization of the calculation of bit charge and energy charge factors used in the multitone carrier system modem shown in Figure 4.

Figure 2; shows, in schematic form, an overview of the systems: system controller interface used in the multitone carrier system modem shown in Figure 4.

Figure 22 shows, in schematic form, the manner in which two of the multitone carrier system modems shown in Figure 4 are interconnected to create a multitone carrier transmission system.

Figure 23 shows, in schematic form, the vector management system used in the multitone carrier system modem shown in Figure 4.

Figure 24 shows BSI length.

Figure 25 shows, in schematic form, NU SC load distribution for BSI- 50 Un 10 20 h) Un 30 643 10 interruptions for the multitone carrier system modem shown in Figure 4.

Figure 26 shows the SUS pattern of the multitone carrier modem shown in Figure 4.

Figure 27 shows the DAS pattern in schematic form, for the multitone carrier modem shown in Figure 4.

Figure 28 shows, in schematic form, “wake-up” signaling for the multitone carrier system modem shown in Figure 4.

Figures 29 to 31 show the set-up sequence of the multitone carrier system modem shown in Figure 4.

Figure 32 shows, in schematic form, a network overview of a network interface for a VDSL modem application.

To facilitate the understanding of the present invention, a list of abbreviations used in this patent application is presented below.

ADC: Analog-to-Digital Converter AIS: "Alarm In Signal" ASIC: Application Specific Integrated Circuit BPSK: Binary Phase Shift Keying BSI : Basic Sync Interval (Base Synch Interval) BSI-D: BSI for downlink connection BSI-U: BSI for uplink connection Un 20 30 CCH: CM1: CM2: CMB: CP: DAC: DAS: DFl: DFB: DMT: DWMT: EMC: FEXT: v11 w * i FTTN: 506 643 ll Control channel Carrier mode 1; bit-loaded and used carrier (Carrier mode 1, bit-loaded and used carrier) Carrier mode 2, (removed) masked or disabled carrier (Carrier mode 2, masked out or disabled carrier) Carrier type (mode) 3, carrier arranged for zero bit loading, (Carrier mode 3, zero bit-loading enabled carrier) Cyclic prefix ( Cyclic Prefix} Digital-to-Analog converter (Digital-to-Analog converter) DF3 frame sequence (DF3 frame seq uence) Data frame, random data parallel CCH, (Data frame, random data parallel CCH) Data frame, random data one CCH (Data frame, random data one CCH) Data frame, fully bit loaded one CCH (Data frame, fully bit loaded one CCH Discrete Multi Tone Discrete Wavelet Multi-Tone Electro Magnetic Compatibility Forward Error Correction (Far End Cross Talk) Fast Fourier Transform ( Fast Fourier Transform Fiber To The Node 506 643 WN IQ UI N Gl MUSIC: G2 MUSIC: G3 MUSIC: IFFT: IIR: ISDN: ISI: JTAG: LEX: LP: NT: OFDM: ONU: PGA: POTS: QAM: 12 First generation, prototype system (VME-based) (Generation one, prototype system VME-based) Three + two, ASIC implementation (Three + two ASIC implementation) Two chips' silicon implementation (Two chips silicon implementation) Reverse "Fast Fourir" transformation (Inverse Fast Fourir Transformat ion) Infinite Impulse Response International Standard for Digital Networks Inter-Symbol Interference Joint Test Action Group Local Pass Low Pass Network Termination Network Termination Network Unit Orthogonal Frequency Division Multiplexing Optical Network Unit Programmable Progain Gain Attenuator Conventional Plain Old Telephony Service ) "Quadrature Amplitude Modulation" 20 h) Un 30 SC: SDH: SF: SNR: STB: SUS: SUSl: SUS2: TA: TDMA: UTP: VCXO: VDSL: 506 643 13 System Controller Synchronous digital hierarchy (Synchronous Digital Hierarchy Synch Frame Signal-to-Noise Ratio Set Top Box Synch Frame Sequence SF and DF1 SF and DFl f SF and DF2 frame sequence (SF and DF2 frame sequence) "Time Advance" Multiple access with time division (Time Division Multiple Access) Unshielded Twisted Pair Voltage Controlled Chrystal Oscillator Digital subscriber lines for very high bit rates high bit-rate Digital Subscriber Lines) The system to which the present invention relates is referred to for simplicity as “MUSIC” (Multi-carrier System for the Installed Copper Network).

MUSIC is intended to provide high-speed communication on copper pair cable for telephony to support broadband multimedia services.

The MUSIC system described in this (SE 9603196-8) and the patent specifications arranged in cross-reference 506 645 20 25 30 14 SE 9603187-7, SE 9603188-5, SE 9603189-3, SE 9603190-1, SE 9603191-9, SE 9603192-7, SE 9603193-5, SE 9603194-3, SE 9603195-0, SE 9603197-6 and SE 9603198-4, offer a cost-effective and robust customer implementation with silicon, which provides 26: 2 or l3: 2 Mbit / s asymmetric transmission over copper cable (<130 meters) for use in existing local telephone networks.

The MUSIC system can be accessed using the network concept known as Fiber To The Node (FTTN), which uses optical fiber, which each serves many users, up to a switch cabinet near the users' homes. Thus, the cable length specification for MUSIC can be successfully limited to 1300 meters.

The MUSIC system is mainly intended for transmitting a high bit rate (26 Mbit / s) signal downstream to the subscriber, and a low (2 Mbit / s) signal from the subscriber. bit rate upstream, Figure 1 shows the MUSIC system. A network device, NOW, is connected to the fixed network through an optical fiber link, (FTTN). A network termination, NT, connected to a multimedia application, e.g. video-on-demand, is linked to the NOW via copper cable. The MUSIC system supports a high data rate downstream and a much lower data rate upstream.

In the MUSIC system described here, two fixed bit rates (l3: 2 and 26: 2 Mbit / s) are supported, where the lower bit rate l3: 2 Mbit / s can be implemented as an additional option for use in bad, or extremely long, copper cables.

For network termination (NT), the connection consists of a set of standardized interfaces, such as POTS, ISDN, ATM25 and Ethernet. All transmission protocols are supported by the carried by the data flow in the modem, except the POTS service which is passively filtered out so that it is independent of the (NOW) modem status. The network device terminates in the fixed network.

MUSIC separates up and downlink spectra by passive filtering in the analog parts.

The version of MUSIC described here is intended to provide the opportunity for future functional upgrades. For this reason, the FFT / IFFT block is designed to support full functionality so that it can be reused in future system upgrades.

The MUSIC system is a DMT-based, multi-carrier VDSL system that uses discrete Fourier transform to create and demodulate individual carriers. This is shown in Figure 2, which shows two transceivers each having a receiver, Rx, and a transmitter, Tx, connected to a twisted pair of copper. Data is transmitted between the two transceivers using a plurality of carriers, some of which may not be used, e.g. when the channel quality is extremely poor. The number of bits transmitted by each of the carriers can also vary, depending on the channel quality.

An over-the-air modulation technology such as DMT handles frequency-dependent losses and interference on twisted pair cable efficiently. In the MUSIC system, the available bandwidth of 10 MHz is divided into 1024 carriers with a width of 9.77 kHz each. The assigned transmission power for the individual carriers depends on the interference power and the transmission losses on each of the belts. Each carrier transmits multilevel pulses that can represent up to 12 bits of data (4096 QAMÉ. Signal-to-noise ratio (SNR) is calculated on the receiver side. If the individual carrier's carrier has a high SNR, up to 12 bits are placed on this carrier. For carriers with lower SNR values, 506 645 UI 10 25 30 16 fewer pieces are placed on the carrier. Carriers that are affected by narrow-band interference sources are switched off. Forward error correction and data interleaving are used to mitigate the effects. of temporary bursts of impulse disturbances.

Asymmetric VDSL is implemented in this version of the MUSIC system, which means that the downstream speed is much higher than the upstream speed. The selected speed of two fixed downstream speeds (26/13 Mbit / s) depends on the current cable length (supported by the system; the m) and / or the quality of the channel. The upstream speed is fixed at 2 Mbit / s.

The MUSIC system for separating the downstream channel from different frequency bands can be used in the upstream channel and both from POTS, see Figure 3.

Alternatively, other duplex methods can be used, e.g.

TDMA and / or a method where every other carrier path is dedicated to the downstream and upstream channel.

Figure 4 shows an overview of a MUSIC modem to which the present invention relates. The most important hardware blocks are ADC and DAC, transformation processing, channel value calculation / equalization, synchronization, fourir symbol mapping and detection, coding and decoding with interleaving, network interfaces and system monitors.

The modem can be considered in the form of four principal function blocks, namely: - the digital receiver unit; - the digital transmitter unit; - the analog input (front end); and system controller / PCI.

Ul 10 h) Vu 30 506 643 17 The analog input includes a hybrid transformer connected to an unshielded, twisted pair and POTS. On the receiver side, the hybrid is connected, via a low-pass filter, LP, a programmable gain attenuator, PGA, to an analog to digital (A / D) converter. A voltage controlled crystal oscillator, VCXO, is used to drive the analog to digital converter. On the transmitter side, the hybrid is connected to a digital to analog (D / A) converter via a low-pass filter.

The digital receiver unit includes a "fixed Fourir" rescaling unit, FFT, connected, as shown in Figure 4, to a synchronization unit and a channel estimator. The channel estimator is connected via a symbol detection unit and a "de-interleaving" and decoding unit, to a bit handling unit: and from there to a network application interface.

The digital transmitter unit includes a bit handling unit which is connected to an inverted "fixed Fourir" transformation and scaling unit, IFFT, via an encoding and interleaving unit and a symbol mapping unit.

The system control is connected to different functional units in the digital receiver and digital transmitter, and to the network application interface and a computer interface, as shown in Figure 4.

The network (network) interface connects the higher protocol level to the modem layer one functionality.

This block is responsible for providing the system with data at the configured bit rate, and adds “dummy if required. frames ”(dummy frames) 506 643 20 25 30 18 The data is then channel coded and interleaved. The MUSIC system described here uses a convolutional code combined with interleaving. Using a depth with a plurality of frames, a combined frequency / time interleaving is obtained (see later in this specification).

The symbol mapping block receives input data as an integer vector. This vector is mapped into the configured constellation depending on the current bit load value. The mapping unit uses a Gray coding scheme to reduce the probability of bit errors.

A real (vector) vector multiplication is the first step in the IFFT block. This causes the system to scale the output power level of each carrier. The IFFT block then performs a real 2048 point inverted FFT on the input data, which modulates each carrier. As a final step, an address "wrap around" is performed on the output data, where a copy of the first 128 samples is added at the end of the frame. This is called the cyclic prefix (CP).

The modulated signal goes to a DAC that converts the signal with a minimum true dynamic range of 84 dB. The DAC is clocked by the system sample clock at 20 MHz. To get rid of Nyquist "ghosts", the signal is LP-filtered, and the hybrid provides a balanced interface to the copper cable.

An overview of the signal path of the MUSIC transmitter and receiver is shown in Figure 4. The transmitter part uses the same hybrid design as the receiver.

At the receiver end, the splitter / hybrid transceiver separates the frequencies used by the POTS, from 0 to 4 kHz, from the frequencies used by the system. It also extracts the low-level reception signal from the combined high-level transmission signal and the low-level reception signal.

To reduce Nyquist effects on the signal, the received analog signal is low-pass filtered before being fed into the PGA (Programmable Gain Amplifier).

The PGA is necessary to get the most out of the dynamic range of the ADC. In this system, the dynamic range shall be at least 66 dB.

After the signal is converted to digital format, the synchronization and FFT blocks receive the data.

In the synchronization block, a frame clock (for controlling the FFT buffers) and a control signal for VCXO n are generated.

Initially, the retrieve synchronization block retrieves the frame clock from the sampled signal. The frame clock is then used to calculate the frame synchronization value calculation (frame timing estimate) and is transferred to the "VCXO feed back controller". The VCXO generates the sampling clock (20 MHz).

A sampling clock that is only controlled by a "frame time estimate" is not sufficiently accurate in a DMT system.

Therefore, after the locking sequence, a dedicated pilot carrier is used to achieve a high synchronization precision on the sampling clock.

A BSI signal is also extracted from the pilot carrier.

BSI is the Base Synchronization Interval timing signal used to synchronize the CCH communication of the transmitter and receiver.

One of the new aspects of the MUSIC system is the algorithm used by the synchronization block, which is discussed in more detail later in this specification.

A 2048 point real EFT is performed on the input frames of the FFT block. After this, rescaling is performed 50 20 h) KJ! (Rescaling), which is based on the energy charge parameters, before the data is transferred to the next block.

The channel value calculation and equalization is performed on the output data from the FFT block. All data frames are used to value the channel properties. These are then used to compute a bit charge vector that determines the number of bits to be transmitted on each of the carriers. This information is then transmitted to the transmitter through the upstream control channel (CCH).

In the symbol detection block, a "demapping" is performed for each carrier according to the bit-loading mask. (de- (FEC, on the detected bitstream).

After mapping, “deinterleaving” interleaving and “error correction on reception” are performed. Forward Error Correction The data is then ready for the network (application) / application interface block after bit handling. The dummy frames (dummy frames) are taken live: in this block.

At the heart of the system, shown in Figure 4, is the system controller (SC). The SC is a general purpose processor that interfaces with and controls the various sub-blocks using a local PCI bus. In the version of MUSIC described here, the CPU controller is programmable. An external port is provided, through an JTAG interface on the motherboard (on-board), to facilitate programming.

The main tasks of the SC are to control system start-up and construction during driving time and to perform bit charge and energy charge calculations. It communicates with the remote side of the modem through a dedicated (CCH). control channel This channel transmits data regarding RJ: 20 h) UI 506 645 21 changes in bit / energy charge and other system signaling.

To obtain a cost-effective product for high-volume use, the digital components of the system must be based on at least two ASIC circuits. Figure 5 shows how the system can be divided (partition) for chip design purposes.

One chip contains the FT / IFFT core. A second chip contains frame synchronization, channel value calculation and smoothing, symbol detection and symbol mapping. The analog block and the network interface block can be implemented on a third and a fourth chip, respectively.

The system parameters used by the MUSIC system described here are shown in Tables 1 to 3 attached hereto.

VDSL systems operate in the spectrum from 0 to 40 MHz.

In this band, the MUSIC system, as described here, occupies the lower 10 MHz. A number of traditional bands exist in this spectrum, including PGTS and some see Figure 6. radio amateur bands. Different frequency bands are used in the MUSIC system described here to separate downstream from upstream channels. Since the MUSIC system described here uses 1024 carriers over 10 MHz, each carrier has a bandwidth of 9.77 kHz, where the first two carriers are allocated by the DC level and the POTS service. The last carrier is disabled because it is the Nyquist point.

Other carriers (on radio tapes) may need to be canceled. This is primarily a matter of immunity and charisma on the balanced copper pair.

By passively filtering the POTS spectrum, this service can be done independently of the MUSIC system described here, runtime status, or power supply.

There are two ways to provide ISDN services for a MUSIC modem connection. One way is to let the POTS and ISDN systems exist under the (below) MUSIC 506 645 Un 10 20 30 22 frequency bands. This can be achieved using a similar filtering process for ISDN band spectrum as for POTS. This filtering allows the service to be provided regardless of configuration.

The second way to provide ISDN is to allow ISDN to be a carrier service in the MUSIC system. This solution has the advantage in terms of spectrum efficiency. The use of 1024 carriers over 10 MHz gives each carrier a bandwidth of 9.77 kHz. The ISDN spectrum requires the allocation (150-4) /9.77 = 5, of these carriers. Depending on the channel characteristics, these five carriers must be chosen to have the best SNR in the system. For a standard connection, this gives 5 * l00 = 500 kbit / s bandwidth.

The optimal solution is therefore to use the modem as a carrier, and allocate only 64 kbit / s, compared to 500 kbit / s for the total bandwidth of the 64 kbit / s ISDN service.

The result of the attenuation measurements and FEXT (Far End Cross Talk) performed on a telecommunications operator's network showed that it is possible to achieve bit rates higher than 100 Mbit / s if the cable is shorter than 200-300 meters. For longer cables, attenuation at higher frequencies limits the maximum bit rate.

For cables of around 500 meters, 40 Mbit / s can be achieved, and for a 1 km cable, 15-20 Mbit / s is realistic.

Another factor that reduces performance is EMC, which limits the power used. Some parts of the frequency domain may also need to be excluded.

A typical PSTN can be expected to have the following characteristics in terms of impulse interference: - maximum duration 250 us - median interval 67 ms 20 506 643 23 - maximum peak amplitude 20 mV - main part of the energy below 200 kHz - background interference -107 dBm / Hz The main source for synchronization in the system is the sampling clock. The reference for the sampling clock is located on the NU page and is common to all twisted copper pairs in a secondary cable.

The sampling clock frequency is 20 MHz at 10 ppm, with a "phase jitter" of less than 0.5 ns.

The sampling clock on the NT page is phase locked to the NOW page. The logic of the reading uses the frame synchronization value calculation (frame timing estimation) in a first stage, and then uses the pilot carrier path to produce a fine adjustment of the reading. The locking logic controls the frequency of a VCXO via an 18 bit digital / analog converter. The requirements for the VCXO are 20 MHz in 25 ppm range and 10 ppm / volt sensitivity. The final reading must have a precision of l / 100 samples, with a "phase jitter" of less than 0.5 ns.

The frame clock is l / (2048 + 128) of the sampling clock and controls the start of receiving and sending the frames.

The frame clock, which is used for both transmission and reception, differs in phase on both the NU and NT side.

The frame clock for transmission on the NT side is the master and controls the start of the signal intervals, see Figure 7.

The reception frame clock on the NT side is obtained from the hardware function for the frame synchronization value calculation and controls the start of the frame sampling period, see Figure 7.

The frame clock for transmission on the NT side is the same as the frame clock for reception, but is a TA sample earlier in phase. TA is a parameter measured during system start-up on the 506 645 Us lx) Un 24 NU side and is used to compensate for propagation delay on the copper conductor. This must be done to maintain the orthogonality, across the copper conductor, for the sampled periods, both uplink and downlink. The frame clock for transmission on the NT side controls the start of the signal intervals, see Figure 7.

The frame clock for reception on the NOW page delays a number of sample clock cycles (TA) in relation to the frame clock for transmission, after TA calculation has been performed. The delay before the calculation of TA in the start-up sequence is determined by the hardware timing for the frame timing estimation hardware function and the value is accessible to the control unit.

The frame clock for reception on the NU side controls the start of the frame sampling period, see Figure 7.

The BSI clock is used to synchronize parameter changes between the sending and receiving sides. The parameters can be, for example, bit charge, energy charge or control channel frequency. The parameters are updated by the system controller, on both sides, before the BSI clock initiates the switch for the new set-up.

The BSI clock is 1/8192 of the frame clock. The uplink BSI clock is delayed by half a BSI clock cycle relative to the downlink BSI clock.

A short pseudo-random sequence on the pilot channel is used for BSI synchronization between the transmitting and receiving sides.

The cyclic prefix is an extension of the frames added by the FFT chipset. To maintain orthogonality throughout the signaling period, the last 128 samples of the frame are copied and placed before the actual frame. This arrangement addresses issues related to interference between symbols caused by time dispersion.

It is important that the part of the signaling period sampled only on the receiving side overlaps a signaling period in the other direction, along the entire copper conductor. TA is used to optimize this overlap period. The maximum cable length is limited by TA = corresponds to 1280 meters (if the propagation delay is 5 ns / m). 128 samples = 6.4 ps propagation delay. This The analog interface connects the received and transmitted digital data stream at the C1 chipset to the telephone line. There are also connections to the Tl chipset and the system controller for control purposes.

The analog interface is shown in Figure 8. The cable is connected to a hybrid transformer that is also linked to POTS. On the receiving side of the hybrid, the incoming signal passes through a low-pass filter and a programmable gain attenuator to an analog-to-digital converter, ADC, and from there to the Cl chipset. On the transmitter side of the hybrid, the output digital signal is converted to analog by a digital / analog converter, DAC, and goes from there via a low-pass filter LP to the hybrid transformer. A voltage controlled crystal oscillator, which drives both the ADC and the DAC, is connected to the sync block of the T1 chip.

An OFZM frame is a sum of sinusoidal waveforms modulated in phase and amplitude and spaced in frequency plane: (frequency domain) with a minimum of separation distance between waves. Assumption: that the symbols within the frame are evenly distributed and uncorrelated in relation to each other giving a signal in the time plane with an approximately normally distributed instantaneous amplitude. Thus, there is a small possibility that input data can interact with each other 506 643 UI k) L / l 26 to create pulses with very high peak levels.

However, the maximum amplitude must be limited to a lower amplitude than this so that there are a sufficient number of quantization levels in the DAC to handle average signals. Although the DAC has sufficient resolution to accommodate a high peak level in the transmitter, there are limitations on the receiver side (ADC). However, the consequences on the recipient side do not have to be as serious as they may seem.

A short cable has lower attenuation in the high frequency range than a long cable, see Figure 9. This means that a temporary pulse can occur in the receiver almost unaffected by the cable characteristics. Therefore, a relatively large dynamic range is required in the receiver. However, this can be easily achieved because almost equal damping does not require a large dynamic range. The ADC needs to accommodate the area indicated in Figure 9 by the solid, rough, arrow-marked line.

However, the greater high frequency attenuation of long cables requires a large dynamic range.

The high frequency attenuation also means that several large peaks would be required from the transmitter to build up high amplitudes in the receiver; a case that is even less likely to occur at the ADC input (input) than single peaks. The “free height” (headroom) can therefore be reduced and the ADC should accommodate the area marked by the rough, dashed arrow line in Figure 9.

In summary, performance can be optimized by carefully setting the signal level at the ADC receiver depending on the cable length.

The line splitter / hybrid has two main tasks, namely to: 10 20 IJ UI 30 506 643 27 - split and combine telephony signal (POTS) and VDSL signal frequency bands; and - prevent the transmitted signal from appearing at the receiver on the same device by balancing the cable.

Since each transmission direction has its own frequency band, it is possible to optimize both sides in terms of their respective frequency bands to increase the overall performance.

The purpose of the low-pass filter on the input signal is to reduce aliasing effects on interference above the frequency range used.

The low-pass filter on the output side reduces the transmitted power on the "stop band". These filters can form part of the division / hybrid module.

The best commercially available ADC today is the "Analog Devices AD9042" which has a signal-to-noise ratio of approximately 66 dB. It is recommended that either this ADC, or one with equivalent performance, be used.

For this description, it is assumed that a DAC with 14 bit resolution is used.

The FFT and IFFT algorithms are built up of 1024-point complex FFTs with data reorganization to allow the calculation of two real sequences at the same time.

Consequently, each of FFT and IFFT are effective 2048 points. The hardware realization is based on a radix ~ 32 core that calculates the result in three “transformations” (passes), see Figure 10.

The relationship between the signal-to-noise ratio and the resolution in the algorithm can be expressed as: 2b-vl SNR = 2 506 IQ Un 30 643 28 where b = number of bits, and v = "transformations") 11 (number of effective radix-2 The solution for b gives 17 bit resolution (based on ADC SNR), but since ADC is not the only source of analog signal degradation, 16 bit resolution in the algorithm should be sufficient to maintain the resolution throughout the system.

The VCXO generates the sampling frequency used in the NT part of the system. The control voltage is based on data from the synchronization unit. The clock frequency must be very stable and phase locked to the NU reference clock to maintain orthogonality between symbols.

To take full advantage of the dynamic range of the ADC, a programmable attenuator must be inserted before the ADC.

The attenuation level is mainly a function of the cable length and can be determined by the value for "timing advance synchronization" by the system controller.

Attenuation resolution and range, and the relationship between the value for timing advance and the attenuation level, must be determined. Equalization and variance values can also be used in the calculations for improved results.

In a DMT system, a very precise synchronization between the transmitter and the receiver is necessary, especially when carriers are modulated with large constellations.

In the embodiment described here, a new frame synchronization method is used which is based on correlation properties built into the structure of the received signal.

On the NOW side, a crystal frequency oscillator is used as a reference for generating the sampling clock. On the NT side, a sampling clock is generated by a VCXO (Voltage Controlled Crystal Oscillator) which is read in phase to the oscillator on the NU side. The VCXO is initially controlled by the IQ Un 29 frame timing estimate.

However, the resolution of the frame synchronization value calculation is not sufficient in the present application. Therefore, a dedicated pilot carrier, after a lock-in sequence, is used to achieve a very high accuracy of the sampling clock synchronization. Depending on the long symbol duration of a DMT system, interference between symbols caused by channel time dispersion can be eliminated by means of a guard interval. interval) as a prefix to each frame in the time domain. To maintain the orthogonality of the frames, the contents of each prefix are a copy of the last part of the following frame, making the frames appear to be partially cyclic.

The synchronization method used to estimate the frame synchronization uses the high correlation between a prefix and the corresponding part of a frame. By continuously correlating samples of the received signal, separated in time by the (known) frame length, the passage of a protection interval will cause a peak in the correlation value calculation. Therefore, these peaks will have a known synchronization relationship to the frames and can be used to create a frame start signal. The principle is shown in Figure ll.

The correlator and peak time estimator use a system clock generated by a VCXO. This clock is divided by the total number of samples in one to create a signal with the same period as the correlation peaks. signal interval (a cyclic prefix and a frame), the frame time deviation between these two signals is used as input to a feed-back controller 506 645 Uu 10 20 30 which adjusts the VCXO frequency to the correct sampling frequency. However, the phase of this sampling clock is not accurate enough to be used in a DMT system. Therefore, the frame synchronization value calculation is mainly used for a lock-in operation. It is also used to monitor frame synchronization to detect major anomalies that will necessitate resynchronization.

The correlation of the received data is calculated continuously. The time difference between the two signals is achieved by using a digital delay line on a frame length. The output of the delay line is multiplied by the non-delayed signal and integrated (accumulated) over an interval corresponding to the length of the cyclic prefix. The output from the integrator is the value calculation (estimate) of the correlation function.

Since only the synchronization information of the correlation value calculation is used, a simplified estimator is implemented which only uses the reduced complexity symbol of the input data compared to using it (sign). This hardware implementation has a highly complete sample word length.

Computer simulations have shown that the use of synchronous averaging of several signal intervals reduces the variance of the frame synchronization value calculation. Depending on the reduced dacaord length used in the multiplier part of the correlator, it is possible to implement such averaging function immediately after the multiplier.

A block diagram showing the implementation of the incoming signal X (k) 1024, the correlator is shown in Figure 12. passes through a delay with N = ie a frame, and Id to 508 643 31 to a conjugator. The output of the delay and the conjugator is then multiplied to produce a signal Y (k) averaging, Z (k) which Z (k) signal W (k) goes to an accumulator which gives an output signal C (k). which goes to an average value generator. The output from gár to a subtractor from delayed by L = 128 is subtracted. This gives a The details of the averaging part of the correlator are shown in Figure 13. The averaging comprises a series of delay elements combined with adders, as shown. The output signal can be expressed as MG Z (k) = Y (k-iM) 'O P ll where Y (k) is the input signal and Z (k) is the output signal.

To make the averaging synchronous with the frame structure of the signal, the delays are equal to the signal interval.

A detector for finding the position of the maximum size of the correlation function value calculation is shown in Figure 14. It is implemented using a register (# 1) for the latest max. Value and a comparator.

The register contents and the correlation size are compared, and each time a value greater than the register contents is found, the new value is stored in the register. The current value of a counter that counts sampling intervals (modulo signal intervals) is also passed to a second register (# 2). this second register to contain an index to it When an entire signal interval has passed, the max. value found during this interval. This index is stored in a third register (# 3), once per signal interval, and the contents of the first register (l #) are divided by two (using shift). (using 50 Lll 20 O 643 32 The index stored in register # 3 is interpreted as the deviation between the counter value and the current synchronization of the input signal frames. The feedback controller will cause the mean value of this deviation to converge to zero. The counter value can be generated to the signal interval. be used as a pointer using this counter value to indicate the frame start.

The value calculation of the complex representation of the pilot carrier in the frequency plane is performed using the FFT unit available in the system. The advantage of using this method is that the value calculation will be independent of the varying modulation of other carriers. They are due to the inherent orthogonality between the carriers. In order to achieve a value calculation with acceptably low variance, a certain average value formation is necessary. De: ta is performed using first-order digital IIR filters.

Unfortunately, the value calculation is represented as a complex number in rectangular coordinates, so the argument is not directly available. In the feedback loop, it is necessary to detect very small argument deviations. Therefore, the resolution of the argument must be high. The feedback controller will cause the pilot carrier argument to converge to zero. An approximation of the argument, which is linear only in a small range around zero, is then sufficient to achieve acceptable performance. A useful approximation that is "monotonic" in almost all four quadrants, and also easy to implement in digital logic, is described by the expression: Aä% ¶3 {C} - (l-sgn¶ {C}). ¶ {C} .sgn.3 {CH Un 643 33 where C is the complex pilot carrier value calculation, M is a positive scaling constant, and K is a positive constant that affects the design of the function (here K = 2 is used).

The channel introduces phase shifts on the pilot carrier which can cause “misalignment” between the frame synchronization on the input signal and the pilot argument zero.

To eliminate this problem, the pilot carrier estimator also passes through the frequency domain equalizer. The equalization parameter for this carrier is set during the start sequence, when the frame timing estimate has converged to its final value.

The choice of pilot carrier will be fixed, but logic for choosing other carriers as a pilot can also be provided. The feedback loop actually has two "controllers", each with its own input: al. The two controller outputs are added and fed via a D / A converter to the VCXO that generates the sampling clock. Both "controllers" are of PI type ( Proportional and Integrating).

Figure 15 gives an overview of the signal paths. The received data in the schedule passes through the correlator and the peak position estimator to result in the frame clock. The complex pilot carrier in the frequency plane derived from the equalizer is fed to a pilot argument estimator, the output of which is fed to "feedback controllers" which also receive output from the peak estimator. The output of the “feedback controllers” is then passed to a D / A converter to provide a signal that is used by the VCXO. to control During the start-up sequence, only the frame synchronization controller is active. Once the frame synchronization has stabilized, the 506 645 Un 10 30 34 equalization parameter for the pilot carrier is valued and set (by SC: I'1). updating this parameter is suppressed. After this This is done only once, and further change of equalization parameters, the averaging for the argument estimator is given sufficient setting time.

Finally, the frame synchronization controller is stopped and the pilot argument controller is activated. When the frame synchronization controller is stopped, its last value is locked so that the VCXO frequency remains close to its final value.

The pilot carrier is also used for the transmission of Base Synchronization Interval (BSI) synchronization information. The carrier argument is normally assumed to be constant.

A short pattern is BPSK modulated on the carrier using phases O and n and leaving the carrier on phase O for the remainder of the BSI interval. If this pattern is only a fraction (<1%), the pilot carrier argument value calculation is negligible. One of the BSI range, is the interference of the correlator used to detect the pattern and provide the synchronization signal for the BSI.

(SC) detection for synchronization locking and for monitoring reasons, the "System Controller" must have read access, in order to register that pours the estimator for frame time deviation and pilot argument approximation.

To handle the initial smoothing of the pilot carrier, it is necessary for the SC to read the averaged complex representation of the carrier and write to the smoothing parameter memory.

An offset register to determine the relative synchronization between the input frames and the frame start signal is necessary and must be writable from the SC. This is used on the NT page. The detected BSI event signals, for both reception and transmission, shall be connected to the SC as interrupt inputs.

Alternatively, the pilot carrier can be recovered from the signal in the plane, using a bandpass filter, and used directly for phase locking of a sampling clock oscillator. The frequency plan method described here has the advantage that the pilot carrier estimator is independent of the modulation of the other carriers. Another frame synchronization method would depend on introducing a known pattern into certain frames.

This would reduce system capacity.

The frame length and the length of the cyclic prefixes are fixed in the embodiment described here. The method, described above, is designed to work in a feedback loop with a VCXO. In a device using a specific sampling clock oscillator, the design of the frame synchronization estimator needs to be slightly modified. It is important that the VCXO has a very low phase fault, as the feedback loop is too slow to compensate for such a fault.

A discrete multitone system (DMT) modulates N complex data symbols on N carriers (here we use N = 1,024 carriers). This mapping is calculated as a reverse (inverse) discrete Fourir transform using Inverse (IFFT).

N st carriers of an FFT.

Fast Fourier Transform In the receiver they are demodulated in the modem, described here, FFT and IFFT are performed by the same unit, using the same base (radix) 16, or 32 "cores", in different phases. This process is shown schematically in Figure 16.

The main operation is divided into frames with lengths of 2048 real, or 1024 complex values. For each frame 506 645 UI W 30 36, this unit performs an FFT, IFFT, scaling, descaling, and adding cyclic prefix.

The FFT and the IFFT calculate 2048 points of real FFTs and work with a minimum of 16 bit arithmetic.

For the night terminal side, (NT), there is a requirement for synchronization between the input frame start and the IFFT output start. (A synchronization between the upstream and downstream carriers). The transmitter must be able to start transmitting a frame before it starts receiving a frame, so-called "timing advance".

A scaling should be provided before IFFT.

This scaling is a multiplication between the real coefficients stored in this unit, and the input values from the symbol mapper (SM). The coefficients are 16 bits each.

The coefficient memory consists of two banks of the same (16XlC24 bit). others are updated. Switching is made possible by size One bank is used while a PCI command and executed at the next BSI.

After the FFT, a rescaling must be performed before the data is transferred for smoothing and symbol detection.

This rescaling is a multiplication by the inverted value of the scaling values. The coefficients are represented by 16 bits.

An exponent (resulting in a 4-bit post shift) may also be needed to maintain precision.

The coefficient memory consists of two banks of the same ((16 + 4} X1024 bit). Others are updated, switching is made possible by a PCI size One bank is used while commanding and executed at the next BSI.

(Ju 20 k) Un 506 643 37 At the beginning of each frame a cyclic prefix is added.

This process is shown schematically in Figure 17. The insertion of a cyclic prefix eliminates interference between symbols (ISI), and preserves the orthogonality between the tones, resulting in a simple input / output ratio that makes it possible to consider each carrier as a separate channel. .

This cyclic prefix consists of a repetition of the last part of the frame.

Provided that "timing advance" is used and the maximum cable length is 1300 m, a cyclic prefix of 128 samples will be required. Thus, the output for each frame will be a sample: 1920, l92l, ..., 2046, 2047, 0,1,2, 2046, 2047 For each of the above components there is a FIFO as an interface to the external world with FFT / IFFT input and output memories. Thus, there are a total of 4 FIFOs.

It is recommended that the FIFOs with interfaces to the analog side have a size of 384 words (16 bits) and the FIFOs that interfaces with Tl chips have a size of 448 words (32 bits).

Another DMT technology that does not use "Fourir transformation" is "Discrete Wavelet Multi-tone Transform" (DWMT). This method has been submitted to the ADSL Standardization Committee, which rejected it.

The precision required in this technology depends on the required dynamic range, which in turn is determined by the analog components (especially DAC). The FIFO size will depend on the clock speed differences and the amount of timing advance used. The use of clipping is a compromise between dynamic range (quantization perturbations) and clipping perturbations. 50 10 25 30 O 643 38 Channel value calculation is performed with a "decision-oriented" method, since all (decision directed) data frames are then used to update the channel model. Known data frames are necessary only at start-up. In certain circumstances, interference on the channel can be value calculated using all This is important for early detection of changes in channel transmission quality.

The basic principle of "decision directed" value calculation is that differences between received data and known, channel model.the channel model exactly enough to be used for transmitted data is used to update a At a certain stage of this process is equalization of the received data, and the detector will produce correct data.This output can then be used in the same way as the known data for further updating of the channel model, therefore the predefined data frames are no longer necessary and random data transmitted through the channel is used instead.

By using data taken after the equalizer as input data, and data after the detector as the second input data, an adaptive update algorithm can be designed. It modifies the equalization parameters in small steps in such a way that the equalizer converges towards a model of the "reverse" (inverse) channel. Figure 18 shows a block diagram of such a system. Input in the frequency plane enters the equalizer and is multiplied by the output of an equalizer parameter update unit, EQ. The resulting signal, U, then goes to a detector (quantizer in whose output is Y. Y then goes to a symbol decoder which produces a decoded data bitstream. U and Y also go to an input on the equalization parameter updater and to a variance estimator .

The output of the variance estimator is W. 5Û6 643 39 An adaptive algorithm for calculating the value of the smoothing parameters (EQ), which uses the smoothed data (U) and the quantized data (Y) as input, is described by the following equation: as = Book + “.EQpUkÉ (Yk- "Q | U1 <| 2 where p is a positive constant (p << 1), which affects the adaptation dynamics. A smaller value gives a slower adaptation than a larger value, but it also gives a larger insensitivity when there are disturbances on the input signals.

For implementation reasons, the division shown in the equation should be avoided. The expression p / | UkP has too large a dynamic range to be replaced by a constant. However, it is possible to quantize this expression in a logarithmic way as shown below: P1 / | UK! 2 z 2-integer (2.log2 | Ukh + integertlogzp) The exponent in the above expression can be produced using the absolute value of U¿ as input in a binary priority encoder and change the sign (negating) of the output.

Since the expression is an integer power of two, the multiplication is implemented in the algorithm using a "barrel shifter".

IQ lJl The interference variance of each of the carriers is value calculated using the standard method of integrating the squared deviations from a mean value. In this case, each quantized value, Y, is used as the mean value of the range of data values, U, which is quantized to this Y. This method assumes that the symbol error rate is low enough to associate each 506 Un 20 IQ Un 643 40 data value with the correct mean. However, if suitable constellations are selected for the different carriers, this condition is met.

Figure 18 shows the variance estimator as part of the system. The algorithm used for the value calculation is described by the following equation: Wk + 1 = .Wk-rs. | Yk_UkI2 The integration is here replaced by an exponentially weighted average value filter. The parameter s is a small, positive constant (2 << 1) that affects the dynamic properties of the filter. This is not a critical parameter, and selecting 8 from integer powers of two will suffice :.

If e :: value of e is selected which gives a good variance value calculator (estimator), the algorithm will not be able to detect sudden changes in the interference level.

Therefore, a separa: algorithm, which works in parallel with the variance calculator, may be necessary for this task.

The "System Controller" must have both read and write access to the memory that pours the smoothing parameters.

Initialization of the parameters is necessary at start-up. Monitoring of the parameters is also necessary to detect when they have leveled sufficiently close to their final values.

The channel variance memory must be available for the System Controller's read operations. Initialization of this memory to all zeros can be connected to a system reset.

The parameters that affect the dynamics of the estimators must be: inaccessible for writing from the System Controller.

V: 20 k) (Ju 506 643 41 The method described here assumes a specific start-up sequence, both for the channel and interference value calculation_ During normal execution, it depends on a suitable choice of bit charge that gives a sufficiently low symbol error rate.

It is important that the smoothing parameters are initialized to unit value at the beginning of the start sequence, as the input to the update algorithm passes through the equalizer.

The update algorithm is sensitive to scaling changes in the data path.

Any change in scaling in the transmitter must be compensated for in the receiver. This also requires special care when using the analog gain control on the input side of the receiver.

The symbol mapper (encoder) maps a number of bits to a complex: al (I, Q) which indirectly determines the phase and amplitude of a carrier. The mapping of all values of a certain bit length is called a constellation, and is shown in figure 19. The detection is the inverse function, ie from a complex value the value of the bits sent on the carrier is determined. The number of bits transmitted on a given carrier path is determined by the bit charge factor thereof.

The construction of a specific constellation is aimed at allowing each point to be moved as far as possible from all other points. At the same time, the average energy should be as low as possible. Another restriction is that the mapping and detection units should be as simple as possible. However, the decision as to which constellation to use will affect not only the symbol mapping and detection units, but also the bit charge and possibly the adaptive equalizer. 506 643 N 25 30 42 For a given carrier, the encoder selects an odd integer point (I, Q) from the square-grid constellation based on the b-bits (vb1_vbQ, _, vLv¿). For simplicity of description, these b-bits are identified by an integer label whose binary representation is (vbl, v¿Ql ___ 'vLv2). For example, for b = 2, the four constellation points 0, 1, 2, 3 are "labeled" corresponding to (vpvz) = (0,0), (0,1), (1,0), (1,1), and .

For even values of b, the integer values of I and Q of the constellation point (I, Q) are determined from the b-bits (vb-i, Vb-L H_, v1jQ) as follows. Divide V by VI = (VVL vb3, _H Ivl) and VQ = (vg2, vb4, _A_ | vO). Then apply the reverse Gray code to VI and VQ. This gives I and Q as I = 2GraW (VI) + 1, and Q = 2Gray (VQ) + 1.

Figure 19 shows how the binary pattern for V folders on I and Q when b = 6.

Before these values are sent to the IFFT, they are normalized by shifting so that "msb" of these numbers becomes "msb" on the output (16 - [b / 2] steps left).

For a given carrier, the decoder uses a constellation point (I, Q) to determine the b-bits (vb fl, vb¿ '___, vLv2). For simplicity of reference, these b-bits are identified by an integer label whose binary representation is (vbl, vb2, _H IVLV2).

It is assumed that the values of I and Q are limited by saturation to the range (X, Y). To determine V, Gray is coded, i1.io) f and Q = (q1s, q14,, q1, qo) f and then combined to V as V = (giß, gqß, giN, gqN, ..... ). where the upper b-bits are valid.

LI! 10 30 35 506 643 43 The number of bits each carrier transmits depends on their respective signal-to-noise ratio (SNR).

The signal-to-noise ratio is calculated for each carrier in the receiver. Based on the signal-to-noise ratios, bit charge factors are calculated for each carrier. Thus, the number of bits each carrier must transmit per transmitted symbol is determined.

These bit load factors are calculated in an initial training session and can be updated as needed.

The MUSIC system uses 2-dimensional Quadrature Amplitude Modulation (QAM) on each carrier, with bit load factors ranging from 0-12 bits.

The number of bits transmitted on each carrier can be expressed by: SNRi) ß¿ = bi + log2 (L) = logz (l + (1) l ”where F, the SNR gap, a system margin, and L is the constellation expansion due to modulation, possible coding and depending on ie extra bit needed for coding.

The use of QAM constellations and some form of coding provides; [Q fl ps / MJ * F: - yd + ymargin 3 (dB) (2) where Psär is the desired symbol error rate, 'm is the “gain of coding” in the system, fi mrwm is the system margin. The system margin is a factor used to compensate for non-modeled losses, impulse disturbances, etc. Equation (l) gives a bit charge factor with infinite granularity. 506 643 Un 10 25 30 44 The bit charge factors are rounded to give the supported factors (0 - 12 bits).

The rounding procedure will reduce the performance of the DMT system. If the energy distribution is allowed to vary, energy charge factors can be calculated for each carrier. This provides the ability to tune the energy so that (1) results in a bit charge factor supported by the system. Reconciliation gives: (Qi-1) T Ei_2 i (3) However, this can result in very large differences between carrier energies. In an environment with several different DMT systems, strange effects can occur if the different energies are allowed to vary too much.

The remote crosstalk (FEXT) will vary considerably in such an environment, and some DMT systems may have the full capacity of the cable. To prevent these effects, only small changes in the carrier energies can be allowed. Another limiting factor is the maximum energy allowed on each carrier.

The input data to the bit load algorithm will depend on the selected frequency domain equalizer. If an adaptive DFE is used, SNR is obtained by: sNRi = wi) where Wi is the value-calculated interference variance described above.

For each carrier, a bit charge factor and an energy charge factor are calculated. The bit load factors can be represented by 3 bits, but to prepare the system even for odd bit load factors, 4 bits are recommended. For 20 IQ Un 30 506 643 45 energy charge, n bit is used to give 2 ”- 1 possible factors.

The implementation of the calculations of bit charge and energy charge factors can be done in four steps shown in Figure 20. To achieve a given bit rate, a required SNR can be calculated and the system margin adjusted so that the desired bit rate is achieved. The process, illustrated in Figure 20, includes the following steps: - First, the SNR is calculated using (4).

In the second step, four comparisons are performed, ie one for each of the four bits representing the bit load factor.

The thresholds depend on L and F, and can be pre-calculated. The first comparison determines if the bit charge factor is greater than 7, and the result of this comparison controls the first of the four bits representing the bit charge factor; it also controls the threshold for the next comparison. Similarly, this comparison controls the second bit and the threshold for the next comparison. After the four comparisons, the bit charge factor is determined.

- The third step is to value the scaling factor for the transmitted energy so that the channel is used more efficiently. The energy is scaled according to equation (3).

Finally, the scaling factor is quantized to n bits.

It should be noted that in order to implement a system of constant energy charging, only the first two steps are necessary. 506 643 Un 20 30 46 The energy charge and shift performed for normalization in the symbol mapping determines the scaling and rescaling factors sent to the IFFT / FFT processor.

Thus, the method of determining an appropriate constellation (bit charge) for each separate carrier is based on a value calculation of a value of SNR for each carrier and the maximum allowable symbol error rate. The simplified method of the present invention takes advantage of the fact that all signals are normalized to the same scale at the input and output of the symbol detector (see Figures 4 and 5).

For each individual carrier, the symbol detector associates the input signal in an area around a specific (bit charge) to the distinct value of this point, which is the output value of the detector. constellation point The variance of the deviations of the input signal values from the distinct output value is a measure of the disturbance on the carrier and proportional to the SNR.

The constellations are usually described as a pattern of points in a complex domain. Assuming that the disturbances have a Gaussian distribution, the frequency function of the input signal values can be represented by rotating symmetrical "Gaussian clock shapes" centered at each distinct constellation point. The shape of the watch is determined by the variance.

The upper limit of the variance for a given constellation (bit charge) is determined by the maximum allowable symbol error rate. The lower limit of the variance will be equal to the upper limit of the second largest constellation. 10 20 30 506 643 47 The standard deviation 0, which is the square root of the variance, can be represented by a circle centered at each constellation point. The shortest distance d between neighboring points is a parameter for the constellation. A special maximum symbol error rate then gives a minimum ratio of d / 0.

Averaging the squared differences between the detector's input and output signals for a given carrier gives a value calculation of the variance oz which is represented in the same unit as the squared distance dzfor the constellation used. The range limits for d / G can also be given as dz / G2. maximum symbol frequency, and dzär a definite parameter for These limits will be constant for a given a certain concellation. This provides a very simple method for determining which constellation to use for a particular carrier, regardless of the different scaling in the multi-carrier transmission system in general.

Thus, the method for determining bit charge according to the present invention is: - independent of signal scaling in a multi-carrier transmission system; and - takes advantage of the fact that the symbol error rate is a function of the ratio between a variance that can be valued using the detector's input and output signals and a determined constellation parameter.

The purpose of channel coding is to reduce the bit error rate. The type of coding that should be used depends on the error pattern characteristics. Expected sources of error include random noise (which induces random bit errors), impulse disturbance (which induces error bursts) and clipping (which induces error bursts). 506 643 l0 15 20 h.) LJI 30 48 Faults caused by impulse disturbances mainly affect one or two bits per carrier. The probability of a single bit error on a carrier is always higher than the probability of 2 bit errors, which in turn is higher than the probability of 3 bit errors, and so on. This depends on the way in which the bits in the symbol are coded (ie Gray coding).

All coding depends on a synchronization to determine the start bit for the codewords and / or the "interleaving" blocks. In a system such as the MUSIC modem, "simple dead reckoning" will suffice, since a data flow slip can never occur without loss of frame synchronization, or bit set error setting. These errors necessitate a partial, or complete, system startup.

Channel coding will also include "interleaving" to increase the ability to correct burst errors.

"Interleaving" should be as deep as possible to obtain optimal function. The limiting factor in depth is the time delay introduced into the system.

The difference between time and frequency interleaving is of little importance because the coding and interleaving function is not sensitive to frame boundaries.

Reed-Solomon codes have the disadvantage that they are mainly burst error correction over a small number of bits (usually eight), a so-called symbol. Impulse bias errors generally introduce a "single-bit error" in some of the symbols.To take advantage of Reed Solomon codes, most bits must be concentrated on the "error prone" one, or a few , of the Reed-Solomon symbols. u: 20 30 (fl CD O \ 49 The system margin as such (in itself) is a kind of coding that uses the margin of each carrier as the redundancy of the symbol. This redundancy per symbol must be converted into a "split" ( the higher redundancy that can be used by a coding speed that this introduces can be used by certain types of convolutional codes.

The use of a convolutional code combined with "soft" information is therefore the optimal solution for a system with MUSIC channel characteristics. The convulsion code must be combined with interleaving.

It is possible to use a "top-level" Reed-Solomon code, or some other burst error correction code, eg Fire codes, to: detect / correct the remaining bit errors.

This is especially useful because these errors occur in bursts as a result of the decoding of the convolutional code.

The "System Controller" is based on a "micro controller", or signal processor, depending on capacity requirements, for the MUSIC system the processor can be placed externally.A PCI bus interface is used to connect the System Controller and the various ASICs that make up the modem.

The function of the System Controller is shown schematically in Figure 21, which shows the paths of interaction over a PCI bus, between the System Controller and the FFT chip, the data mapping and detection chip, and the coding and decoding chip. Functions performed by the system controller are: - management of "Control Channel Signaling"; - calculation of bit charge and energy charging factors; - updating of system parameters in real time; and - system monitoring. 506 643 Ix) k !! 50 System Controller, used for it The modem described here is programmable and accessible through an on-board JTAG interface.

As shown in Figure 22, in a modem context with modem described here, the two data paths operate independently on the same physical copper cable, terminating in the network (NU) unit (NOW) on the network side, and (NT) and the receiver Rx controlled by System Controllers. the night termination on the user side. Both the transmitter Tx System Controller calculates and updates, after start-up, the bit charge and energy charging factors_ This update must be done simultaneously with start from the same frame, on both the transmitter and receiver side.

The calculations are made and the update is initiated on the receiving page. The control channel, combined with the BSI clock, is used to ensure the synchronization of the update.

System Controller also monitors the system. Indications of system errors include the control channel starting to indicate errors, or receiving too many errors from the decoding channel unit.

The System Controller can initiate reboot at different levels; for example, go back to "idle mode", or do a full boot.

The control channel is a selected carrier that is only used for signaling between the two modems. The constellation on the carrier is initially 4 QAM and the data rate is approximately 16 kbit / s. The bit load can be changed to another constellation to increase the data rate.

The protocol on the control channel is partly based on HDLC for the physical layer. This means that the messages are packaged as a number of octets using "flag UI 20 I ~ J Un 506 643 51 sequence" and "bit-stuffing". A 16-bit "frame check sequence" guarantees that each message is received correctly.

"Flag-sequence", "bit-stuffing" and "frame check sequence" are handled in the hardware on the mapping and detection chip. The contents of the messages are handled by the System Controller.

The maximum message length is limited to 64 octets depending on the size of the buffers on the mapping and detection chip.

Higher level protocols may be based in part on CCITT Q.92l recommendations.

In MUSIC modem SC several different vectors are handled; these are shown schematically in Figure 23.

For the transmitter part, there is the bit charge and energy scaling vector. Corresponding on the receiver side is the bit charge, rescaling and equalization vector.

As previously described, the pilot carrier delivers a transmitter / receiver synchronization by transmitting and detecting a specific pattern. This clock is used by the system to synchronize changes in the transmitter and receiver vectors.

The time between the pilot synchronization patterns is called the Base Synchronization Interval (BSI) and is determined by the system response time, as shown in Figure 24.

This ESI is hardware dependent. Its length will not change, as the response time always remains the same.

When the system is running, there will be synchronization between the uplink generator and the receiver, through "base 506 643 20 52 sync interval uplink" (BSI-D), the same length but shifted half a BSI interval.

(ESI-U) and “base sync interval downlink” see Figure 25. These BSIs are of the exact SC at the NOW, or NT, will receive interrupts for both BSI-U and BSI-D .

For the NOW, there will be a transmission BSI-D interruption and a reception BSI-U interruption. By switching BSI-U with BSI / 2, the SC charge will be better distributed over the BSI period.

The bit charge vector provides the system with the modulation pattern for each carrier path. This is a vector that needs to be kept and updated at the exact same time to provide a flawless connection. By using ESI n, the vector changes synchronously on the transmitter and receiver side to on the receiver and transmitter side.

The bit charge factors, constellations used on each carrier path, are handled by two memories for reception and two memories for transmission on the mapping and detection chip. Each of the four memories contains a 4-bit word for each carrier (l024x4).

The System Controller identifies which of the memories to use to send and which to use after receiving from the next BSI interval.

The bit load factor can have values between 0 and 12, where 0 indicates an unused carrier; l-12 indicates the number of bits in the constellation 2 for 4QAM, 4 for 16QAM, 10 for 1024 QAM). (for example.

The energy vector pours information about how the carriers are scaled / scaled down on energy. This is a vector that needs to be updated synchronously, otherwise it will generate a distorted channel value calculation and bit error. will also use the scaling vector Un 20 506 643 53 as a pattern (mask) for canceled (canceled) carriers.

Scaling of the various carriers on the transmitter side is handled by a memory area on the FFT chipset. The memory consists of a 16-bi word for each carrier (1024 x 16). These values are multiplied by the vector of each carrier in the frequency domain (I and Q are multiplied by the value separately).

The memory is doubled to ensure a synchronous update. The System Controller points out which of the two memories will be used from the start of the next BSI interval.

A corresponding memory (doubled) is implemented on the receiver side: to rescale the carriers before symbol detection. If these memories contain a complex value for each carrier (32 bits / carrier), only the I value will be used for rescaling.

The scaling and rescaling factors have values between 0.5 and 2.0. The value O is used for carrier cancellation.

The smoothing vector is used to smooth the received frame according to the channel characteristics. This vector is updated periodically, independently of the other side, as the channel value calculation is calculated by the receiver.

Depending on the specific transmission characteristics of the carrier, it will be assigned one of the following modes: - standard carrier - this carrier transmits data according to the calculated bit charge value and is "transmitter scaled" and "receiver scaled"; 506 643 UI 20 k) Un 30 54 - canceled carrier - no energy is transmitted at this frequency and the scaling vector is therefore set to zero; or - poor carrier; SNR is too low to transmit any data and the bit charge is therefore set to zero.

In carrier mode 1 (CMI), the system operates normally.

The receiver continuously equalizes the channel.

Equalization changes are made for each new value calculation. Using the characteristic, the SC calculates the optimal bit charge factor. This value is transmitted to the transmitter using CCH, and a synchronous change is made.

In carrier mode 2 (CM2), the energy scaling / rescaling value is set to 0 to disable all output / input energy. The value of the bit charge vector is also set to zero to indicate that the carrier is disabled. For this carrier path, no channel value calculation can be made.

(CM3) zero for the bit charge factor. On the transmitter side, this means that in carrier mode 3, the receiver has calculated that no data can be transmitted, and therefore no channel value calculation can be made at the receiver. To avoid this, the corresponding carrier value is transmitted from the synchronization frame and makes it possible to perform a channel value calculation at the receiver. The scaling / rescaling value can be used to lower the output power.

The carrier modes are presented briefly in Table 4.

The basic functionality of the "startup" sequence in the system, ie "cold" and "warm" boot, will now be treated.

Initially, the power supply in the system is considered to be switched off at one or both ends, NU and NT. This occurs if the power supply is lost due to a power failure, UI 10 20 30 596 643 55 or by the user disconnecting (unplugging) the NT equipment. The most important thing to consider when starting up is, in addition to the connection function, to minimize the level of interference for other modems that use adjacent cables.

The different frame types used by the system are discussed below. 1. The synchronization frame is used for channel value calculation. This frame pours a specific modulation pattern for each carrier and thus enables easy channel value calculation. By having the modulation pattern described by a "random sequence", the cross-correlation within the frame is kept low, so that the frame correlation used for synchronization is improved 2. Data frame 1, (DF1), transmits "random data" on all carriers except four predefined ones carriers that transmit the control channel (CCH) in parallel. It is used in "start-up" when the CCH carrier is indeterminate and enables the receiver to select the least disturbed carrier, thereby guaranteeing the CCH connection. 3. Data frame 2 (DF2) conveys " random data ”on all (CCH). is used when the CCH carrier has been determined, and carriers other than one carrying the control channel The bit charge factors have not yet been set.

(DF3) bit load function to maximize bandwidth. Data frame 3 conveys data and uses A carrier is always dedicated to the control channel (CCH).

The system uses a special frame sequence, shown in Figure 26, at start-up and in idle mode, called the Start-Up Sequence (SUS). 506 645 Un 10 20 30 56 SUS can be composed using the different DFI and DF2. In the SUS frame sequence, the synchronization frames for the data frames are used, which are consequently called SUS1 and SUS2. channel value calculation.

After start-up, the synchronization frames are replaced with data frames, as shown in Figure 27, and the channel value calculation process shifts from the use of the synchronization frames to the use of the data frame. The type of data frame for this sequence is DF3.

At system startup, neither side of the modem, NOW and NT, sends any energy across the copper pair.

The default setting for each side is at this stage to drive the receiver, leaving the transmitter "dead".

The receiver tries, on each side, to perform a frame correlation to detect a frame start. This correlation is run through a threshold function that gives the receiver a distinct indication of when the other side starts transmitting. It is this indication that serves as a "wake-up" signal.

The “wake-up” signal is used only by the NT side. If the start-up decision is made on the NU page, the system goes directly as described to the set-up sequence below.

This part of the start-up procedure is subjected to "time out" if a transition to the set-up sequence is not detected.

The basic wake-up signaling for the modem is shown in Figure 28. Initially, both modems search for frame correlation. One of the modems, on the right in Figure 28, sends a "wake-up" signal in the form of an SUS1. The second modem detects frame correlation and starts the set sequence described below.

IO 20 30 506 643 57 When the "wake-up" state has been passed, the web page (NOW) initiates the set-up sequence.

The set-up sequence will now be processed. This set-up sequence starts after the web page has detected a wake-up signal, or the network initiates the set-up.

The first step in the set-up sequence is shown in Figure 29. In this phase, NOW starts to send the SUS1 pattern. NOW repeatedly sends a timing advance (TA) setting, with TA = 0, on the CCH. The master clock in the system is now the NOW transmitter frame and the sample clock in NOW.

The pilot is sent continuously.

The NT-mc :: agar page, which looks for frame correlation, detects frames and can retrieve the frame and sample clock. It now starts the channel value calculation, which at the current speed of synchronization frames makes an accurate value calculation within 300 ms. Using this value calculation, the receiver starts polling the predefined CCH carriers and, at message receive, selects this carrier for the CCH. The NT transmitter now starts with TA = 0 for local synchronization and sends receipt (acc.) On The CCH carrier for each received TA selection message, repeating the received TA value, also shifts the outgoing pilot with BSI / 2 from the incoming pilot, so that the SC charge is distributed over time.When the NOW detects the frame correlation, the transition is made to step 2 of the set-up sequence.

Thus, step 1 of the set-up sequence begins with transmitting a SUS1 and a At the transmitter, in the network modem, TA message with TA = 0 in periodic intervals. receiving this, the receiver in the terminal modem will: - perform frame correlation and retrieve the frame clock; 506 643 IJ UI 30 58 - start FFT treatment; enable pilot decoding; - recover the BSI; enable channel value calculation; - select a CCH; and - decoding the TA selection message.

The transmitter in the terminal unit then sends a receipt (acc.), SUS1, a TA = O message and a pilot shifted by BSI / 2. The receiver in the network unit waits for the frame correlation Step 2 in the set-up sequence, see Figure 30, begins with the NU page now calculating a "timing advance" value (TA) .The CCH message changes to the new, corrected TA value.

When the NT page receives the new TA value: it changes the local synchronization and continues to send the acknowledgment message, with a new TA value, for each TA selection message.

In the NOW receiver, the frame clock is lost, due to the NT transmitter changing the frame clock, and the unit. After the frame clock has been recovered, the CCH is decoded and, in the event of acknowledgment, needs to correlate. detection, which contains the new TA value, the system terminates the TA message and goes to the third step of the set sequence.

Thus, step 2 of the set-up sequence starts with the transmitter in the network unit, NOW, transmitting a TA message containing the correct TA t, together with a SUS1, transmitted from the transmitting terminal. say X, in response to SUS1 and the TA = 0 message that 59 Terminal Unit, NT: - receives the new TA message; - corrects the outgoing frame clock; and - sends a receipt SUS1 and TA = X. 5 The network unit, NOW: - performs frame correlation; - retrieves the frame clock; - starts FFT data processing; - enables pilot decoding; w - BSI resets; - enables channel value calculation; - selects a CCH; and - decodes the message.

The last set-up sequence, step 3, see Figure 1:31, handles the CCH selection for uplink and downlink. For the uplink, the NU receiver has selected the most suitable carrier and sends a CCH message containing this selection to the NT page. The message is sent repeatedly until it receives a receipt (acc.). On the NT side, the receiver decodes the CCH message and terminates SUS1 and transmits an SUS2, ie terminates parallel CCH transmission by transmitting only the CCH on the selected carrier.

The uplink CCH carrier has now been configured. For the downlink, the same step is performed in parallel, initiated by the NT 506 643 60 page after receiving the first CCH selection message from NOW.

Thus, in step 3, the network unit will: - send the selected CCH for the uplink; - wait for a receipt; and - exit the CCH message.

The terminal unit: - receives the CCH selection for the uplink; - terminates SUS1; and starts SUS2; - acknowledges each CCH selection.

The network unit: receives the CCH selection for the downlink; terminates SUS1; starts SUS2; and acknowledges each CCH selection.

The terminal unit: - sends the selected CCH for the downlink; - waiting for a receipt; - ends the CCH message.

When these steps have been taken, the modem has reached idle mode, the bit load factors now change according to channel characteristics transmitting SUS2. Using CCH, DAS transmission can also be started.

Un IQ UI 506 645 61 The VDSL modem can have interfaces to different network elements, depending on the physical location of the modem, ie in the space for access nodes or in customer premises. In the customer premises, the VDSL modem can interface with an active network termination equipment_ At the access node, the VDSL modem will interface with an access-specific interface, see Figure 32, which shows a logical view of the network elements that interface with the VDSL modem.

The VDSL modem can be physically integrated with the network termination equipment, and the VDSL modem at the access node can be physically located in the cabinet in which the access node is located.

NT (interface A1) and the access node (interface A2) require a layer 1 frame format of the VDSL modem. Integrated in the layer l-frame, there are, apart from the frame head and the payload, a number of information fields for handling and control information. These handling and control fields include various alarm indicators, such as SDH alarms, e.g. AIS (valid only if SDE is taken all the way to customer premises) bit error rate measurements for performance monitoring, indications of whether synchronization is poor, or lost, equipment management alarm for loss of power and too high temperature, etc. Management fields also include activation of various loop tests on the modem, for operation and maintenance purposes. 506 645 62 TABLE l System parameters for the overall system Orthogonality between modems No Duplex method Separate bands Frequency intervals between up- Depending on duplex filter / down data flow characteristics Net bit rate, - upstream 2 Mbit / s - downstream 13 or 26 Mbit / s, Coding dependency - downstream Coding dependence Cable length <1300 meters Cable bandwidth 10 MHz Modulation, single carrier - upstream O-4096 QAM - downstream O-4096 QAM Number of carriers, total 1024 Bandwidth for each carrier 9, 77 kHz Cyclic prefix DMes Module (sample VDSL Signal power -60 dBm / Hz 586 643 63 Bit error rate 104 Interleaving delay 0.5 ms (Interleaving delay) System margin 6 dB CCI-I - bandwidth l carrier, minimum 16 kbit / s - protocol HDLC 'Sample clk 20 MHz il0ppm Frame clk 20 Mz / (2048 + ll2) = 9.19 kHz 506 643 64 TABLE 2 System parameters for the Transmitter Interleaving - depth 2 x frames - delay 0.5 ms DAC resolution 84 dB Clip Clipping algorithm No (Clipping algorithm) IFFT - type Real - points 2048 - resolution 16 bit LP filter LP 10 MHz Biz charge Yes, 0, 2, 4, 6, 8, 10, 12 bit Energy charge Yes, 4 bit BSI distance 1 s TABLE 3 506 645 65 System parameters for the Receiver ADC resolution 66 dB FFT - type Real - points 2048 - resolution 16 bit LP filter LP 10 MHz Synchronization - jitter <0.5 ns VCXO 125 ppm, 10ppm / V sensitivity - DAC 18 bit, range 0-5 V - resolution l / 100 of a sample 506 645 66 TABLE 4 Carrier mode Mode Transmitted Bit charge Smoothing Scaling CMl Data 2 - 12 Yes Yes cM2 No o No o CM3 Sync info 0 Yes, sync Yes, low

Claims (23)

U | 10 506 643 67 PATENT REQUIREMENTS A multi-carrier transmission system using high order QAM constellations orthogonal carriers for transmitting a plurality of bits per carrier and symbol, said system including digital receiver and transmitter units, said receiver unit including a symbol detection unit, characterized said system is arranged to determine a parameter for each individual carrier, said parameter indicating a deviation of a received signal from a corresponding constellation point; comparing the parameter with an upper and lower limit; and, if the parameter is outside said limits, changing the constellation used to modulate the carrier to an adjacent constellation. A multi-carrier transmission system according to claim 1, characterized in that said symbol detection unit is used to determine the parameter of a carrier. A multi-carrier transmission system according to claim 1, characterized in that where d is the shortest distance between adjacent constellations, o is 1, or claim 2, said parameter is a ratio dz / oz, a standard deviation, and ozär the variance of the deviations of input and output the output values in said symbol detection unit. A multi-carrier transmission system according to claim 3, characterized in that said symbol detection unit is arranged to associate an input signal applied thereto, in an area around a specific constellation point, to a distinct value for this point, said distinct value being the value of the output from said symbol detection unit, and measuring 506 645 UA l0 25 68 differences between said input and output signals for the carrier, wherein said differences are squared and averaged to provide a calculation of a value of 02. A multi-carrier transmission system according to claim 4, characterized by being ozrepresented in the same units as the squared distance dz of the constellation used to modulate the carrier. A multi-carrier transmission system according to claim 4, or claim 5, characterized in that the differences between the input and output values in said symbol detection unit vary in value, and in that said variance is a measure of the carrier disturbance and is directly related to SNR. A multi-carrier transmission system according to claim 6, characterized in that said variance is proportional to SNR. A multicarrier transmission system according to claim 6, or claim 7, characterized in that an upper limit of said variance for a specific constellation is determined by a maximum allowable symbol error rate, and in that a lower limit of said variance is equal to an upper limit for the next largest constellation. A multi-carrier transmission system according to claim 8, characterized in that a specific maximum symbol error rate gives a minimum ratio of d / G. A multicarrier transmission system according to any one of the preceding claims, characterized in that said system is a DMT system. Un 10 * Ju Ik) Un 506 643 69 A multi-carrier transmission system according to any one of the preceding claims, characterized in that said system is a DMT-based VDSL system. In a multi-carrier transmission system using orthogonal carriers with high order QAM constellations for transmitting a plurality of bits per carrier and symbol, said system including digital receiver and transmitter units, said receiver unit including a symbol detection unit, a method of determining a constellation (bit charge) for each individual carrier path, characterized by the steps of determining a parameter for each individual carrier path, said parameter indicating a deviation of a received signal from a corresponding constellation point; comparison of the parameter with an upper and lower limit; and, if the parameter is outside said limits, changing the constellation of the carrier to an adjacent constellation. A method according to claim 12, characterized in that the determination of the parameter for each individual carrier path is performed by said symbol detection unit. A method according to claim 12, or claim 13, characterized in that the parameter is a ratio dz / G2, where d is the shortest distance between adjacent constellations, U is a standard deviation, and ozar the variance of the deviations of the input and output signal values in said symbol detection unit. A method according to claim 14, characterized in that the determination of the parameter of a carrier path includes the steps of associating an input signal to said symbol detection unit, in an area around a specific constellation point, to a distinct value for this point, where said distinct value is the value of the output 506 lJ | In said symbol detection unit; and measuring differences between said input and output signals of said symbol detection unit for the carrier, wherein said difference is squared and averaged to provide a calculation of the value of O 2. A method according to claim 15, characterized in that it is ozrepresented in the same units as the squared distance d2 of the constellation used to modulate the carrier. A method according to claim 15, or claim 16, characterized in that the differences between the input and output signal values in said symbol detection unit vary in value, and in that said variance is a measure of the disturbance on the road and is directly related to SNR. 15. k ä n n e t e c k n a d that the said variance is A method according to claim 17, proportional to SNR. 19. characterized in that an upper limit of said method of claim 17, or claim 18, the variance of a specific constellation is determined by a maximum allowable symbol error rate, and in that a lower limit of said variance is equal to an upper limit of the second largest constellation. A method according to claim 19, characterized in that a specific maximum symbol error rate gives a minimum ratio of d / 0. A method according to any one of claims 12 to 20, characterized in that said system is a DMT system. 506 643 '71 A method according to any one of claims 12 to 21, characterized in that said system is a DMT-based VDSL system. A multi-carrier transmission system using orthogonal carriers with high order QAM constellations to transmit a plurality of bits per carrier and symbol, said system including digital receiver and transmitter units, said receiver unit including a symbol detection unit, characterized by said system uses a method according to any one of claims 12 to 22 to determine a constellation (bit charge) for each individual carrier.