WO2009018980A2

WO2009018980A2 - Radio device with new cifdm modulation method

Info

Publication number: WO2009018980A2
Application number: PCT/EP2008/006382
Authority: WO
Inventors: Oliver Bartels
Original assignee: Oliver Bartels
Priority date: 2007-08-03
Filing date: 2008-08-01
Publication date: 2009-02-12
Also published as: WO2009018980A3; EP2174440A2

Abstract

The invention relates to a radio device having a novel comb-interleaved frequency division multiplex (CIFDM) modulation method that combines the advantages of known OFDM modulation with the advantages of conventional carrier modulation. In each case, the data stream to be transmitted is divided into two multi-carrier blocks, which are sent in an interlocked and frequency- and time-offset fashion. For each single sub carrier of a block, a classic modulation with a matched filter system is used; the guard interval is omitted. The spectral efficiency is considerably increased by an artificial zero crossing being introduced with each modulation symbol. During each zero crossing, the other block is transmitted and separated again in the receiver via a synchronous switch.

Description

RADIO UNIT WITH RENEWABLE CIFDM MODULATION PROCEDURE

Short Description:

The invention relates to a radio with a novel Comb Interleaved Frequency Division Multiplex (CIFDM) modulation method, which combines the advantages of the known OFDM modulation with the advantages of conventional single-carrier modulations. In any case, the data stream to be sent is split into two multicarrier blocks, which are sent into each other with frequency and time offset. For each sub-subcarrier of a block, a classic modulation with a matched filter system is used; the guard interval is eliminated. The spectral efficiency is significantly increased by introducing an artificial zero crossing in each modulation symbol. During this zero crossing of the other block is transmitted and separated again via a synchronous switch in the receiver.

The invention further relates to a modulator for generating a multi-tone signal for the novel Comb Interleaved Frequency Division Multiplex (CIFDM) modulation method in the specific expression of a largely hexagonal arrangement of the subcarriers. The modulator makes it possible to achieve the optimum hexagonal packing density of the subcarrier symbols in order to achieve high spectral efficiency while maintaining extremely high immunity of the multitone signal.

The invention also relates to a modem which uses a multitone transmission method with band-limiting Gauss or Matched Filterbänken, with an additional device for measuring and reducing crosstalk interference by using specially modulated especially with pseudo random or gold codes subcarrier for detecting crosstalk.

Description:

The invention has for its object to transmit a broadband data stream as a radio signal as trouble-free as possible over a greater distance. In this case, in particular reflections caused by multipath signals (multipath) can be reconverted undisturbed in the original data stream again. A task comparable to this task consists in the transmission of a broadband data stream via a simple copper cable, such as e.g. a subscriber line, also on this there are reflections on discontinuities such. Terminal blocks.

For data transmission by radio, the data is usually modulated on a carrier, the associated methods are well known from the literature, it is here by way of example on the state of the art very detailed descriptive work of Karl-Dirk Kammeyer, Messaging, 3rd edition, Teubner 2004, referenced.

In addition to the classical methods of modulation of a single carrier, which are advantageous in terms of short latencies, but have the disadvantage that they with reflections and multipath reception over long distances with high data rates only with great effort, for. can handle many coefficients by an adaptive receive filter (equalizer), the OFDM method has largely prevailed in the current state of the art.

The abbreviation OFDM stands for Orthogonal Frequency Division Multiplex, whereby a multicarrier signal with a large number of subcarriers is used, which is generated by means of an orthogonal transformation - generally an inverse Fourier transformation - the transformation takes place in the context of a digital signal processing. The result of the transformation can then be mixed analogously with a main carrier, yet only the subcarriers synthetically generated by mathematical transformation show up in the transmission signal.

The problem of reflections is solved in OFDM in that on the one hand the transmission of the transformed data blocks in each case a guard interval (protection area) is connected upstream, which is typically a repetition of the last sample values of the transformed data block to be transmitted.

As a result, it is ensured during the transmission that, when viewing a single subcarrier, a continuous transition between the protected area and the actual transformed data block is ensured, which does not trigger any new step responses in the time level as a result of a change in modulation or new echoes. This leaves only the step response generated from the transition of the previous symbol to the new guard interval or the echoes arising at this discontinuity, which dead-end within the guard interval.

The problem of remaining interference can then be remedied in a second step in the OFDM receiver by a greatly simplified equalization. Since OFDM usually uses an orthogonal inverse Fourier transform, there is, of course, exactly one complex number per subcarrier after backtransformation in the receiver. Because the transformation is linear and multipath is within the system boundaries is merely a linear superposition of information transported over different paths, and also because the time shift due to the convolution theorem and the design of the guard interval after the return Fourier transform into the frequency plane corresponds to only one scaling, can be achieved by a single complex multiplication per subcarrier a substantial equalization be made. The determination of the coefficients for equalization can be carried out, for example, by means of pilot carriers; likewise, certain subcarriers can preferably be selected, see, for example, DE 19827514Al.

The OFDM method impresses with its simplicity and effectiveness and is therefore used today in the state of the art in many modern radio systems such as IEEE 802.11 a / g WLAN, IEEE 802.16d / e WiMAX or wired for ADSL.

The disadvantage here, however, that on the one hand, the method in echoes whose duration exceeds the length of the guard interval, can fail completely, which is why for transmission over longer distances an increasing proportion of bandwidth must be sacrificed to the guard interval, and on the other hand it is not a matched filter system, which is why the achievable throughput is not optimal with poor signal / noise ratio.

Furthermore, there is the problem of latency, which results from the necessary summation of the data to an OFDM symbol, because the OFDM symbol duration should be a multiple of the duration of the guard interval for reasons of spectral efficiency. Especially with larger transmission distances with longer transit time differences and long echo propagation times, this means a considerable undesired increase in the latency time.

Already at the time of introduction of the OFDM method in the prior art, made possible by providing the fast digital circuits of digital signal processing suitable for fast calculation of the necessary Fourier transformation, the direct generation of real multicarrier signals by means of filter banks was discussed as an alternative. An example is provided by DE4208808A1. From US4131766 and US4799179 (to the theory see also Norbert fly, multi-rate signal processing, Teubner 1993, chapter 8.6) are also DFT polyphase filter banks are known, with which a realization of a variety of similar FIR filters extremely efficient means of a multiplexer, a set of FIR filters, the cost compared to a pure FIR filter bank is reduced both by the rate of the samples as well as the number of coefficients by the multiplex factor, and a fast inverse Fourier transform can be. In EPl Ol 621 IBl the application is shown in an analogue receiver.

By using efficient filter banks to be calculated, the data stream is divided according to this method to a plurality of individual carriers, of which modulates in contrast to the OFDM each individual with a low data rate, but still in the sense of a classic single-carrier transmission system conventional and continuous without guard interval becomes. As a result, runtime differences due to multipath reception act directly as intersymbol interference on the transmission, but since each carrier operates at a very low data rate, correspondingly large temporal differences are required, so that the intersymbol interference is felt at all, it can also with an adaptive filter with only small coefficients or Trellis encoding with Viterbi evaluation can be easily eliminated.

A particular problem with this, however, is the poorer utilization of the spectrum by the multicarrier signal thus generated in addition to the higher computational effort, which is only incidental to today's circuit technology.

In contrast to the OFDM signal, which provides the orthogonal Fourier transformation for each complex amplitude and phase specification of a single carrier in the frequency plane exactly a complex - corresponding upper and lower sideband - or two real-valued samples in the time plane and thus without guard interval, the Nyquist In the case of multibarrel transmissions with filter banks, a conventional modulation is used per carrier, which concentrates the signal power in certain areas of the spectrum and covers other areas, especially in the sidebands, with far less or almost no power. As a result, there is sub-critical sampling and analysis on the receiver side with the associated loss of spectral efficiency.

WO 2004 / 014034A2 likewise discloses an approach for the division of a QPSK signal into at least two sub-channels with mutually offset frequency bands with classical channel filters. Although this already results in improved QPSK signal spectral utilization, but this far from the known OFDM approaches, this method is therefore primarily economically only for transmission paths such as satellite connections, in which the use of a purely phase-modulated signal due to the special Properties of the transmission path is mandatory. The object of the invention is therefore to find a modulation method which combines the advantages of the high spectral efficiency of the OFDM method with the robustness of a multicarrier method with conventional modulation of the individual carriers.

The problem is solved by the invention described in claim 1 devices or assemblies whose function will be explained below with reference to an embodiment. Figure 1 shows the appropriate signal processing in graphical form, whose individual elements in Figure 1 are referenced below with numbers in parentheses.

A microprocessor or digital signal processor according to dependent claim 14 takes over the data to be transmitted from the source and divides them by software into individual data blocks (1). The data blocks are provided with a fail-safe forward error correction code according to dependent claim 8, e.g. a turbo code which adds redundant information to the actual data bits for later error correction in the receiver.

Then, the blocks are processed by the processor for modulation (2), this is done, for example, a selection of two bits for a QPSK modulation according to dependent claim 6 per single carrier. From each one bit group, a complex number is selected as the baseband sample corresponding to a table or formula, which is then fed as a transmit sample into the channel filter bank. If this is designed as in the picture for four subcarriers, results in the example, an information amount of two times four bits per data block minus according to dependent claim 10 to be transmitted sub-carrier pilot or spectral zero or backing.

Then, the data blocks are prepared for transmission, for this purpose, in each case the even data blocks are given to one and the odd data blocks to another filter bank (3). According to dependent claim 3 is in a particularly advantageous embodiment of the invention by a multiplexer, reduced FIR filters and inverse fast Fourier transform formed polyphase filter bank, which simultaneously makes the necessary upsampling to the Ausgangsabtastrate and thus the classification of the subcarriers in the spectrum. This allows large filter banks for e.g. 64 to 1024 subcarriers realize.

Since the filter banks have in contrast to a pure inverse Fourier transformation over an arbitrarily adjustable impulse response, according to the dependent claim 4, a matched filter system be formed. This allows an optimization of the system behavior with noise present in the channel or other interference levels while at the same time largely avoiding intersymbol interference. For the correct modulation of the subcarriers, it is sufficient to give the individual baseband samples of successive (even or odd depending on the filter bank in the example) data blocks that belong to a subcarrier, as input values in the polyphase filter bank and, if necessary, to adjust the sampling rate in addition To add zero samples, the necessary interpolation and signal shaping take over the FIR filters, which also act as Gedächnis the filter bank.

At the output of the respective filter bank for the even or odd data blocks, the respective multicarrier signal is now directly in the time plane. According to the main claim, one of these signals is first frequency-shifted by half a subcarrier frequency spacing, for example by complex multiplication of the samples with a local oscillator signal (NCO) numerically generated by direct digital synthesis in the complex number space (4). The latter is obtained by sampling successively taken from a sine or cosine table, since in the specific case usually an integer division ratio is present, the signal generation is limited to a purely sequential cyclic removal without phase accumulator.

Furthermore, the output signal of the other filter bank is now offset in time by half a symbol duration (5), this can be done, for example, by means of a ring buffer or a FIFO queue.

Thereafter, the output signals are added up to form a data stream, resulting in a multidimensionally intermeshed transmission signal according to the invention (FIG. 2).

In a particularly advantageous embodiment of the invention according to dependent claim 5, each output signal before the summation (7) via a for example from a table (6) generated periodic function is scaled so that the output power of one signal is then a maximum, if the other at zero lies. However, this need not be handled with each signal in such a way, for example, according to dependent claim 11, a dynamic selection of the function can be made. In addition to requirements regarding the channel response, there is also a choice for reducing the ratio of the peak transmission power to the average transmission power (PAPR, peak to average power ratio). To realize the latter, for example, various scaling functions can be applied in parallel to the output signals to realize a conceivable variant, the results are then time-delayed to perform an evaluation of the variants according to the best PAPR and then the selected one of the time-delayed signals on the Transmitter or line driver to give. Due to the time delay, the selection then takes effect almost retroactively.

Optionally, both the individual data streams and their sum prior to transfer to the D / A converter (8) for - as in conventional systems, for example, a Lokaloszilator (9), vector modulator (10) and (PA) transmission amplifier (11) with existing antenna - transmitter still subjected to a predistortion or another upsampling to an intermediate frequency, the latter is then advantageously separated for the individual data streams of the individual filter banks vornehmbar if the necessary frequency offset between the two filter banks together with the conversion to the intermediate frequency and a last bandpass filtering is realized in one step, corresponding modules for an up-conversion of baseband data streams (Digital Up Converter) are commercially available in the prior art.

Furthermore, it is conceivable that the scaling functions should also be based on the meaning of the data to be transmitted or on its freedom from error or on the available redundancy.

The embodiment according to dependent claim 12 provides for this purpose that, in the case of a packet-oriented (IP) data transmission, either the data packets belonging to a low quality service class are in turn assigned transmit data blocks, which are transmitted via a filter bank, which with reduced power into the transmission signal received. Alternatively, it is possible to timestamp packets which indicate the desired arrival time of the packet at the receiver or to derive these time relationships from the position of the packet in a queue.

In the second case, the sender can then attempt to pre-transmit on transmission data blocks with reduced power or transmission quality those packets which are not yet ready for transmission according to the time assignment. If the packet reaches the receiver and the error control code evaluates it as error-free or correctable, this can be communicated to the sender and so a later retransmission via transmit data blocks with full power and high transmission quality are avoided, thereby bandwidth is saved in the case of good radio transmission ratios and the latency shortened.

Based on a timestamp within the packets can then be simulated to the actual customer of the data, a continuous flow of data according to dependent claim 13, in which the packets are cached in the receiver and delivered only at the specified time. As a result, a high data throughput is ensured especially with window-oriented protocols such as TCP / IP.

In order to be able to receive and demodulate the data thus transmitted again, a special demultiplexer is used in the receiver according to the particularly advantageous embodiment of dependent claim 2, which is synchronized to the symbol clock. This makes a soft weighting of the incoming samples, which, and this is the crucial trick at this point, simulates a zero crossing in the example QPSK symbol whenever the incoming samples are each destined for the other analysis filter bank.

Since, in principle, a sequence of 180 degree phase jumps of the individual subcarriers in a matched filter system is properly transmitted, the power drop resulting from a 180 degree transition can also be artificially realized for all other phase jumps, thereby ensuring that at this point the spectrum is available for transmission of the other multicarrier symbol class from the other filter bank. Since the individual (unmodulated) subcarriers of the one class are placed in the filter edges of the matched filter of the other class by the frequency offset, they contribute little to the interference between two multicarrier symbol classes due to the temporal and frequency offset. The effect can be further enhanced by inserting zero sample values in a subcritical filter bank or by the already described weighting of the output signals of the filter banks according to dependent claim 5.

At this point, the special performance of this transmission method compared to the prior art shows: On the one hand, a very high spectral efficiency comparable to classical OFDM methods offered by the nesting of multiple multicarrier symbol classes in the transmission signal, it can in this sense quite by an orthogonal frequency division although a more complex transformation across multiple symbols is used. Figure 2 shows the composite of two partial spectra resulting spectrum. On the other hand, in this transmission method can be dispensed with the guard interval, as a classical equalization is only a little expensive due to the long in relation to the signal delay symbol duration, according to dependent claim is sufficient for this purpose an ordinary trellis-Viterbi combination completely. The bandwidth for the guard interval is thus completely available for data transmission, there is also no risk of failure of the transmission method for large propagation time differences in multipath reception and interference on adjacent carriers are easily eliminated by the frequency response of the classical filter, while in the ordinary OFDM as a result the sin (x) / x function of the filter curve of a single carrier, which inherently results from the Fourier transform, can penetrate directly to even more distant carrier.

A particular advantage of the invention is the possible choice of differently wide filters for individual subcarriers for the purpose of different symbol rates for data of different service classes according to subclaim 9.

For example, a set of high symbol rate carriers may be used for the bandwidth allocation protocol in a point-to-multipoint transmission system, thereby speeding up the allocation of bandwidth and reducing overall latency. Another possibility is the rapid transmission of information for confirmation of successfully transmitted data or for the rapid transmission of repeat requests (ARP / HARP). If necessary, the data sent again if necessary can then be combined with the originally transmitted data for the purpose of error correction; such a soft overlay is particularly suitable for turbo codes.

If further pilot carriers are transmitted according to subclaim 10, they can also be used as an alternative for preamble in packet-oriented transmission or cyclically inserted sync word for exact synchronization of the symbol clock between transmitter and receiver, which for the unit according to dependent claim 2 for separating the input data streams in the receiver necessary is.

In this case, it is advisable during the transmission of the preamble or the synchronous word at least on selected subcarriers to dispense with a parallel emission of more multicarrier symbols of other classes to use the resulting power dip in (possibly synthetically enforced) zero crossing to recognize the symbol boundary can. Furthermore you can According to dependent claim 10 also orthogonal pilot subcarriers can be used in another class, for example, to correct an I / Q imbalance or an I / Q offset. A use of the pilot subcarriers for Doppler correction is also conceivable. Since a pure frequency offset is produced by the Doppler effect, the correction can be carried out directly by reversing the frequency shift of a symbol class in the receiver and the optionally combined transmission of the intermediate frequency into the baseband by means of Digital Down Converter.

In addition, the use of the pilot subcarrier offers for a correction of the channel impulse response, for this purpose, the coefficients of the FIR filter of an analysis filter bank constructed according to dependent claim 3 can also be corrected adaptively. Of course, this method does not preclude a subsequent correction of the samples of the individual subcarriers according to complex multiplication as in classical OFDM.

Alternatively, a differential modulation according to dependent claim 6 can be used so as not to give rise to the need for an absolute phase reference.

Radio devices according to the present invention can be combined with single-frequency radio transmission methods, the symbols are simply transmitted by different transmitters synchronously with the same data on the transmitter side, the correction takes place virtually automatically due to the long symbol duration. In the reverse direction of transmission, for example, the procedure according to DE102004013701A1 find use.

It is likewise possible to carry out a combination of the invention with MIMO transmission methods and thus to radiate different classes of multicarrier symbols via different antennas or on different polarization planes or by beamforming by suitable selection of relative phases of corresponding symbol classes and to recombine them in the receiver via a suitable transformation or matrix multiplication to lead.

Based on the pilot information, this transformation can then be optimized mathematically for the respective transmission conditions.

The use of the invention is not limited to the wireless transmission, the use for data transmission on copper subscriber lines for a DSL service is just as conceivable as the transmission of data under water by means of ultrasound. The invention further relates to a modulator for CIFDMultiton modulation methods with hexagonal symbol arrangement.

The invention relates to a modulator for generating a multi-tone signal for the novel Comb Interleaved Frequency Division Multiplex (CIFDM) modulation method in the specific expression of a largely hexagonal arrangement of subcarriers. The modulator makes it possible to achieve the optimum hexagonal packing density of the subcarrier symbols for the purpose of achieving high spectral efficiency while maintaining an extremely high immunity of the multitone signal.

For data transmission by radio, the data are usually modulated on a carrier, the associated methods are well known in the literature, it is here by way of example on the state of the art very detailed work by Karl-Dirk Kammeyer, news, 3rd edition, Teubner 2004, referenced.

In addition to the classical methods of modulation of a single carrier, which are advantageous in terms of short latencies, but have the disadvantage that they with reflections and multipath reception over long distances with high data rates only with great effort, for. can handle many coefficients by an adaptive receive filter (equalizer), the OFDM method has largely prevailed in the current state of the art. For further details regarding the advantages and disadvantages of the prior art reference is made to the work of Kammeyer and to DE 102007036828.

In DE 102007036828 the novel modulation method CIFDM is introduced, which is characterized in that it has the advantages of the high spectral efficiency of the OFDM method the robustness of a multicarrier method combined with conventional modulation of the single carrier. A three-dimensional representation of the multicarrier signal generated by this method can be found in Figure 3. It shows that an optimal packing density of the subcarrier results in any case if they result in a hexagonal arrangement of equilateral triangles in the time and frequency axis as in Figure 4.

In this case, this distance statement refers to units which are related to one another according to the time-bandwidth product and which correspond to corresponding samples in a data stream representing the signal. It is obvious that in this case, unlike a conventional Fourier transform for the known OFDM method, the fixed number of transformed samples results in a symbol duration divided by the number of samples transformed at the block, rather it is indicated by the indicated one Geometry required to adjust the symbol duration to the vertical in the triangles in Figure 4.

The object of this invention is to construct a suitable modulator for this, the difficulty is to be considered that the number of samples at the output of the modulator per symbol is a non-integer multiple of the input values of the transformation due to the adaptation of the symbol duration.

The problem is solved according to the invention by the modulator described in claim 16, whose function will be explained below with reference to an embodiment. Figure 5 shows the appropriate signal processing in graphic form, whose individual elements in Figure 5 are referenced below with numbers in parentheses.

A microprocessor or digital signal processor according to dependent claim 24 takes over the data to be transmitted from the source and divides them by software into individual data blocks (101). The data blocks are provided, if necessary, with a fail-safe code for forward error correction, which adds to the actual data bits redundant information for later error correction in the receiver.

The blocks are then prepared by the processor for modulation (102), this being done, for example, a selection of several bits per subcarrier for a QPSK or QAM modulation. From each bit group, a complex number is formed as a baseband sample corresponding to a table or formula. This number is an input value for a subsequent orthogonal transformation, in particular inverse Fourier transformation provided.

According to a particularly advantageous embodiment according to dependent claim 21, a phase adaptation of the baseband sample values corresponding to the respective start time of a symbol now takes place. According to the subclaim 17, the sub-symbols of the two symbol combs in the example are to be distributed in a hexagonal pattern according to FIG. It follows that the number of samples in the time plane is a non-integer multiple of the number of input values. However, an input value-activated sinusoidal tone of a subcarrier as a result of the inverse Fourier transformation (104) would be disturbed by this time offset in each case discontinuously and thus sensitively.

The phase matching provides that the sample at the start time of a symbol corresponds to the phase of the sine wave that would result if the sine tone of the last symbol had continued until then, assuming, of course, a constant input value.

The phase matching is thus achieved in the example by setting the extension of the symbol duration to the symbol length resulting from the transformation length - where a ratio of one corresponds to a phase angle of 360 degrees -, hence the necessary phase angle for the correction of the fundamental frequency of the Fourier transformation is derived, this multiplied by the index of the input value - with index start value zero for the DC component - and then the 360 degrees modulo is formed.

According to dependent claim 22, a complex number can now be calculated from the resulting phase value by means of the CORDIC algorithm; the input value for the inverse Fourier transformation is now multiplied by this number in the complex number plane. In this case, the CORDIC algorithm, which is inherently incremental in nature, can also be integrated into the calculation of the phase adjustment values; the phase adaptation can be carried out both on the basis of absolute and, for example, the previous symbol of relative phases. Furthermore, it is advisable to buffer the phase adjustment values for recurring phases.

The thus phase-corrected input values are now supplied to the orthogonal transformation (104), preferably an inverse Fourier transformation in the form of the fast Fourier transformation (FFT) is used here. In addition to the FFT method of Cooley-Tukey, the FFT configuration of Pease with a fixed permutation is particularly well suited for fast hardware implementation. The optionally statically or dynamically scaled within the inverse FFT arithmetic unit output values are then stored according to the invention in a memory.

In a simple case, the storage takes place in each case with a gap of one block spacing, as indicated in FIG. 5 (105), in order to be able to mask the read pointer for the following multiplexed FIR filter according to subclaim 18.

The masking causes a repetition of the data, but with the application of new coefficients, with equally progressive coefficients and data pointers of the filter. Only then is it possible to correctly synthesize the number of total output values corresponding to the number of input values and thus also output values of the inverse FFT in a non-integer ratio in the sense of the FIR filter equations.

Of course, advanced implementations can reuse the resulting memory gap, for example, by adjusting the bits of the read data pointer above the mask by means of a barrel shifter so that when the FFT output data is stored, they can be continuously written.

Alternatively, according to dependent claim 19, a modulo arithmetic unit can be used to calculate the read pointer. In both cases, the masking as well as the modulo arithmetic unit, further offset values can, for example, be added for different send combs or channels before the read or write memory access, if required.

In the example, the output values of the FFT thus stored are now fed to a multiplexed complex FIR filter which, according to dependent claim 23, can use the complex coefficients to incorporate the frequency offset of the combs required for this modulation method against each other in one go.

The filter receives its coefficients, as shown in FIG. 5, block-by-block from a plurality of data blocks obtained by inverse FFT, one complex sample each comprising one data block with a precisely defined coefficient from the coefficient memory (106). multiplied (107), the sum of these products is formed in an adder (108) and is then available at the modulator output as the output value for digital-to-analog conversion with or without further processing. Further processing, for example, depending on the subsequent analog or high-frequency module, the combination with other symbol combs, frequency offset, global filtering or upsampling or the formation of an analytical signal by delayed Hilbert transformation in question.

In order to reduce the number of complex multipliers required (107) and to have a high degree of flexibility in the design of the length of the data blocks for a limited number of read ports of the buffer (105) and the coefficient memory (106), it is appropriate for the Output FIR filter to perform necessary calculations sequentially and this to design the FIR adder (108) as an accumulator. This is cleared at the beginning of a FIR arithmetic cycle and returns the previous sum as new summands for the next cycle with each multiplication addition clock. Of course, this calculation can also be made according to the prior art in a pipeline with optional stall. After the required number of cycles, the output value is output and an increment of the base read pointers.

According to the particularly advantageous embodiment of the invention according to dependent claim 20, the read pointer can be formed for this purpose by means of a bit reversal operation. For this purpose, an index is first formed, which is set to zero with the clearing of the accumulator and increased with each addition cycle. This index is now bit-inverted so that the lowermost bit corresponds to the address offset equal to half the distance of the fixed-distance hardware read pointers. The second lowest bit corresponds to the quarter distance, etc. The value thus formed is added to the offset (base read pointer) corresponding to the position of the current sample or coefficient in the first data block or coefficient block.

As a result, it is surprisingly possible to use any power of two as a block size despite a fixed concatenation of the hardware read pointers with a specific offset, which may be necessary due to the fixed allocation to specific sub-blocks of the buffer.

The computation operations required according to the invention can be implemented both in hardware and by suitable configuration of an FPGA, for example as a result of a VHDL description as well as a computer program which controls a signal processor as part of the modulator. The modulator according to the invention is preferably used in radios, in particular base stations or radio modems or mobile devices of a radio network, but also in radio links. Furthermore, in particular, the use in a DSL modem offers, whereby a particularly interference-resistant and long-range signal is generated, which significantly reduces the known crosstalk problems with common leadership with other DSL and non-DSL signals in a wiring harness. Furthermore, an application for the control of optical modulators for fiber optic cables is conceivable to produce a signal which is immune to physically induced dips in the optical transmission spectrum.

It is also conceivable to reverse the process described here according to claim 25 for demodulation, the FIR filter is then input side, the buffer (105) with the input values of the A / D converter of the receiver (possibly after formation of an analytical signal by delayed Hilbert transformation or by frequency offset and filtering), whose output values are then collected in blocks in a further memory and fed to a Fourier transform or FFT following. This is followed by the phase adjustment of the output values and then the decoding. Alternatively, the demodulation of such a hexagonal symbol pattern is also possible in a classical manner with one or more polyphase feeder banks, the input filters of which receive the data from a ring buffer each with the necessary offset. This does not necessarily have to be a multiple of the transformation block length here, the choice of the sampling time is non-binding in the case of analysis.

The invention further relates to a modem with an additional device for measuring and reducing crosstalk attenuation.

The invention relates to a modem which uses a multitone transmission method with band-limiting Gauss or Matched Filterbänken, with an additional device for measuring and reducing crosstalk interference by using specially modulated especially with pseudo random or gold codes subcarrier for detecting crosstalk.

The invention is based on the object of providing a broadband data stream as a radio signal in a point-to-multipoint system with a central unit, typically in the case of wired transmissions, this is a DSL connection multiplexer (DSLAM), in wireless transmissions the radio base station, and peripheral units, typically the customer modems, if possible trouble-free transmission over a greater distance. In particular, the capacity-limiting crosstalk between different transmission directions, ie individual cables in a cable bundle or spatial directions, which are served by separate antennas or by phased array beam forming, should be reduced by a suitable choice of transmission parameters.

DE 102007036828 introduces the novel modulation method CIFDM, which is distinguished by combining the advantages of the high spectral efficiency of the OFDM method with the robustness of a multicarrier method with conventional modulation of the individual carriers.

Nevertheless, the signal / noise ratio of this novel transmission method also suffers from crosstalk disturbances, for example on a cable bundle, due to the signals of other subscribers.

Task of this invention is to reduce these. In EP 1422835 Bl a compensating circuit is proposed for this purpose, the disadvantage here is that this is economically feasible only in a few cables in the bundle. On the other hand, EP 1283655 B1 envisages limiting the total bit rate permitted on a cable bundle, which is disadvantageous in that this limitation is only an estimated value and possibly far from the actually realizable capacity of the cable bundle. In US 6987800, the attempt to extinguish the far end comes into play, disadvantageous is the high effort of many participants.

A direct measurement and allocation of the subcarrier fails so far on the coupling properties of the commonly used OFDM / DMT method due to the unfavorable sin (x) / x filter form in the frequency domain.

The problem is solved by the invention described in claim 26 devices whose function will be explained below with reference to an exemplary embodiment.

Given is a DSL transmission system consisting of a DSL connection multiplexer (DSLAM) and the customer modems, which are connected via a cable bundle with single pairs with the DSLAM.

In order to reduce mutual interference in the uplink now, each modem is requested to send a test signal on one or more dedicated subcarriers, according to the main claim. In the advantageous embodiment according to dependent claim 29 is a pseudo random noise signal or gold codes, wherein the polynomial and the initialization data from the DSLAM are specified as a central unit on a control channel. The DSLAM now performs a test measurement on each subcarrier on each incoming pair and makes a correlation with the given codes.

At this point, the special capability of the invention to combine such a measurement method with such a multi-tone transmission method, which works with band-limiting filters or matched filters or filter banks and thus allows continuous transmission of pseudo random noise signal, in contrast to the known OFDM modulation with blockwise transmission: The correlation can also be continuous over long periods of time and thus allows a very high accuracy in the determination of crosstalk, at the same time takes place by the transmission of the test signal no interference other payload channels, as worked here with clear band-limiting Matched or Gauss filters becomes.

In addition, a transmission can even take place on a useful signal subcarrier, a slow additional modulation of the useful signal with a pseudo random noise signal of low amplitude will hardly affect the useful signal, but is easily through the long continuous transmission time accurate cross-correlation possible. The assignment of the codes is preferably carried out according to the criterion of the subclaim 30 in order to ensure a certain strength of the measurement against reflections.

According to dependent claim 27, a crosstalk matrix is calculated from the measurement data obtained in this way, or a bipartite graph is calculated to reduce the memory requirement, and the identifiers of the subscriber modems on the one hand and the identifiers of the high over talk as contiguous on the other hand Subcarrier has. For third-party disturbances by third-party modems not included in this system, a further total entry is conceivable.

According to dependent claim 28, an optimal allocation of the subcarriers and transmission powers with suitable optimization algorithms is now calculated from these data, these are well known from the literature, the optimization can also be based on economic specifications such. the bandwidths booked by the customer, a uniform distribution according to current or predicted data throughput needs is to be striven for. In addition, a suitable strategy can be chosen for the occupancy of subcarriers, which are particularly covered by third-party bugs. The use for pilot carriers is conceivable.

The thus calculated allocation is then transmitted as transmission release to the respective modems from the DSLAM. According to dependent claim 33, this calculation and allocation can be made continuously, in particular it makes sense for a newly einbuchendes first modem only to allow the transmission of weak test signals and initially make a very conservative assignment of subcarriers, which does not bother other participants.

Another use for the measured values is to introduce them into a forward error correction method according to subclaim 32 as a priori information in order to specify the degree of uncertainty of the soft bits obtained from individual subcarriers. In the simplest case, a lowering of the soft bit values from subcarriers affected by high crosstalk towards an undecidable mean value for turbo decoders or LDPC decoders, also a more detailed introduction into the correction process, such as the selection of the result which is least dependent on unsafe subcarriers, is conceivable.

In addition, according to dependent claim 31, the test signals can also be used simultaneously as an alternative to the modulation on user data carriers as a pure pilot carrier, which at the same time the correction of reflections through an equalizer in the frequency domain.

A further possibility of using the crosstalk measurement results is in the control of the line driver, here it is advisable, in addition to the transmission power whose impedance electronically controlled so that the crosstalk or reflections or the operating attenuation are reduced by reflections. For this purpose, in the simplest case, a test transmission can be carried out with different configurations of the line driver and the configuration with the best transmission characteristics can be selected. A frequency-dependent setting is also conceivable, for example, by a suitable selection of filters in an analog feedback path for active impedance control. The same applies to the configuration of a line receiver.

In large networks, the control and monitoring of the sub-carrier allocations is centralized according to claim 35 via an external computer system with a suitable control software, in this case can be retrieved from a customer database profiles for the confirmed and variable data rates of a customer and incorporated into the decision-making process.

Claims

claims:

1. A radio system or radio or radio modem or modem for wired transmission or message transmission device or module or integrated circuit of such a device consisting of at least one transmitter or receiver or a combination thereof, wherein the transmitter digital or digitized by him information for transmission in data blocks is divided, wherein the transmission of data is a multi-carrier method is used, the multi-carrier distributed by modulation - which can also be implemented as a complex multiplication - at least one data block on at least two subcarriers, the entirety of the so in a clock cycle to be transmitted data of a block is a multi-carrier symbol denotes, the clock is referred to as a symbol clock, characterized in that

(1) a division of the data blocks to be transmitted continues to take place on at least two classes of multicarrier symbols, which are respectively transmitted in parallel on the same transmission medium, the classes differing in that

(2) the multicarrier symbols of at least two classes are transmitted offset in frequency with respect to each other, so that the areas of mostly high spectral power density of one class each fall into those areas with largely low spectral power density of at least one other class;

(3) the multicarrier symbols of at least two classes are transmitted offset in time from each other, so that the sampling time of a multicarrier symbol of a class falls within the temporal proximity of a power minimum of one or all subcarriers of the multicarrier symbol of at least one other other class.

2. A radio system or radio device or radio modem or modem for wired transmission or message transmission device or module or integrated circuit of such a device consisting of at least one transmitter or receiver or a combination thereof, wherein the transmitter digital or digitized by him information for transmission in data blocks is divided, wherein the transmission of data is a multi-carrier method is used, the multi-carrier distributed by modulation - which can also be implemented as a complex multiplication - at least one data block on at least two subcarriers, the entirety of the so in a clock cycle to be transmitted data of a block is a multi-carrier symbol denotes the clock is referred to as a symbol clock, furthermore, the frequency spacing of the unmodulated center frequencies of two immediately adjacent Subcarrier of a multicarrier symbol referred to as subcarrier frequency spacing, characterized in that in parallel on the same transmission medium in overlapping frequency ranges at least two sequences of multicarrier symbols are transmitted with the same symbol clock frequency, which are interleaved by at least one subcarrier frequency of at least one first sequence against at least one subcarrier frequency of at least one second sequence is offset in frequency by half of a subcarrier frequency spacing occurring in at least one of the two sequences, and that the symbol clock of that first sequence, and thus its sampling instant, is time offset from the symbol clock of the second sequence by half a symbol clock period, for the subcarrier frequency offset and the symbol clock frequency. Time offset each small - the intersymbol interference not significantly increasing - deviations from the ideal half offset are permissible and wobe i between the symbol clock frequencies of the two sequences also small - the intersymbol interference not significantly increasing - fluctuations are allowed.

3. Device or module or circuit according to claim 1 or 2, characterized in that the receiver has a synchronized with the symbol clock unit, which divides the received signal to the analysis filter for each multicarrier symbols by the intended for the respective analysis filter Sampling values are weighted with at least one periodic function whose period corresponds to the period of the symbol clock or is an integral or fractional-rational multiple thereof, wherein as analysis filter here each stage of the signal processing is called, which - in particular using at least one discrete FIR or HR Filter, at least one Fourier transform or inverse Fourier transform or wavelet transform or other orthogonal transform or a combination of these - is able to decompose the multicarrier signal back into its subcarrier components or immediately to demodulate.

4. Device or module or circuit according to claim 3, characterized in that a polyphase filter bank is used as the analysis filter, which is in particular realized by a combination of an input demultiplexer, FIR polyphase sub-filters and an inverse digital or fast Fourier transform.

5. Device or assembly or circuit according to one of claims 1 to 4, characterized in that for the synthesis of the respective multicarrier symbol a polyphase filter bank is used, which together with the analysis filter forms a matched filter system with respect to the respective individual carrier.

6. Device or module or circuit according to one of claims 1 to 3, characterized in that the transmission side in the modulation of the subcarriers of a multicarrier symbol artificially a power reduction between two symbols by the resulting from the synthesis of a multicarrier symbols samples prior to merging with samples of others Multi-carrier symbols are weighted with at least one periodic function whose period corresponds to the period of the symbol clock or is an integral or fractional multiple thereof, this function being selected so as to reduce the intersymbol interference between two symbol sequences, in particular involving the transmission channel characteristics or avoiding transmit power peaks.

7. Device or module or circuit according to claim 1 or 2, characterized in that for the modulation of the individual subcarriers, a BPSK, QPSK, nPSK, MSK or QAM modulation is used, which is preferably designed differentially so as not to require an absolute phase reference.

8. Device or module or circuit according to claim 1, 2 or 7, characterized in that for correcting interference between subcarriers or multicarrier symbols or individual components of these trellis coding with Viterbi decoding is used.

9. Device or module or circuit according to claim 1, 2 or 7, characterized in that to correct an interference between subcarriers or multicarrier symbols or individual components of these forward error correction means turbo code, low density parity code, Reed Solomon Code or BCH code use place.

10. Device or module or circuit according to claim 1 or 2, characterized in that for prioritizing individual parts of the data stream one or more subcarriers are transmitted with an increased symbol clock rate to shorten the latency, for this purpose by increasing the channel filter bandwidth of the frequency range at least two subcarriers with the basic symbol clock rate is used to maintain the orthogonality, where inclusion of more distant carriers in the use of classical band-limited channel filters, in particular of the Gauss type, is not required.

11. Device or module or circuit according to claim 1 or 2, characterized in that are used to determine the channel impulse response pilot subcarriers, which are distributed over the spectrum to be transmitted, and if necessary, are orthogonal to adjacent pilot carriers of another multicarrier symbol sequence.

12. Device or module or circuit according to claim 11 and 3 or claim 11 and 6, characterized in that the periodic weighting function to the prevailing transmission conditions, in particular based on the information obtained from the pilot subcarriers, or to the requirements of the transmitter operation as the minimization of Ratio of the peak transmission power to the average transmission power is adjusted.

13. Device or module or circuit according to claim 1 or 2, characterized in that in the case of transmission of packet-oriented data different packet classes with different quality of service or packets whose transmission is to take place at different times, suitable different multi-carrier symbol sequences are assigned, while the Modulation or signal power of the different multicarrier symbol sequence can be varied in accordance with transmission quality and security requirements; in particular, transmission of packets that would be necessary at a later time according to schedules or timestamps can be tentatively preferred on a low quality multicarrier symbol sequence. If, however, an error-free transmission succeeds and is confirmed, bandwidth in the area of a multicarrier symbol sequence of high quality is thus saved.

14. Device or module or circuit according to claim 13, characterized in that data packets which were transmitted prematurely in a different class of service or multicarrier symbol sequence, nevertheless cached in the receiver prior to delivery to the customer until specified in the timestamp or calculated from this delivery time in order to simulate a continuous flow of data to the recipient of the packets and to ensure optimum data throughput, in particular with window-oriented protocols such as TCP / IP.

15. Device or module or circuit according to claim 1 or 2, characterized in that for the realization of the signal processing stages, a microprocessor, microcontroller, digital signal processor or an FPGA is used.

16. A modulator in a radio or wireless modem or modem for wired transmission, on copper as on fiber optic cables, or in a device for communication or in an assembly or in an integrated circuit for such devices, which at least one multi-carrier signal or at least one composite multicarrier signal from at least two intercommunicating groups of sub-carriers, which differ from one another by a time and frequency offset, characterized in that

(1) to generate at least one group of subcarriers a Fourier transform or inverse Fourier transform or wavelet transform or inverse wavelet transform or orthogonal transform is used which generates the necessary samples in the time plane to form at least one comb of subcarriers,

(2) the result of the transformation is stored in a buffer,

(3) the sampled values from this buffer are fed in groups to at least one FIR filter or at least one HR filter, resulting in the effect of a filter bank or a polyphase filter,

(4) repeatedly supplied to the FIR filter or HR filter to generate a sample duration independent of the number of samples.

17. A modulator according to claim 16, characterized in that the output signal of the modulator is a data stream or signal, which distributes the individual sub-symbols of the sub-carriers into a hexagonal grid in the time / frequency plane.

18. A modulator according to claim 16, characterized in that for the repeated selection of samples from the buffer, the access address is formed on this buffer by using a data word formed by bitwise AND masking or OR masking the coefficient index of an FIR filter. Unused memory areas resulting from masked-out bits can hereby be made usable again by a subsequent partial shift operation of individual fields of the data word, in particular by means of a barrel shifter.

19. A modulator according to claim 16, characterized in that the access address to this buffer is formed by means of a modulo arithmetic unit for the repeated selection of samples from the buffer.

20. A modulator according to claim 16, characterized in that at least one access address is formed on a buffer or coefficient memory using a bit reversal operation.

21. A modulator according to claim 16, characterized in that for adapting the phase position of the samples in the time plane prior to the transformation of these - in particular by means of inverse Fourier transform - a complex multiplication of the input values of the transformation in the frequency plane for the purpose of phase rotation is performed. The phase rotation is thereby updated in blocks, it is necessarily dependent on the index and thus the subcarrier frequency of the input value to be transformed, in the case of inverse Fourier transform, the calculation of the increment is carried out in particular by multiplying the fundamental phase rotation with the frequency index of the input value and subsequent modulo operation. The conversion of the phase rotation into the complex multiplication factor can be carried out by means of the Euler formula; it can also be integrated into the process of symbol constellation generation in the case of a phase modulation of the subcarriers.

22. A modulator according to claim 21, characterized in that for determining the factor of the complex multiplication from a phase value of the CORDIC algorithm is used, in particular also in incremental form.

23. A modulator according to claim 16, characterized in that in the FIR or HR filter complex coefficients and complex multiplies are used, whereby a frequency offset of the output signal can be realized.

24. A modulator according to claim 16, characterized in that for the realization of the signal processing stages, a microprocessor, microcontroller, digital signal processor or an FPGA is used.

25. Demodulator in a radio or wireless modem or modem for wired transmission, on copper as on fiber optic cables, or in a device for communication or in an assembly or in an integrated circuit for such devices, which at least one multi-carrier signal or at least one composite multicarrier signal from at least two interleaved groups of subcarriers, which differ by a time and frequency offset from one another evaluates, using the signal processing stages described in claim 16, characterized in that the realization by reversing the order of the signal processing steps described in claim 16 FIR or IIR filters and Fourier transformation or inverse Fourier transformation or wavelet transformation or inverse wavelet transformation or orthogonal transformation, wherein the input values of the demodulator are stored in the buffer and the output values of the FIR or IIR filter are the input values of the transform, which are optionally latched prior to application of the transform.

26. A radio or modem or modem for wired transmission, on copper such as fiber optic cables, or assembly or integrated circuit for such devices, which at least one multi-carrier signal or at least one composite multicarrier signal from at least two nested groups of subcarriers, which by a time and Distinguish frequency offset from one another, generated, which are generated or evaluated by means of band-limiting filter or matched filter or filter banks, which is used in a point to multipoint system, with at least one central unit - in particular DSL connection multiplexer or radio base station - and at least two peripheral units - in particular customer modems or customer radio equipment -, wherein the central unit via at least one transmission line - in particular a cable bundle of subscriber lines or a passive optical network or radio with multiple n Sectors or polarization planes - communicates with the peripheral units, which basically has a certain separation of the signals to different peripheral units, but which nevertheless has an undesirable frequency-dependent crosstalk between different transmission directions, characterized in that

(1) for detecting the occurrence of crosstalk interference in certain frequency ranges at least one unit, which may also be the central unit itself, requested by the control unit signal from the central unit a coded test signal on at least one subcarrier sends and

(2) at least one further unit makes a test measurement to determine if and at what level this test signal can be received by it, even though it has been transmitted in a transmission direction that is not intended to be received by that unit, and hence undesirable crosstalk this has arrived.

27. Apparatus according to claim 26, wherein at least one crosstalk matrix or at least one bipartite graph or at least one other data representation representing the couplings is formed from the information thus obtained by measurement, whereby third-party perturbations can be detected via a sum entry.

28. Apparatus according to claim 27, characterized in that from this information at least one with respect to minimal mutual interference of the transmissions optimized assignment of subcarriers and allocation of subcarrier transmission powers calculated by Wasserfüll-, greedy, flow or other known optimization algorithms and by control signals the units are sent for execution.

29. Device according to one of claims 26 to 28, characterized in that the test measurement of the subcarriers is carried out by correlation by at least one test signal for correlation suitable code, in particular pseudo random noise code or gold code is modulated and the measuring unit against this one Cross or autocorrelation performs, the modulation can also be done on a payload carrier in order to use this simultaneously as a test signal can.

30. Apparatus according to claim 29, characterized in that the codes are assigned by the central unit so that there is a low cross-correlation between them, which can be introduced as an additional criterion that the cross-correlation should be low even with reflections on the transmission medium ,

31. Device according to one of claims 26 to 30, characterized in that the test signals or codes are used simultaneously as a pilot carrier, after their evaluation, a correction - in particular multipath correction - of the received useful signal in the time or frequency plane or a deduction of Reflections or avoidance of use due to destructive interference due to reflections or multipath disturbed subcarriers make.

32. Device according to one of claims 26 to 30, characterized in that the evaluation results of the test signals or the codes or the thus obtained attenuation and crosstalk measurement results or the crosstalk matrix or the bipartite graph are introduced as a quality standard in an error correction method for the user data stream.

33. Device according to one of claims 26 to 30, characterized in that the test measurements are carried out continuously to make a change in the assignment of subcarriers and allocation of subcarrier transmission powers in case of a change in the transmission conditions or booking new units in the network, in particular to compute a first allocation for newly posted units prior to their first payload transmission, which avoids disturbances to existing data transfers, which allocation can later be changed to a uniform distribution of the total channel capacity.

An apparatus according to any one of claims 26 to 30, characterized in that the evaluation results of the test signals or the codes or the thus obtained attenuation and crosstalk measurement results or the crosstalk matrix or the bipartite graph are used to determine the output impedance of a line driver or line receiver frequency-dependent or frequency-independent optimization so that the crosstalk or reflections or the operating loss can be reduced by reflections.

35. Device according to one of claims 26 to 34 or computer with software for operating a network with such devices, characterized in that individual necessary calculations and evaluations are made in the control software or the FPGA code of one of the units or an external computer system.