WO2001095582A1

WO2001095582A1 - Multi-carrier data transmission method in which the individual carrier frequencies are directly occupied in the useful band

Info

Publication number: WO2001095582A1
Application number: PCT/EP2001/006233
Authority: WO
Inventors: Klaus Lehmann; Jürgen Weith; Stefan Wimmer; Rainer Fuchs; Rainer Maul
Original assignee: Gds Gesellschaft Für Digitale Signalverarbeitung Mbh
Priority date: 2000-06-05
Filing date: 2001-06-01
Publication date: 2001-12-13
Also published as: DE10027301A1

Abstract

The invention relates to a broadband-efficient multi-carrier data transmission method, in which no orthogonal modulation channels are necessary. The individual complex carrier frequencies are directly occupied in the useful band. The generated complex time signals are only formed as real parts or imaginary parts or by the addition of real and imaginary parts. Using correlation reception, implemented as synchronously clocked discrete Fourier transformation on the receive side, the symbols combined on the transmit side can be completely recovered, whereby the provision of filters which is normally required in modulation and demodulation with orthogonal channels is eliminated. Another advantage of the invention is that it is not necessary to synchronize the carrier frequency according to frequency and phase.

Description

METHOD FOR MULTI-CARRIER DATA TRANSFER, IN WHICH THE INDIVIDUAL CARRIER FREQUENCY IS DIRECTLY ASSIGNED IN THE USER TAPE

The invention relates to a method for orthogonal frequency multiplex modulation and a modulator, a demodulator and a communication system, which are based on the method.

The use of so-called discrete multi-tone methods for bandwidth-efficient transmission is known from the prior art. Such multi-tone methods are also referred to as "orthogonal-frequency division multiplex" (OFDM). With this multi-carrier concept, the orthogonality of temporally successive and spectrally adjacent symbols is fulfilled.

The OFDM process is, for example, in the specialist books of Prof. Dr. Kammeyer, "Messaging", BG Teubner Verlag GmbH, Stuttgart, 1992 and in "The Mobile Communications Handbook", Jerry D. Gibson, CRC Press, Inc. 1996 and in US 3,488,445.

An OFDM system known from the prior art is explained below with reference to FIGS. 1 and 2. FIG. 1 shows a transmitter 100 in which the OFDM method is implemented. The input data to be transmitted by the transmitter 100 are first subjected to a serial-parallel conversion in the serial-parallel converter 102. The parallelized data is then channel encoded by channel encoder 104 and interleaved. The carrier assignment is then carried out in the module 106 by mapping. The data symbols obtained in this way are then subjected to an inverse discrete Fourier transformation (IDFT) in module 108. After parallel-to-serial conversion by the parallel-to-serial converter 110, the coefficient values previously obtained by the inverse DFT are subjected to a subsequent digital-to-analog conversion in the digital-to-analog converter 112 and form the analog transmit signal.

FIG. 1 also shows a receiver 114 for receiving the analog transmission signal emitted by the transmitter 100. The analog received signal is first converted by the analog-digital converter 116, which is followed by a serial-parallel conversion in the serial-parallel converter 118. The digitized and Parallelized data is then subjected to a discrete Fourier transform (DFT) in module 120. The values transformed in this way are subjected to an inverse mapping operation in module 122. After inverse mapping, a de-interleaver operation follows in module 124. Finally, the modules 126 and 128 are used for channel decoding or parallel-serial conversion.

In the OFDM system shown in FIG. 2, a so-called “guard interval” is added before the transmission of the transmission signal via the transmission channel during the signal generation as a periodic continuation of the complex transmission signal, in order to ensure that the signal is reconstructed on the reception side to be able to compensate for the settling caused by the transmission channel. The guard interval is added in module 130. The samples of the complex-generated time signal output by module 130 are then subjected to modulation with carrier functions f orthogonal to one another, both the time signal components I and Q and the modulation results being orthogonal to one another. As is known, the addition or subtraction of the modulated partial signals leads to a real bandpass signal which, after appropriate amplification, serves as a transmission signal.

On the receiving side, after appropriate bandpass filtering for interference suppression and a gain control (AGC = Automatic Gain Control) in module 134, the bandpass signal is converted into components that are orthogonal to one another again by modulation with the orthogonal carrier frequencies in the modulator 136, which then undergoes a Fourier transformation in the OFDM receivers are subjected to the image symbols in the state space can be obtained from them again.

In this known multicarrier method, two implementation methods are possible for signal conditioning in modulator 132 on the transmitter side and for signal recovery in modulator 136 on the receiver side. The modulation with sin (ω _M t) and cos (ω _M t) (mixture) can be carried out both analog and digital. With analog mixing, complex circuits for multiplying and generating the mutually strictly orthogonal carrier signals sin (α) _M t) and cos (ω _M t) are necessary. If, on the other hand, the mixing is done digitally, the I and Q basebands must be oversampled, complexly interpolated, multiplied by the digitally generated carrier signals and then added. This extensive and intensive effort is necessary so that the data symbols can be calculated back with an acceptable bit error rate after the Fourier transformation.

The OFDM method and its uses are also described in “Data Transmission by Frequency-Division Multiplexing Using the Discrete Fourier Transform ,,, S.B. Weinstein and Paul M. Ebert, IEEE Transactions on Communication Technology, vol. COM-19, No. 5, pages 628 to 634, October 1971, “Performance of an Efficient Parallel Data Transmission System ,,, Burton R. Saltzberg, IEEE Transactions on Communication Technology, Vo. COM-15, No. 6, pages 805 to 811, December 1967, and in “Digital Sound Broadcasting to Mobile Receivers ,,, Bernhard Le Floch, Roselyne Halbert-Lassalle, Damien Castelain, IEEE Transactions on Consumer Electronics, Vol. 35, No. 3, pages 493 to 503, August 1989, and in EP 840 485, EP 932284 and EP 975 097.

A disadvantage of the known OFDM method is the relatively large implementation effort for the transmitter and receiver-side mixtures and complex signal filtering.

The invention is therefore based on the object of creating an improved method for orthogonal frequency multiplex modulation, and of creating an improved modulator and demodulator or a communication system based thereon.

The object on which the invention is based is achieved with the features of the independent claims. The dependent claims indicate preferred embodiments of the invention. The invention allows the multi-frequency signal to be transmitted to be formed on the transmitter side without further modulation or shifting into the bandpass range. This eliminates the need for a complex mixture.

The invention also makes it possible to generate the real transmission signal from the image area by summing the real and imaginary parts of the Fourier-transformed data symbols. Alternatively, it is also possible to use only the real part or only the imaginary part of the transformed data symbols for generating the transmission signal.

The invention is particularly advantageous for using a transmission channel with a plurality of bandpass areas, only those carriers which lie within the respective bandpass areas being used for the use of the OFDM method. The frequency components outside the bandpass ranges are set to zero. This enables flexible and inexpensive use of the OFDM method for such channels.

This allows, in particular, a cost-effective implementation of the so-called powerline technology with a frequency band from 9 kHz to 148.5 kHz, in which an energy supply network is used for the transmission of communication data. This applies in particular when the frequency bands of the power line technology are expanded into the megahertz range.

However, the invention can also be used with any other transmission channels, in particular with radio, light wave or infrared transmission.

An embodiment of the invention is shown in the drawing and is described in more detail below:

Show it

1 shows an OFDM transmitter and receiver according to the state of the

Technology, FIG. 2 shows a detailed view of the transmitter and receiver of FIG. 1,

Figure 3 is a flow diagram of an inventive

Modulation method,

FIG. 4 shows the frequency assignment according to the example in FIG. 3,

FIG. 5 shows a block diagram of an OFDM transmitter according to the invention,

6 shows the range of magnitudes of the transmission signal in the case of a

Bandpass range,

7 shows the magnitude spectrum of the transmission signal in the case of several

Bandpass areas and

FIG. 8 shows the block diagram of a receiver according to the invention.

In the method shown in FIG. 3, a number of M data are transmitted per clock cycle, where M <N. The M data is first subjected to a digital modulation method in step 300. A known modulation rule, e.g. xPSK or xQAM, in particular 4PSK, differential PSK or 16 QAM, are used. In turn, M result in data symbols transformed by the modulation. These data symbols generated from the modulation in step 300 are assigned to specific carrier frequencies by a mapping operation:

In step 310, the desired frequency spectrum of the signal to be generated is first defined. This frequency spectrum can be specified, for example, as the frequency characteristic of the transmission channel to be used. Such a frequency spectrum is shown as an example in step 310. The Frequency spectrum has the bandpass regions 312 and 314. The sampling frequency is designated as frequency f _c on the frequency axis. A suitable choice of the number of frequency points N to be used for the method results in a frequency interval Δf = f _c / N.

This results in the definition of N frequency points starting from the frequency zero on the frequency axis, which are spaced apart by the frequency interval Δf. Representing the N equidistant frequency points between the frequency zero and the frequency f _c , the frequency point Fj is shown in FIG. 3, where 0 <i <N.

The frequency points F _L> F _{L + 1} and F _{L + 2} or F _κ , F _{κ + 1} and F _{κ + 2} are located in the bandpass regions 312 and 314.

Such a definition of the frequency spectrum and subdivision into equidistant frequency points results in a number m of frequency points that lie in the respective passband - in this example, the frequency points F _L , F _{L + 1} , F _{L + 2} and F _κ , F _{κ + 1} and F _{κ + 2} . These frequency points in the bandpass regions 312 and 314 thus form a subset of all defined frequency points Fj

In step 320, the data symbols obtained in step 300 are assigned to the m frequency points in the bandpass areas 312 and 314. The frequency points F, lying outside the bandpass ranges are set to zero. This results in a data vector with a number N of data, the data being zero except in the bandpass areas. In the bandpass areas, however, the data of this data vector are defined by the result of step 300 and the subsequent assignment to the m frequency points. This will be shown in detail in FIG. 4.

In step 320, the data vector obtained in this way is subjected to an inverse Fourier transformation, in this example an inverse Fast Fourier transformation. This results in a data vector of 2 N values with a frequency spectrum of the carrier frequencies corresponding to that in step 310 predetermined frequency spectrum. These values are output in output step 340 as a result of the modulation.

FIG. 4 shows the assignment of the M data symbols to the m frequency points of the bandpass ranges.

There is one data symbol D ₀ , D ^ D ₂ D ₅ per time cycle

Data sequence for processing in step 320. To carry out the so-called mapping, the elements of the data vector 410 are defined as follows:

Those positions i in the data vector 410, the frequency points F | on the frequency axis outside of bandpass ranges 312 and 314 are set to zero. The positions in the data vector 410 which correspond to the frequencies F _L , F _{L + 1} , F _{L + 2} and F _K) F _{κ + 1} and F _{κ + 2} in the bandpass areas 312 and 314 are marked with the corresponding data symbols D ₀ applied to D ₅ .

A data vector 410 with N = 14 data then results from the N = 14 frequency points of the exemplary embodiment in FIG. 3. The leftmost date in the data vector 410 corresponds to the frequency zero; the rightmost date corresponds to the frequency point F ^, ie on the frequency scale that frequency point immediately below the sampling frequency f _{c. In} general, a date i of the data vector corresponds to the frequency Fj, where 0 <= i <N.

In the example under consideration, the data on the vector positions between i = 0 and i = 4 are approximately zero, since the corresponding frequency points lie below the first bandpass range 312. The subsequent frequency points F _L ,

(ie between i = 5 and i = 7) the data D ₀ , D _{1 (} D _{2) are} assigned. The procedure for the other vector positions is between i = 8 to i = 13. The resulting data vector 410 is then input to a microprocessor 420 per clock cycle, which carries out the inverse Fast Fourier transformation.

FIG. 5 shows a transmitter 500 according to the invention. The data to be transmitted are first parallelized in the serial-parallel converter 502, after which they are encoded in the encoder 504. This results in a number of M complex data symbols per time cycle, which are mapped into a data vector of length N (cf. step 320 of FIG. 3). This is input to the microprocessor 420 to perform the inverse Fast Fourier transform of length N.

The resulting 2-N values per clock cycle are in turn transformed in the parallel-serial converter 506 into a real and an imaginary serial signal x _R (k) or x, (k). Depending on whether switches S1 and S2 are closed or open, the real and imaginary components are summed in summer 508 or only one of the two components is used for the transmission. Depending on the switch position, the output signal x (k) of the summer 508 is either equal to the real part x _R or equal to the imaginary part x, or the sum of real and imaginary parts. The signal x (k) is then digital-analog converted, filtered and amplified, so that the transmission signal y (t) results.

Depending on the assignment of the inputs of the IFFT block, the two serial time signals are of the form

_μ -cos (ju- Gr ₀ -kT _c ) with n = number of frequency lines

and

_A , -sinCα- ar ₀ . Ä -2)

ie the calculated harmonic signals only appear in the integer grid (orthogonality).

The further processing now takes place without complex modulation and without further signal filtering either by (see FIG. 5) a) only the real part x _R (k) (only switch S1 closed), or only the imaginary part (only switch S2 closed), is subjected to a D / A conversion, analog post-filtering (reconstruction filter) and amplification. b) the real part x _R (k) and the imaginary part x, (k) are summed and further processed according to a) (both switches S1, S2 closed).

In summation (case b)), the following discrete signal arises from the real part and imaginary part samples by trigonometric conversion:

x (k) = x _R (k) + x _I (k)

L + nl. , x (k) - ^ - cosltf- βτ ₀ -k -T _c -φ _μ ± π) μ = L

During the summation, a sinusoidal oscillation with a different amplitude c _μ and a phase shift φ _μ per generated carrier frequency ω _μ (ω _μ = μ-ω ₀ ) is generated again.

The summation has the advantage of a higher signal energy. An additional phase shift that arises does not have a disadvantageous effect due to the usual differential method.

FIG. 6 shows an example of the magnitude spectrum of the transmission signal in the event that the transmission channel has a bandpass characteristic between the frequencies f _L and f _{L + n} _. A further bandpass range is present in the magnitude spectrum of the transmission signal in FIG. 7. As already explained with reference to FIG. 3, such a transmission characteristic can also be optimally used without additional effort by the modulation method according to the invention. If, as is necessary, for example, in powerline technology, bandpass signals are to be generated for the transmission signal, proceed as follows:

The useful band range is, for example, in the range from f _L = 100 kHz to approx. F ^^ = 120 kHz. The choice of the sampling frequency with fc = 307.2kHz and the determination of the IFFT length with N = 256 leads to a distance of Δf = 307.2kHz / 256 = 1200Hz in each case orthogonal carrier frequencies. The first 84 values in the calculation of the IFFT are set to zero, from the 85th value to the 100th value, M = 16 frequencies are applied with a known digital modulation method (4PSK, 8PSK, etc.). The values from 101 to 256 are reset to zero. This results in a bandpass signal in the range from f _L = 100.8 kHz to ^ ₊ "., = 118.8 kHz, ie with a bandwidth of 18 kHz. The symbol time is T _s = N-1 / fc = 833.33μs. For the transmission via a real channel, the guard interval time has to be added for the transient response, so that for this example, T _s ^* »1ms can be expected. When using a 4 PSK modulation per carrier in the useful band, a data rate of 16-2-1 / ms «32kbit / s can be achieved. 2 = k is the ordinal number for the modulation type 4 PSK with 2 ^k = 4, the factor 16 represents the number of carriers. The resulting time signal for the transmitter is similar to noise and is without further modulation, i.e. without sin (ω _M t) and cos (ω _M t) (see FIG. 2).

One advantage of the method is that a multiband signal, as shown in FIG. 7, can easily be generated only by software change, and also without orthogonal modulation, without the otherwise required pre-filtering of the transmission signal and without complex low-pass filtering of the back-modulated reception signal such as otherwise required in the prior art (see FIG. 2).

By zeroing or assigning frequencies using conventional digital modulation methods (PSK, QAM, etc.), it is possible to advantageously allow permitted ranges in a given transmission channel, for example in powerline technology according to the currently applicable standards (EN50065-1) within a band up to exploit half the sampling frequency fc. Alternatively, the transmit signal can also be obtained by Fourier synthesis from the individual carrier frequencies with the corresponding start phases. The carrier frequencies can be read from a memory (sine table) or can also be generated using the Goertzel algorithm, for example.

FIG. 8 shows a receiver 800 according to the invention. The signal x (t) generated according to the method according to the invention is first converted in the analog-digital converter 802; At the output of the downstream serial-parallel converter 804 there are again the values x _R or x, or x _R + x, before-depending on the position of the switches S1 and S2, which are connected in the transmitter 500 (cf. FIG. 5 ). The complex symbols X are recovered by the subsequent Fast Fourier transformation of these values in the microprocessor 806. After their decoding in module 808 and parallel-serial conversion in converter 810, the received data are recognized and output by decision maker 812.

The signal arriving at the receiver is distorted by the properties of the transmission channel and has additive interference signals such as noise, pulse interference and sine interference. A correlation receiver is optimal for such a signal. As can easily be shown, an FFT filter bank (FFT = Fast Fourier Transform) with the known computing time equal to the symbol time T _s or T _s ^{* is} such a correlation receiver. The components real part x _R (k) and imaginary part x, (k) of the transmission signal, which were added on the transmission side, or are only available as real parts or imaginary parts, have to be broken down again into the orthogonal components for the FFT.

An essential advantage according to the invention is the possibility of recovering the data signals with an FFT via symbol clock T _s or T _s \ as shown in FIG.

In the further simplified it is assumed that the undisturbed transmission signal i + nl x (k) = a _μ -cos (/ - σ ₀ -kT _c ) + jb _μ -cos (jU-aτ ₀ -kT _c ) the Fast Fourier Transformation μ = L is subjected. The signals X (μ) arise which correspond to the image area of the original coded data signals.

With the abbreviations of the individual arguments ω _{μ c} -k = Ω _μ k of the harmonic functions, the following applies to the FFT if the sum signal is interpreted as a real part:

ΛΓ-I .2Λ-

X (μ) = ∑χ (k) - _e - ^J ^ ^k k = 0

N-L L + N-L

= ∑ (a _t , -cosΩ- v-k + b _y -smΩ- vk) - (cosΩ.- -k-jsmΩ-jU-k) i = 0 μ = L with Ω = 2π / N. Ω-v are the frequencies transmitted in the integer grid v and Ω- μ are the comparison frequencies for the FFT acting as a correlator bank. The coefficients a _v and bv represent the former transmission signal amplitudes a _μ and b _μ arriving at the receiving _location .

The following partial terms are obtained by trigonometric conversion:

With the correlation over the symbol time, the entire relationship is reduced to

that is, at the correlator output (according to the FFT), the information a _v and bv contained in the transmitter carrier frequencies are recovered with μ in ascending order. The only requirement is the synchronization of the receiver to the sequence x (k), which can be carried out using various known methods. A further otherwise necessary synchronization to the angular frequency and the phase of orthogonal modulation and demodulation frequencies is not necessary, since these are not required at all in the method according to the invention.

On the transmitter side, the method can e.g. by suitable programming of a Motorola 56002 digital signal processor (DSP), on the receiver side using Motorola 56303 DSP.

Claims

claims

1. A method for orthogonal frequency division multiplex modulation (OFDM), characterized in that only those carrier frequencies are used for the modulation that are in a pass band (312, 314) of a channel.

2. The method according to claim 1, characterized in that

(a) a data sequence to be modulated contains a first number (M) of data per time cycle,

(b) a second number (N) of equidistant frequency points are defined between the frequency zero and the cut-off frequency, a third number (m) of frequency points lying as a subset of the second number of frequency points in a pass band of the channel,

(c) the data are each assigned to one of the frequency points of the subset.

3. The method according to claim 2, characterized in that the channel has two or more bandpass areas, wherein in one or more of the bandpass areas there are frequency points of the subset to which data are assigned.

4. The method according to any one of the preceding claims, characterized in that an inverse Fourier transformation, preferably an IFFT, is carried out per time cycle.

5. The method according to claim 4, wherein only the real components or only the imaginary components or the sum of the real and imaginary components of the coefficients resulting from the inverse Fourier transformation are used for generating a transmission signal.

6. modulator for performing the method according to one of claims 1 to 5 with a microprocessor, preferably a digital signal processor, for performing the assignment of the carrier frequencies.

7. A method for demodulating a signal modulated according to a method of one of claims 1 to 5, characterized in that the signal is subjected to a Fourier transformation and the data are recognized by correlation over the symbol time, the method being carried out synchronously with the data sequence ,

8. Demodulator for performing the method according to claim 7 with a microprocessor, preferably a digital signal processor, for performing the Fourier transformation.

9. Communication system with at least one modulator according to claim 6 and a demodulator according to claim 8, wherein the modulator and the demodulator can be connected to one another via a channel, the channel preferably being a network for electrical energy supply.

10. Computer program for performing a method according to one of claims 1 to 5 and / or according to claim 7.