WO2004079953A1

WO2004079953A1 - Method and arrangement for transmitting signals according to a multiple access method

Info

Publication number: WO2004079953A1
Application number: PCT/DE2004/000348
Authority: WO
Inventors: Ralf Irmer; Rene Habendorf
Original assignee: Technische Universität Dresden
Priority date: 2003-03-07
Filing date: 2004-02-25
Publication date: 2004-09-16
Also published as: DE112004000305D2

Abstract

The invention relates to a method for transmitting signals according to a multiple access method, during which data symbols to be transmitted are spread in transmitter by using a spread code and are transmitted in the form of a transmit signal from the transmitter to the receiver(s) via a transmission link between the transmitter and one or more receivers by means of a radio communications system. The invention also relates to an arrangement for transmitting signals according to a multiple access method, comprising a transmitting unit, which is situated inside a transmitter, and comprising processing units that are connected up from the transmitting unit. The aim of the invention is to reduce the bit error probability inside the receivers in a targeted manner and with a low amount of computing. To this end, the inventive method involves carrying out a simulation with a transmit signal by using a model that represents the transmission link, and the transmit signal is modified until a receive signal with a minimal bit error probability is simulated on the receiver/s. The inventive arrangement fulfills the aim of the invention in that means for simulating the transmission link are situated inside the transmitter that are actively connected to at least one of the processing units.

Description

Method and arrangement for sending signals after a

Multiple access methods

The invention relates to a method for transmitting signals according to a multiple access method, in which data symbols to be transmitted are pretreated in a transmitter and sent by the transmitter as a transmission signal to the recipient or recipients via a transmission link between the transmitter and one or more receivers by means of a radio communication system become. The transmission signal can be received by the receiver as a reception signal, the transmission signal consisting of transmission signal elements and the reception signal consisting of reception elements.

The invention also relates to an arrangement for transmitting signals according to a multiple access method, with a transmission unit arranged in a transmitter and processing units connected upstream of the transmission unit.

In the case of signal transmission via radio channels (for example in the case of digital mobile radio), a method with outstanding importance is code division multiple access (CDMA). In the case of multipath propagation in the channel, interactions arise between the codes of the different users of a system which are referred to as single-user interference (Inter Symbol Interference, IST.) Or multi-user interference (Multiple Access Interference, MAI). Appropriate preprocessing of the transmission signal in the transmitter can reduce these interferences and at the same time increase the strength of the desired signal at the respective receiver. As a result, improved data estimates can be achieved at the receivers or the transmission power required to achieve a specific quality target (Quality of Service, QoS) can be reduced.

CDMA signals can be detected by so-called RAKE receivers without special calculation or predistortion of the transmission signal being provided. Interferences occurring at the detector in the RAKE receiver result in an increased number of bit errors per number of bits sent, which is referred to below as the bit error probability (BER), in particular in the case of a large number of active users.

Numerous methods for reducing the interference or adapting to the reception channel in the receiver are known for CDMA signals. These receiver-based methods are summarized under the term multi-user detection (MUT). These methods are described, for example, in S.Verdu: "Multiuser Detection", Cambridge University Press 1998 and in the special issues on MUT: IEEE J.Select. Areas Commun, vol. 19, no. 8, August 2001 and vol. 20, no. 2, February 2002. A disadvantage of the receiver-based methods is that they are mostly only suitable for the uplink (mobile station in the direction of the base station, uplink) or for the downlink (base station in the direction of the mobile station, downlink) in the mobile station, which should be done for reasons of complexity and energy is usually not possible.

Transmitter-based methods can be summarized with the generic term multiuser transmission (MUT). Methods for determining the transmission signals and for pre-distorting the transmission signals in the transmitter are particularly suitable for the downlink, i.e. these methods are then used in the base station. In contrast to multi-user detection, knowledge of the radio channel must also be available in the transmitter, for example an estimate of the channel impulse response and a

Estimation of the noise power at the receiver. The known methods of multi-user transmission differ in their optimization goals and in the achievable ones Bit error probability.

The aim of methods for complete interference elimination (zero-forcing (ZF) joint transmission, transmitter precoding) is to completely suppress the interference which occurs on the detectors of the individual users. They are described, for example, in patent applications US 5,461,610, DE 101 41 809 AI, DE 100 42 203 AI, in Andre Noll Barreto: “Signal Pre-Processing in the Downlink of Spread- Spectrum Communications Systems, VDI Progress Reports Series 10, No. 687, or in M. Meurer, P.W. Beier, T.Weber, Y.Lu and A. Papathasiou: "Joint transmission: advantageous downlink concept for CDMA mobile radio systems using ti e division duplexing". IEE Electronics letters, 36: 900-9001, May 2001. To derive these methods the criterion of the smallest mean squared error (MMSE) is also sometimes used.

These methods for determining the transmission signal have in common that the bit error rate (BER) at the mobile receivers is not directly minimized, but that the bit error probability is indirectly achieved by minimizing other targets, e.g. MMSE or complete interference cancellation (IF) takes place. Furthermore, the exact knowledge of the data symbols in the transmitter is not used to calculate the preprocessing coefficients or matrices. However, knowledge of the exact transmission symbols is available in the transmitter. The disadvantage of the methods mentioned is the large number of bit errors that arise in the receivers. This disadvantage is particularly present when there is high additive noise in the receivers at the same time. A further disadvantage is the considerable computational effort which is based on the methods mentioned and which has to be applied in the base station.

The object of the invention is to determine the bit error probability in the receivers in a targeted manner and with little computing effort to reduce.

On the procedural side, this problem is solved in that a simulation with a transmission signal is carried out in the transmitter by means of a model representing the transmission link, and the transmission signal is changed until a reception signal with a minimal bit error probability is simulated at the receiver or the receivers of the model. This is done on the condition that the transmission power permitted in the communication system is not exceeded. The changed transmission signal, which leads to the simulated minimum bit error probability, is then sent to the receiver or receivers over the transmission link.

This makes it possible to take advantage of the fact that the data symbols to be transmitted are precisely known in the transmitter. Furthermore, the characteristic of the transmission link, e.g. from knowledge of a channel impulse response, it being assumed that the characteristic of the transmission link remains essentially constant at least for the time from the beginning of the simulation to the actual transmission.

A base station is particularly suitable as a transmitter, and the receivers are, in particular, mobile stations that can also be referred to as users. An application for one or more users is possible.

However, the invention is not limited to an application in the mobile radio or W-LAN area (W-LAN (wireless local area network) = wireless local area network).

In a first variant of the method according to the invention it is provided that data symbols to be transmitted are spread in a transmitter with a spreading code and despread in the receiver and that each data symbol intended for a receiver is predistorted in the transmitter by a predistortion coefficient specific for this data symbol. By choosing a targeted Predistortion, the transmission signal can then be prepared for transmission over the transmission channel in such a way that the best possible transmission result is achieved, ie the probability of bit errors at the receiver or receivers is minimal. Taking advantage of the knowledge of the data symbols to be transmitted, it becomes possible to achieve this goal with relatively simple means and with little computational effort. This also makes it possible, for example, to implement this method in a mobile station when it becomes the transmitter.

In principle, it is possible to determine the predistortion coefficient in a type of control loop, with an initial predistortion coefficient being changed until a corresponding simulation result is achieved. However, the predistortion coefficients of all symbols of the transmitter for all receivers can also be determined together in such a way that the mean bit error probability at all receivers (BER) is minimal, whereby

a) the bit error probability is calculated by predicting the distances of the received symbols on a detector of the receiver from the decision thresholds of the detector for each transmitted bit of a symbol and the estimated variance of the additive noise at each receiver,

(b) the reception symbols are calculated from the transmission symbols for all receivers and through the entirety of the transmission links of all receivers, which consists in particular of predistortion, spreading, prefiltering, radio channel, reception filtering and despreading, and

c) the totality of the coefficients is determined in such a way that the average total power at the output of the transmitter does not exceed a permissible level due to the predistortion. All predistortion coefficients can expediently be determined with an iterative nonlinear optimization with a secondary condition.

It proves expedient to use sequential quadratic programming (SQP) as an iterative non-linear optimization method.

Various termination conditions for the calculation of the predistortion coefficients are specified below, which can act individually or at least partially together.

On the one hand, provision is made for the calculation of the predistortion coefficients to be terminated if the predicted bit error probability falls below a predetermined limit.

On the other hand, it is provided that the calculation of the pre-distortion coefficients is terminated if a specified maximum number of iterations is exceeded.

Thirdly, it is provided that the calculation of the coefficients is terminated if the gradient of the bit error probability with respect to all predistortion coefficients falls below a specified limit.

In a preferred embodiment of the first variant of the method according to the invention, in which the predistortion coefficients are calculated, it is provided that the distances between the reception symbols at the detector and the decision thresholds of the detector are calculated for each transmitted bit of a symbol in that

a) the partial code correlation matrices of the spreading codes of all recipients are calculated,

b) a symbol change for a specific realization of the radio channel and certain transmission filters and reception filters interaction matrix is calculated, which characterizes the influence of each possibly predistorted transmission symbol of each receiver on each reception symbol of each receiver,

c) the reception symbols on the detectors of the receivers are calculated therefrom with the transmission symbols and the predistortion coefficients, and

d) the distance from the decision thresholds of the detector for each transmitted bit of a symbol is thereby calculated for each transmitted bit of a symbol.

One embodiment of the method according to the invention is characterized in that within the iterative optimization method for determining the predistortion coefficients, the first derivative of the bit error probability is calculated analytically based on the predistortion coefficients.

Another embodiment of the method according to the invention provides that within the iterative optimization method for determining the predistortion coefficients, the second partial derivatives of the bit error probability are calculated analytically based on the predistortion coefficients.

In a second variant of the method according to the invention, it is provided that the transmission signal is not necessarily generated depending on the data symbols to be transmitted and is changed during the simulation until a reception signal is simulated at the receiver or the receivers of the model, the reception symbols of which correspond to the correspond to sending data symbols with a calculated predicted bit error probability. The transmission signal corresponding to this state of the simulation is then sent to the receiver.

There are no requirements regarding the structure of the transmission signal generation. Rather, such a transmission signal is generated which most likely causes a reception signal at the receiver, the reception symbols of which correspond to the transmission symbols. This inventive approach is only possible in the transmitter in which the data symbols to be transmitted are known. The introduction of the transmitter-integrated simulation of the transmission makes use of the knowledge of the data symbols to be transmitted in order to generate reception symbols in the simulation which largely correspond to the data symbols to be transmitted. With the transmission signal obtained in this simulation, the reception symbols that are also to be generated are then generated with the greatest probability in the receiver.

It is expedient here that the bit error probability

a) the prediction of the distances of all bits of the reception symbols at the detector to the decision thresholds of the detector and

b) and an estimated additive noise variance at each receiver is calculated,

wherein the receive symbols from the elements of the transmit signal

(Chips) and a system matrix are calculated, which contains the effect of all elements of the transmission signal (chips) on the reception symbols of all users by the total of the transmission links, which consists in particular of radio channel, reception filtering and despreading.

In the same way as in the first variant, it is also provided in the second variant of the method according to the invention that the transmission signal is determined using an iterative optimization method with a secondary condition in such a way that the predicted bit error probability is reduced and the transmission signal does not exceed a predetermined power. It is particularly expedient that sequential quadratic programming (SQP) is used as the iterative non-linear optimization method.

Various termination conditions for calculating the transmission signals are specified below, which can act individually or at least partially together.

On the one hand, it is provided that the calculation of the transmission signal is terminated if the predicted bit error probability falls below a defined limit.

On the other hand, it is provided that the calculation of the transmission signal is terminated if a predetermined maximum number of iterations is exceeded.

Thirdly, it is provided that the calculation of the transmission signal is terminated when the gradient of the bit error probability with respect to all transmission signal elements falls below a predetermined limit.

An embodiment of the second variant of the method according to the invention is characterized in that within the iterative optimization method for determining the transmission signal, the first derivative of the bit error probability is calculated analytically based on the transmission signal.

In a further embodiment it is provided that within the iterative optimization method for determining the transmission signal, the second partial derivatives of the bit error probability are calculated analytically based on the transmission signal.

In a further preferred embodiment, the phases of the transmission signal are used as an optimization variable.

The following advantageous configurations are possible in both variants of the method according to the invention. It is possible to send several data streams with different spreading codes simultaneously within one transmitter. According to a favorable embodiment of the method, these different data streams are treated like different transmitters. Several logical transmitters then exist within a physical transmitter, the method according to the invention then being used for each logical transmitter. This enables the method according to the invention to be suitable for a so-called multi-code transmission, that is to say that one or more users simultaneously send several data streams with different spreading codes and these different data streams are treated like different users

Furthermore, the method according to the invention can be designed by using several antennas in the transmitter or in one or more receivers. This means that several transmit and / or receive antennas can be used to further reduce the probability of bit errors.

The method according to the invention is intended in particular the CDMA method (CDMA = Code Division Multiple Access) as a multiple access method. Here are particularly suitable

Direct sequence CDMA (DS-CDMA) or multi-carrier CDMA (MC-CDMA).

A preferred embodiment of the method according to the invention provides that the signal processing takes place with a digital signal processor or microprocessor or a system-specific circuit.

On the arrangement side, the task according to the invention is achieved in that means for simulating the transmission link are arranged in the transmitter, which have an effect connection with at least one of the processing units.

The means for simulation make it possible to transmit the

To estimate the effects of the transmission signal in one or more receivers. Through the operative connection to the processing unit, the transmission signal is then influenced depending on the simulated effect.

In an expedient refinement of the invention, it is provided that the means for simulating the transmission link as a controlled system are part of a control loop, the controlled variable output of which is formed by a simulated receiver input and via at least one processing unit as a controller, the setpoint input with a bit error probability to be achieved or with a transmit power to be reached can be applied, is fed back to the manipulated variable input of the controlled system, the manipulated variable input is simultaneously connected to the at least one processing unit and the processing unit is connected to the transmitting unit via a decision unit. With such an arrangement, it becomes possible to set either the transmission signal itself or the predistortion coefficients by means of feedback so that the best simulation result is achieved. The best simulation result is achieved if either the bit error probability is minimal for a given transmission power or the transmission power is probably minimal for a given bit error. As soon as this occurs, the transmission signal is influenced accordingly, i.e. either the "regulated" transmission signal is transmitted or the transmission signal with the "regulated" predistortion coefficients is predistorted in a targeted manner. This is done with the decision unit. Termination conditions are implemented in the decision unit, which specify when the result of the simulation is used for the transmission of signals.

Accordingly, a first embodiment of the arrangement according to the invention is characterized in that the processing unit executes the transmission signal in a pre-distorting manner and the control system is acted upon by the corresponding pre-distorted transmission signal as a manipulated variable. Likewise, in a second embodiment of the arrangement according to the invention it is provided that the processing unit is designed to generate the transmission signal and the control system is acted upon by the transmission signal of the processing unit as a manipulated variable.

The invention is to be explained in more detail below with the aid of two exemplary embodiments. In the accompanying drawings

1 shows a CDMA transmission path from a base station to three receivers, for example mobile users, with a predistortion unit according to the invention,

2 shows a detailed example of the unit according to the invention for controlling the predistortion,

3 shows a QPSK reception constellation and the calculation of the code correlation matrix

4 shows exemplary simulation results in which the bit error probability of methods of the prior art (RAKE without predistortion, IF predistortion) is compared with that of the method according to the invention for different values of the signal-to-noise ratio,

5 shows a transmitter (base station) with two users and three transmitting antennas according to the prior art,

6 shows a transmission path from a base station with three transmitting antennas to two mobile stations (users), each with two receiving antennas

7 shows a unit according to the invention for calculating the transmission signal according to the invention,

8 shows a QPSK reception constellation and 9 shows simulation results in which the bit error probability of methods of the prior art

(RAKE without predistortion, IF predistortion) is compared with that of the method according to the invention for different values of the signal-to-noise ratio.

In the exemplary embodiments, the values are generally complex. Vectors are highlighted by small bold formula symbols, matrices by large bold formula symbols and scalars by italic formula symbols. A Hermitsche of a matrix or a vector is identified by a superscript H, the transpose of a matrix or a vector is identified by a superscript T, and the conjugate complex operation is identified by a superscript star. The real part formation is marked with 5R and the imaginary part formation is marked with 3. With blockdiag (x) the

Formation of a block diagonal marked. The description is in the baseband, that is, discrete. The data are transferred in blocks.

A first exemplary embodiment shows the implementation of a targeted influencing of the transmission signal by means of predistortion. This is shown in FIGS. 1 to 4.

The first exemplary embodiment reproduces a simplified transmission system in an embodiment according to the invention, which is described below. A simplified TDD-CDMA system similar to the 3GPP-TDD or TD-SCDMA standard is used as the CDMA transmission path, for example, which is described in Esmailzadeh, Nakagawa: “TDD-CDMA for wireless communications”. Artech House, London 2003 ,

In the first exemplary embodiment, the downlink for U simultaneously active mobile users 1 is described, in FIG. 1

U - 3 users 1 shown. A data block which is intended for a user 1 as a receiver consists of K symbols. QPSK is used as the modulation. A data symbol k of the bottom user 1 has, for example, the QPSK constellation

where e

ine

The assignment of the bits to the symbols is done by Gray labeling. BPSK modulation can be seen as a special case of QPSK modulation. 3 shows this signal constellation 2 in the complex plane. The data symbols 3 of a wth user 1, as shown in FIG. 1, can be arranged in a vector d _u = (l), .., d _u (K) ^~ , the vector of the data symbols

3 of all users 1 is d = [d [, .., dy | , In the predistortion unit

4, each data symbol 3 d _u (k) of each user 1 is multiplied by a specific complex predistortion factor _u {^). The determination of these coefficients is part of the invention and will be explained later with reference to FIG. 2. The spreading code of the wth user 1, which contains the scrambling and channeling code, is represented as a vector c _u = [c _M (l), .., c _H (G) l. The elements c _κ (m) are as

Called chips. G is called the spreading factor or spreading gain. The energy of the spreading codes is standardized, in

Following with an example

The spreading codes are in the spreading code matrix C _u = block diagram (c _H , .., c _u ) of size

(GKxK) arranged. The data symbols 3 are spread with the spreader 5. In one embodiment, the filtering with a user-specific FIR transmission filter 6 can be carried out with the

Impulse response _{Tx u} , which is also referred to as pre-RAKE, take place. The spread signal of the wth user 1 with the

Length (GK) is s _κ = C _κ d _M. The transmission signal at the transmitter 7 is added up for u all users 1, s = ^ s _M and then from the transmission antenna 8 κ = l radiated

The frequency-selective transmission channel for each user 1 is modeled below as a tapped delay line at the chip spacing. The maximum length of the channel impulse responses is L chips. In the following, the channel is assumed to be constant or almost constant for the duration of the data transmission of a data block. The channel impulse response from the transmitter 7 to the receiver

9 of the wth user is h _{I (} =

(l), .., h _u (L)]. The Toeplitz structured channel convolution matrix can be used for this

be formed. In addition, an additive noise at the input of each receiver 9 is modeled on the adder 10. The reception signal at the receiver of the wth user 1 is then

This is then optionally followed by a RAKE receiver or a simple despreader 12. In one variant, the RAKE receiver is used as a channel-adapted FIR filter 11 with the

Channel impulse response h _ÄA . , followed by the despreader 12. In the despreader 12, the received signal is correlated with the complex conjugate spreading code sequence and im

Symbol clock sampled. In the despreader 12 this is due to the multiplication by the conjugate complex transposed

Spreading code matrix C symbolizes. At the output of the detector 13, the reception symbols d _u (k) (which are subject to noise and interference) are then present, followed by a following Demodulation bits can be assigned again.

In the unit for calculating the code correlation matrices 14, two partial code correlation functions φf _v {n) and ψi _{, v} {n) are carried out from the spreading codes of all users 1 involved

"-1

^< P _u l (n) = ∑c _v (Gn + i) c _u {ή and i = 0

calculated, where n are the relative delays (in chips) between the code sequences of the u-th and the v-th user 1. In the exemplary embodiment, the assumption is that only short

Codes (short codes) of length G are used and the

Channel influence length does not exceed a symbol length, applies to n = 0..G.

3 shows the calculation of the partial code correlation

15 clarifies.

In the unit for calculating the symbol interaction matrix 16, the interaction γ of the transmission symbols of the different users 1 with each other is calculated at different times, which occurs at the receiver 9. The transmission channel, the transmission filter 6 and the reception filter are also taken into account, as are the spreading codes of the different users 1.

The calculation of this matrix is shown below as an example for a transmission system with RAKE receivers 11. It is assumed that in this exemplary embodiment the channel impulse response is smaller than the symbol duration (L <G). This has an effect on a symbol of a user 1 from the previous, simultaneous and subsequent symbols of all users 1. The influence of the previous symbol of the vth user 1 to the current symbol of the wth user 1

L is the index of the RAKE finger of the RAKE receiver 11 of the wth user 1 and m is the channel coefficient index of the vth user 1. Furthermore, the influence of the current symbol of the vth user 1 the current symbol of the wth user 1

The influence of the next symbol of the vth user 1 on the current symbol of the wth user 1 is

L ι- \ rc, _v , u = ∑∑ {ι) h {™) φ ^{ *} {G- + rn-) l = \ m = \ The calculated values are independent of the data and the predistortion vector. They therefore only have to be calculated once per transmission block, or only in the event that the channel coefficients change. In the event that the channel influence length is longer than a symbol duration {L> G), further symbol interaction matrices must be calculated accordingly.

Each possibly complex data symbol 3 d _u (k) to be sent is multiplied by a possibly complex coefficient CX _u (k). These coefficients are in unit 17 as

Vectors a _u = [ _u (l), .., _u (K) and

provided.

The reception symbols on the detectors 13 of the individual mobile receivers 9 are predicted in the unit for calculating the reception symbols 18 subject to interference. That in the

Receivers 9 additive noise (AWGN) occurring first not considered. The predistortion coefficients α, the data symbols d to be sent and the symbol interaction matrices γ must be present at the input to unit 2.3. Then the reception symbol with interference

In the unit for calculating the distances to the decision threshold 19, the distances to the decision-making thresholds of the detector 13 are calculated in the mobile receivers 9. It is

the distance to the decision threshold of the in-phase or I-

Component of the QPSK symbol. With negative ζ _l (k) there is a wrong decision even in the absence of additive noise. The distance to the decision threshold of the quadrature phase or Q component is also calculated accordingly

In unit 19, the distances for all symbols k = l..K of a transmission block for all users w = 1 .. are calculated.

In the unit (not shown) for calculating the noise variances at the receiver 9, the variances of the additive noise at the individual detectors 13 are calculated taking into account the respective structure of the receiver 9. The noise variance per symbol at the input of the wth

Receiver 9 is σ. In the event that only a simple despreader 12 is used in the receiver 9, the result is σ = σ ' _u ² .

In the event that a RAKE receiver 11 is used, the variance at the detector 13 is calculated as follows: σ _u ² = σ ^a _u ∑K {m) h _u (m). m = l

The bit error probability, which is also referred to as BER or P _e , is predicted in the unit for predicting the bit error probability 20:

In the unit for calculating the transmission power 21

Performance of the overall radiated transmission signal ^'is calculated for all users and compared with the predetermined transmission power E _Bl of a transmission-symbol block. This results in the performance constraint g (α) = s ^H sE _ß; .

which is zero if the secondary condition is met. For orthogonal codes and if no Pre-RAKE 6 is used, the following advantageous execution of this calculation is possible: g () = ^H -E _m .

In the iterative optimization unit 22, the

Objective function _e (α) is minimized under the constraint g (α), ie the vector α is determined which is the

Bit error probability -P _e (α) minimized and sends at the same time with the specified transmission power. The optimization task is

a _opt = arg min P _e (α) with constraint g (α) = 0 α

In an advantageous embodiment, the optimization tasks can be transformed into a problem without secondary conditions by Lagrangian multipliers,

F (, λ) = P _e {) + λg (). Methods for nonlinear optimization with secondary conditions can be used, which are part of the prior art. The procedures provide a local minimum, a global minimum can usually not be guaranteed. Possible methods are described, for example, in M. Papageorgiou: "Optimization", Oldenbourg-Verlag Munich, 1996. In an advantageous embodiment of the invention, the method of "Sequential Square Programming", SQP is used, which is described, inter alia, in the reference mentioned.

It may be necessary to calculate the analytical first derivatives of the objective function and the secondary condition according to α in the unit for calculating the Hesse matrix 23 and to make them available to the unit 2.7. This can improve the computing speed and the convergence behavior of the iterative nonlinear optimization algorithm.

It may be necessary to calculate the analytical second derivatives of the target function and the secondary condition according to α in unit 23, which are referred to as the Hessian matrix, and to make them available to unit 22. This can improve the computing speed and the convergence behavior of the iterative nonlinear optimization algorithm.

The unit 22 supplies a new predistortion coefficient vector α in each iteration, which can be used in the unit 17 for predistortion. From this, new values of the objective function can then be calculated in units 18 19 and 20, optionally new first and second partial derivatives of the objective function in unit 23. The process continues until a desired termination criterion is met. A maximum number of iterations or a maximum number can be provided as the termination criterion of function value calculations or to use a minimum possible value of the gradient or a minimum possible value of the target function.

In an advantageous embodiment of the invention, the SQP method is used in the unit for iterative optimization 22. The SQP method requires a start vector for the optimization variable α. Any vector can be chosen for this, for example a random vector, a vector in which each element is one or a vector which is the solution to another method for calculating transmission signals, for example a method which was mentioned in the explanation of the prior art ,

In an advantageous embodiment variant of the invention, the analytical first derivatives of the target function and / or the secondary condition can be calculated in the unit 23 according to the vector of the predistortion coefficients α:

In this way, all elements of the so-called Jacobi vector p of the first derivatives are calculated,

p (v, k). The Jacobi vector

p is provided at the outlet of unit 23 and can be used as

Serve input of unit 22. The first derivatives of the objective function can be calculated as follows:

These derivatives in a vector

p = [q (l, ϊ), .., q (l, K), .., q (U, K)] and are arranged at the exit of the

Unit 24 provided and can serve as the input of unit 22.

In an advantageous embodiment of the invention, the elements of the Hesse matrix of the second derivative of the target function can additionally be in the unit 23

, , „,, are calculated and in a matrix A d _u {k) d _v {k ')

to be ordered.

4 shows exemplary simulation results in which the bit error probability of prior art methods (RAKE without predistortion, IF predistortion) is compared with that of the method according to the invention for different values of the signal-to-noise ratio. Spreading codes from document 3GPP TS 25.223, V.4.0.0, 2001 http are used. //www.3gpp.org. with the spreading factor G = 16, and QPSK modulation. The channel with coefficients in the chip spacing is constant during a block and has a power

Delay profile of [0, -3, -6, -9] eÜ? , so the channel influence length is = 4 chips. An example is shown with 3 active users 9. It can be seen that the invention (Proposal Min BER) compared to the conventional RAKE receiver and has an improved bit error probability compared to the linear IF channel equalizer. For three users 1, the proposal has a bit error probability comparable to that of a conventional 1-user system without any interference (single user in FIG. 4).

In a variant of the digital transmission system, a channel-matched filter (Pre-RAKE) can also be used in the transmitter. Then the RAKE receiver in the mobile receiver unit can be replaced by a simple despreader. The essential elements of the invention are not influenced by this, but the unit for calculating the symbol interaction matrix 15 must be modified, which is described below.

The influence of the previous symbol of the vth user 1 on the current symbol of the wth user is

Here, the index of the channel w-th user is 1 and m is the pre-RAKE filter coefficient index of the v-th user 1. Furthermore, the influence of the current symbol of the v-th user is FIG. 1 on the current one Icon of the wth user 1.

J ^ J ^ _» .,,. φ ^ (ml) for m> ll = \ m = l φ ^{1) (Gl + ml) ffit m ≤ l

L ll r _{c, v, u} = ∑∑K (l) K HφS (G-l + ml).

The calculated values are independent of the data and the predistortion vector. They therefore only have to be calculated once per transmission block, or only in the event that the channel coefficients change. In the event that Channel influence length is longer than a symbol duration (L> G), further symbol interaction matrices must be calculated accordingly.

In a further advantageous embodiment of the invention, other methods of nonlinear optimization with secondary conditions can be used in the iterative optimization unit 22, which belong to the prior art. In particular, this includes the method of optimization with conjugate gradient and penalty function (method of conjugate gradient with penalty function).

In a further advantageous embodiment of the invention, a plurality of antennas 8 are used on the transmitter 7 and / or on one or more receivers 9. When using several transmit antennas 8, the transmit filter 6 is replaced by a spatiotemporal filter, which in an advantageous embodiment is a spatiotemporal pre-RAKE. If several receiving antennas are used in one receiver, unit 12 is replaced by a spatial-temporal filter, which in an advantageous embodiment is a spatial-temporal RAKE. For both cases mentioned, the unit 16 for calculating the symbol interaction matrix must then be modified such that all influences from all antennas are taken into account.

In a further embodiment variant of the invention, the modulation used is not QPSK, but another digital modulation, for example QAM or PSK. Then the unit 19 is to be modified so that the distances to the respective

Decision thresholds are calculated. The unit 20 is to be modified such that the bit error probability for the modulation used is calculated, possibly by an intermediate calculation of the symbol error probability. In a further embodiment variant, long codes are used for the spreading in the spreader 5 and for the despreading in the despreader 12, that is to say codes in which the spreading code length exceeds the symbol duration. The symbol interaction matrix in the unit 16 must then be recalculated for each data symbol k and the unit 18 modified accordingly.

In a further embodiment variant, one or more users 1 simultaneously send a plurality of data streams with different spreading codes, which is referred to as multi-code transmission. Then the different spreading codes of a user must be treated like different users in the invention.

A second exemplary embodiment is shown in FIGS. 5 to 9. The same assemblies are shown with the same reference numerals as in the first exemplary embodiment.

This shows the generation of transmission symbols which are suitable for generating 9 reception symbols with the lowest bit error probability at the receiver. The data are transferred in blocks. Pulse-shaping filters in the transmitter and receiver are not considered in the exemplary embodiment, but can be integrated into the channel model.

5, a conventional transmission link (downlink, downlink) is first described. This shows a conventional formation of the transmission signal in the transmitter 7 (base station). 6 shows the further transmission path with channel and mobile users (receivers 9). U mobile users 9 are active at the same time. The base station has K transmission antennas 8, K = 1 being explicitly included. The receivers 9 each have Q receiving antennas 24, with β = 1 being expressly included. A data block intended for a receiver 9 consists of N symbols. QPSK is used as the modulation. A data symbol n of the wth user 1 has, for example

QPSK constellation ^ e ^ ^' 1 _{+ j.} -1 --1 _{+ j} . ^ 1, ^ 1 - j. 1 1 j.11

where the bits are assigned to the symbols by gray labeling. BPSK modulation can be seen as a special case of QPSK modulation. This signal constellation 25 is shown in the complex plane in FIG.

The data symbols 3 of a wth user 1 can be in a vector d _u =

be the vector of

df, .., dyj. Subsequently, data predistortion can take place in a predistortion unit 4 (for example joint transmission, Viennese transmission filter) according to the prior art, or no data predistortion takes place. The spreading code of the wth user 1, which contains the scrambling and channeling code, is represented as vector c, ₍ = _u (l), .., c _{i (} (G) l. The elements c _u (m) are represented as

Following with an example

The spreading codes are arranged in the spreading code matrix C _H = block diagram (c _H , .., c _H ) of size (GNXN). The data symbols are spread in the spreader 5. In one embodiment, the filtering can be carried out with a user and transmitter antenna-specific FIR transmission filter 6 with the impulse response h _{Tx ux} and the maximum filter length L _j . _χ take place. In an advantageous embodiment variant, this transmission filter 6 is a space-time pre-RAKE. The FIR transmit filters 6 are combined in the block-toeplitz-structured channel convolution matrix H _Tx , analogously to the formation of the channel matrix described below. The spread and pre-filtered transmission signal i _k with the length (GN) is located on the k-th transmission antenna 8,

s _ft = ∑C _κ d _w h ^^ an. This signal consists of individual chips.

The total broadcast signal is called s

calculated.

The frequency-selective transmission channel for each user 1 from each transmission antenna 8 to each reception antenna 24 is modeled below as a tapped delay line at the chip spacing. The maximum length of the channel impulse responses is L chips. In the following, the channel is assumed to be constant or almost constant for the duration of the data transmission of a data block. The channel impulse response from the kth transmission antenna 8 of the transmitter 7 to the qth reception antenna 24 of the receiver 9 of the wth

User 1 is h _u ^ _k = h _{u ι} (l), .., h _u ^ _k L) \. The Toeplitz structured channel convolution matrix can be used for this

be formed. This becomes the channel matrix of the overall system

educated. In addition, an additive noise 2.5 is modeled at the input of each antenna 24 of each receiver 9. The reception signal at the q-th reception antenna 24 of the w-th user 1 is then ^T u, _q ~ __ι "-u, _q , k ^S k ⁺ Αu, _q k = l

The entire received signal of all users 1 with all receiving antennas 24 is combined into a vector r = r ₁ , .., T _Q .., r _u , .X _QU j.

In the receivers 9, spatial-temporal reception filters 11, followed by despreaders 12, are used. The spatial-temporal reception filter 11 with the filter length L _Rx have the

Filter coefficients h _ft with the Toeplitz-structured convolution matrix H ^. These can be found in the receive filter matrix for the entire system

^H _Ä , _H = [ ^H * _C , «, I •• ^H Λ *,«, _Q ]> ^ü _H x = ^bl ockdiag (H _Rx .., ^' Α _RXtU )

to be ordered. The spatial-temporal reception filter 11 can be channel-matched filters (space-time RAKE receiver) in one embodiment variant, or be absent in another embodiment variant (formally h ^ = 1) or contain any filter coefficients in another embodiment variant.

In the despreader 12, the received signal is correlated with the complex conjugate spread code sequence and sampled at the symbol clock. In the despreader 12, this is due to the

Multiplication by the conjugate complex transposed

Spreading code matrix (despreading matrix) symbolizes C for the wth user 1 and the entire despreading matrix C ^H formed therefrom for all users 1.

The symbol vector d on the receivers 9 contains both interference and additive noise.

The method according to the invention for Determination of the transmission signals s described in the transmitter 7. The basic principle consists in a simulation of the entire transmission path with a prediction of the bit error probability at the receivers 9, the transmission signal being selected that minimizes the bit error probability and thereby does not exceed the predetermined total transmission power.

The received data symbols with interference, but without additive noise, are d = C ^H H _Rx Iϊs = As, whereby the system matrix A = C ^H H _Rx H, Ae ^NVxKGN with the elements a _pm the influence of each

KGN

Transmit chips on each receive symbol d _un = ∑ a, _ _N ^ _nm s _m describes. m = \

In the unit 2.1 the system matrix A with its elements is a _pm calculated by A = C ^H H _Rx H, Ae ^NUXKGN _{^ in E} _ _ner advantageous embodiment be of the transmitter 7 effort efficiently determined only nonzero elements, and Symmetrienen and band structures of the split dies utilized , The system matrix is independent of the data to be sent and only needs to be changed if the transmission channel or the send or receive filter configurations change. In the unit for calculating interference

KGN

Receive symbols 18, the receive symbols d _un = ∑ a, _l , _{N + nm} s _m m = l are calculated from A and the current transmission signal s in this iteration.

In the unit for calculating the distances to decision threshold 19, the distances become

Decision thresholds of the detector 13 are calculated in the mobile receivers 9. It is

the distance to the decision threshold of the in-phase or I-

Component of the QPSK symbol. With negative ζ, _u (k) there is a

Wrong decision even in the absence of additive noise. The distance to the decision threshold of the quadrature phase or Q component is also calculated accordingly

The distances for all symbols = 1.N of a transmission block for all users w = 1 .. [/ are calculated in the unit for predicting the bit error probability 20.

In the unit for calculating the noise variances 26 at the receivers 9, the variances of the additive noise <7 ² at the individual detectors 13 are calculated taking into account the respective receiver structure. The noise variance per symbol at the input of the wth receiver 9 is σ ^a _u . In the event that only a simple despreader 12 is used in the receiver 9 with only one receiving antenna 24, σ _u ² = σ ^a _u results. In the event that a space-time RAKE receiver is used, the variance at the detector is calculated as follows:

The bit error probability, which is also referred to as BER or P., is predicted in the unit for predicting the bit error probability 20:

In the unit for calculating the transmission power 21, the power of the entire transmitted transmission signal is for everyone User 1 calculated and compared with the predetermined transmission power E _{m of} a transmission symbol block. This results in the performance constraint

g (s) = s ^H s- E _{ß /} ,

which is zero if the secondary condition is met.

In the iterative optimization unit 22, the

Objective function P _e (s) minimized under the constraint g (s), i.e. H . the vector of the transmission signal s is determined, which minimizes the bit error probability P _e (s) and transmits at the same time with the predetermined transmission power. The optimization task is

s _opt = arg min P _e (s) with constraint g (s) = 0 s

F (s, λ) = P _e {s) + λg (s).

Methods for nonlinear optimization with secondary conditions can be used, which are part of the prior art. The procedures provide a local minimum, a global minimum can usually not be guaranteed. Possible methods are described, for example, in M. Papageorgiou: "Optimization", Oldenbourg-Verlag Munich, 1996. In an advantageous embodiment of the invention, the method of "Sequential Square Programming", SQP is used, which is described, inter alia, in the reference mentioned.

The analytical unit 23 can be used to calculate the first derivatives of the objective function and the secondary condition according to s and to make them available to the unit 22. This can improve the computing speed and the convergence behavior of the iterative nonlinear optimization algorithm.

It may be necessary to calculate in unit 23 the analytical second derivatives of the objective function and the secondary condition according to s, which are referred to as the Hessian matrix, and to make them available to unit 22. This can improve the computing speed and the convergence behavior of the iterative nonlinear optimization algorithm.

The unit 22 supplies a new transmission signal vector s in each iteration, which is used in the unit for generating the transmission signals 27 and in the following units for recalculating the bit error probability. From this, new values of the target function can then be calculated in units 18, 19 and 20, optionally new first and second partial derivatives of the target function in unit 23.

The process continues until a desired termination criterion is met. Provision can be made to use a maximum number of iterations or a maximum number of function value calculations or a minimum possible value of the gradient or a minimum possible value of the target function as the termination criterion.

After the iterations have been terminated according to an abort criterion, the resulting transmission signal s is used for transmission from the base station 7.

In an advantageous embodiment of the invention, the SQP method is used in the unit for iterative optimization 22. Iterative optimization methods, such as the SQP method, require a start vector for the optimization variable s. Any vector can be selected for this, for example a random vector, a vector in which each element is one or a vector which is determined by another method for determining the transmission signal. For example, a method can be used that was mentioned in the explanation of the prior art, such as joint transmission. In an advantageous embodiment, the signal s = H _Tx Cd is calculated as the start vector. In a further advantageous embodiment, the signal s = Td, T = (CH _Rx H) ^_1 is used as the start vector.

In an advantageous embodiment of the invention, the analytical first derivatives of the target function and / or the secondary condition can be calculated in the unit 26 according to the vector of the transmission signals s:

The elements p _m are combined in a so-called Jacobi vector p of the first derivatives. p is provided at the output of the unit 23 and can serve as the input of the unit 22.

The first derivatives of the objective function can be calculated as follows:

_ dg {) _ b_ = = 2s "ds"

These derivatives b are arranged in a vector b and are provided at the output of the unit 23 and can serve as the input of unit 22.

In an advantageous embodiment variant of the invention, the elements of the Hesse dP ² (s)

Matrix of the second derivative of the objective function - ^e calculated

and are arranged in a matrix B, can serve as the input of the unit 22.

In an advantageous embodiment of the iterative optimization unit 22, the complex quantities s, p, and B are converted into stacked real quantities with a real part and an imaginary part. Iterative optimization methods for real parameters can therefore also be used with complex parameters.

8 shows exemplary simulation results in which the bit error probability of methods of the prior art (conventional RAKE without predistortion, IF predistortion, Wiener filter / MMSE predistortion) is compared with that of the method according to the invention for different values of the signal-to-noise ratio. Spreading codes from document 3GPP TS 25.223, V.4.0.0, 2001 http are used. //www.3gpp.org. with the spreading factor G = 16, and QPSK modulation. The channel with coefficients in the chip spacing is constant during a block and has a power

Delay profile of [0, -3, -6, -9] dB, i.e. the

Channel influence length is L = 4 chips. An example with three active users 1 is shown. It can be seen that the invention (Min BER chip) has an improved probability of bit errors compared to the conventional RAKE receiver and compared to the linear MMSE channel equalizer (Vienna transmit filter).

In an embodiment variant of the digital transmission system, a channel-matched filter (Pre-RAKE) can also be used in the

Transmitter 7 can be used. Then the reception filter 11 in the mobile receiving unit 9 are omitted, consequently only a simple despreader 12 is used.

In an embodiment variant of the digital transmission system, a channel-matched filter 6 (pre-RAKE) can also be used in the transmitter 7.

In an embodiment variant of the digital transmission system, a so-called self-precoder can also be used as the transmission and reception filter 6, 11. The transmission filter 6 is used only to calculate the initialization transmission signal.

In a further advantageous embodiment of the invention, other methods of non-linear optimization with secondary conditions that belong to the prior art can be used in the unit for iterative optimization 22. In particular, this includes the method of optimization with conjugate gradient and penalty function (method of conjugate gradient with penalty function).

In a further advantageous embodiment of the invention, a plurality of antennas 8, 24 are used on the transmitter 7 and / or on one or more receivers 9. When using multiple transmit antennas 8, the transmit filter 6 is replaced by a spatiotemporal filter, which in an advantageous embodiment is a spatiotemporal pre-RAKE. When using a plurality of receiving antennas 24 in a receiver 9, unit 11 is replaced by a spatial-temporal filter, which in an advantageous embodiment is a spatial-temporal RAKE. For both cases mentioned, the unit for calculating the symbol interaction matrix 16 must then be modified such that all influences from all antennas 24 are taken into account.

In a further embodiment variant of the invention, an initial transmit signal s = s _{0 is} replaced by another Process determined, for example by conventional spreading, a pre-RAKE, or a joint transmission or a Sende-Wiener filter. In the subsequent optimization of the BER, it is not the complex transmission signal s that serves as an optimization variable, but only its phase, which is represented in the vector as the phase angle z. Accordingly, the

Initialization vector of optimization z ₀ = arg (s ₀ ). The

Any condition of the limited transmission power is fulfilled by any z, so if it is already fulfilled by s ₀ , it can be omitted. This simplifies numerical optimization, since optimization methods can now be used without constraints, such as the preconditioned conjugate gradient (PCG) method. In addition, the number of real optimization variables is halved. The resulting transmission signal is s = | s ₀ per ^yz . The modified first

Derivation according to the optimization variable is now

^Pm dz _m 2NUJ2ä £ ϊä σ _H '(ul) N + n, m

In a further embodiment variant of the invention, the modulation used is not QPSK, but another digital modulation, for example QAM or PSK. The unit 19 is then to be modified such that the distances to the respective decision thresholds are calculated and the unit 20 is to be modified such that the bit error probability for the modulation used is calculated, possibly by an intermediate calculation of the symbol error probability.

In a further embodiment variant, long codes are used for spreading in unit 5 and for despreading in unit 12, that is to say codes in which the spreading code length is the same Symbol duration exceeds. Then the calculation of the system matrix in unit 16 must be modified accordingly so that the interaction of each transmitter chip is represented with each receive symbol.

In a further embodiment variant, one or more users simultaneously send several data streams with different spreading codes, which is referred to as multi-code transmission. Then the different spreading codes of a user must be treated like different users in the invention.

Method and arrangement for sending signals after a

Multiple access methods

LIST OF REFERENCE NUMBERS

User signal constellation data symbol predistortion unit spreader transmission filter (FIR filter) transmitter transmission antenna receiver adder reception filter (RAKE receiver) despreader detector unit for calculating the code correlation matrices calculating the partial code correlation unit for calculating the symbol interaction matrix unit for vectoring unit for calculating the receiving symbols subject to interference Unit for calculating the distances to the decision threshold Unit for predicting the bit error probability Unit for calculating the transmission power Unit for iteratively optimizing Unit for calculating the Hesse matrix Receiving antenna signal constellation Unit for calculating the noise variances Unit for generating the transmitted signals

Claims

Method and arrangement for sending signals according to a multiple access method

1. Method for sending signals according to a multiple access method, in which data symbols to be sent are pretreated in a transmitter and sent by the transmitter as a transmission signal over a transmission link between the transmitter and one or more receivers to and by the receiver by means of a radio communication system Receiver can be received as a reception signal, the transmission signal consisting of transmission signal elements and the reception signal consisting of reception elements, characterized in that a simulation with a transmission signal is carried out in the transmitter by means of a model representing the transmission path, and the transmission signal under the condition that the Communication system permissible transmission power is not exceeded, it is pretreated until a receiver signal with a minimal bit error probability is simulated at the receiver or the receivers of the model and this changed transmission signal via the transmission is sent to the recipient (s).

2. The method according to claim 1, characterized in that data symbols to be transmitted are spread in a transmitter with a spreading code and despread in the receiver and that each data symbol intended for a receiver is predistorted in the transmitter by a predistortion coefficient specific for this data symbol.

3. The method according to claim 2, characterized in that the predistortion coefficients of all symbols of the transmitter for all receivers are jointly determined in such a way that the mean bit error probability at all receivers (BER) is minimal, whereby

b) the reception symbols are calculated from the transmission symbols for all receivers and through the entirety of the transmission paths of all receivers, in particular from predistortion, spreading, prefiltration,

Radio channel, reception filtering and despreading exists, and

c) the entirety of the coefficients is determined in such a way that the average total power at the output of the transmitter is a permissible value due to the predistortion

Dimension does not exceed.

4. The method according to claim 2 or 3, characterized in that all pre-distortion coefficients are determined with an iterative non-linear optimization method with a constraint.

5. The method according to claim 4, characterized in that the sequential quadratic programming (SQP) is used as the iterative non-linear optimization method.

6. The method according to any one of claims 2 to 5, characterized in that the calculation of the pre-distortion tion coefficient is aborted if the predicted bit error probability falls below a specified limit.

7. The method according to any one of claims 2 to 6, characterized in that the calculation of the predistortion coefficients is terminated when a predetermined maximum number of iterations is exceeded.

8. The method according to any one of claims 2 to 7, characterized in that the calculation of the coefficients is terminated when the gradient of the bit error probability with respect to all predistortion coefficients falls below a predetermined limit.

9. The method according to any one of claims 2 to 8, characterized in that the distances of the reception symbols at the detector to the decision thresholds of the detector for each bit of a symbol transmitted are calculated in that

b) a symbol interaction matrix is calculated therefrom for a specific realization of the radio channel and specific transmission filter and reception filter, which characterizes the influence of each possibly predistorted transmission symbol of each receiver on each reception symbol of each receiver,

d) the distance to the decision thresholds of the detector for each transmitted bit of a symbol is thereby determined for each transmitted bit of a symbol. be counted.

10. The method according to any one of claims 4 to 9, characterized in that within the iterative optimization method for determining the predistortion coefficients, the first derivative of the bit error probability is calculated analytically based on the predistortion coefficients.

11. The method according to any one of claims 4 to 10, characterized in that within the iterative optimization method for determining the predistortion coefficients, the second partial derivatives of the bit error probability are calculated analytically according to the predistortion coefficients.

12. The method according to claim 1, characterized in that the transmission signal is not necessarily generated depending on the data symbols to be transmitted and is pretreated in the simulation until a reception signal is simulated at the receiver or receivers of the model, the reception symbols of which are to be transmitted correspond to the data symbols with a calculated predicted bit error probability and that the transmission signal corresponding to this state of the simulation is then sent to the receiver.

13. The method according to claim 12, characterized in that the bit error probability

b) and an estimated additive noise variance at each receiver is calculated, wherein the reception symbols are calculated from the elements of the transmission signal (chips) and a system matrix which determines the effect of all elements of the transmission signal (chips) on the reception symbols of all users through the entirety of the transmission links, which consists in particular of radio channel, reception filtering and despreading, contains.

14. The method according to claim 12 or 13, characterized in that the transmission signal is determined using an iterative optimization method with a secondary condition in such a way that the predicted bit error probability is reduced and the transmission signal does not exceed a predetermined power.

15. The method according to claim 14, characterized in that the sequential quadratic programming (SQP) is used as the iterative non-linear optimization method.

16. The method according to any one of claims 12 to 15, characterized in that the calculation of the transmission signal is terminated when the predicted bit error probability falls below a predetermined limit.

17. The method according to any one of claims 12 to 16, characterized in that the calculation of the transmission signal is terminated when a predetermined maximum number of iterations is exceeded.

18. The method according to any one of claims 12 to 17, characterized in that the calculation of the transmission signal is terminated when the gradient of the bit error probability with respect to all transmission signal elements falls below a predetermined limit.

19. The method according to any one of claims 14 to 18, characterized in that within the iterative optimization process for determining the transmission signal, the first derivative of the bit error probability is calculated analytically based on the transmission signal.

20. The method according to any one of claims 14 to 19, characterized in that within the iterative optimization method for determining the transmission signal, the second partial derivatives of the bit error probability are calculated analytically after the transmission signal.

21. The method according to any one of claims 12 to 20, characterized in that the phases of the transmission signal are used as an optimization variable.

22. The method according to any one of claims 1 to 21, characterized in that several data streams with different spreading codes are sent simultaneously and these different data streams are treated like different transmitters.

23. The method according to any one of claims 1 to 22, characterized in that several antennas are used in the transmitter or in one or more receivers.

24. The method according to any one of claims to 23, characterized in that direct sequence CDMA (DS-CDMA) or multi-carrier CDMA (MC-CDMA) is used as the multiple access method.

25. The method according to any one of claims 1 to 24, characterized in that the signal processing is carried out with a digital signal processor or microprocessor or a system-specific circuit.

26. Arrangement for sending signals after a multiple Subject access method, with a transmitter unit arranged in a transmitter and processing units upstream of the transmitter unit, characterized in that means for simulating the transmission link are arranged in the transmitter, which are operatively connected to at least one of the processing units.

27. Arrangement according to claim 26, characterized in that the means for simulating the transmission link as a controlled system are part of a control loop, the controlled variable output of which is formed by a simulated receiver input and via at least one processing unit as a controller, the setpoint input with a bit error probability to be achieved or with of a transmission power to be achieved, each of which can be acted upon as a setpoint, is fed back to the manipulated variable input of the controlled system, the manipulated variable input is simultaneously connected to the at least one processing unit and the processing unit is connected to the transmitting unit via a decision unit.

28. The arrangement according to claim 27, characterized in that the processing unit executes the transmission signal pre-distorting and the controlled system is acted upon with the corresponding pre-distorted transmission signal as a manipulated variable.

29. Arrangement according to claim 27, characterized in that the processing unit is designed to generate the transmission signal and the control system is acted upon by the transmission signal of the processing unit as a manipulated variable.