EP1166393B1 - Beamforming method and device - Google Patents

Description

The invention relates to a method for beam shaping for an adaptive antenna group containing several antenna elements in the downlink of frequency duplex systems, where for the antenna elements antenna weights for transmission in the Downlink based on direction information of the uplink be determined.

Furthermore, the invention relates to a device for Beamforming for one containing several antenna elements adaptive antenna group in the downlink of frequency duplex systems, with a signal processor unit for determining Antenna weights for the antenna elements for transmission in the Downlink based on direction information of the uplink.

It is known to consist of several individual antennas Change the directional characteristics of group antennas electronically, to adapt them optimally to the respective channel situation adapt.

For example, from US 5 778 324 A (which EP 755 090 A corresponds to) a technology for controlling the radio beam in the downlink of a mobile radio system with the help of adaptive Antennas known, the beam control based on directional information the uplink, namely the detection of the relative amplitudes of the multiple beams of the uplink is realized; this involves two groups of antenna elements used, the distances between the antenna elements in Comparison to the wavelength in one group small and in the other group are large, and being the signals of the uplink be weighted.

Adaptive antennas were first used in radar technology, and for some time now its use in cellular examined. The use of adaptive antennas can be one Reduction of the received interference by directed reception, to reduce the interference generated by directed Send and a reduction in the time dispersion of the mobile radio channel and thus reducing the intersymbol interference that the Bit error rate significantly determined, lead.

These improvements can be used to increase capacity Increase in spectral efficiency, to reduce the necessary Transmission power corresponding to the gain of the antenna group, for better transmission quality (lower bit error rate), for a data rate increase and for a range increase be used.

Even if not all advantages are used at the same time could, some of the above could Improvements can be achieved. It would be very important that it with adaptive antennas is possible, the available frequency spectrum to use more efficiently and with the same Frequency band and the same number of base stations capacity and thus the possible number of users in one Increase cell.

Cellular cellular networks are generally interference limited, i.e. that spatial reuse of one and the same Radio channel on the one hand and spectral efficiency on the other is limited by co-channel interference. A radio channel is by its frequency and / or its time slot (in time multiplex - TDMA - Time Division Multiple Access) or its Code (in code division multiplex - CDMA - Code Division Multiple Access) established. To supply more than one participant one and the same radio channel in TDMA and FDMA (Frequency Division Multiple Access) systems have been proposed on the spatial separability and the directionally selective Receive in the uplink (mobile station sends, base station receives) and the directionally selective transmission of the subscriber signals in the downlink (base station sends, mobile station receives) are based (so-called SDMA - Space Division Multiple access system; Multiple Access System). The Directionally selective sending / receiving can also be done in CDMA systems used to determine the possible number of participants to increase a frequency and thus the spectral efficiency and to increase the capacity of a cellular mobile radio system. It with constant interference, the possible number the subscriber on a traffic channel raised by the base station with the linear, adaptive antenna group in the Uplink detected and supplied in the downlink can be.

For the separation of the signals of the individual participants Co-channel interference suppression and its detection are three basic methods known: (1) procedures based on the Knowledge of the spatial structure of the antenna group (so-called spatial-reference method), cf. R. Roy and R. Kailath, "ESPRIT - Estimation of Signal Parameters via Rotational Invariance Techniques ", IEEE Trans. Acoust., Speech and Signal Processing, Vol. 37, July 1989, pp. 984-995; (2) Methods based on knowledge of a known signal sequence (so-called temporal-reference method), cf. in S. Ratnavel, A. Paulraj and A.B. Constantinides, "MMSE Space-Time Equalization for GSM Cellular Systems ", Proc. IEEE, Vehicular Technology Conference 1996, VTC'96, Atlanta, Georgia, pp. 331-335; and (3) so-called "blind" methods, the well-known structural Use signal properties for signal separation and detection, see. in A-J. van der Veen, S. Talwar, A. Paulraj "A Subspace Approach to Blind Space-Time Signal Processing for Wireless Communications Systems ", IEEE Transactions on Signal Processing, Vol.45, No. 1, January 1997, pp. 173-190.

Different methods are used for the downlink, based on different estimates of the cellular channel build up. Basically, either the directions of incidence the signals of the mobile stations (see e.g. US 5 515 378 A or US 5 778 324 A EP 755 090 A) is used, or it becomes the spatial covariance matrix (spatial correlation matrix) used for beam shaping (cf. US 5,634,199 A).

The different carrier frequencies pose a difficult problem in frequency division duplex systems (FDD systems). In FDD systems use the signals in the uplink and in the downlink broadcast on different frequencies, and thereby the necessary separation between sent and received data at the mobile and at the base station. On The antenna directional diagram is due to the frequency difference Using the same physical antenna group and the same Antenna weights (amplitude and phase) with different Frequencies different. That is why it is not advisable same antenna weights for sending and receiving at the base station to use a cellular mobile radio system. The exclusive use of those estimated in the uplink With this frequency offset there is no direction of incidence Problems, however, limits beam shaping to a single one discrete direction of incidence, reflecting the physical nature of the cellular channel contradicts and therefore to a limited capacity gain through the adaptive antenna. The usage of the spatial covariance matrix of the uplink, however the problem of frequency offset with it.

For the compensation of this frequency duplex distance in the spatial covariance matrix have been different approaches described. So it is proposed in the uplink Direction of incidence, the signal power and the associated angular spread (Angular Spread) of each individual participant estimate, cf. T. Trump and B. Ottersten, "Maximum Likelihood Estimation of Nominal Direction of Arrival and Angular Spread Using an Array of Sensors ", Signal Processing, Vol. 50, No. 1-2, April 1996, pp. 57-69. From this estimate for the uplink is an estimate of the spatial covariance matrix for formed the downlink, cf. also P. Zetterberg, "Mobile Cellular Communications with Base Station Antenna Arrays: Spectrum Efficiency, Algorithms and Propagation Models ", Dissertation, Royal Institute of Technology, Stockholm, Sweden 1997. However, this method only works if every mobile station just a single nominal direction of incidence with respect to the base station owns. By reflections on mountains in rural Areas or large building complexes in urban areas this requirement is often not met, and therefore is then this approach is not applicable.

Another proposal in the prior art goes in that Base station two different, with the wavelength used scaled antenna groups for sending and receiving in to use a frequency duplex system, cf. G. G. Rayleigh, S.N. Diggavi, V.K. Jenes and A. Paulraj, "A Blind Adaptive Transmit Antenna Algorithm for Wireless Communication ", Proceedings IEEE International Conference on Communications (ICC'95), IEEE 1995, pp. 1494-1499, or the corresponding WO 97/00543 A. However, here the two "adjusted" Antenna groups are manufactured and calibrated very precisely and be placed in exactly the same position. Besides, is a second antenna group is necessary, which is disproportionately expensive elevated.

According to the already mentioned US 5 634 199 A, the spatial covariance matrix of the downlink by sending of test signals from the base station and returning the measured signals are measured by the mobile station (cf. also WO96 / 37975, where likewise on the sending of test signals is pointed out). However, this test signal method requires System capacity for this feedback process and reduced hence the possible increase in capacity. Furthermore, the Standard of existing mobile radio systems are changed, since so far in no cellular mobile radio system Feedback of the mobile station of this type is provided.

No. 5,848,060 A describes that the spatial covariance matrix is estimated from the received signals of the uplink; the occurring relative phases of the matrix elements are then scaled with the ratio of the transmission frequency to the reception frequency (f _S / f _E ). Due to the multipath propagation of the individual signals, however, the frequency is non-linear in the phase relationship of the individual antenna elements. Therefore, this application is limited to cases with a direct line of sight between the transmitter and receiver without reflections from different directions, such as in satellite communication.

In order to obtain a covariance matrix for the downlink, it has also been proposed to apply a rotation matrix to the covariance matrix for the uplink, which corrects the phases of a wave incident from a specific direction with the ratio of the transmission frequency to the reception frequency f _S / f _E , cf. the aforementioned GG Rayleigh, SN Diggavi, VK Jenes and A. Paulraj, "A Blind Adaptive Transmit Antenna Algorithm for Wireless Communication", Proceedings IEEE International Conference on Communications (ICC'95), IEEE 1995, pp. 1494-1499. However, only the phase relationship of an incident direction with respect to the base station is corrected correctly here. This method fails when several different directions of incidence occur, and it is therefore only applicable to rural areas with a dominant direction of incidence.

In the above-mentioned dissertation by P. Zetterberg, "Mobile Cellular Communications with Base Station Antenna Arrays: Spectrum Efficiency, Algorithms and Propagation Models", Royal Institute of Technology, Stockholm, Sweden 1997, the proposal also includes a compensation matrix on the covariance matrix the upward route. This compensation matrix is only valid for very small relative duplex spacings 2 (f _S -f _E ) / (f _S + f _E ) and is averaged over the entire application angle range of the adaptive antenna. This method does not correct the frequency difference, but only reduces the deviation and "smears" the spatial structure of the mobile radio channel contained in the covariance matrix over the entire angular range. For this reason, this method cannot be used under any circumstances.

Finally, the covariance matrix has already been proposed to decompose the uplink into Fourier coefficients and at to restore the transmission frequency, cf. J.M. Goldberg and J.R. Fonollosa, "Downlink beamforming for spatially distributed sources in cellular mobile communications ", signal processing Vol.65, No.2, March 1998, pp. 181-199. This method tries the exact phase relationship of the individual signal paths on the transmission frequency to restore, but also smears the spatial structure of the covariance matrix.

The aim of the invention is now a method and a device to be provided as indicated at the beginning, with the or in an efficient manner with FDD systems Beamforming in the downlink allows the Interference also with those transmitted by the base station the signals received by the mobile stations are reduced and an increase in the number of users that can be served, i.e. Mobile stations is made possible.

For this purpose, the method according to the invention is the one mentioned at the beginning Kind characterized in that the antenna weights for transmission in the downlink based on the power angle spectrum the uplink of each user are determined, the power angle spectrum by Hiding unwanted areas is modified.

The device according to the invention is correspondingly of the type mentioned above, characterized in that the Signal processor unit for determining the antenna weights for the transmission in the downlink on the basis of the power angle spectrum the uplink of each user under its modification by hiding unwanted Areas is set up.

In the technique according to the invention, beam shaping is thus used in the downlink the power angle spectrum of the The upward link of the individual user is used, whereby undesirable angular ranges in this power angle spectrum are hidden, i.e. any interferers are in the power angle spectrum hidden for optimal alignment of the Main club in the direction of the respective user. According to the invention, it becomes the important, useful Extracted areas of the power angle spectrum and the Determination of the antenna weights for beam shaping in the Downward route used as a basis. Have investigations shown that it is possible to get particularly good results in terms of interference suppression if only one dominant part in the power angle spectrum cut out of this becomes.

It is advantageous if the power angle spectrum using a known signal sequence of the transmission signal, such as spreading code, midamble etc. is estimated. It is also from Advantage if the power angle spectrum of the uplink based on the spatial covariance matrices of the uplink of the individual users or, if applicable, of mean values the same is estimated. Furthermore, it has proven to be cheap proven if based on the modified power-angle spectrum of the individual user or of its mean value the respective spatial covariance matrix of the downlink is determined becomes. Finally, it is advantageous if the spatial covariance matrix the downlink or its mean value for calculation the antenna weights are used for transmission.

• Schätzung der räumlichen Kovarianzmatrix der Aufwärtsstrecke;
• Bestimmung des Leistungs-Winkelspektrums mit Methoden der spektralen Suche auf der Empfangsfrequenz;
• Rekonstruktion der räumlichen Kovarianzmatrix der Abwärtsstrecke unter Verwendung des geschätzten, modifizierten Leistungs-Winkelspektrums auf der Sendefrequenz; und
• Berechnung der Antennengewichte für jeden Benutzer des physikalischen Kanals besteht.

A beam shaping of the spatial properties of the mobile radio channel in relation to the spatial covariance matrix is thus preferably carried out, which consists of the four steps

• Estimation of the spatial covariance matrix of the uplink;
• Determination of the power angle spectrum using methods of spectral search on the receiving frequency;
• Reconstruction of the spatial covariance matrix of the downlink using the estimated, modified power-angle spectrum on the transmission frequency; and
• Calculation of the antenna weights for each user of the physical channel exists.

The invention Technology is unrestricted by the spreading conditions of electromagnetic waves applicable. It is subject to no restrictions on a single dominant Direction of incidence per participant and is without additional Hardware expenditure can be used. There are no assumptions about that Frequency difference between transmission and reception, and therefore the technique described here also works independently of relative duplex spacing. It is neither expensive iterative approximation methods or high-resolution direction estimation algorithms needed, so a very computationally efficient Solution is achieved.

Fig. 1 in einem Schema eine adaptive Antenne mit Strahlformung in der Abwärtsstrecke;

Fig. 2 schematisch eine lineare Antennengruppe mit einer einfallenden Welle, zur Veranschaulichung von Wegunterschieden;

Fig. 3 schematisch eine Vorrichtung zur Strahlformung, wobei eine Basisstation und mehrere Mobilstationen gezeigt sind;

Fig. 4A ein Antennendiagramm bei einer Aufwärtsstrecken-Frequenz;

Fig. 4B ein entsprechendes Antennendiagramm bei der Frequenz der Abwärtsstrecke;

Fig. 5 in einem Ablaufdiagramm die Ermittlung der Antennengewichte für die Strahlformung in der Abwärtsstrecke;

Fig. 6 in einem detaillierten Ablaufdiagramm den Vorgang bei der in Fig. 5 gezeigten Frequenztransformation;

Fig. 7 ein Leistungs-Winkelspektrum eines Benutzers zusammen mit "Störern";

Fig. 8 ein zu Fig. 7 gehöriges Antennendiagramm noch vor der Modifikation;

die Fig. 9 und 10 den Fig. 7 und 8 entsprechende Diagramme des Leistungs-Winkelspektrums und der Antennencharakteristik, nun jedoch nach dem Ausblenden eines Störers; und

Fig. 11 schematisch die Struktur der Signalprozessoreinheit zur Berechnung der Antennengewichte für die Strahlformung.

The invention is further explained below by means of examples and with reference to the drawing. Show it:

1 shows a diagram of an adaptive antenna with beam shaping in the downlink;

2 schematically shows a linear antenna group with an incident wave, to illustrate path differences;

3 schematically shows a device for beam shaping, a base station and a plurality of mobile stations being shown;

4A is an antenna diagram at an uplink frequency;

4B shows a corresponding antenna diagram at the frequency of the downlink;

5 shows in a flow chart the determination of the antenna weights for the beam shaping in the downlink;

FIG. 6 shows a detailed flowchart of the process for the frequency transformation shown in FIG. 5;

7 shows a power angle spectrum of a user together with "interferers";

FIG. 8 shows an antenna diagram belonging to FIG. 7 even before the modification;

9 and 10 the diagrams corresponding to FIGS. 7 and 8 of the power angle spectrum and the antenna characteristic, but now after the suppression of an interferer; and

11 schematically shows the structure of the signal processor unit for calculating the antenna weights for beam shaping.

The task of beamforming in the downlink from cellular mobile radio systems with adaptive antennas at the base station consists of the signals of each user to send from the base station that most of the energy from the desired one Users receive and as little energy as possible to others To users who appear as interference becomes. A radiation formation in the downlink that this Requirements met, guaranteed for every user sufficient interference ratio and thus adequate transmission quality (Bit error rate BER). To achieve this goal you have to put the main lobe of the antenna pattern into the Direction of the desired user and zeros in the antenna diagram towards those users who are on the same Frequency are supplied. This principle is illustrated in Fig. 1.

An adaptive antenna is shown schematically in detail in FIG. 1 1 shown with beamforming in the downlink, with a Signal processor 2, the individual antenna elements 1.1, 1.2 to 1.M controlled with different phases and amplitudes thus the desired antenna pattern 3 or 4 is generated. The Main lobes 5 and 6 of antenna pattern 3 and 4 show in Direction of a respective user 7 or 8, with zeros 9 or 10 in the antenna diagram 3 or 4 in the direction of the show other user 8 or 7.

Due to the different weighting of the individual elements the antenna group 1 is the shape of the antenna pattern 3 or 4 set. This is illustrated below using a linear example Antenna group explained with reference to FIG. 2. 2 shows schematically one from a direction  on the antenna elements 1.1, 1.2, 1.3 ... 1.M incident wave.

In Fig. 2, d is also the distance between the individual Antenna elements and with ΔL the path difference of the wave from an antenna element, e.g. 1.2, to the next antenna element, e.g. 1.3. The distance d lies here for example in the order of the wavelength and is preferably less than the wavelength (e.g. approximately the same half the wavelength).

The path difference ΔL of the electromagnetic wave from one antenna element to the next corresponds to a phase difference of the received signal, which can be written as follows Δϕ = 2π d · f c * Sin () and which depends on the wavelength of the transmitted signal. In this respect, f further denotes the carrier frequency of the transmitted signal and c the speed of light. Because of this relationship, the group response of the adaptive antenna 1 to this incident wave, which is also called "array steering vector" a (, f ), a (, f ) = [1 e j · 2π d · f c * Sin () ... e j · 2π d · f c · ( M - 1) sin () ].

As can be seen from this relationship, the group response is the antenna group 1 both from the direction of incidence of the wave as well as depending on the carrier frequency.

There is not just one in cellular cellular networks Propagation path, but there occurs multipath propagation on. That means multiple propagation paths with different Path length and different directions between the base station and the mobile station exist. systematic this multipath propagation is sketched in FIG. 3.

In detail, a base station 11 with a adaptive antenna 1 with nine antenna elements 1.1 ... 1.9 and with multipath propagation between base station 11 and mobile stations (MS) 7, 8 illustrates where multipath propagation for example due to reflections on buildings 12 comes about.

The individual signals overlap in the uplink on the antenna elements 1.1 to 1.9 of the linear antenna group 1 and in the downlink on the antenna of the respective mobile phone 7, 8. Whether the individual signals overlap constructively or destructively depends on the phase relationship of the individual waves to one another from. Since different carrier frequencies are used in an FDD system for the uplink and the downlink, the phase relationships of the waves to one another also change. For this reason, the shrinkage (the constructive and destructive overlay) in the up and down sections is absolutely uncorrelated. But not only the fading also the antenna pattern changes due to the frequency shift. Both the position of the main lobe and the position of the zeros and their shape in the group directivity change very strongly, as illustrated in FIGS. 4A and 4B. FIG. 4A shows an antenna diagram for the frequency of the uplink and FIG. 4B shows a corresponding antenna diagram for the frequency of the downlink. As can be seen from FIG. 4A, the signals come from the directions -20 ° and 40 ° for a user B1 and from the directions -50 ° and 10 ° for a user B2. On the other hand, when using the same antenna weights in the downlink (see FIG. 4B), the main lobes are at -18 ° and 35 ° for user B1 and at -45 ° and 8 ° for user B2. (The following values were used as carrier frequencies: f _E = 1920MHz, f _S = 2110MHz.)

As can be seen from FIGS. 4A and 4B, both the Zeroing as well as the main lobes in their direction on the ground of different frequencies shifted. The influence on The main mace is not as strong as it is anyway are wide and therefore only a maximum of 0.5dB smaller Antenna gain results. The zeros in the direction of each other users, however, are very narrow and in use the same antenna weights for the downlink as for the Upward path is the generated disturbance for the other User increased dramatically. Because of this, it is not advisable to use the same antenna weights for receiving and for to use the transmission at the base station 11.

Because of the frequency shift, the fading is between Sending and receiving case uncorrelated, and there is a different antenna pattern when using the same antenna weights.

The uncorrelated shrinkage cannot be compensated for all path lengths should be known, which is impossible. With a suitable beam shaping, however, the influence of Carrier frequency on the antenna pattern can be compensated, which consequently the disturbance generated for the other users reduced and the transmission quality and system capacity increased becomes.

A signal processor unit is used in base station 11 for this signal shaping 2 used, cf. Fig. 3, based on the received signals for driving the antenna elements 1.1 up to 1.M antenna weights especially for the downlink determined. For example, in the mobile radio system K users B1 to BK supplied simultaneously, and the antenna group 1, generally formulated, consists of M antenna elements 1.1 to 1.M. The received signals are at 13 band-limited (filtering with channel selection filter) and at 14 mixed into the baseband, amplified at 15 and digitized at 16, and in the signal processor unit 2, the signals detected using adaptive algorithms. In the downlink the signals are then weighted and modulated accordingly (at 14) and radiated by the antenna 1. 3 is the signal exchange between the base station is schematic 11 and the access network 17 additionally illustrated.

In Fig. 5 a flow chart is shown that the evaluation from the input signals to the determination of the antenna weights for the desired beam formation in the downlink schematically illustrated.

As shown in FIG. 5, a matrix X of noisy input signals of a plurality of co-channel signals serves as an input data set which is to be processed further in the signal processor unit 2. The matrix X contains N samples with critical sampling (sampling rate 1 / T) of K co-channel signals which are derived from the M individual elements of the array antenna 1, and interference signals from neighboring cells which use the same frequencies. With the help of a known signal sequence S _k (block 31 in Fig. 5) of the transmitted signal, with k = 1 to K, such as the spreading code in CDMA systems or the preamble or middle name in TDMA systems, the channel impulse responses of each of the K users B1 to BK estimated on each antenna element 1.1 to 1.M in step 30 ("subscriber recognition"). The channel impulse responses of each subscriber B1 to BK can be estimated independently of one another using methods known per se (for example by correlation with the known signal sequence S _k ) or all in one step (for example using the method of least squares).

darin bezeichnen h_k(t,τ) und S_k(t) die zeitvariante Impulsantwort zum Zeitpunkt t und das gesendete Signal des k-ten Benutzers; und N(t) bezeichnet den Vektor mit dem thermischen Rauschen an den Antennenelementen 1.1 bis 1.M. Die Summation berücksichtigt, dass man die Signale von allen K Benutzern B1 bis BK empfängt. Aus dieser Beziehung kann man nun die Kanalimpulsantworten der Benutzer B1 bis BK schätzen.The channel impulse responses are estimated in more detail from the received data X and the known signal sequence S _k (preamble, midamble in TDMA, or spreading code in CDMA systems), the received signal can be represented as follows:

h _k (t, τ) and S _k (t) denote the time-variant impulse response at time t and the transmitted signal of the k-th user; and N (t) denotes the vector with the thermal noise on the antenna elements 1.1 to 1.M. The summation takes into account that the signals are received from all K users B1 to BK. From this relationship, the channel impulse responses of users B1 to BK can now be estimated.

In TDMA systems you can use the mentioned preambles or midambles - either for all users at the same time (joint estimate) or separately for each user. The separate estimation can also be carried out using the least squares method, which can be represented in time-discrete notation as follows:

The joint estimate can be made as follows:

This corresponds to a common estimate using the method least squares. Forming pseudo inverses a matrix is denoted by "#".

In CDMA systems the output signal one is used the spread code used for the signal-adapted filter. This matched filter is a standard receiver component of CDMA systems; a description of the corresponding relationships for the estimation can be superfluous here.

The channel impulse response matrices H _k with k = 1 to K (for users B1 to BK) contain all the information required for the beam shaping process. The channel impulse response matrices have the following structure H k = [ H k (0) H k ( T ) ... H k (( L - 1)· T )], where h _k (t) is the vector of the channel impulse response at time t. This representation assumes that the channel impulse response is L samples long.

With the help of these channel impulse responses, the spatial Covariance matrices of the uplink of the individual user calculated, see Step 40 in FIG. 5.

A signal that is incident on antenna group 1 from a direction  results in a group response that is equal to the array steering vector a (, f) already mentioned. The spatial covariance matrix R (f) of this signal is in this case as R ( f ) = e { a (, f ) · a H (, f )} Are defined. There are usually many propagation paths with different reception services. For this reason, the spatial covariance matrix can be represented as follows

The channel impulse response contains all signals with the group responses and the associated signal strengths. For this reason and by replacing the formation of the expected value with the temporal mean (in the discrete-time mean of the samples), the spatial covariance matrix can be represented as follows

With this relationship, the covariance matrices of the uplink of users B1 to BK are therefore estimated. The spatial covariance matrix R _k is also frequency-dependent. The spatial covariance matrix R _{k of} the uplink is generally used to calculate the complex antenna weights for reception with adaptive antennas. However, the use of these antenna weights for the downlink shifts the zeros, as already explained. For this reason, one must try to transform the spatial covariance matrix R _k from the reception frequency f _{E of} the base station to the transmission frequency f _S in order to be able to calculate the antenna weights for the downlink.

This frequency transformation is indicated in FIG. 5 at step 50, the frequency transformation the spatial structure of the mobile radio channel, which is contained in the spatial covariance matrix R _k , from the reception frequency of the base station (frequency of the uplink) f _E to the transmission frequency of the base station (frequency the downlink) f _{S is} transformed. This technique is shown in more detail in Fig. 6 and will be described in more detail below.

The estimated spatial covariance matrices R _{k of} the K users of the downlink are formed so that they are hermitic. This means that all directions of incidence are considered to be independent of each other. The covariance matrices R _k (f _S ) at the transmission frequency f _S , which are obtained at the end of step 50, are used to calculate the optimal antenna weights for transmission in the downlink. This is done in step 60 in FIG. 5. All beam shaping algorithms based on knowledge of the spatial covariance matrix can be used for this. The signals for the individual users are now multiplied (weighted) by their antenna weights and sent by the base station 11.

(1) Mittelung der Kovarianzmatrizen bei der Empfangsfrequenz (Aufwärtsstrecke)

(2) Mittelung des Leistungs-Winkelspektrums (nach Schritt 52 in Fig. 6)

(3) Mittelung der Kovarianzmatrizen bei der Sendefrequenz (Abwärtsstrecke)

For the frequency transformation (step 50) according to FIG. 6, the following must be carried out in detail: As already described, the fading (the phase relationship) of the individual signal paths in the downlink and uplink is uncorrelated. Only the directions of incidence of the individual partial waves and their mean signal strength (power) are the same in the up and down sections. The estimated power-angle spectrum is therefore used for beam shaping in order to reconstruct the spatial covariance matrix. The power angle spectrum contains the power that is received from the respective angular range. Exactly this parameter is the same in the downward and upward path. For this reason, all of the information that can be used for transmission in the downlink is contained again in the reconstructed covariance matrix. Since only the mean signal strength remains the same and not the current one, a temporal averaging can be included. The time averaging can be carried out in three places:

(1) Averaging the covariance matrices at the reception frequency (uplink)

(2) Averaging the power angle spectrum (after step 52 in FIG. 6)

(3) Averaging the covariance matrices at the transmission frequency (downlink)

In principle, it doesn't matter where the averaging takes place - it has Studies have shown that the averaging of the covariance matrix delivers particularly good results at the reception frequency.

6 shows the power-angle spectrum estimate at block 52, starting from the covariance matrices R _k (f _E ) of the uplink for the k-th user. In principle, all known spectral search methods can be used in this power-angle spectrum estimation.

The power angle spectrum APS _k (Azimuthal Power Spectrum) can be determined by the maximum likelihood method (also called minimum variance method or Capon's method, shown in DH Johnson, DE Dudgeon, "Array Signal Processing - Concepts and Techniques", Prentice Hall, Inc. , Englewood Cliffs (New Jersey), 533 p.) Can be estimated as follows: APS k = P k () = 1 a H (, f e ) * ( R k ( f e )) -1 · a (, f e ) ,

In this regard, a (, f _E ) is the "array steering vector" of the uplink, which depends on the reception frequency f _E , the inter-element spacing d of the linear antenna group with M elements and the direction  as follows: a ( , f e ) = [1 e j · 2π d · f e c * Sin () ... e j · 2π d · f e c · ( M - 1) sin () ].

This means that if the geometry of the uniform, linear antenna group 1 is known (ratio of the antenna element spacing d to the reception wavelength λ _E , ie d / λ _E ), the power angle spectrum APS _{k of} each of the K users is estimated. Of course, this step can also be carried out using other, similar methods of spectral search. The power angle spectrum APS _k does not contain any phase relationships of the individual signal paths of the mobile radio channel to one another, which is neither necessary nor meaningful, since the fading and the phase relationships due to the multipath propagation are absolutely uncorrelated due to the different transmission and reception frequencies in a frequency duplex system.

FIG. 7 shows an example of an estimated power angle spectrum APS _{k of} a user Bk who is in the + 10 ° direction from the base station 11. The dashed line in Fig. 7 outlines the estimated power-angle spectrum of some co-channel interferers, which are at -30 °, + 12 ° and 50 °.

In step 54 in FIG. 6, the dominant regions of the power angle spectrum APS _k are then extracted. The entire power angle spectrum APS _k does not necessarily have to be used to reconstruct the spatial covariance matrix, but only those angle ranges can be used from which the majority of the signals in the uplink are received, with the antenna lobes therefore being directed into these angle ranges are or in relation to the interference only in such angular ranges zero points are placed in the antenna diagram. This technique of masking out some angular ranges, e.g. in order to only set zeros in the direction of the dominant interferers or to avoid zeros in the direction of those interferers that lie approximately in the same direction as the desired user and thereby negatively influence the antenna diagram, is exemplified in FIG 8 (in conjunction with Fig. 7) and illustrated in Figs. 9 and 10. While FIG. 7 shows the estimated power angle spectrum of the desired user and the interferer, FIG. 8 illustrates the antenna directivity for this scenario.

In Fig. 7 it can be seen that an interferer and the desired one User roughly in the same direction (+ 12 ° or + 10 °) are. If you try in the direction of this one jammer transmitted energy, which is at + 12 ° from the base station seen to decrease, the main lobe does not show exactly in the direction of the desired user.

To suppress this effect, it is possible to suppress the proportion of one interferer in the power angle spectrum, and thereby prevent the main lobe from shifting. This application of the modification of the power angle spectrum is shown in Fig. 9 and Fig. 10 shows that accordingly modified antenna pattern.

When using the modified power angle spectrum for beam formation in the downlink now shows the Main lobe in the antenna diagram (Fig. 10) again in the direction of desired user (+ 10 °). Especially in CDMA systems (the Systems of the 3rd generation of mobile communications such as UMTS are all based on CDMA) with many users who are served on one channel, can the separability of users at an angle (multiple users do not lie in the same direction, which is a minimum distance the angle at which the users lie, by far) cannot be guaranteed. For this reason, the one shown here Case occur frequently in CDMA systems.

Estimation errors in the covariance matrices of the users or the Interferers amplify the effect shown here. In real operating Systems is the possible hiding of certain areas therefore often required in the power angle spectrum.

Thereafter, in step 56 of FIG. 6, the estimated, modified power angle spectrum APS _{k, mod is used to} reconstruct the spatial covariance matrix (correlation matrix) R _k (f _S ) of the mobile radio channel of the downlink of the K users. This is done according to the following procedure:

P_k,mod() bezeichnet hier das modifizierte Leistungs-Winkelspektrum des k-ten Benutzers.The power angle spectrum can of course not be determined continuously, but only discretely with a certain angular resolution. Extensive computer simulations have shown that a resolution of around one degree is sufficient. This means that the above integral can be replaced by a discrete sum with relatively few summands. The discrete total looks like this:

P _{k, mod} () here designates the modified power angle spectrum of the kth user.

The method described is characterized in that the entire direction information of the mobile radio channel for the Beamforming is used in the downlink without one To commit errors by the duplex frequency, and therefore the same Gain in the downlink of cellular cellular systems with frequency duplex is possible as in time duplex systems. there make no assumptions about the number of discrete directions of incidence or used small duplex spacing, and therefore the described technique can be used without restrictions. Furthermore, the spatial covariance matrix or the channel impulse responses used to shape the downlink beam, which are also required for detection in the uplink and therefore do not have to be charged extra.

At the output of the frequency transformation according to block 50, the covariance matrices R _{k of} the downlink (R _k (f _S )) for the k-th subscriber are thus obtained, and these are finally obtained in step 60 according to FIG. 5 of the beam shaping, ie the determination of the antenna weights for the downlink. As already mentioned, all known algorithms for beam shaping can be used which are based on the knowledge of the spatial covariance matrix. In the following, an algorithm is explained as an example, which is a standard algorithm in the literature for calculating the antenna weights in the uplink (see, for example, P. Zetterberg, and B. Ottersten: "The Spectrum Efficiency of a Basestation Antenna Array System for Spatially Selective Transmission ", IEEE Transactions on Vehicular Technology, Vol. 44, pp. 651-660, August 1995).

If one knows the covariance matrix of the individual users and the interferers, one can calculate the antenna weights from this information. R _k (f _S ) denotes the covariance matrix of the kth user and Q _k (f _S ) the covariance matrix of the interference for the kth user at the transmission frequency f _S. The weight vector is calculated from this information as the dominant generalized eigenvector of the matrix pair [R _k (f _S ), Q _k (f _S )]. When receiving in the uplink, this method maximizes the ratio of received interference power ratio SNIR _k . In the downlink, the ratio of signal power generated for the desired user to interference power generated for the other users is maximized. This problem can be represented mathematically as follows:

The covariance matrices are used for detection in the uplink at the receiving frequency and for computing the antenna weights for the downlink the frequency-transformed Covariance matrices (at the transmission frequency of the base station). However, it uses the same algorithm for calculation the complex antenna weights for receiving and transmitting used with the adaptive antenna 1. Because of this, and because the spatial covariance matrix for reception in the uplink Generally used is this procedure for beam shaping for the downlink of systems with frequency duplex very simple, and compared to Upward distance only the frequency transformation of the spatial Additional covariance matrix, as shown schematically in FIG. 11 at 70 is shown.

11 shows the structure of the signal processor unit 2 for calculating the antenna weights for the adaptive antenna 1, the received signals being indicated schematically at 71. At 72 the unit for estimating the upward covariance matrices R _{k is} illustrated and the beam shaping unit is shown at 73. The antenna weights determined are designated W _k (f _S ) for the downlink and W _k (f _E ) for the uplink.

Claims (14)

1. A beamforming method for an adaptive antenna array including several antenna elements (1.1. to 1.M) in the downlink of frequency duplex systems, wherein antenna weights (W_k(f_S)) are determined for the antenna elements (1.1 to 1.M) for downlink transmission on the basis of directional information of the uplink, characterized in that the antenna weights (W_k(f_S)) for downlink transmission are determined on the basis of the power angle spectrum (APS_k) of the uplink of the individual users (B1 to BK), wherein the power angle spectrum (APS_k) is modified by masking out undesired angular regions.
2. A method according to claim 1, characterized in that the power angle spectrum (APS_k) is estimated using a known signal sequence (S_k) of the transmission signal, such as, e.g., spread code or midamble.
3. A method according to claim 1 or 2, characterized in that the power angle spectrum (APS_k) is estimated on the basis of the spatial covariance matrices (R_k(f_E)) of the uplink of the individual users (B1 to BK).
4. A method according to claim 3, characterized in that the power angle spectrum (APS_k) is estimated on the basis of mean values of the spatial covariance matrices (R_k(f_E)) of the uplink of the individual users.
5. A method according to any one of claims 1 to 4, characterized in that the respective spatial covariance matrix (R_k(f_S)) of the downlink is determined on the basis of the modified power angle spectrum (APS_k) of the individual users.
6. A method according to claim 5, characterized in that the spatial covariance matrix (R_k(f_S)) of the downlink is determined on the basis of the mean value of the modified power angle spectrum (APS_k).
7. A method according to claim 5, characterized in that the mean value of the spatial covariance matrix (R_k(f_S)) of the downlink is used to calculate the antenna weights (W_k(f_S)) for transmission.
8. A beamforming device for an adaptive antenna array including several antenna elements (1.1. to 1.M) in the downlink of frequency duplex systems, comprising a signal processing unit (2) used to determine antenna weights (W_k(f_S)) for the antenna elements (1.1. to 1.M) for downlink transmission on the basis of directional information of the uplink, characterized in that the signal processing unit (2) is arranged to determine the antenna weights (W_k(f_S)) for downlink transmission on the basis of the power angle spectrum (APS_k) of the uplink of the individual users (B1 to BK) upon modification of the former by masking out undesired angular regions.
9. A device according to claim 8, characterized in that the signal processing unit (2) is fed with a known signal sequence (S_k) of the transmission signal, such as, e.g., spread code or midamble, to estimate the power angle spectrum (APS_k).
10. A device according to claim 8 or 9, characterized in that the signal processing unit (2) is arranged to estimate the power angle spectrum (APS_k) on the basis of the spatial covariance matrices (R_k(f_S)) of the uplink of the individual users (B1 to BK).
11. A device according to claim 10, characterized in that the signal processing unit (2) forms the mean values of the spatial covariance matrices (R_k(f_S)) of the uplink.
12. A device according to any one of claims 8 to 11, characterized in that the signal processing unit (2) is arranged to determine the respective spatial covariance matrix (R_k(f_S)) of the downlink on the basis of the modified power angle spectrum (APS_k) of the individual users (B1 to BK).
13. A device according to claim 12, characterized in that the signal processing unit (2) forms the mean value of the modified power angle spectrum (APS_k) to determine the respective spatial covariance matrix (R_k(f_S)) of the downlink.
14. A device according to claim 12, characterized in that the signal processing unit (2) forms the mean value of the spatial covariance matrix (R_k(f_S)) of the downlink to calculate the antenna weights (W_k(f_S)) for transmission.

Images