JPH1075110A - Transmit beamforming method - Google Patents

Abstract

(57) [Summary] [Problem] An average reception SIR at a mobile station can be maximized, and information on arrival directions of desired and interference wave paths is not required. A by sampling the received signals of the antenna elements E-1~E-N, to obtain a snapshot vector X _k, the correlation vector P _j of the X _k and the known pattern D (4), the P _j Searching for the correlation matrix R _p from M (6), determine the eigenvector y corresponding to the maximum eigenvalue of R _p (8), and from the L X _k Searching for the correlation matrix R _x (10), W = R ^-1 _{x
^{y / (y H R -1 x}} y) by seeking a transmission weight vector W (12), is multiplied to a corresponding one of the signals to be transmitted to the element from the E-1~E-N (16) .

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention is applied to a mobile communication base station, and controls a transmission directivity pattern by controlling a amplitude and a phase by multiplying a signal to be transmitted to a plurality of antenna elements by a weight vector. The present invention relates to a transmission beam forming method for obtaining a weight vector from a received signal in an array antenna, and more particularly to a transmission beam forming method in a base station for maximizing an average signal power to interference wave power ratio of a signal received by a mobile station.

[0002]

2. Description of the Related Art A feature of mobile communication is that a radio propagation environment is a multipath propagation path. Considering the upstream (mobile station transmission, base station reception) communication path, a bundle of transmission element waves affected by scattering, diffraction, and reflection around the mobile station is directly or after being reflected at a further distance, Arrives at the base station. Therefore, the base station receives the transmission signal divided into a plurality of components having different arrival angles. These components corresponding to each path are composed of the bundle of the above-mentioned transmission rays,
Since the process of creating this bundle (scattering, diffraction, reflection, etc.) is different for each pass, each pass is subject to independent fading. Such a geometrical positional relationship between a plurality of propagation paths is reversible between uplink and downlink (base station transmission, mobile station reception). That is, if each component is transmitted in the direction in which each component is received by the base station, the signal always reaches the mobile station. However, in this case, if the uplink and downlink frequencies are different, the instantaneous fading variation does not match.

It is known that an interference wave can be canceled by receiving a signal using an antenna having a plurality of elements at a receiving station and weighting and combining the received signals of each element. Such an adaptive array antenna weight control algorithm, for example, minimizes the mean square error between the combined signal and the ideal signal pattern, or maximizes the combined signal power to interference power ratio. Work to meet the norms. The optimal weight vector that satisfies these synthesis criteria differs depending on the amplitude and phase of the signal corresponding to each path of the desired wave and each path of the interference wave. Therefore, when the amplitude and phase of each of these paths fluctuates due to fading. , The optimal weight vector changes according to the fading variation. However, in a system using different frequencies for uplink and downlink, different fadings occur as described above, so that the uplink optimal weight vector does not match the downlink optimal weight vector. Furthermore, since the amplitude and phase of the downlink signal received by the mobile station cannot be known by the base station, it is impossible to determine the optimum weight vector of the base station transmitting antenna for the instantaneous amplitude and phase at the mobile station.

[0004] However, it is not impossible to obtain a transmission beam pattern for obtaining the optimum in the average sense, instead of obtaining the instantaneous optimum in the mobile station. Although the instantaneous amplitude and phase of the received wave fluctuate drastically due to fading, the position of the mobile station itself changes at most at its traveling speed, and therefore the arrival direction of each path component (DOA: Di
rection of Arrival) changes only to that extent. Therefore, the transmission beam pattern that maximizes the average reception SIR (signal wave interference ratio) of the mobile station averaged in a time interval in which the DOA is considered to be unchanged from the base station maximizes that of the base station. Can be regarded as the same as the transmission beam pattern to be transmitted. That is, the average reception S of the base station
If a pattern is formed so as to maximize IR, this is also the optimal pattern for transmission.

One simple method for this purpose is to detect the direction of arrival (DOA: Direction of Arrival) of each path component, direct the peak of directivity in the direction of arrival of the desired wave, and direct the peak in the direction of arrival of the interference wave. Is to direct the gender Null (zero).
As a method of detecting DOA of each path component, a conventional MUS
IC algorithm (references S. Haykin, J. Litva and TJS
hepherd, “Radar Array Processing”, Springer-Verla
g, pp.64-72, 1993) and ESP
RIT algorithm (Ref. SJOrfanidis, “Optimum
Signal Processing ”, Macmillan Publishing Co., p.
p.370-372, 1988, etc.) are known. These algorithms can detect the direction in which a path exists (signal energy arrives), but when applied to a multipath environment, the problem is which component corresponds to the desired wave and which component corresponds to the interference wave. Occurs. If this association is incorrect, Null may be directed in the direction in which the desired wave arrives, or the peak may be directed in the direction in which the interference wave arrives. However, the error rate of this association cannot be reduced. Therefore, methods that rely on DOA are not practical. Average reception S at mobile station, independent of DOA
A method for base station transmit antenna beam pattern formation that maximizes IR has not been previously known.

[0006]

SUMMARY OF THE INVENTION An object of the present invention is to obtain an optimum weight vector for forming a base station transmitting antenna beam pattern that maximizes the average reception SIR at a mobile station, and to determine different frequencies for uplink and downlink. It is an object of the present invention to provide a transmission beam forming method which can be optimized for a system to be used and which is not DOA-dependent.

[0007]

Means for Solving the Problems] According to the present invention, the correlation vector P _j snapshot vector obtained by sampling the received signal of each antenna element and its known pattern, the correlation matrix of M that P _j R _p
, The eigenvalue problem of the matrix R _p is solved, the eigenvector y corresponding to the eigenvalue having the largest absolute value is obtained, and the correlation matrix _Rx of the snapshot vector is obtained,
A weight vector that maximizes the reception SIR is obtained from _Rx and y, and each element of the weight vector is multiplied by a signal transmitted by the corresponding antenna element to control a transmission directivity pattern. In this way, a weight vector that maximizes the average reception SIR is obtained. The reason will be described below.

[0008] Generally, the reception SIR can be maximized by optimally weight vector ^{_{W 0 = R -1 x P /}} (P H R -1 x P) (1) ( Reference, GGRaleigh, SNDigga
vi and A. Paulraj, “ABlind Transmit Antenna Algori
thm for Wireless Communication ”, Conf. Rec. of IE
See EE G'Com 95, pp. 1494-1499, Seatle). here,
R _x is a snapshot vector X _k created from the received signal sample values of each of the N antenna elements
Can be estimated for a sufficiently large L ₁ by R _x = (1 / L ₁ ) ΣL ¹ _{k = 1} X _k X ^H _k (2) (L ₁ is the sample sequence length). Also,
P is a correlation vector between the pattern D _k known on the transmitting side and the receiving side and the snapshot vector X _k, and is given by P = <X _k D ^* _k > (3) This is also called a steering vector. Here, <> represents an ensemble average. Physically, P includes DOA information of each path component of the desired wave. As the known pattern _Dk, for example, a synchronization pattern that is transmitted periodically, or an entire received frame in which no error is detected by decoding an error detection code can be used.

[0009] Now, the received desired wave is divided into path L _d, the j-th path components is assumed to come from the direction of theta _j to the antenna array. Further, the m-th exist U number of interferers is received is divided into path L _um, further its b-th path components is assumed to come from the direction of theta _b to the antenna array. In this case, the snapshot vector X _k is ^{_{X k = Σ Ld j = 1}} a (θ j) · d j (kT) + Σ U m = 1 Σ Lum b = 1 a (θ b) · u mb of combined reception signal (KT) + g (kT) (4) Where T is a sampling interval, a
(Θ) is the antenna array response to the angle of arrival θ and is determined by the geometric shape of the antenna array. d _j (t)
Is the j-th path component of the desired wave at time t, u _mb (t) is the b-th path component of the m-th interference wave, and d _j (t) = z _Dj (t) D (t) (5) u _mb (t) = z _umb (t) U _m (t) (6) Here, z _Dj (t) is the fading complex amplitude of the j-th path component of the desired wave, D (t) is the transmission waveform of the desired wave,
D (kT) = D _k , z _mb (t) is the fading complex amplitude of the b-th component of the m-th interference wave, U _m (t) is the transmission waveform of the m-th interference wave, and U (kT) = U _k is there. Also, g (t)
Is a noise vector, and g (t) = [g ₁ (t) g ₂ (t)... G _N (t)] ^t (7), and g _i (t) is noise added to the i-th element. is there.
g _i (t) are mutually independent zero mean Gaussian variables, and <g (t)> = 0 (8.1) <g * (t) g t (t)> = σ 2 g I N (8.2) Become. Here, _IN is an N-th order unit matrix. ^* Is complex conjugate,
^t is transposition, and σ _g is the average power of the desired signal.

Now, the expression (3) in a fading environment
Of the ensemble mean appearing in Now, to obtain this ensemble average physically,
It needs to be replaced with a time average. The purpose is average reception S
Since the IR is maximized, the sampling frequency (1 / T) for time averaging needs to be sufficiently lower than the maximum Doppler frequency of fading, and fluctuation due to fading needs to be sufficiently averaged. In this case, the correlation vector P is replaced by the time _{average, P = (1 / L 2} ) Σ L2 k = 1 X k D * k = (1 / L 2) Σ L2 k = 1 D k {Σ Ld j = ₁ a (θ _j ) ・ d _j (kT) + Σ ^U _{m = 1} Σ ^Lum _{b = 1} a (θ _b ) ・_mb (kT) + g (kT)｝ → 0, (L ₂ → ∞) (9) As a result, the optimum weight vector w cannot be obtained.

[0011] Conversely, when the sampling frequency (1 / T) is sufficiently higher than the maximum Doppler frequency of _{fading, P = (1 / L 2} ) Σ L2 k = 1 X k D * k → Σ Ld j = _{1 a (θ j) · z} Dj (kT), (L 2 → ∞) and becomes (10), the fading is not averaged, the complex amplitude z _Dj (t) is left will. However, since it is phase modulation,
_DkD ^* _k = 1.0, and since there is no correlation between the desired wave and the interference wave, < _DkD ^* _k > = < _UkD ^* _k >
= 0 was assumed. Therefore, the process of finding the correlation vector P at a sufficiently short sampling interval is repeated M times more sufficiently (the process is given a number indicating the order and k
Th and P _K a Motoma' was P in the process of), we obtain the matrices from their _{R p = (1 / M)} Σ M k = 1 P K P K H (11). Furthermore, the eigenvalue problem with R _p, solving the _{R p y = λy (12)} , the eigenvector to the maximum eigenvalue lambda _max y
Ask for. y represents an average signal space formed by the sum of the array responses of each desired path component to the DOA. This _{is, R p = (1 / M} ) Σ M k = 1 P K P K H → σ 2 g I N + Σ Ld j = 1 | z Dj (kT) | 2 · a (θ j) a H ( θ _j ), (L → ∞) (13). For these, ESP
A description of the RIT algorithm is given on page 346 of the reference. Therefore, using this y instead of P in formula (1), using the ^{_{W o = R -1 x y /}} (y H R -1 x y) W o which is obtained by (14) as a weight vector,
A base station transmit antenna beam pattern that maximizes the average received SIR at the mobile station can be formed.

[0012]

FIG. 1 shows an embodiment of the present invention expressed in complex baseband. The received combined signals are received by N antenna elements E-1 to E-N, respectively, and the combined signals 1-1 to 1-N received by these antenna elements E-1 to E-N are sampled by the sampler 2. -1
２−2-N to obtain a snapshot vector 3. The sample values 3-1 to 3-N of the combined signals 1-1 to 1-N are input to the correlation vector calculator 4 and the first correlation matrix calculator 10. The correlation vector calculator 4 receives the known pattern D, performs the calculation of Expression (10), and outputs a correlation vector 5 replaced with a time average. However, in this case, sampling is sufficiently fast so that the fading complex amplitude does not change between sample values. Also repeated for Formula processing (10) (composed of the sample L ₂ amino) variation is sufficiently independent of the considered degree temporally distant block fading complex amplitude. The correlation vector 5 is input to the second correlation matrix calculator 6, performs the operation of Expression (11), and outputs the second correlation matrix 7. The eigenvalue eigenvector calculator 8 solves the eigenvalue problem of the equation (12) for the second correlation matrix 7, outputs an eigenvector 9 for the maximum eigenvalue, and outputs the eigenvector 9 to the beamformer 1.
Supply to 6.

On the other hand, the first correlation matrix calculator 10 performs the calculation of equation (2) and outputs the first correlation matrix 11. In this case, the sampling is sufficiently slow so that the fading complex amplitudes between the sample values can be regarded as independent.
The optimum weight calculator 12 calculates the equation (14) using the eigenvector 9 and the first correlation matrix 11 and outputs the optimum weight vector W _o13 . The transmission information generator 14 outputs the downlink transmission waveform 15 to the beamformer 16 (in the case of a service provider system, the downlink information of the user corresponds to this).

FIG. 2 shows an example of the configuration of the beamformer 16 expressed in complex baseband. Inputs are an optimal weight vector W _o 13 and a downlink transmission waveform 15. Signals are transmitted from the vector W _o elements 13-1 to 13-N respectively multipliers 18-1 to 18-N respectively are multiplied in the transmission waveform 13 by the antenna element E-1~E-N 17-1~ 17-N is output. As a result, an optimum beam pattern is obtained. However, as described above, since FIGS. 1 and 2 are expressed in complex baseband, carrier waves are not shown. In an actual transmitter configuration, a similar optimum beam can be formed by controlling the amplitude and phase of the carrier modulated by the transmission waveform 15 instead of the downlink transmission waveform 15 itself, instead of multiplying by the weight vector _Wo13. Needless to say.

[0015]

As described above, according to the present invention, a base station transmitting antenna beam pattern which maximizes the average reception SIR at a mobile station without using DOA information of each path of a desired wave and an interference wave is formed. it can. Since the fluctuation of the received wave due to fading is averaged in the process of obtaining the optimum weight vector, the present invention can be applied to a system using different frequencies for uplink and downlink. Since the average reception SIR is improved by using the base station transmission antenna beam pattern according to the present invention, the repetition distance of the same frequency can be reduced, and the frequency utilization efficiency of the system can be improved.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an example of a functional configuration of a device to which the method of the present invention is applied in complex baseband.

FIG. 2 is a diagram illustrating a beam former 16 in FIG. 1 in a complex baseband.

Claims (1)

[Claims] 1. A radio communication transmitting and receiving apparatus having N (N is an integer of 2 or more) antenna elements, wherein a signal input to an i-th (i = 1, 2,..., N) antenna element is Transmission weight vector W = (w1, w2,
, WN) to control the antenna directivity pattern, and transmit using the controlled directivity pattern. In the transmission beam forming method, each received signal of the antenna element is sampled. a process of obtaining a snapshot vector X _k for the k-th sample, and the process of obtaining the X _k D _k by multiplying the known pattern D _k in transmission and reception both in the snapshot vector X _k, L ₁ piece (L ₁ is R _x = (1 / L ₁ ) ΣL ¹ _{k = 1} X _k X ^H _k for a snapshot vector X _k of (an integer of 2 or more) to obtain a matrix R _x , and ( ^H is a conjugate transpose ), L ₂ (L ₂ is an integer of 2 or more) multiplication results X _k D _k
For _{P j = (1 / L 2} ) Σ L2 k = 1 D k X H k and the process of calculating to obtain the vector P _j and the M vectors _{P j R p = (1 /} M) Σ M j _{= 1} and P _j P ^H _{process j} a by computing seek matrix R _p, the eigenvalue problem of the matrix R _p: the process of obtaining the eigenvector y absolute value by solving R _p y = [lambda] y corresponds to the largest eigenvalue, the matrix R _x, eigenvector y for _{^{W = R -1 x y / (}} y H R -1 x y) and calculates the transmission weight vector W = (w1, w2,
.., WN).