Advanced Codebook for Multi-Antenna Transmission Systems

FIELD OF THE INVENTION

The present invention relates to a method, system, transmitter apparatus, receiver apparatus, and computer program product for providing feedback signaling in a multi-antenna transmission system, such as multiple-input multiple-output (MIMO) system.

BACKGROUND OF THE INVENTION

In wireless communication system, multiple antennas can be used to improve link reliability and/or increase transmission rate. Generally, multiple-antenna techniques can be categorized into open loop mode and closed loop mode depending on the availability of channel state information at the transmitter. Meanwhile, close loop methods, such as precoding or beamforming, may lead to better performance at the expense of a requirement to feed back some form of channel state information (CSI) to the transmitting end.

The required CSl at the transmitting end can be maintained via feedback from the receiver at FDD (Frequency Division Duplex) mode or through the reciprocity prin- ciple at TDD (Time Division Duplex) mode. Alternatively, at FDD mode the receiving end might decide on the transmit strategy, i.e., antenna weighting, and feed back this information via a feedback channel after proper quantization.

Transmit beamforming or precoding and receive combining are simple methods for exploiting significant diversity available in multiple-input and multiple-output (MIMO) wireless systems. In such MIMO systems antenna arrays are used to enhance bandwidth efficiency. MIMO systems provide multiple inputs and multiple outputs for a single channel and are thus able to exploit spatial diversity and spatial multiplexing. Further information about MIMO systems can be gathered from the IEEE specifications 802.1 1 n, 802.16-2004 and 802.16e, as well as 802.20 and 802.22 which relate to other standards. Specifically, MIMO systems have been introduced to radio systems like e.g. WiMAX (Worldwide Interoperability for Micro-

COWFiRIVIATION COPY

wave Access) and are currently standardized in 3GPP for WCDMA (Wideband Code Division Multiple Access) as well as 3GPP E-UTRAN (Evolved Universal- Mobile Telecommunications System (UMTS) Terrestrial Radio Access Network), such as LTE (Long Term Evolution) or 3.9G.

Unfortunately, in transmission systems where forward and reverse channels are not reciprocal, this requires coarsely quantization of the channel and beamforming vector to accommodate the limited bandwidth of the feedback channel. To support such limitations of the feedback channel, codebooks of possible beamforming vec- tors can be used, which are known to both transmitting and receiving ends. The codebook is restricted to have fixed cardinality and may be designed off-line. The receiving end (e.g. mobile station) is assumed to select from the available code- book the best beamforming vector or matrix and to convey it over the feedback channel to the transmitting end (e.g. base station). More specific, the receiving end learns the CSI from received DL information and selects a transmit beamforming vector or matrix from the available codebook. An index of the selected beamforming vector or matrix is then fed back to transmitting end. Having received the index, the transmitting end looks up the corresponding codebook and selects the beamforming matrix or vector according to the index. The selected matrix or vector can then be used for MIMO precoding operation.

In current WCDMA-based 3GPP standard TS 25.202, for precoding or beamforming of two transmission antennas, Mode 1 and Mode 2 are defined, corresponding to a 2-bit and 4-bit codebook, respectively, which may as well be extended to a case of four transmission antennas , e.g., a 6-bit codebook for Mode 1. Furthermore, D. J. Love and R. W. Heath, "Grassmannian beamforming for multiple-input multiple-output wireless systems", IEEE Transactions on Information Theory, vol. 49, No.10, pp. 2735-2747, Oct. 2003 discloses Grassmannian packing as an optimum solution for the finite-rate feedback problem from a perspective of outage probability and SNR maximization, which leads to a so-called "Grassmannian codebook". Additionally, a system unitary construction method is proposed in B. M. Hochwald, T.L. Marzetta, T.J. Richardson, W. Sweldens, and R. Urbanke, "Systematic design of unitary space-time constellations", IEEE Transactions on Information Theory, vol. 46, No. 6, pp. 1962-1973, Sep. 2000 to design a unitary space-time constellation for non-coherent transmission. This method can also be used to construct precoding or beamforming weights, which leads to phase-only weighting and has circular correlation property.

Moreover, Intel et al, "Compact codebooks for transmit beamforming in closed- loop MIMQ", IEEE C802.16e-05/050r6 disclose codebook for four transmission antennas, which is based on a Household transform and has been standardized into IEEE standard 802.16e-2005, "part 16: Air interface for fixed and mobile broadband wireless access systems". In addition, P. Xia and G. B. Giannakis, "Design and analysis on transmit-beamforming based on limited-rate feedback", IEEE Transactions on Signal Processing, vol. 54, No. 5, pp. 1853-1863, May 2006 suggests using a modified Lloyd algorithm to design the codebook.

In practice, the feedback mechanism may lead to imperfect or partial CSI at the transmitting end. Feedback delay, channel estimation errors etc. may influence the accuracy of weights available at the transmitting end. Another important imperfection is the bandwidth constraint over the feedback link. For instance, in 3GPP WCDMA specification, only one bit for feedback of precoding or beamforming weights is transmitted in each slot, resulting in a 1500 bps signaling overhead. Therefore a basic problem or question related to precoding/beamfoming is how to design the codebook, i.e. how to quantize the channel state information or precoding information so that good performance with low feedback overhead can be achieved.

SUMMARY

It is therefore an object of the present invention to provide a method and system for advanced feedback signaling with low feedback overhead in multi-antenna transmission systems.

This object is achieved by a method comprising:

• maintaining at a receiving end of a multi-antenna transmission channel a codebook comprising an indexed set of beamforming elements;

• selecting at said receiving end at least one of said beamforming elements based on at least one predetermined parameter of said multi-antenna transmission channel; and

- A -

• feeding back an index information of said at least one selected beamform- ing element to a multi-antenna transmitting end of said multi-antenna-trans-- mission channel,

• wherein said codebook comprises a first subset of elements for phase-only antenna control, and at least one of a second subset of elements for antenna subset selection and a third subset of elements for single antenna selection.

Furthermore, the above object is achieved by a method comprising:

• maintaining at a multi-antenna transmitting end of a multi-antenna transmission channel a codebook comprising an indexed set of beamforming elements;

• receiving at said multi-antenna transmitting end a data stream which comprises an index information fed back from a receiving end of said multi- antenna transmission channel, said index information indicating a beam- forming element selected from said codebook; and

• controlling beamforming at said multi-antenna transmitting end based on said indicated selected beamforming element,

• wherein said codebook comprises a first subset of elements for phase-only antenna control, and at least one of a second subset of elements for antenna subset selection and a third subset of elements for single antenna selection.

Additionally, the above object is achieved by a receiver apparatus comprising:

• a maintaining unit for maintaining a codebook comprising an indexed set of beamforming elements;

• a selecting unit for selecting at least one of said beamforming elements based on at least one predetermined parameter of said multi-antenna transmission channel; and

• a feedback unit for feeding back an index information of said at least one selected beamforming element to a multi-antenna transmitting end of said- multi-antenna transmission channel,

• wherein said codebook comprises a first subset of elements for phase-only antenna control, and at least one of a second subset of elements for antenna subset selection and a third subset of elements for single antenna selection.

Moreover, the above object is achieved by a transmitter apparatus comprising:

• a maintaining unit for maintaining a codebook comprising an indexed set of beamforming elements;

• at least one receiving unit for receiving an index information fed back from a receiving end, said index information indicating a beamforming element selected from said codebook; and

• a control unit for controlling beamforming at said transmitter apparatus based on said indicated selected beamforming element,

• wherein said codebook comprises a first subset of elements for phase-only antenna control, and at least one of a second subset of elements for antenna subset selection and a third subset of elements for single antenna se- lection.

Further, the above object is achieved by a transmission system comprising at least one transmitter apparatus as defined above, and at least one receiver apparatus as defined above.

In addition, the above object is achieved by respective computer program products comprising code means for producing the steps of the above methods when run on a computer device.

Accordingly, the combination of elements of the at least one of the second and third codebook subsets with the first codebook subset helps the proposed codebook to improve the performance in uncorrelated channel in addition to phase-only

weighting from first codebook subset. A structured approach is thus used to generate the codebook. This structured approach allows generation of the codebook when necessary, which means that codebook elements (e.g., codeword, vectors or matrices) do not have to be stored all the time, which is advantageous over some random-searched codebooks, e.g., Grassmannian and Xia's codebooks.

In an embodiment, the third subset of elements for single antenna selection may comprise a number of elements corresponding to the number of transmission antennas. The second subset of elements may comprise elements for antenna sub- set selection with L different relative phase rotations φ _t = (i - ϊ) -2π/L, l ≤ i ≤ L between selected antenna elements. The first subset of beamforming elements may comprises a Hochwald-type codebook with a circular correlation property.

According to a specific implementation example, the multi-antenna transmitting end may comprise four antennas. The second subset of elements may then comprises twelve elements for antenna subset selection of two selected antennas with 0 or π relative phase rotation. The Hochwald-type codebook of the first subset may have a size of 48 elements.

Further advantageous modifications or developments are defined in the dependent claims.

BRIEF DESCRIPTION OF THE DRAWINGS

The present invention will now be described on the basis of various embodiments with reference to the accompanying drawings in which:

Fig. 1 shows a schematic diagram of a multi-antenna transmission system according to an embodiment;

Fig. 2 shows a schematic block diagram of a mobile transceiver unit according to the embodiment;

Fig. 3 shows a schematic block diagram of a base station device according to the embodiment;

Fig. 4 shows a schematic representation of a codebook structure according to the embodiment;

Fig. 5 shows a more detailed example of a codebook structure according to the embodiment;

Fig. 6 a flow diagram of a receiver-side feedback procedure according to the embodiment;

Fig. 7 a flow diagram of a transmitter-side feedback procedure according to the embodiment;

Fig. 8 shows a diagram indicating SNR gains of different Hochwald-type code- books according to various embodiments;

Fig. 9 shows a diagram indicating SNR gains of a specific Hochwald-type codebook according to the embodiment in comparison with other codebook types; and

Fig. 10 shows a schematic block diagram of a computer-based implementation of the embodiment.

DESCRIPTION OF EMBODIMENTS

An embodiment will now be described based on a wireless multi-antenna trans- mission system, such as - but not limited to - a MIMO system with a general UL feedback scheme for MIMO DL transmission for an exemplary case of four available Tx antennas at a transmitter unit of a base station device, such as a Node B. However, it will be apparent from the following description and is therefore explicitly stressed that the present invention can be applied to any other network archi- tecture with different radio access technologies involving multi-antenna transmitter devices (e.g. base station devices, access points or other access devices) capable of being operated in different operating modes.

Fig. 1 shows an exemplary multi-antenna system according to the embodiment, in which a mobile station (MS) 10 (or UE in 3G terminology) is radio-connected to a base station device (BS) 20 (or Node B in 3G terminology) which comprises four

Tx antennas 201 to 204 for transmitting a respective DL radio transmission 42 to-

wards the MS 10. The MS 10 transmits an UL transmission 50 towards the base station device 20 which provides access to a radio access network 30, such as ^~ an E-UTRAN or the like. The UL signal may be received at the BS 20 by the same antennas 201 to 204 or an additional reception antenna may alternatively be pro- vided. The MS 10 might alternatively have more than a single antenna available that could be used for dual-antenna or multi-antenna transmission in UL direction and/or dual- or multi-antenna reception of DL radio transmissions 42.

According to the present embodiment, a new 6-bit codebook is suggested for the four Tx antennas 201 to 204 to provide better performance than conventional codebooks, considering different correlation and scenarios. The proposed code- book comprises a combination of a first codebook (or first codebook subset) for phase-only transmission control and at least one of two other codebooks (or second and third codebook subsets) for antenna subset selection and single-antenna selection, respectively. As an example, the first codebook may be a Hochwald codebook or any other type of codebook which provide phase-only transmission control of the transmission beams generated by the Tx antennas.

Additionally, in a specific example of four Tx antennas, a size-48 Hochwald code- book may be used, which is enhanced by a size-16 codebook comprising the second and third codebook subsets. The second codebook subset may comprise 12 codebook elements (e.g., precoding or beamforming vectors) for antenna subset selection, and the third codebook may comprise 4 codebook elements for single antenna selection.

It is however apparent to the skilled person that corresponding other codebook sizes are reasonable for a different number of Tx antennas as well as considering the tradeoff between overhead and performance, etc.

The transmitter and receiver could maintain or store a common codebook, i.e. a finite collection of precoding vectors (codewords). Then the receiver decides which vector/vectors are selected to be used from the codebook and then feeds back its index to the transmitter via a feedback channel. After receiving the codeword index, the transmitter chooses the corresponding beamforming or precoding vec- tor/vectors for data transmission. The selection of proper beamforming or precoding weights from the codebook may follow some criterion, such as maximizing the

post-processing SINR or maximizing the sum of the throughput of all streams, as non-limiting examples.

Fig. 2 shows a schematic block diagram of a transmit and receive unit according to the embodiment, such as the MS 10, which is configured to support or implement the suggested advanced feedback signaling with mode indicator. Access to the radio access network is provided by a transceiver unit 14 capable of receiving and transmitting RF signals via at least one antenna. As an alternative the transceiver unit 14 may comprise or may be replaced by separate transmitter and receiver units with separate transmission and receiving paths.

The transceiver unit 14 is connected to a signal processing stage 12 which is responsible for receiver-related processing, such as demodulating, descrambling, decoding etc. of received DL data, and for transmitter-related processing, such as modulating, scrambling, coding etc. of UL data to be transmitted, and which is additionally configured to add a feedback information for precoding or beamforming to the UL data stream. This feedback information comprises a index to an element of a codebook 18 maintained or stored an UL feedback circuit 16 which generates the UL feedback index information 70 based on a corresponding control informa- tion issued by the signal processing stage 12. The UL feedback index information 70 comprises an index to an element of the codebook 28 which corresponds to the enhanced codebook described above. The UL feedback index information is then added, e.g. as a binary control word, to the UL stream and transmitted via the UL transmission 50 towards the radio-connected BS 20.

Fig. 3 shows a schematic block diagram of a base station device, e.g. the BS 20, according to the embodiment, with four antennas 201 to 204 for transmitting and receiving data. In the present example, all antennas 201 to 204 are connected to a single transceiver unit 22 capable of processing four transmission and reception streams. Of course, each of the four antennas 201 to 204 can be connected to a single dedicated transceiver unit. As a further alternative, all antennas 201 to 204 may be pure Tx antennas, while at least one separate reception antenna may be provided for receiving an UL data stream with the feedback index information 70. Furthermore, a feedback extraction unit 27 is provided, to which the received UL data is supplied in order to extract or derive the feedback index information 70 and possible other feedback information. The transceiver unit 22 is further connected to a signal processing stage 26 which is responsible for receiver-related process-

ing, such as demodulating, descrambling, decoding, etc. for received UL data, and for transmitter-related processing, such as modulating, scrambling, coding, beam- forming, user selection etc. for DL data to be transmitted. The signal processing stage 26 is controlled by a codebook element 75, which may be a codeword, vec- tor or matrix, indexed by the feedback index information 70 in a codebook 28 maintained or stored at the feedback extraction unit 27. The codebook 28 corresponds to the codebook 18 of the transmit and receive unit of Fig. 2, so that the indexed codebook element corresponds to the one selected at the transmit and receive unit. The signal processing stage 26 controls beamforming for multi- antenna transmission based on the indexed codebook element 75, e.g., by applying corresponding real and/or complex weights indicated by the indexed codebook element 75 to transmission signals transmitted via the antennas 201 to 204.

As a modification, the codebook 28 may be maintained or stored at the signal processing unit 26, wherein the feedback index information 70 is supplied by the feedback extraction unit 27 to the signal processing unit 26.

The generation of the enhanced codebooks 18 and 28 is now described with reference to Figs. 4 and 5.

Fig. 4 shows a general representation of an example of the codebooks 18 and 28 shown as a table of codebook elements which consist of weights wi to Wι_ arranged as columns and which are indexed by numbers 1 to i+k+l indicated in the top row of Fig. 4. These index numbers directly or indirectly correspond to the above feedback index information 70. Thus, a total number of i+k+l codebook elements is provided and separated into a first subset 110 of i elements, a second subset 120 of k elements and a third subset 130 of I elements. It is noted that the arrangement of the three subset 110 to 130 may of course vary substantially and any possible interleaved or even non-regular structure could be used, as long as the location of codebook elements of each subset is known and be indexed by the associated index number. It is however pointed out that the first subset 110 may as well be enhanced by only one of the second and third subsets 120, 130, so that the codebook only comprises two subsets.

According to the embodiment, the first subset 110 of codebook elements consist of weights for phase-only antenna control, i.e., complex weights which only affect the phase of the transmission signal transmitted via the associated antenna. In con-

trast thereto, the weights of the codebook elements of the second subset 120 are used for antenna subset selection control, which means that they can used to transmit the transmission signal only via a corresponding subset of all antennas 201 to 204, e.g., only two antennas. Finally, the weights of the third subset 130 of elements are configured to provide single antenna selection control, which means that the codebook elements of the third subset 130 serve to transmit the transmission signal only via a single one of the antennas 201 to 204.

Fig. 5 shows a more specific example of a codebook according to the embodi- ment. Here, the first subset 110 of the codebooks 18 and 28 consists of a size-48

Hochwald codebook {w, w ₂ -wj which has a circular correlation property and can be generated for example similar to the initially mentioned publication "Systematic design of unitary space-time constellations" by B. M. Hochwald, T.L. Mar- zetta, T.J. Richardson, W. Sweldens, and R. Urbanke, with the following relation- ship:

W ₇ = Q^ ¹ W ₁ , 1 = 2,3,-,L

where L is size of Hochwald codebook (48 in the present example of four anten- nas 201 to 204), Wi is the first element, which can be chosed to be one column of M _t X M _t IDFT (Inverse Digital Fourier Transformation) matrix, e.g.,

where M _t is the number of transmit antennas, and the above rotation matrix Q is a diagonal matrix constructed by an integer rotation vector u = [u _] u ₂ - - -u _Mι \, 0 ≤ u _] , u ₂ ,-- -,u _Mι ≤ L -l ,

,2π J—"i

0

Q =

.2π

J _T ^U M _t

0

The choice of the rotation vector should minimize the maximum correlation between elements in the codebook. The exemplary 48 elements all lead to phase-

only adaptation from the 4 antennas 201 to 204, which provides good performance in strong correlated channel.

Additionally, the last 16 elements in the codebook cover the second and third sub- sets 120 and 130 and which include 12 elements of the second subset 120 for antenna subset (e.g., antenna pair) selection with 0 or π relative phase rotation, and additional 4 elements of the third subset 130 for single antenna selection. This proposed selection of codebook elements helps the proposed codebook to improve the performance in uncorrelated channel in addition to phase-only weighting achieved by the incorporated Hochwald codebook of the first subset 110.

The above exemplary codebook of Fig. 5 can be denoted as "48+12+4" since it includes a size-48 Hochwald codebook (first subset 110) and a size-12 two antenna selection codebook (second subset 120) and a size-4 single antenna selec- tion codebook (third subset 130). It thus provides both phase-only weighting (via the first subset) and amplitude-only weighting (in the third subset) as well as the combination of both (in the second subset) which leads to good tradeoff as correlation changes.

As more general examples, improved Hochwald or improved other phase-only adaptation codebooks combine the Hochwald-type or other phase-only adaptation codebooks with antenna subset selection with phase rotation. In a general expression "x+y+z" means a codebook including size-x Hochwald or phase-only adaptation codebook, y elements of two or more antenna selection and z elements of single antenna selection. In the specific but non-limiting case of a 6-bit codebook, the sum of x, y and z is 64. The number z of single-selection codebook elements is thus always the number of transmit antennas, while the number y of antenna- subset selection codebook elements depends also on the number of possible relative phases given two or more selected antennas. I.e., for a "48+12+4" codebook, two antennas are selected and the two relative phases are 0 or π, so that y=2*C(4,2)=12. For a "42+18+4" codebook, the three different relative phases are

0, 2*π/3 and 4*π/3. For a "36+24+4" codebook, relative phases are 0, π/2, π and 3*π/2 and so on. This can be basically written as phases φ _t = (i -Y) - 2π/L, l ≤ i ≤ L having L different phase states.

As another example, "(64-m)+y+z" means a codebook including a size-64 Hochwald codebook, in which m elements have been left out, y elements of two

anteπna selection and z elements of the single antenna selection. The sum of 64- m, y and z is 64, while y and z have the same meaning as in the above "x+y+z" codebook.

The above "48+12+4" codebook example provides weights for both single antenna selection and antenna subset selection. It includes 24 orthogonal pairs (18 pairs from weights for antenna subset selection and 6 pairs from weights for single antenna selection) which can be used for two stream transmission. The number of additional orthogonal pairs in the first subset of the codebook is dependent on the selected phase-only adaptation codebook or Hochwald codebook. Pairs of weights for single antenna selection can be used e.g. for 4x2 S-PARC (Selective-Per Antenna Rate Control) systems. The weights for antenna subset selection corresponds to a generalization of a 1-bit TxAA mode 1 and antenna selection. Some orthogonal pairs of weights for antenna subset selection can also be used for Double TxAA, DSTTD-SGRC (Double STTD - Sub Group Rate Control) or GS- PARC (Group Selective Per Antenna Rate Control).

However, needless to mention that other combinations according to Fig. 4 with or without Hochwald-type or other phase-only adaptation codebook subsets are pos- sible as well.

As mentioned above, a structured approach is used to generate the codebook. Therefore the generation of the codebook can be done when necessary, which means that there is no need to store the codebook elements all the time. This is advantageous over some random-searched codebooks, e.g., Grassmannian and Xia's codebook.

Better performance is achieved for the proposed codebook considering different correlation and scenarios, compared with other codebooks in literatures (e.g., ini- tially mentioned references).

Moreover, the above systematically generated (structured) codebook is easier to be implemented compared with computer generated codebooks.

Figs. 6 and 7 show flow diagrams of the basic processing steps at both radio communication ends of a MIMO transmission system with multiple transmission

antennas according to an implementation example with the proposed advanced feedback signaling according to the embodiment.

The processing at the receiving end, e.g., at the MS 10, is shown in Fig. 6 and comprises a first step S101 of receiving a multi-antenna DL signaling. Then, a desired codebook element for optimized transmission is selected in step S102 from the codebook 18 and a corresponding index number is derived. In step S103, the feedback index information 70 is added to the UL transmission stream and forwarded to the transmitting end of the MIMO system.

Furthermore, the processing at the transmitting end, e.g., at the BS 20, is shown in Fig. 7 and comprises a first step S201 of receiving an UL stream with the incorporated feedback index information 70. Then, the incorporated feedback index information 70 is extracted in step S202. Finally, in step S203, the extracted feedback index information 70 is used to access the codebook 28 in order to derive the indexed codebook element 75. Thereby, the transmitting end is capable of controlling beamforming for multi-antenna transmission based on the derived codebook element 75.

Figs. 8 and 9 show diagrams indicating simulation results of SNR gains over a flat fading channel of different Hochwald-type codebooks according to various embodiments in dependence on different Tx correlation factors.

Fig. 8 shows simulation results for different Hochwald and improved Hochwald codebooks, which reveals that codebook-type "x+y+z" is better than codebook- type "(64-m)+y+z" basically, except the two cases with only single antenna selection, i.e., types "60+4" and "(64-4)+4". In particular, it can be gathered that codebook type "48+12+4" provides the best overall performance.

Additionally, it can be gathered that two antennas selection achieved by the above second codebook subset 120 (i.e., parameter y) is also important for improving performance in weak and medium correlated channel in addition to single antenna selection.

Fig. 9 shows simulation results for improved Hochwald codebook type "48+12+4" in comparison with other conventional codebook types TxAA Mode 1 , Grassman-

nian, Intel, Hochwald, and Xia. As can be gathered, the proposed codebook-type "48+12+4" is better than the conventional codebook types.

Fig. 10 shows a schematic block diagram of a software-based implementation of the proposed advanced feedback transmission system. Here, the transmitter shown in Fig. 3 and the receiver shown in Fig. 2 each comprise a processing unit 210, which may be any processor or computer device with a control unit which performs control based on software routines of a control program stored in a memory 212. Program code instructions are fetched from the memory 212 and are loaded to the control unit of the processing unit 210 in order to perform the processing steps of the above functionalities described in connection with the respective Figs. 6 and 7 or with the respective blocks 12, 16 and 18 of Fig. 2 or blocks 26, 27 and 28 of Fig. 3. These processing steps may be performed on the basis of input data Dl and may generate output data DO, wherein at the receiver end the input data Dl may correspond to the received DL data and the output data DO may correspond to the feedback index information 70. On the other hand, at the transmitter side, the input data may correspond to the received UL data and the output data may correspond to control information (e.g. weights) required to control beamforming of the multi-antenna transmission.

To summarize, methods, a system, a transmitter apparatus, a receiver apparatus, and computer program products for providing advanced feedback signaling in ar multi-antenna transmission system have been described, wherein a codebook comprising an indexed set of beamforming elements is adapted to comprise a first subset of elements for phase-only antenna control, and at least one of a second subset of elements for antenna subset selection and a third subset of elements for single antenna selection. Thereby, better performance can be obtained for different correlations and scenarios, and systematic codebook generation simplifies implementation.

It is to be noted that the present invention is not restricted to the embodiments described above, but can be implemented in any network environment involving multi-antenna transmission controlled by feedback signaling. Any signaling format or means can be used for feeding back the feedback information, which may be an index information or even the codebook element itself. Moreover, any kind of codebook structure may be used for arranging the first codebook subset and the at

least one of the second and third codebook subsets. The embodiment may thus vary within the scope of the attached claims.