BACKGROUND
[0001] Embodiments of the invention relate to a method for determining beamforming parameters in a wireless communication system, and to a wireless communication system. More specifically, embodiments of the invention may be used for improving the transmission in wireless communication systems and may be particularly interesting for mobile radio systems and wireless millimeter wave transmission systems.
[0002] For improving the performance of wireless communication networks or radio systems, multi antenna techniques using group antennas (antenna arrays) at the transmitting side and at the receiving side may be used. One approach is called beamforming, and in accordance with this approach a signal is split at the transmitter and multiplied by a complex weighting factor (having a magnitude and a phase) for every transmitter antenna individually. At the receiver, the signals of the individual receiving antennas are also weighted with complex factors and added. Weighting the signals of a group antenna is implemented by a beamformer. If the weights all have constant amplitude and differ only in phase, this is referred to as equal-gain beamforming or as a phased array. Contrary to the beamforming signal processing, in MIMO signal processing (MIMO = Multiple-Input Multiple-Output), not only complex weightings but also costly digital signal processing operations need to be performed in every branch. The MIMO operations may each have a different effect on certain portions of the antenna signals (samples in time or frequency), whereas in beamforming all signal portions are weighted identically. Equal-gain beamformers may be implemented in analog circuitry with relatively little effort and are hence particularly interesting when a large number of antennas is used. In contrast, systems using MIMO signal processing require a higher effort in the analog and digital circuitry and are hence generally limited to moderate numbers of antennas, e.g. to only 2 or 4 antennas.
[0003] Fig. 1 shows a schematic equivalent baseband representation of a unidirectional wireless communication system comprising
M antennas at the transmitter and
N antennas at the receiver. The system 100 comprises a transmitter 102 having an input 104 at which an input data signal
d_{s} to be transmitted in the wireless communication system or radio system 100 is received. The transmitter comprises a plurality of antennas 105
_{1}, 105
_{2}, ... 105
_{M}, i.e. the transmitter 102 comprises
M antennas. The input data signal received at the input 104 is processed by a transmitter signal processing unit 106 which outputs a signal x to be transmitted. The signal x received at the beamformer input 107 is distributed via a transmit beamformer 108 to the respective antennas 105
_{1} to 105
_{M}. The beamformer 108 comprises a dividing or splitting circuit 109 and a plurality of weighting elements 110
_{1}, 110
_{2}, ... 110
_{M} applying to the input signal x received at the beamformer input 107 respective weighting factors
w_{1}, w_{2}, ...,
w_{M}. The weighted input signals are transmitted from the antennas 105
_{1} to 105
_{M} via a radio channel 112 to a receiver 114. The receiver 114 comprises a plurality of receive antennas 116
_{1}, 116
_{2}, ..., 116
_{N}. The signals received from the respective antennas 116
_{1} to 116
_{N} are fed into a receive beamformer 118. The receive beamformer 118 comprises a plurality of weighting elements 120
_{1}, 120
_{2}, ... 120
_{N} that are provided for applying to the respective signals received from the antennas 116
_{1} to 116
_{N} the respective weighting factors
z_{1}, z_{2}, ...
z_{N} and an adding circuit 122. The adding circuit adds the weighted receive signals to form the output signal
y of the beamformer 118 that is provided at an output 124. The signal
y is fed into the receiver signal processing unit 126 providing the received data signal
d_{r} at the output 128 of the receiver 114. In case beamforming is done at the transmitter and at the receiver, a beamforming system comprises a transmit beamformer, transmit antennas, receive antennas and a receive beamformer. For example, the transmit beamformer 108, the transmit antennas 105
_{1} to 105
_{M}, the receive antennas 116
_{1},..116
_{N} and the receive beamformer 118 shown in Fig. 1 form a beamforming system. When beamforming is only applied at the transmitter, the beamforming system comprises the transmit beamformer, the transmit antennas, and the receive antennas. Alternatively, when using beamforming only at the receiver, the beamforming system comprises the transmit antennas, the receive antennas, and the receive beamformer.
[0004] At the transmitter 102
M beamforming branches are formed, each of the beamforming branches comprises one of the weighting elements of the beamformer 108 and one of the antennas of the transmitter. For example, a first beamforming branch is formed by the weighting element 110
_{1} of the beamformer 108 and the antenna 105
_{1}. Likewise, at the receiver 114
N beamforming branches are formed, the respective branches comprises one of the weighting elements of the beamformer 118 and one of the antenna elements of the receiver. For example, a first beamforming branch at the receiver 114 is formed by the antenna element 116
_{1} and the weighting element 120
_{1} of the receive beamformer 118.
[0005] By beamforming at the transmitter 102, the power radiated in certain space directions is increased, while it is reduced in other space directions. Beamforming at the receiver 114 has the effect that signals from certain space directions are received in an amplified manner and from other space directions in an attenuated manner. Because the transmission attenuation increases with rising transmission frequencies, beamforming is considered as promising and inexpensive means for increasing the performance of systems having high transmission frequencies, e.g. future 60 GHz systems.
[0006] The weighting factors
w_{1}, w_{2}, ..., w_{M} or
z_{1}, z_{2}, ..., z_{M} for the individual antennas 105
_{1} to 105
_{M} or 116
_{1} to 116
_{N} at the transmitter 102 or at the receiver 114 may each be combined into one beamforming vector. Fig. 1 shows an example of an unidirectional wireless communication system allowing for a transmission using
M beamforming branches at the transmitter 102 and
N beamforming branches at the receiver 114. The adjustment of the signals provided by the transmitter 102 using the transmit beamformer 108 is described by the transmit beamforming vector w:
[0007] The adjustment of the signals received at the receiver 114 using the receive beamformer 118 is described by the receive beamforming vector z:
[0008] In the case of using the equal-gain beamforming, the elements of the beamforming vectors have a constant modulus. If the magnitude of the beamforming vectors is defined to 1, the beamforming vectors are given as follows:
and
wherein
- ϑ_{m}
- = phase values ϑ_{m} ∈ [0,2π] for the transmitter 102, and
- ϕ_{n}
- = phase values ϕ_{n} ∈ [0,2π] for the receiver 114.
[0009] Many known systems may use discrete (quantized) phase values only, so that the number of possible beamforming vectors is limited.
[0010] The wireless transmission between the antenna groups 106 and 116 at the transmitting side 102 and at the receiving side 114 is performed via the radio channel 112 including all possible connection paths between all transmitting antennas 106
_{1} to 106
_{M} and all receiving antennas 116
_{1} to 116
_{N}. The radio channel 112 is defined using a matrix, the so called channel matrix
H.
[0011] The presented beamforming techniques are considered for a unidirectional transmission between a transmitter 102 and a receiver 114. Conventionally, wireless communications systems are provided for a bidirectional transmission between stations. Each station needs to be provided with a transmitter and a receiver. Both in the transmitter and in the receiver beamforming techniques may be used. Fig. 1(a) depicts a bidirectional, wireless beamforming transmission system 900 having two stations 902 and 904. Each station is provided with a transmitter 906, 910 and a receiver 908, 912 having a structure as described in Fig. 1. Up to four beamforming vectors may be involved in case of such a bidirectional transmission between the two stations, station 902 and station 904: for a transmission from the station 902 to the station 904 the beamformer 914 at the station 902 may use for a transmitting beamforming at station 902, and the beamformer 916 at the station 904 may use for a receiving beamforming at station 904; and for a transmission from the station 904 to the station 902 the beamformer 918 at the station 904 may use for a transmitting beamforming at station 904, and the beamformer 920 at the station 902 may use for a receiving beamforming at station 902. Since a bidirectional transmission can always be split into two unidirectional transmissions in opposite directions, with respect to the beamforming techniques it is sufficient to consider a unidirectional transmission and a unidirectional transmission system, respectively, including one transmitter and one receiver.
[0012] A problem for the operation of a multi-antenna system using beamforming is the adaptive (dynamic) adjustment of the beamforming vectors for maximizing the transmission quality in dependence on the propagation conditions. The methods for determining beamforming vectors may be divided into two categories: Methods with explicit beamforming channel knowledge, and methods without beamforming channel knowledge. In the former case, beamforming channel knowledge means that the radio channel 112 between any transmitting beamformer antenna element 106
_{1} to 106
_{M} and any receiving beamformer antenna element 116
_{1} to 116
_{N}, i.e. the beamforming channel matrix, is known. In the latter case, estimating the channel matrix presents a significant additional challenge. In bidirectional transmission, in general, two channel matrices are to be considered: one for the forward direction and one for the backward direction, and they have to be acquired in practice by a beamforming channel estimation.
[0013] The following problems occur when using a beamforming system with beamforming signal processing according to Fig. 1:
- 1. Determining optimal beamforming vectors at transmitter and receiver without explicit channel knowledge.
- 2. Estimating a multi-antenna channel in systems with beamforming signal processing.
- 3. Determining suitable beamforming vectors at the transmitter and at the receiver using channel knowledge for Systems with pure beamforming signal processing.
The following problem occurs when using a hybrid MIMO beamforming system with MIMO signal processing and beamforming signal processing according to Fig. 5:
- 4. Determining suitable beamforming vectors at the transmitter and at the receiver in hybrid MIMO beamforming systems.
[0014] For determining suitable beamforming vectors, known methods without explicit channel knowledge provide for a training phase, during which test signals or training symbols are transmitted and evaluated within a training frame at different suitably selected beamforming vectors (see e.g.
ECMA-387 Standard: High Rate 60 GHz PHY, MAC and HDMI PAL, 2008, Ecma International). The temporal sequence of beamforming adjustments may be described by a matrix (a training matrix), which consists of the respective beamforming vectors. In a bidirectional radio system using two-way beamforming in the transmitting and receiving branches, transmission of training frames is performed in both directions, Optimizing the beamforming vectors is obtained by repeating the alternating transmission several times and iteratively adapting the beamforming vectors.
[0015] At present methods for determining the beamforming channel matrix are only known for systems where a group antenna is used only on one side (at the transmitter or at the receiver). In such a case, the beamforming channel matrix transitions into a beamforming channel vector, which is calculated using side information. The side information relate to the direction of incidence of the receive signal or the desired transmitting direction of the transmit signal and the geometry of the group antenna. This requires the presence of definite a-priori directional information and only little multipath propagation may exist in the radio channel (a typical field of such an application is the communication to a geostation-ary satellite, a communication from a vehicle, or a target tracking radar). Estimating the directional information for the receiver merely from the receive signals without a-priori information is possible, requires, however, MIMO signal processing see e.g.
Chung, Pei-Jung and Bohme, J.F., "Recursive EM and SAGE-inspired algorithms with application to DOA estimation" Signal Processing, IEEE Transactions on, 53(8):2664--2677, 2005;
Schmidt, R., "Multiple emitter location and signal parameter estimation", Antennas and Propagation, IEEE Transactions on, 34(3):276--280, 1986; or
Stoica, P. and Sharman, K.C., "Maximum likelihood methods for direction-of-arrival estimation", Acoustics, Speech and Signal Processing, IEEE Transactions on, 38(7):1132--1143, 1990).
[0016] Methods for determining a beamforming vector on the transmitter side or on the receiver side using channel knowledge from the directional information have been known for a long time for phased-array applications. However, these direction-based methods may only be applied with little or non-existing multipath propagation. Methods for determining the optimal beamforming vectors on the transmitter side and on the receiver side using channel knowledge - also with multipath propagation - have so far only been known for systems having MIMO signal processing (see e.g.
Heath, R.W., Jr. and Paulraj, A., "Multiple antenna arrays for transmitter diversity and space-time coding", Communications, 1999. ICC '99. 1999 IEEE International Conference on, pages 36--40 vol.1., 1999). For MIMO systems, different approaches for determining the channel matrix are known. Transferring such techniques to systems having only beamforming signal processing has not been possible so far, since, on the one hand, channel knowledge without side information (directional information) was not available for these systems and, on the other hand, it was unclear how a common beamforming vector is to be determined for all possibly different signal portions (in time and frequency).
[0017] For hybrid methods the principle of combining beamforming and MIMO signal processing is described e.g. by
Dammann, A. and Raulefs, R. and Kaiser, S., "Beamforming in combination with space-time diversity for broadband OFDM systems", Communications, 2002. ICC 2002. IEEE International Conference on, pages 165--171, 2002. Smart antennas are controlled via an adaptive antenna processor. The aim of beamforming is the transmission of the signal via several ideally statistically independent propagation paths. On the transmitting side, the data stream is split into several sub-streams based on the diversity principle, and combined again on the receiving side. Among others, space-time coding (STC) as a form of MIMO signal processing is suggested as method. Further, when using beamforming at the transmitter and receiver, a mutual allocation of the transmitting and the receiving antenna groups may be performed, wherein every group generates one data channel. However, a method for the allocation is not presented by
Dammann, A. and Raulefs, R. and Kaiser, S., "Beamforming in combination with space-time diversity for broadband OFDM systems", Communications, 2002. ICC 2002. IEEE International Conference on, pages 165--171, 2002, Further, it is assumed that the antenna processor provides the directions into which the beams are to be formed. Methods for determining the beamforming vectors are not discussed. In
Morelos-Zaragoza, R.H. and Ghavami, M., "Combined beamforming and space-time block coding with a sparse array antenna", Wireless Personal Multimedia Communications, 2002. The 5th International Symposium on, pages 432--434 vol.2, 2002, beamforming is also considered in the context of STC. The research focus lies on the influence of a correlation between different antenna beams on the performance of the system. Methods for determining suitable beamforming vectors are not considered.
[0018] Heath, R.W., Jr. and Paulraj, A., "Multiple antenna arrays for transmitter diversity and space-time coding", Communications, 1999. ICC '99. 1999 IEEE International Conference on, pages 36--40 vol.1., 1999 examine what gains may be obtained with different transmitting side diversity technologies in combination with beamforming, and what effect beamforming vectors deviating from the optimum have. The considerations are limited to a system having several antenna groups at the transmitter and one antenna at the receiver (MI-SO) and only apply under the assumption that only one propagation path exists between one antenna group and the receiver. Further, the research relates to a single user, wherein it is noted that in a multi-user system beamforming is not only to be used for maximizing the received power for the desired user, but at the same time for reducing interference for other users. This principle is also described by
Wu, Sau-Hsuan and Chiu, Lin-Kai and Lin, Ko-Yen and Chung, Shyh-Jong, "Planar arrays hybrid beamforming for SDMA in millimeter wave applications" Personal, Indoor and Mobile Radio Communications, 2008. PIMRC 2008. IEEE 19th International Symposium on, pages 1--6, 2008;
Wu, Sau-Hsuan and Lin, Ko-Yen and Chiu, Lin-Kai, "Hybrid beamforming using convex optimization for SDMA in millimeter wave radio", Personal, Indoor and Mobile Radio Communications, 2009 IEEE 20th International Symposium on, pages 823--827, 2009; and
Smolders, A.B. and Kant, G.W., "THousand Element Array (THEA)" Antennas and Propagation Society International Symposium, 2000. IEEE, pages 162--165 vol.1, 2000, where hybrid beamforming is considered. It is to be noted that the term "hybrid" refers to the combination of beamforming in the baseband and in the RF-range. The approach does include a transceiver architecture having several parallel transmitting and receiving branches in the digital baseband, however, no MIMO signal processing but beamforming signal processing is performed on the branches. Hence, the same are no hybrid methods in the sense of the above definition.
[0019] WO 2009/093870 A2 describes a method for transmitting a signal in Multiple Input Multiple Output (MIMO) system which includes transmitting a training signal based on a predetermined sequence via at least one beamforming antenna group, the beamforming antenna group including a plurality of antennas, receiving first information in not dictating at least one available beamforming antenna group from among the beamforming antenna groups, and transmitting second information and antenna weight information, the second information indicating beamforming antenna group being determined on the basis of the first in not formation, the antenna weight information being associated with a signal transmitted from the antenna group indicated by the second information.
[0020] US 2009/189812 A1 describes a system and a method for multi-stage antenna training of beamforming vectors. The method comprises acquiring a beamforming pattern in a wireless communication system, receiving a first plurality of signals having different transceiver sector patterns, measuring first indicators of link quality corresponding to the first plurality of signals, selecting at least one transceiver sector pattern based on the first indicators of link quality, receiving a second plurality of signals having different transceiver beam patterns, each transceiver beam pattern associated with the selected at least one transceiver sector pattern, measuring second indicators of link quality corresponding to the second plurality of signals, and selecting at least one transceiver beam pattern based on the second measures of link quality.
[0021] EP 2 037 594 A2 describes a method and apparatus for communicating in a wireless personal area network, comprising using adaptive beamforming configured for a low-rate mode for reliable low-rate communications and a high-rate mode for high-rate communications and using a fast algorithm to perform antenna beamforming for the high rate mode, wherein the fast algorithm includes training performed on a block-by-block basis with decision feedback from a receiver (RX) to a transmitter (TX) about the usefulness of further training stages.
[0022] US 2009/232240 A1 describes methods for beamforming that achieve beamforming optimality criterions. Some beamforming techniques are based on antenna directions with multiple resolutions.
SUMMARY OF THE INVENTION
[0023] It is an object of the present invention to provide for an improved approach for determining beamforming parameters in a wireless communication system or network including hybrid MIMO beamforming systems with MIMO branches at a transmitter and at a receiver.
[0024] This object is achieved by a method of claim 1, and by a wireless communications system of claim 3. While several embodiments and/or examples are disclosed in this description, the subject matter for which protection is sought is strictly and solely limited to those embodiments and/ or examples encompassed by the scope of the appended claims. Embodiments and/or examples mentioned in the description that do not fall under the scope of the claims are useful for understanding the invention.
[0025] The present invention provides a method for determining beamforming vectors for a transmitting station in a wireless communication system and beamforming vectors for a receiving station in the wireless communication system, wherein the transmitting station or the receiving station comprises a hybrid MIMO beamforming configuration including a plurality of MIMO branches, each MIMO branch comprising a plurality of antennas, the method comprises splitting the hybrid MIMO beamforming system into a plurality of subsystems, the splitting of the hybrid MIMO beamforming system comprising splitting the system into asymmetric subsystems comprising only one MIMO branch on the transmitting side or on the receiving side.
[0026] The present invention provides a wireless communication system comprising:
a transmitting station, and
a receiving station,
wherein the transmitting station or the receiving station comprises a hybrid MIMO beamforming configuration including a plurality of MIMO branches, each MIMO branch comprising a plurality of antennas, and
wherein the system is configured to split the hybrid MIMO beamforming system into a plurality of subsystems, the splitting of the hybrid MIMO beamforming system comprising splitting the system into asymmetric subsystems comprising only one MIMO branch on the transmitting side or on the receiving side.
[0027] In accordance with an embodiment of the fourth aspect of the invention, the subsystems may be considered as beamforming systems and the beamforming parameters for each subsystem may be determined in accordance with one or more of the methods of the first, second and third aspects described in detail below.
BRIEF DESCRIPTION OF THE DRAWINGS
[0028]
- Fig. 1
- is a schematic representation of an unidirectional wireless communication system using beamforming comprising M antennas at the transmitter and N antennas at the receiver.
- Fig. 1(a)
- is a is a schematic representation of a bidirectional wireless communication system using beamforming.
- Fig. 2
- is a flow diagram of a first aspect for determining beamforming vectors for the transmitting and receiving station.
- Fig. 3
- is a flow diagram showing the respective steps of a method in accordance with a second aspect for determining a beamforming channel matrix of a channel between the transmitting and receiving stations.
- Fig. 4
- is a flow diagram of a third aspect for determining beamforming vectors for the transmitting and receiving station.
- Fig. 5
- is an example of a unidirectional hybrid MIMO beamforming system having two MIMO branches at the transmitter and at the receiver each having associated therewith two beamforming branches.
- Fig. 6
- is a flow diagram representing the steps of a method in accordance with a fourth aspect of the invention.
- Fig. 7
- shows examples for the static allocation of subsystems of the 2x2 MIMO system of Fig. 5.
- Fig. 8
- shows examples for the asymmetric splitting of a 2x2 MIMO system as it is for example described in Fig. 5.
DESCRIPTION OF THE EMBODIMENTS
[0029] In the following the different aspects will be described. It is noted that the respective aspects, while being described separately may be used in combination, e.g. in a wireless communication system the beamforming parameters may be determined applying one or more of the subsequently described aspects (approaches).
[0030] In the subsequent description, the following notation is used: Small letters in italics (e.g.
a) describe complex- or real-valued quantities, capital letters in italics (e.g.
A) describe complex- or real-valued constants, bold small letters (e.g.
a) describe complex or real valued vectors, and bold capital letters (e.g.
A) describe complex- or real-valued matrices. The dimensions of a matrix having
N rows and
M columns is
N x
M. The k-th element of vector
a is indicated by [
a]
_{k}, and [
A]
n,m is the element of the n-th row and m-th column of matrix
A. A row vector having
K elements that are each 1 is described as
1_{1,K}, a matrix having
N rows and
M columns that are each 1 is describe as
1_{N,M}. The transpose and hermitian of a matrix
A are symbolized by
A^{T} and
A^{H}. The Kronecker product between matrices or vectors is represented by ⊗. The abbreviation
A^{-1} represents the inverse matrix of
A. A diagonal matrix having the values of the vector
a on the diagonal is generated by diag(
a). The operation vec(
H)=[[
H]
_{1,m}K[
H]
_{N,m}]
^{T} with
m=[1K
M] generates a column vector of length
N·M from the joined rows of matrix
H.
1^{st} aspect: training with the help of complete training matrices
[0031] In the following, a first aspect will be described. The first aspect concerns the training of a beamforming system, as it is for example depicted in Fig. 1, using complete training matrices. A beamforming vector
w for the antenna group 106 of the transmitter 102 shown in Fig. 1 is determined. Also, a beamforming
z of the antenna group 116 of the receiver 114 of the wireless communication system 100 of Fig. 1 is determined. The transmitter 102 and the receiver 114 use respective codebooks, each including a plurality of predefined beamforming vectors. The transmit codebook of the transmitter 102 may be stored in a memory provided by the beamformer 108 of the transmitter 102. Alternatively, the codebook may be provided at another location inside or external from the transmitter 102. Likewise, a receive codebook for the receiver 114 may be stored in a memory of the beamformer 118 or may be provided somewhere else in the receiver 114 or may be provided from an external source. The wireless communication system 100 as shown in Fig. 1 is configured to perform a method for determining a beamforming vector for the respective antenna groups of the transmitting and receiving stations, as it is depicted and described in the following with regard to Fig. 2. The respective method steps may be implemented in the control circuitry of the overall system or may be part of the control circuitry of the respective beamformers 108 and 118.
[0032] Fig. 2 shows a flow diagram of the first aspect for determining a beamforming vector for the transmitting and receiving stations, in a first step S100 a test transmission from the transmitting station to the receiving station using a test signal or a test symbol and a beamforming vector pair is performed. The beamforming vector pair includes a beamforming vector selected from the codebook of the transmitting station 102 and a beamforming vector selected from the codebook of the receiving station 114. Following the test transmission in step S100, in step S102 a transmission characteristic, for example a receive power, a signal-to-noise ratio (SNR), a signal-to-interference ratio (SIR) or a signal to interference-plus-noise ratio (SINR), of the test transmission is determined at the receiving station 114. At step S104, following the determination of the receive power, the SNR, the SIR or the SINR for a test transmission, it is determined as to whether all possible test transmissions were performed or not. In case not all possible test transmissions were performed, a new beamforming vector pair is selected at step S106 and the method returns to step S100 for performing a test transmission using the new beamforming vector pair and the test signal. Thus, by means of steps S104 and S106 the test transmission and the determination of the transmission characteristic are repeated using different beamforming vector pairs. In accordance with the first aspect, the beamforming vectors in the beamforming vector pairs are selected such that each beamforming vector from the codebook of the transmitting station encounters all beamforming vectors from the codebook of the receiving station. Once all possible test transmissions were performed, e.g. all possible combinations of beamforming vectors from the transmitting station and from the receiving station were used for performing the test transmission the method proceeds to step S108, in accordance with which the beamforming vectors for the transmitting and receiving stations are determined from that beamforming vector pair for which the transmission characteristic, for example the receive power, the SNR, the SIR or the SINR at the receiver 114 had a predefined value, for example which of the beamforming vector pairs resulted in a maximum receive power, SNR, SIR or SINR at the receiver 114. In accordance with other examples, it is not necessary to evaluate the receive power, SNR, SIR or SINR after each step. Rather, the received test symbols may be recorded at the receiver and the evaluation of some or all of received test symbols and the selection may be done after all or a predefined number of test symbols has been transmitted.
[0033] The thus determined beamforming vectors are used for a transmission from the transmitter 102 to the receiver 114.
[0034] As just described, the 1
^{st} aspect relates to the use of suitable training matrices
T, i.e. a specific selection and temporal sequence of beamforming vectors for the training. The method may operate without knowledge of the beamforming channel matrix. It is assumed that the beamforming is performed based on codebooks. A codebook C is the (finite) magnitude of all possible and allowable beamforming vectors. Basically, an individual codebook may be defined for every antenna group in the system, however, group antennas having the same number of antenna elements may also use the same codebook. The codebook may also be expressed as codebook matrix C, into which the beamforming vectors of the codebooks are entered column by column. The selection of the codebook may be arbitrary, as long as the rows and columns of the codebook matrix are not linearly dependent. The maximum diversity gain, visible in the maximum increase of the bit error frequency curve for large signal/interference power intervals is obtained for unitary codebook matrices (see e.g.
Love, D.J. and Heath, R.W., Jr., "Equal gain transmission in multiple-input multiple-output wireless systems", Communications, IEEE Transactions on, 51(7):1102--1110, 2003). In a unitary codebook matrix, all beamforming vectors (columns) are pairwise orthogonal and have the norm one (orthonormal). By adding further non-orthonormal beamforming vectors, additionally, antenna gain may be realized, which is expressed in an improvement of the signal-to-noise ratio. Further optimization criteria for codebooks are, for example, minimal phase numbers for equal-gain beamformers (see e.g.
ECMA-387 Standard: High Rate 60 GHz PHY, MAC and HDMI PAL, 2008, Ecma International).
[0035] In the following, the codebook matrices for the transmitter 102 and the receiver 114 are referred to by
C_{T} and
C_{R}, respectively. For a unidirectional transmission between the two stations 102 and 114 two different training matrices are provided: The matrix
T_{T} for beamforming at the transmitter 102, and the matrix
T_{R} for beamforming at the receiver 114. The matrices
T_{T} and
T_{R} form a matrix pair. Each of the matrices of the matrix pair has the same number of columns. If the codebook of the transmitter 102 includes
K_{T} vectors and the codebook of the receiver 114 includes
K_{R} vectors, the training matrices
T_{T} and
T_{R} will each have
K_{T}·K_{R} columns. The vectors of the codebook of the transmitter 102 are included
K_{R}-times in the training matrix
T_{T} of the transmitter. In the same way, for the receiver 114, the training matrix
T_{R} includes the vectors of the codebook of the receiver 102
K_{T}-times.
[0036] For the training, the beamforming vectors for the transmitter 102 and for the receiver 114 are each taken column by column, starting with column 1, successively from the respective training matrices, and the test transmission using suitable training signals or training symbols is performed. Hence, for the training,
K_{T}·K_{R} beamforming configurations and test transmissions are required. The method in accordance with this aspect is based on selecting the order of beamforming vectors in the training matrices such that every vector from the codebook of the transmitter 102 encounters all vectors from the codebook of the receiver 114 - and vice versa. A simple design rule for obtaining the training matrices may be stated using the Kronecker product. A matrix pair
T_{T},
T_{R} may be calculated as follows:
[0037] Equations (1) and (2) may be exchanged, which means
T_{T}=
C_{T}⊗
1_{1},
_{KR},
T_{R}=
1_{1,KT}⊗
C_{R}. Also, simultaneously exchanging columns in
T_{T} and T
_{R} is possible.
[0038] In a bidirectional transmission, the method may be performed for both directions according to the duplex method used in the system. Every station requires both transmitting and receiving beamforming vectors and a training matrix
T_{T} for the transmitter or for the receiver
T_{R}. The training is then performed separately for both directions of transmission, wherein the respective matrix pairs of training matrices are used. A station may use the same antennas for transmitting and receiving. In such a case the beamforming vectors for one direction of transmission may be determined and used also in the other direction of transmission.
[0039] After the complete run of all test transmissions, those beamforming vectors on the transmitter side and on the receiver side for which the highest received power, SNR, SIR or SINR has been obtained during the training phase are obtained. These beamforming vectors are optimal for the selected codebooks at transmitter 102 and at the receiver 114, independent of the used data transmission method. The optimization may take place at the receiver 114 so that the determined transmitting beamforming vector (codebook entry) has to be transmitted to the transmitter 102. The method in accordance with the first aspect determining optimal beamforming vectors for the transmitter 102 and for the receiver 114 in a single training phase - without any iterative feedbacks from the receiver 114 to the transmitter 102.
[0040] The first aspect is advantageous, since for the training of the transmitting and receiving beamformers 108 and 118 (unidirectional), the transmission of training symbols in one direction is sufficient using the respective training matrix pair. Consequently, for a complete training of the beamformers 108, 118, for both directions of transmission (bidirectional), only a single transmission in each direction (station 102 to station 114 as well as station 114 to station 102) is required. The method allows not only determining particularly suitable adjustments but allows for the determination of optimal beamformer adjustments with respect to the codebooks and the chosen optimization criterion (e.g. received power, SNR, SIR, SINR). Optimizing the beamformer weights by several transmissions in both directions and iterative adoption of the weights is omitted. The training is simplified, accelerated and the performance of the data transmission system is maximized. If small-scale codebooks are used, the method is also interesting for mobile applications with quickly changing radio channels.
2^{nd} aspect: estimating the beamforming channel matrix
[0041] Subsequently, the second aspect will be described, in accordance with which a beamforming channel matrix is estimated without side information and without MIMO signal processing. Again, a wireless communication system 100 as depicted in Fig. 1 is assumed, and a beamforming channel matrix is to be determined which describes the radio channel 112 between the transmitter 102 and the receiver 114. The transmitter 102 and the receiver 114 comprise the respective antenna groups 106
_{1} to 106
_{M} and 116
_{1} to 116
_{N}. Further, as already described above with regard to the first aspect, respective codebooks for the transmitter 102 and 114 are provided, each of the codebooks comprising a plurality of predetermined beamforming vectors for the antenna group of the transmitter 102 or for the antenna group of the receiver 114.
[0042] Fig. 3 is a flow diagram showing the respective steps of a method in accordance the second aspect. In a first step S200 a test transmission from the transmitter 102 to the receiver 114 using beamforming vectors for the receiver and for the transmitter and using a test symbol is performed. At step S202 the beamforming vectors at the receiver and at the transmitter are varied in accordance with a scheme, wherein the scheme allows for a variation of the beamforming vectors at the transmitting station and at the receiving station on the basis of a transmit estimate matrix and a receive estimate matrix, wherein each element of an estimate matrix defines the beamforming weight for a specific antenna form the antenna group used during a specific test transmission. Basically the approach in accordance with the second aspect is similar to the first approach except that other matrices are used. For each new beamforming setup a test symbol is transmitted. At step S204 it is determined as to whether a variation of the beamforming vectors in accordance with a scheme was completed. In case it was not completed, the method returns to step S200 and performs the next test transmission on the basis of the varied beamforming vectors. Otherwise, in case the beamforming vector variation was completed, the method proceeds to step S206 and the beamforming channel matrix is determined from the test transmissions.
[0043] As just described, the 2
^{nd} aspect relates to a method for estimating the beamforming channel matrix
H without side information and without MIMO signal processing. In a system with
M transmitting antennas 106 and
N receiving antennas 116,
N·M test transmissions are required. The beamforming channel matrix
H has
M columns and
N rows, and the element [
H]
_{n,m}=
h_{n,m} describes the transmission from the transmitting antenna
m to the receiving antenna
n:
[0044] The beamforming channel matrix
H is estimated by performing several test transmissions, i.e. transmitting several estimation symbols subsequently, while varying the beamforming vectors at the transmitter 102 and at the receiver 114 according to a specific scheme. The scheme may be described mathematically using two matrices, a transmitting estimation matrix
E_{T} and an associated receiving estimation matrix
E_{R}. The transmitting estimation matrix
E_{T} includes the beamforming vectors for the transmitter 102 as column entries in chronological order, beginning with column 1.
E_{T} is derived from a base transmitting estimation matrix
B_{T}. For a beamforming system having
M transmitting antennas, the base transmitting estimation matrix
B_{T} has the dimension
M x
M. The transmitting estimation matrix
E_{T} may be defined using the Kronecker product:
[0045] It follows that the transmitting estimation matrix
E_{T} has the dimension
M x
NM, and that the element [
E_{T}]
_{m,k} (=value in the
m-th row and at the
k-th column of the transmitting estimation matrix
E_{T}) describes the beamforming weight for the
m-th transmitting antenna 106 in the
k-th of
N·M transmissions.
[0046] The receiving estimation matrix
E_{R} includes the beamforming vectors for the receiver 114. The receiving estimation matrix
E_{R} is also derived from a base receiving estimation matrix
B_{R}, wherein the base receiving estimation matrix
B_{R} has the dimension
N x
N for a system having
N receiving antennas:
[0047] It follows that the receiving estimation matrix
E_{R} has the dimension
N x
NM, and that the element [
E_{T}]
_{n,k} (=value in the
n-th row and at the
k-th column of the receiving estimation matrix
E_{R}) describes the weight for the
n-th receiving antenna 116 in the
k-th transmission. The base estimation matrices
B_{T} and
B_{R} for the transmitter 102 and for the receiver 114 may be selected to be the same or different.
[0048] In the
k-th transmission, a training symbol
x_{k}=
[x]_{k} is transmitted, and the symbol
y_{k}=[
y]
_{k} is received at the receiver 114.
[0049] The beamforming channel matrix may be estimated as follows. All transmissions may be presented in matrix vector notation as follows:
wherein
X = diag(
x) includes the transmitting vector x. The matrix
S includes the base estimation matrices according to
and
h = vec(
H) is the vectorized channel matrix
H. Using the transmission coefficient
d_{k}=[
d]
_{k} for every training symbol
equation (6) reads as follows
[0050] The estimation of the beamforming channel matrix or is performed by:
[0051] Since the matrix
S is independent of the channel
H and previously known,
S^{-1} may be calculated and stored in advance, such that equation 10 may be implemented efficiently. Unitary matrices may be used for the base estimation matrices, and S is also unitary and the following applies for equation 10:
[0052] For the base estimation matrices
B_{T} and
B_{R}, basically, any square matrices or codebooks may be used, as long as the elements are valid settings for the beamforming weights
w_{m}=[
w]
_{m} or
z_{n}=[
z]
_{n} and the matrices have full rank, i.e. the rows or the columns are not linearly dependent. In the case of equal-gain beamforming, the beamforming weights only differ in phase and the implementation of the beamforming signal processing allows only a discrete number of equidistant phase states. For channel estimation matrices, only certain phase states are possible and low phase numbers are generally advantageous.
[0053] Regarding the magnitude, unitary matrices may have the same eigenvalues and are hence optimal with respect to a low estimation error in error-prone transmissions. Above that, unitary matrices, may be inverted in an particularly easy manner. The following unitary matrices for equal-gain beamforming may be used:
- 1. Hadamard matrices having the two phase states {0, π} corresponding to beamformer weights {1, -1}. Hadamard matrices of the dimensions N x N are known for N=2 orN=4k with k ∈ N.
- 2. Matrices having four equidistant phase states {0, π, π/2, -π/2} corresponding to beamformer weights {1,-1,j,-j} and
These matrices cannot only be stated for N = 4k with k ∈ N but also for further even N. - 3. Matrices having
equidistant phase states may be stated for all N forming a square number by the cyclical shift (Töplitzmatrix) of a minimum-phase uniform sequence having perfect periodical autocorrelation. - 4. Matrices having N equidistant phase states may be constructed for all N. For this, DFT matrices may be used.
[0054] The unitarity of the base estimation matrices may be abandoned and suitable matrices may be determined by selecting certain rows and columns from suitable larger matrices. For example, from a Hadamard matrix that is larger than the desired base estimation matrix, smaller matrices having two phase states may be derived.
[0055] Basically, the estimation method may also be performed using non-square base estimation matrices. In such a case,
B_{T} and
B_{R} include more columns than rows and the determination of the vector
h is performed with an over determined equation system. The over determined equation system may be solved according to the criterion of least error squares, and the estimation accuracy may be improved in case the transmission is interfered by noise. However, correspondingly more test transmissions are required.
[0056] If beamforming is used for both directions of transmission, the method is generally performed for both directions, i.e. one beamforming channel matrix for each direction is determined. In case the same antennas are used for transmitting and receiving at a station, estimating the channel matrix for one direction of transmission may be sufficient. The beamforming channel matrix derived may be used both for transmitting and for receiving.
[0057] In the following some examples for unitary base estimation matrices are given:
Matrices having two phase states:
[0058] Lists of Hadamard matrices for
N = 4
k with
k ∈ N are available in mathematical literature, example for
N =
8:
Matrices having four phase states:
[0059] Matrices having four phase states are described for all even
N with
N ≤ 16 (see e.g. http://chaos.if.uj.edu.pl/∼karol/hadamard/), example for
N=8:
Matrices having phase states:
[0060] Matrices having
phase states may be generated by cyclically shifting from Frank sequences (see e.g. Frank, R. and Zadoff, S. and Heimiller, R. Phase shift pulse codes with good periodic correlation properties.
Information Theory (Corresp.), IRE Transactions on, 8(6):381--382, 1962), example for
N=16:
Matrices having N phase states:
[0061] The descriptive matrices of the discrete Fourier transformation (DFT matrices) may be used as base estimation matrices having
N phase states, by cyclically shifting or by permuting rows and columns, further base estimation matrices having the same characteristics may be generated. Example for
N=8:
[0062] The second aspect of the invention is advantageous, since for estimating the beamforming channel matrix for one direction of transmission, one transmission phase using the respective estimation matrix pair is sufficient. For estimating the channel matrices for both transmission directions, one transmission phase for each direction (station 102 to station 114 as well as station 114 to station 102) is required. The number of estimation symbols to be transmitted may be minimized and consequently also the period for which this transmission phase is required.
3^{rd} aspect: determining the beamforming vectors by using the beamforming channel matrix
[0063] In the following, the third aspect will be described in further detail. The beamforming vectors may be determined using the known beamforming channel matrix. Again, a wireless communication system as described with regard to Fig. 1 is assumed, and a beamforming vector of the antenna group 105 of the transmitter 102 as well as a beamforming vector of the antenna group 116 of the receiver 114 is determined. Again, as described above, the transmitter 102 and the receiver 114 comprise a codebook including a plurality of predefined beamforming vectors.
[0064] Fig. 4 depicts a flow diagram of the just mentioned third aspect. In a first step S300 from the codebook of the transmitter 102 or from the codebook of the receiver 114 the beamforming vector is determined that yields a first predetermined result when applying the beamforming weights defined in the beamforming vector to a known beamforming channel matrix describing the radio channel 112 between the transmitter 102 and the receiver 114. In a subsequent step S302 for the receiver or the transmitter the beamforming vector is selected from the codebook, wherein that beamforming vector is selected that yields a second result when applying the weights of the selected beamforming vector to a combination of the beamforming channel matrix and the beamforming vector determined in the preceding step S300.
[0065] As just described, the 3
^{rd} aspect relates to the determination of the optimal beamforming vectors using beamforming channel matrix knowledge. The beamforming vectors for transmitter and receiver are determined based on the beamforming channel matrix
H. Any codebooks C may be used. The system may use equal-gain beamforming at the transmitter 102 and at the receiver 114. In the following, the codebook for the transmitting beamformer 108 is referred to as C
_{T}, and the codebook for the receiving beamformer 118 is referred to as C
_{R}.
[0066] The beamforming vectors may be determined in two steps. In the first step, a beamforming vector
w_{CH} is determined for the transmitter 102, which optimizes the expression
Hw according to the criterion of the L1 Norm ∥ ∥
_{1} (also named Taxi Cab Norm or Manhattan Norm):
[0067] For determining
w_{CH}, an optimization method may be used or a search across all vectors of the codebook C
_{T} may be performed.
[0068] In a second step, the beamforming vector
z for the receiver 114 is determined. The term
z^{T}Hw_{CH}=
z^{T}h_{w,CH} is maximized according to the criterion of the largest absolute value. This may again be performed by a search across all vectors of the codebook C
_{R}:
[0069] Alternatively,
may be determined and the vector having the maximum correlation with
z_{H} may be selected from the codebook:
[0070] The order in which the beamforming vectors are determined may be changed. In such a case, a reciprocal system is considered and the transmitting and receiving beamforming vectors are exchanged in the equations (exchanging
w and
z) and the transposed beamforming channel matrix (
H →
H^{T}) is used. In that manner, at first, a suitable receiving beamforming vector may be determined without considering the transmitting beamforming vector, and subsequently, a suitable transmitting beamforming vector considering the determined from the receiving beamforming vector.
[0071] In a bidirectional transmission system using beamforming in both directions of transmission, the method may be performed for both directions. In case the stations use the same antennas and beamformers for transmitting and receiving, determining the beamforming vectors for one direction of transmission may be sufficient. A beamforming vector may then be used both for transmitting and for receiving.
[0072] In a multi-carrier system having K subcarriers (K spectral components), basically, for every subcarrier k, a beamforming channel matrix
H^{(k)}) may be determined:
[0073] In a system using MIMO signal processing, individual beamforming vectors would be determined for every subcarrier and adjusted separately. This is not possible in a system with beamforming signal processing, since only one beamforming vector may be adjusted for all spectral components. Therefore, an easy-to-calculate solution is suggested. For determining the beamforming vectors the method is based only on the beamforming channel matrix having the highest modulus sum norm (sum of the absolute values of matrix entries):
with
[0074] For reducing the effort further, only a sub range of the K channel matrices may be considered in (23):
with
[0075] In that way, for example, only every second subcarrier may be considered. The further steps for determining the beamforming vectors starting from
H correspond to the above described method.
[0076] The third aspect is advantageous, since suitable beamforming vectors are determined based on the estimated beamforming channel matrix. The gain obtainable for a given system by beamforming depends, apart from the hardware, on the performance of the algorithms used. With a suitable algorithm, not only particularly suitable but optimal beamformer settings with respect to the codebooks may be determined. A time-consuming training phase and/or iterative optimization of the beamforming vectors by repeated transmission of training symbols are not required. Hence, the method may operate quickly and transmission resources are saved. It is of particular interest for large-scale codebooks, since the number of required estimation symbols does not depend on the scale of the codebook but merely on the number of the transmitting and receiving beamformer branches.
[0077] Beamforming vectors may also be determined for a multi-carrier system based on an individual beamforming channel matrix. This reduces the computational overhead. At the same time, the selection criterion ensures that the vectors are optimized for a suitable carrier. In that way, higher gains are possible by beamforming, compared to the use of a fixed carrier. After the selection of the channel matrix, any method may be used for determining the beamforming vectors that are based on the knowledge of the channel matrix.
4^{th} aspect: determining the beamforming vectors in a hybrid MIMO beamforming system
[0078] The fourth aspect relates to determining the beamforming vectors in a hybrid MIMO beamforming system. Fig. 5 shows an example of a hybrid MIMO beamforming system having two MIMO branches at the transmitter and at the receiver each having associated therewith two beamforming branches. To be more specific, Fig. 5 shows a unidirectional radio system comprising a transmitter 502 having an input 504, and a receiver 506 having an output 508. The transmitter 502 comprises a transmitter signal processing unit 510, beamformers 512
_{1} and 512
_{2} having inputs 514
_{1} and 514
_{2}, and antennas 516
_{1} to 516
_{4}. The antennas 516
_{1} and 516
_{2} are connected to the beamformer 512
_{1} and form a first antenna group. Together with the beamformer 512
_{1} the antennas 516
_{1} and 516
_{2} form a first MIMO branch 511
_{1} of the transmitter 502. The antennas 516
_{3} and 516
_{4} are connected to the beamformer 512
_{2} and form a second antenna group. Together with the beamformer 512
_{2} the antennas 516
_{3} and 516
_{4} form a second MIMO branch 511
_{2} of the transmitter 502. The beamformer 512
_{1} comprises a splitting circuit 520
_{1} and two weighting elements 522
_{1} and 522
_{2}. Also the beamformer 512
_{2} comprises a splitting circuit 520
_{2} and two weighting elements 522
_{3} and 522
_{4}.
[0079] The data signal
d_{s} fed to the input 504 of the transmitter 502 is processed by the transmitter signal processing unit 510. In the transmitter signal processing unit 510 also the MIMO signal processing of the transmit signal takes place. The output transmission signal
x^{(1)} of the first MIMO branch 511
_{1} is fed via the input 514
_{1} into the beamformer 512
_{1}, and is split using the splitting circuit 520
_{1}. The split signal is subsequently weighted using the weighting elements 522
_{1} and 522
_{2}, and is forwarded to the antennas 5161 and 5162. Likewise, the output transmission signal
x^{(2)} of the second MIMO branch 511
_{2} is fed via the input 5142 of the beamformer 512
_{2} into the beamformer 512
_{2}, and is split using the splitting circuit 5202. The split signal is subsequently weighted using the weighting elements 522
_{3} and 522
_{4}, and is forwarded to the antennas 516
_{3} and 516
_{4}.
[0080] The signal radiated by the antennas 516
_{1} to 5164 is transmitted via a radio channel and is received by a receiver 506. The receiver 506 comprises a receiver signal processing unit 526, beamformers 528
_{1} and 528
_{2} having outputs 530
_{1} and 530
_{2} and antennas 532
_{1} to 5324. The antennas 532
_{1} and 532
_{2} are connected to the beamformer 528
_{1} and form an antenna group. Together with the beamformer 528
_{1} the antennas 532
_{1} and 532
_{2} form a first MIMO branch 521
_{1} of the receiver 506. The antennas 532
_{3} and 532
_{4} are connected to the beamformer 528
_{2} and also form an antenna group. Together with the beamformer 528
_{2} the antennas 532
_{3} and 532
_{4} form a second MIMO branch 521
_{2} of the receiver 506. The beamformer 528
_{1} comprises an adding circuit 536
_{1} and two weighting elements 538
_{1} and 538
_{2}. Also the beamformer 5282 comprises an adding circuit 5362 and two weighting elements 538
_{3} and 538
_{4}.
[0081] The signals received via the antennas 532
_{1} and 5322 are fed to the beamformer 528
_{1}, are weighted by the weighting elements 5381 and 5382, and are added using the adding circuit 536
_{1}. At the output 530
_{1} of the beamformer 528
_{1} the signal
y_{1} of the first MIMO branch 521
_{1} is present, which is input into the receiver signal processing unit 526. Likewise, the signals received via the antennas 532
_{3} and 532
_{4} are fed to the beamformer 528
_{2}, are weighted by the weighting elements 5383 and 5384, and are added using the adding circuit 536
_{2}. At the output 530
_{2} of the beamformer 528
_{2} the signal
y_{2} of the second MIMO branch 521
_{2} is present, which is input into the receiver signal processing unit 526. In the receiver signal processing unit 526 the signals
y_{1} and
y_{2} are processed. In the receiver signal processing unit 526 also the MIMO signal processing of the receive signals occurs. At the output 508 of the receiver 506 the received data signal
d_{r} is present.
[0082] As just described, the 4th aspect of the invention relates to a multi-antenna radio system with MIMO signal processing. In every MIMO branch beamforming signal processing can be applied using a beamformer and an antenna group, which results in a hybrid MIMO beamforming configuration. The system has
P MIMO branches at the transmitter 502 and
Q MIMO branches at the receiver 506. Every MIMO transmitting branch
p, p = 1....
P, comprises
M_{p} beamforming transmitting branches. Every MIMO receiving branch
q, q = 1....
Q consists of
N_{q} beamforming receiving branches. Fig. 5 shows an example of a hybrid MIMO beamforming system 500 having the two MIMO branches 5111, 5112 at the transmitter 502 and the two MIMO branches 521
_{1}, 521
_{2} at the receiver 506, wherein each of the beamformers 512
_{1}, 512
_{2} and 528
_{1}, 528
_{2} has two branches.
[0083] Fig. 6 depicts a flow diagram representing the steps of a method in accordance with the fourth aspect. The method is implemented, for example, by a system as described with regard to Fig. 5 and comprises as a first step S600 the splitting of the hybrid MIMO beamforming system into a plurality of subsystems. In a subsequent step S602 for each subsystem the beamforming vectors are determined separately.
[0084] In accordance with the fourth aspect, a method for determining suitable beamforming vectors for hybrid MIMO beamforming systems having any number of MIMO branches at the transmitter and at the receiver is described. The basic idea is to suitably split the hybrid MIMO beamforming system into several subsystems. Then, for every subsystem, any known method for determining suitable beamforming parameters may be used. If the subsystems are beamforming systems or are considered as beamforming systems one or more of the above described methods in accordance with the 1
^{st} to 3
^{rd} aspects for determining suitable beamforming parameters may be used. Considering a subsystem with more than one MIMO branch at one side as beamforming system, means to assume for the optimization that the beamformers of all MIMO branches at this side form one large beamformer and that only beamforming signal processing can be applied.
[0085] For splitting an overall system, two approaches may be used.
[0086] In accordance with examples, the first approach comprises a fixed or static allocation, where the hybrid MIMO beamforming system is split into a plurality of beamforming subsystems. Every MIMO transmitting branch is allocated to one MIMO receiving branch, and every MIMO receiving branch is allocated to one MIMO transmitting branch. For optimizing the performance of the system, the following allocation rules are defined:
- 1. The allocation is performed "as evenly as possible", i.e. the number of MIMO receiving branches allocated to a MIMO transmitting branch or the number of MIMO transmitting branches allocated to a MIMO receiving branch is minimized.
- 2. The information on the spatial arrangement of the MIMO transmitting and receiving antennas are considered as follows: If several MIMO receiving branches are allocated to a MIMO transmitting branch (several MIMO transmitting branches to one MIMO receiving branch) those receiving branches are respectively allocated to a MIMO transmitting branch whose MIMO antennas are spatially as far as possible apart from one another (those transmitting branches are allocated to a MIMO receiving branch whose MIMO antennas are spatially as far as possible apart from one another). Dependent on the MIMO signal processing, it may be advantageous to use not the MIMO branches having the most distant antennas but those MIMO branches having their antennas as close as possible.
[0087] Figs. 7(a) and (b) show examples for the static allocation on the basis of the MIMO beamforming system of Fig. 5. As can be seen, the hybrid 2x2 MIMO beamforming system is separated into two beamforming subsystems, and in accordance with Fig. 7(a) a first beamforming subsystem 700 is formed of the MIMO branches 511
_{1} of the transmitter 502 and the MIMO branch 521
_{1} of the receiver 514. The second beamforming subsystem 702 comprises the second MIMO branch 511
_{2} of the transmitter 502 and the second MIMO branch 521
_{2} of the receiver 514. In Fig. 7(b) the first beamforming subsystem 704 comprises the first MIMO branch 511
_{1} of the transmitter 502 and the second MIMO branch 521
_{2} of the receiver 514. The second beamforming subsystem 706 comprises the second MIMO branch 511
_{2} of the transmitter 502 and the first MIMO branch 521
_{1} of the receiver 514.
[0088] By the allocation, the MIMO beamforming overall system is split into max(
P,
Q) beamforming subsystems. Fig. 7 illustrates the two options for splitting a hybrid MIMO beamforming system having two MIMO branches at the transmitter and at the receiver. For every beamforming subsystem, suitable beamforming vectors for the transmitter and the receiver are to be determined. For this purpose, the methods in accordance with the 1st to 3rd aspect described may be used.
[0089] In methods with training matrices (see the first aspect), the different beamforming subsystems are considered sequentially in any order. Thereby only the MIMO branches of the currently considered subsystem are active. Transmit branches that do not belong to the currently considered subsystem are turned off and receive branches that do not belong to the currently considered subsystem remain unconsidered. Only the training matrices of the currently considered subsystem are used.
[0090] In methods for determining beamforming vectors having channel knowledge (see the third aspect), only those beamforming channel submatrices describing each of the channels between the allocated MIMO branches have to be known. For the channel estimation, the subsystems may be considered sequentially in any order. Thereby, on the transmitter side, only the MIMO branch of the currently considered subsystem is active. Transmit branches that do not belong to the currently considered subsystem are turned off and receive branches that do not belong to the currently considered subsystem remain unconsidered.
[0091] In case there are more MIMO branches in the transmission system on the transmitter side than on the receiver side, or vice versa, for determining the beamforming vectors some MIMO branches are considered several times. For example, in case of three MIMO branches at the transmitter and two MIMO branches at the receiver, one MIMO branch of the receiver is considered twice. Since for each beamformer, finally, only one beamforming vector is used, two different approaches are suggested:
- 1. The beamforming vector determined while considering the MIMO branch for the first time is maintained. In case the same MIMO branch is considered again, this beamforming vector will only be used for optimizing the beamforming vector on the opposite side.
- 2. At the beginning no attention is paid to the fact whether MIMO branches are considered multiple times. Thus, for some MIMO branches a plurality of beamforming vectors are determined. From these vectors for each branch that beamforming vector is selected and maintained which provides the best performance in accordance with a predefined optimization criterion. Subsequently a new optimization of all those MIMO branches on the opposite side is done, which are associated with the currently considered MIMO branch and which in accordance with the predefined optimization criterion showed a worse performance. On the basis of the already determined beamforming vector only the respective beamforming vector on the opposite side is optimized.
[0092] Using the static allocation has little complexity, since the MIMO branches are firmly allocated to one another and the beamforming subsystems have, at the most, the dimension
[0093] In accordance with examples, the second approach comprises asymmetric splitting of the beamforming system. The basic idea of this approach is splitting the hybrid MIMO beamforming overall system into asymmetric subsystems, each having only one MIMO branch on the transmitter side or on the receiver side. The method may be divided into two steps:
Step 1: The hybrid MIMO beamforming overall system is split into P
M_{p} subsystems. For every subsystem, a suitable transmitting beamforming vector is determined. By considering the subsystem as beamforming system, the methods described above can be used.
Step 2: The hybrid MIMO beamforming overall system is split into Q
subsystems. For every subsystem, a suitable receiving beamforming vector is determined. By considering the subsystem as beamforming system, the methods described above may be used. (Method without considering the transmission beam, determining the beamforming vectors in reverse order).
[0094] After these steps, all transmitting and receiving beamforming vectors are determined. If the methods described above are used, step 1 and step 2 may be considered independently of one another and are exchangeable. Further, the subsystems may be considered in any order.
[0095] Fig. 8 shows an example of the asymmetric splitting of a hybrid 2x2 MIMO beamforming system as it is for example described in Fig. 5. In Fig. 8(a) the first subsystem 800 which is an asymmetric system, comprises the first MIMO branch 511
_{1} of the transmitter 502 and the first and second MIMO branches 521
_{1} and 521
_{2} of the receiver 514. The second asymmetric subsystem 802 comprises the second MIMO branch 511
_{2} of the transmitter 502 and the two MIMO branches 521
_{1} and 521
_{2} of the receiver 514. In Fig. 8(b) an asymmetric subsystem 804 comprises the MIMO branches 511
_{1} and 511
_{2} of the transmitter 502 and the first MIMO branch 521
_{1} of the receiver 514. The asymmetric, subsystem 806 comprises the two MIMO branches 511
_{1} and 511
_{2} of the transmitter 502, and the second MIMO branch 521
_{2} of the receiver 514.
[0096] Fig. 8 illustrates the two steps for splitting a hybrid MIMO beamforming system having two MIMO branches at the transmitter and receiver. In accordance with embodiments, for the specific cases that a hybrid SIMO (single-input multiple-output) or a MISO (multiple-input single-output) beamforming system is treated and the subsystems are considered as beamforming systems, the modifications described below will be considered, which increase the gain obtainable by beamforming.
[0097] For the special case of a SIMO beamforming system (
P=1,
Q>1) first, the
beamforming system is considered and a suitable transmitting beamforming vector is determined. Then, the SIMO beamforming overall system is split into
Q M x N_{q} subsystems. For every subsystem, a suitable receiving beamforming vector is determined. The difference to the basic method is that during determining the receiving beamforming vectors, the previously determined transmitting beamforming vector is considered.
[0098] For the special case of a MISO beamforming system (
P>1,
Q=1), first, the
beamforming system is considered and a suitable receiving beamforming vector is determined without considering transmitting beamforming vectors. Then, the MISO beamforming overall system is divided into
P M_{p} x
N subsystems. For every beamforming subsystem, a suitable transmitting beamforming vector is determined. The difference to the basic method is that during determining the transmitting beamforming vectors, the previously determined receiving beamforming vector is considered.
[0099] In accordance with embodiments splitting a hybrid MIMO beamforming system into several subsystems is advantageous as this allows the determination of suitable beamforming vectors in hybrid MIMO beamforming systems by applying known methods for determining the suitable beamforming vectors in beamforming systems.
[0100] Although some aspects have been described in the context of an apparatus, it is clear that these aspects also represent a description of the corresponding method, where a block or device corresponds to a method step or a feature of a method step. Analogously, aspects described in the context of a method step also represent a description of a corresponding block or item or feature of a corresponding apparatus.
[0101] Depending on certain implementation requirements, embodiments of the invention can be implemented in hardware or in software. The implementation can be performed using a digital storage medium, for example a floppy disk, a DVD, a CD, a ROM, a PROM, an EPROM, an EEPROM or a FLASH memory, having electronically readable control signals stored thereon, which cooperate (or are capable of cooperating) with a programmable computer system such that the respective method is performed. Some embodiments according to the invention comprise a data carrier having electronically readable control signals, which are capable of cooperating with a programmable computer system, such that one of the methods described herein is performed.
[0102] Generally, embodiments of the present invention may be implemented as a computer program product with a program code, the program code being operative for performing one of the methods when the computer program product runs on a computer. The program code may for example be stored on a machine readable carrier.
[0103] Other embodiments comprise the computer program for performing one of the methods described herein, stored on a machine readable carrier. In other words, an embodiment of the inventive method is, therefore, a computer program having a program code for performing one of the methods described herein, when the computer program runs on a computer. A further embodiment of the inventive methods is, therefore, a data carrier (or a digital storage medium, or a computer-readable medium) comprising, recorded thereon, the computer program for performing one of the methods described herein. A further embodiment of the inventive method is, therefore, a data stream or a sequence of signals representing the computer program for performing one of the methods described herein. The data stream or the sequence of signals may for example be configured to be transferred via a data communication connection, for example via the Internet.
[0104] A further embodiment comprises a processing means, for example a computer, or a programmable logic device, configured to or adapted to perform one of the methods described herein. A further embodiment comprises a computer having installed thereon the computer program for performing one of the methods described herein. In some embodiments, a programmable logic device (for example a field programmable gate array) may be used to perform some or all of the functionalities of the methods described herein. In some embodiments, a field programmable gate array may cooperate with a microprocessor in order to perform one of the methods described herein. Generally, the methods are preferably performed by any hardware apparatus.
[0105] The above described embodiments are merely illustrative for the principles of the present invention. It is understood that modifications and variations of the arrangements and the details described herein will be apparent to others skilled in the art. It is the intent, therefore, to be limited only by the scope of the impending patent claims and not by the specific details presented by way of description and explanation of the embodiments herein.