Method and apparatus for performing spatial-division multiple access

[0001] The present invention relates generally to multi-access spatial division, and in particular to a method and apparatus for performing multi-access spatial division in a multi-node communication system.

Background of the Invention [0002] Broadcast beamforming (sometimes called transmission adaptive transmit matrix (TXAA)) increases the effective signal-to-noise seen by receiver devices by creating a range pattern that is usually directional (i.e. not uniformly broadcast). This is achieved by using multiple antennas on the transmitting side and loading each antenna so that the combined transmissions result in a shaped beam pattern providing maximum power / energy to the receiver. When broadcasting multiple streams to multiple receivers (i.e., multi-access spatial division or SDMA ( spatial division multiple access)) it would be beneficial to calculate antenna load so as to achieve minimum crosstalk or interference between users. In other words, in addition to choosing an array of antennas (i.e. loads for each antenna) that increases the power for the desired receiver, it would also be beneficial to minimize power for unwanted receivers. Achieving this, however, requires full channel knowledge of the transmitter, which is difficult to obtain in practical communication due to the large amount of feedback required to get full knowledge of the channel. Thus, there is a need for a method and apparatus for performing multi-spatial division access in a multi-node communication system that minimizes crosstalk or interference for unwanted receivers using a practical feedback method.

[0003] US 2005/286663 A1 discloses a system in which stations in a multiple-input-multiple-output (MIMO) network are searching in a code book of code words to determine which code word is closest to the desired precoding matrix on the Grassmann collector. The index or indexes corresponding to the code word are transmitted from the receiver to the transmitter to identify the code word to be used for the transmitting beam.

Brief description of the figures [0004]

FIG. 1 is a block diagram of a communication system using multi-access with spatial division.

FIG. 2 is a block diagram of a node used in the communication system of FIG. 1.

FIG. 3 is a flowchart illustrating the operation of the node of FIG. 2 during transmission.

FIG. 4 is the block diagram of the node.

FIG. 5 is a flowchart illustrating the operation of the node of FIG. 4 when calculating channel subspace and zero subspace.

PZ / 3044 / AG EP 1 985 125 B1

Detailed description of the figures [0005] To meet the above-mentioned need, a method and apparatus for performing multi-access spatial division in a communication system is provided here. During operation, the nodes will return information about zero subspace together with information about their subspace. The zero subspace is similar to the channel subspace except that the information about the zero subspace directs the base station to the matrix in the previously specified code book, which results in the node receiving minimum power. The zero subspace for each node is considered when communicating with a specific node.

The present invention includes a method comprising the steps of transmitting pilot data to the first and second nodes from multiple antennas, receiving first and second channel subspace information from the first and second nodes in response to broadcasting, and receiving first and second zero subspace information from the node in response to broadcasting. The first channel subspace information is used along with the second zero subspace information to determine the antenna load for the first node. In addition, the subspace information of the second channel is used along with the information about the first zero subspace to determine the load on the antennas for the second node. Finally, data is transmitted to the first and second nodes simultaneously, using the first and second antenna loads, respectively.

[0007] The present invention further includes a method comprising the steps of receiving pilot data from a base station that is transmitted through a plurality of antennas, determining channel subspace information based on the received pilot data, determining zero subspace information based on the received pilot data, and transmitting channel subspace information and zero subspace information for use by the base station for determining the antenna load.

[0008] The present invention further includes a device comprising a transmitter transmitting pilot data, a receiver receiving first and second channel subspace information from the first and second nodes, and first and second zero subspace information from the first and second nodes, and a stream load calculation circuit using first channel subspace information and second zero subspace information to determine the first antenna / stream load for the first node, a stream load calculation circuit using additionally the second channel subspace information and the first zero subspace information to determine the second antenna load / stream for the second node.

[0009] The present invention further includes an apparatus comprising a receiver receiving pilot data from a base station, which are transmitted through a plurality of antennas, a logic circuit determining channel subspace information based on the received pilot data and determining zero subspace information based on the received pilot data, and the transmitting circuit, transmitting channel subspace information and zero subspace information for use by the base station for determining the antenna load.

[0010] Before describing the method and apparatus for performing multi-access spatial division, the following definitions were provided to provide the context needed to utilize the preferred embodiment of the present invention.

• Vector space - a set consisting of all linear combinations of a specific set of vectors.

• Subspace - a vector space that is part of a larger vector space.

• Zero Subspace - a subspace for a specific receiving node, the use of which by the base station to calculate transmission loads will result in zero power / energy received by the node. Mathematically, it is the "zero space" of the channel response matrix, where the term "zero space" is well defined in matrix theory. The channel response matrix includes the channel response between each transmitting antenna and each receiving antenna.

• Channel subspace - a subspace for a specific receiving node, the use of which by the base station to calculate transmission loads will result in the maximum power / energy received by the node. Mathematically, it is the "range" of the channel response matrix, where the term "range" is well defined in matrix theory. The channel response matrix includes the channel response between each transmitting antenna and each receiving antenna.

[0011] Referring now to the figures in which the numbers designate similar elements, Fig. 1 is a block diagram of the communication system 100. In a preferred embodiment of the present invention, the communication system 100 uses the Orthogonal Frequency Division Multiplexed architecture - OFDM) or based on multiple carriers. In alternative embodiments of the present invention, this architecture may include the use of spreading techniques such as multi-carrier CDMA (MC-CDMA), multi-carrier direct sequence (MC-) CDMA DS-CDMA), Rectangular Frequency and Code Division Multiplexing Orthogonal Frequency and Code Division Multiplexing (OFCDM) with dispersion in one or two dimensions, or it can be based on simpler techniques of multiple access / multiplexing with time and / or frequency division, or a combination of these different techniques.

[0012] As shown, the communication system 100 comprises a base station 101 and a plurality of mobile or stationary nodes 102-103. Although only one base station and two nodes are shown, one of ordinary skill in the art will recognize that such a communication system typically includes multiple base stations 101 in communication with multiple nodes 102-103. During operation, the communication system 100 uses SDMA to shape the beam between base station 101 and nodes 102-103. As described above, beamforming is achieved by using multiple antennas on the transmitting side and loading each antenna so that the combined transmissions result in a shaped beam pattern with maximum power received by receivers with minimal crosstalk or interference between users. Crosstalk minimization is very beneficial for SDMA because SDMA involves broadcasting to two or more nodes simultaneously on the same channel resources (e.g. time frame, subcarrier, subchannel, etc.).

[0013] To determine the appropriate antenna load for the user, in a preferred embodiment of the present invention, the base station 101 transmits piloting from each of its antennas that are received by nodes 102-103. Nodes 102-103 perform channel estimation based on received pilotage and determine the channel subspace that will result in the maximum received power by individual

PZ / 3044 / AG mobile devices when the base uses channel subspace to calculate transmission loads. In a preferred embodiment of the present invention, the channel subspace includes a code book index that directs the base station to the corresponding antenna matrix in the predetermined code book divided by the base station and the nodes. However, in alternative embodiments of the present invention, the channel subspace may comprise the actual channel matrix itself. When the node determines the appropriate channel subspace, the channel subspace information is transmitted to the base station 101, where it is used (along with other information) to determine the appropriate antenna load for specific nodes.

[0014] As described above, it would also be beneficial for the base station 101 to calculate the antenna load so as to achieve the minimum crosstalk and interference between users. In other words, in addition to choosing an antenna matrix (i.e., loads for each antenna) that maximizes power for the desired receiver, it would also be beneficial to minimize power for unwanted receivers. Thus, for example, transmitting to node 102, it would be beneficial if, along with power maximization for node 102, antenna loads for transmission to node 102 further minimized the power received by node 103, especially in the case of SDMA transmission.

[0015] To address this problem, in a preferred embodiment of the present invention, many nodes will return zero subspace information along with their channel subspace information. The zero subspace is similar to the channel subspace except that the zero subspace information directs the base station 101 to the matrix in the predetermined code book, which results in the minimum power received by the node. The zero subspace for each node is considered when communicating with a specific node. Thus, for example, if both nodes 102 and 103 transmitted their channel subspace and zero subspace information to base station 101, base station 101 will calculate the load on antennas used to communicate with mobile device 102 based on both channel subspace returned by mobile device 102, as well as the zero subspace returned by the mobile device 103. This will have the effect of maximizing the power to the mobile device 102 while minimizing the power to the mobile device 103. More specifically, the zero subspace of the mobile device 103 is used to create a projected matrix that will project the matrix (or vector) onto the zero subspace associated with the mobile device 103 . The tared matrix is ​​applied to the channel subspace of the mobile device 102 to create the projected channel subspace for the mobile device 102. Then, SDMA transmit loads (i.e. antenna loads) for the mobile device 102 are calculated using this projected channel subspace. The power for the mobile device 103 is minimized by applying a projection matrix to the channel subspace for 102. Power for device 102 is maximized by designing loads that are singular vectors associated with the largest singular values ​​of the projected channel subspace for mobile device 102. A detailed explanation of how this is achieved is described below.

[0016] The Spatial Division Multiple Access SDMA strategy involves designing transmission filters (or antenna loads) for each user.

When considering the system of K users (nodes), Mt transmitting antennas in the base, Mr receiving antennas for

For each mobile device (user or node) and Ms data streams transmitted to each mobile device, the system can be described by K equations as:

κ y_fc = H ^ⁱⁿ<^xand + ⁿfc ^fc = !.....^κ (1) fc = 1 where H_k is the matrix of the M channel_rM *_t for user k, W_and is an SDMA sending filter (i.e. antenna load) of the M matrix_tM *_s for each mobile device i (designed to minimize crosstalk to other mobile devices by maximizing the signal strength to the mobile device i), xi is a symbolic vector M_s* 1 for a mobile device i (e.g. QPSK symbol vector), and n_k is the connection noise for the k-th user. The goal is to design W_and (i = 1, ..., K) so that the average BER (plotted bit error rate) K of users is minimized.

[0017] It should be noted that the assumption that each mobile device has Mr receiving antennas is illustrative only, and that each mobile device may have a different number of receiving antennas.

SDMA Based on Medium Subspace with delivered zero subspace and channel subspace [0018] The following text assumes that the base station has full knowledge of the channel (i.e. the base has perfect channel state information (CSI)) to all mobile devices and designs SDMA broadcast loads. The main purpose of the SDMA broadcast loads described here is to introduce the concept of subspace averaging that can be used by the base station to calculate SDMA broadcast loads when the base station has a channel subspace to mobile device 102 and a zero subspace to mobile device 103 (instead of full knowledge of the channel to both). The high-level idea of ​​subspace averaging is to improve the SDMA broadcasting load by allowing a certain level of crosstalk (i.e., interference from transmission intended for mobile device 102 received by mobile device 103), which is irrelevant due to the noise power of mobile device 103. By allowing these crosstalks, the receiving power of mobile device 102 may be increased compared to SDMA broadcast loads that do not allow crosstalk.

[0019] The following introduces a method of calculating the antenna load based on the subspace averaging. Consider K users for SDMA with channels Hi, ..., H_K and suppose that Ms data streams are broadcast to each user. The goal is to determine the corresponding SDMA W1,., WK loads. Assuming that the database has information about the channel space (e.g. zero subspace and channel subspace obtained by the sound channel or channel feedback) and Mt> KMs, the calculation of Wk is obtained in the following stages:

Medium Subspace SDMA [0020]

1) Let's define zero {H -, ..., H_K}: Suppose M_t> KM_s and let Hi, ..., H_K will be K channel matrices for each mobile device, each with size M * M- with complex elements. Let's define the KM matrix_rM *_t J = [Hf | Hf ... | H "]^AND . (Mt-KMs) right singular vectors J corresponding to

Peculiar zeros are set in columns to create an orthonormal matrix

M_t* (M_rKMs) defined as ze / O {Hi, ..., H_K}.

2) Let's define V = av (V₁V₂): Suppose that V₁, V₂ are M orthonormal matrices_tM *_s and define the "subspace distance" between the subspaces spanned on the Vi and V orthonormal matrices₂ as dist (V₁V₂) = || v₂vf - where || * ||₂ is a 2-matrix norm as well defined in matrix theory. The orthonormal matrix V = śr (V1, V2) specifies that the subspace spanned on the orthonormal matrix V has an equal distance from the subspaces spanned on V1 and V1. Matrix V can be calculated as follows. Own vectors (V₂V ^ - V, V,^H) dominating in M_s, set in columns of the matrix M_tM *_s form Wed (V₁V₂). It should be noted that this average subspace can be extended to the average of any number of subspaces represented by its bases by V1, V2 ,. and can also be extended to a weighted average.

3) For the user k, calculate Nk = ze / o (H1,., Hk-1, Hk + 1,., HK) and the rectangular projected matrix Pfc = N, N ".

4) Let for the user k V_k will be an orthonormal matrix M_tM *_s formed by right singular vectors Hk dominating in Ms.

5) Let's calculate Wk = śr (Vk, PkVk).

[0021] Note that Wk = Vk implies that interference is bypassed, Wk = PkVk implies interference reduction or zeroing (i.e., zero crosstalk for mobile device 103 if user k is mobile device 102). When using subspace averaging for mobile device 103, a certain level of crosstalk is present, however, the power supplied to mobile device 102 is increased. As long as the crosstalk level for the mobile device 103 is below the noise power, the performance of the mobile device 103 remains unchanged.

SDMA Based on Medium Subspace with Quantized CSI (i.e., Code Book Based CSI) [0022] Now that the concept of subspace averaging has been introduced, it will be used to design the SDMA broadcast station baseload loads when the base station only knows channel subspace and subspace zero for any mobile device. Suppose there are K mobile devices (each with M_r receiving antennas) with downlink channels H₁,., H_K M + M. which are measured by the mobile device based on pilot signals sent from each of the Mt base station transmitting antennas. The idea is to plan a quantization strategy based on a code book, which enables a practical way of sending the zero and channel subspaces through a mobile device to the base station. For each user, the feedback consists of the quantized zero subspace Q (Nk) and the quantized subspace Q (Vk) determined by the following algorithm. Note that quantization can be done by selecting an item from the B-matrix codebook (the matrix dimension is given below). Code books can be designed as known in the art, DJ Love, RW Heath, Jr., and T. Strohmer,

PZ / 3044 / AG EP 1 985 125 B1 "Grassmannian Beamforming for Multiple-Input Multiple-Output Wireless Systems," IEEE Transactions on Information Theory, October 2003.

Code Book Quantization for K Mobile Users supporting Narrow Band SDMA [0023] Let's define the channel subspace quantization for each mobile device: Q (V₁) = ^arg"^ || HiV || 2, through Q (VK) =^arg vecl | H £ V || 2, where C is the code book M_s dimensional subspace C^mt .

Let's define the zero subspace quantization for each mobile device Q (N_X) = ^argvec "l^Hand^vl², 0(ⁿk) = ^argVec "l^H^^vl², C "is a code book (Mt-Mr) of dimensional subspace C^mt . [0024] The mobile device u communicates its quantized channel subspace to the base station via a code book matrix index feedback that maximizes || H * V || 2, where VeC. A mobile device u transmits its quantized zero subspace to a base station via a codebook matrix index feedback that minimizes || H * V ||₂where V e C '.

[0025] Codecodes C and C 'can be designed by uniformly spacing points on a Grassmann collector using any algorithm available in the art. The SDMA transmit load can be calculated by the base station in the following steps.

Calculation of Medium Subspace SDMA Loads with Quantization [0026] In this algorithm, it was assumed that the base station would simultaneously transmit to K mobile devices. The following stages are used to calculate the SDMA broadcast load:

1) Let's calculate the matrix of the rectangular projection for the mobile device u as:

l '"= iT and U (N,) U (N,)".

2) Let's calculate the SDMA load for the mobile device u as Wu = śr (Q (Vu), PuQ (Vu)).

[0027] The base station will use transmit loads (i.e. antenna loads), Wu, to shape the data beam to the mobile device and these loads will maximize power to one mobile device, providing low interference to other K-1 mobile devices.

SDMA for MIMO-OFDM with perfect Channel State Information

- CSI) [0028] The next step is to apply the above algorithms developed for MIMO with flat decay (e.g. OFDM) to broadband MIMO systems. In the following, the MIMO-OFDM system is considered with the entire band assigned to all users and the average BER is selected as the performance metric. With total channel knowledge (i.e. perfect CSI) for the transmitter, SDMA loads can be calculated independently for each subcarrier using a medium subspace solution.

SDMA for MIMO-OFDM with quantized CSI [0029] As in the case of flat decay channels, we assume that there are K users with channels represented as H₁(f) by H_K(f), f = 1, ..., N, where f is the subcarrier index and N is the integer

PZ / 3044 / AG EP 1 985 125 B1 subcarriers. Then information about zero subspace and channel subspace for each subcarrier can be quantized, specifically Q (N₁(f)), Q (Vi (f>), ..., Q (N_K(f)), Q (V_K(f)), and the SDMA-based medium subspace solution listed above can be applied independently to each subcarrier. In this case, the feedback load is increased N-fold compared to the flat decay MIMO channel. However, since there is usually a correlation between channels on neighboring carriers, the amount of feedback information can be reduced as follows.

[0030] Due to the fact that the number of multi-path decomposable elements is usually smaller than the number of subcarriers, adjacent subcarrier channels are correlated, which can be used to limit feedback information. Adjacent subcarriers can be grouped to form a cluster sharing a common SDMA load and provide common feedback for channel subspace and zero subspace. Considering an exemplary cluster of F subcarriers (F <N) and marking subcarriers in a cluster by 1, ..., F, common feedback information can be determined as follows.

Coding Book Quantization for K Mobile Users for OFDM SDMA based on grinding [0031]

1) Define the "grape channel subspace" for mobile device u as: Q (V ") = ^arg ^^ ™ Σ / · ιΗΗ = * (/) ν |₂, where C is the M code book_s-dimensional subspaces in C^mt (i.e. the space formed by Mt-dimensional complex-value vectors).

2) Define the "grape zero subspace" for mobile device u as: Q (N ") = ^argv ™ X / III =^H£ (y) V || 2, where C'is a code book (M_t-M_r) -dimensional subspaces in C^mt.

[0032] Code books can be assigned as in the case of flat decay presented above. Note that in the above definitions, the subcarriers in the cluster are from 1 to F, but in general the cluster can be any group of F subcarriers from N subcarriers carrying data.

[0033] The mobile device u communicates a quantized cluster of channel subspaces to a base station via a code book matrix index feedback that maximizes Σ / = ιΗΗ * (/) ν |₂where VeC. The mobile device u communicates a quantized zero subspace cluster to the base station via a codebook matrix index feedback that minimizes Σ / = ιΗΗ * (/) ν |₂where VeC '. Each mobile device will return a quantized cluster of channel subspaces and zero subspace for each cluster in N subcarriers carrying data for which the base station will request feedback.

[0034] Special grape cases include a cluster size of one (i.e. F = 1), which means that each subcarrier is treated independently or a cluster size of F = N when only one SDMA load is used for all subcarriers. Let us remind you that in the case of Ms = Mr, either channel subspace or zero subspace information is sufficient, because one of the two subspaces can be obtained from the other.

[0035] Fig. 2 is a block diagram of a base station 101. The base station 101 includes a circuit that calculates the load on stream 201, the inverse Fast Fourier Transform (inverse) circuit

Fast Fourier Transform - IFFT) 203, optional cyclic prefix circuit 207 and transmitter 209. During operation, data stream s (f), f = 1.2, ... N enters the circuit that calculates the load on stream 201 (where N is the number of subcarriers) . Note that the load calculation operation performed by the stream load calculation circuit will be repeated for each of the Ms data streams for each of the K nodes (mobile devices). The load stream calculation circuit 201 outputs a plurality of loaded data streams, in particular one loaded data stream per antenna. Each loaded data stream (alternatively referred to as "antenna stream") is appropriately loaded in the frequency domain by the antenna specific load vn (f), where n = 1, 2,. Mt, where Mt is the number of antennas 211. It should be noted that the loads may or may not be different for each shaped subcarrier. Assuming that vm (f) is the load for the m subcarrier antenna f, then the circuit calculating the load on stream 201 gives the output data stream / antenna xm (f) = vm (f) s (f) for the m antenna. In the case where no data beam shaping is to be carried out on the subcarriers, the data streams / antenna s (f) for these carriers are directly fed to the fth subcarrier as input to IFFT. In other words, on these subcarriers vm (f) are effectively set to one.

[0036] The IFFT circuit 203 performs an inverse Fast Fourier Transform on each loaded data stream, converting the frequency domain data stream into a time domain data stream. The optional operation of the cyclic extension is then performed through the circuit of cyclic extension 207 on the cyclically shifted antenna streams. In particular, a cyclic prefix or protective space is added. The cyclic prefix is ​​usually longer than the expected maximum channel spread delay. As the average specialist will recognize, a cyclic extension may contain a prefix (prefix), postfix (suffix) or a combination of prefix and postfix. The cyclical extension is an integral part of the OFDM communication system. The cyclic prefix introduced means that the usual convolution of the multi-channel transmit signal appears as a cyclic convolution when the channel impulse response is in the range of 0 to LCP, where LCP is the length of the cyclic extension. Finally, properly loaded and cyclically shifted antenna data streams are OFDM modulated and transmitted by transmitters 209 from antennas 211. [0037] It is obvious that the base station 101 further includes a receiver 213 for receiving channel subspace and zero subspace information transmitted by the nodes. For example, the node 102 will transmit subspace information 105 and the node 103 will transmit subspace information 106 that will be received by receiver 213. This information will be forwarded to the stream load calculation circuit 201 to determine the appropriate stream / antenna load for use in communication with a particular node.

[0038] Fig. 3 is a flow diagram illustrating the operation of the base station of Fig. 2.

Logical flow begins at step 301, where pilot data is transmitted to at least the first and second nodes. Transmission of pilot data includes the usual transmission of a known sequence from each of the plurality of antennas 211. The known sequence could be different for each antenna and / or could be transmitted on different channel resources from each antenna. In response, receiver 213

PZ / 3044 / AG receives the first and second channel subspace information from the first and second node in step 303. As described above, the channel subspace information includes information on the basis of which the antenna load will be calculated that will maximize the power for mobile device. In a preferred embodiment of the present invention, this information includes an index to code book 215 divided by base station 101 and mobile devices 102-103.

[0039] Continuing, at step 305, information about the first and second zero subspaces is received from the first and second nodes by the receiver 213. As described above, the information about the zero subspace is received in response to the transmitted pilot data and includes information on the basis of which calculated antenna load that will minimize power for the mobile device. In a preferred embodiment of the present invention, this information includes an index to code book 215 divided by base station 101 and mobile devices 102-103.

[0040] In step 307, the first channel subspace information and the second zero subspace information are used by the stream load calculation circuit 201 to determine the first stream / antenna load for the first node, and in step 309 the second channel subspace information and the first zero subspace information are used by the stream load calculation circuit 201 to determine the load of the second stream / antenna for the second node. As described above, the first stream load will maximize the power for the first node, minimizing (or limiting) the power for the second node, while the second stream load will maximize the power for the second node, minimizing (or limiting) the power for the first node. Finally, in step 311, information or data is transmitted to the first and second nodes using loads of the first and second streams, respectively. Transmission to the first and second nodes using the first and second stream loads can take place simultaneously (e.g. in the form of SDMA).

[0041] Fig. 4 is a block diagram of a node 400. As shown, the node 400 includes logic circuit 401, receiver 403, transmitter 405 and code book 215. Logic circuit 401 preferably includes a microprocessor / controller and is used to determine the best code book indexes from the book code 215, which represent the node's subspace and the node's zero subspace based on pilot data received from the base station (pilot data is received by logic circuit 401 from receiver 403). The logic circuit provides codebook indexes to the 405 transmitter to provide index feedback for the base station.

[0042] Fig. 5 is a flow diagram of the operation of the node 400 during the calculation of the channel subspace and the zero subspace. The logic circuit starts at step 501, where the pilot signal (pilot data) is received by the receiver 403. As known in the art, the pilot signal includes a known sequence transmitted at a known time by each of the Mt transmit antennas in the base. Receiver 403 calculates the channel estimate and forwards the channel estimate to logic circuit 401 (step 503). When channel estimation is known, logic circuit 401 determines subspaces that will result in maximum power and minimum power (step 505). Both are determined based on channel estimates. Logic circuit 410 has access to code book 215 and determines the appropriate indexes for channel loads (step 507). At step 509, the channel subspace information and the zero subspace information are transmitted by the transmitter 405 as entries in the code book.

[0043] The invention has been detailed and described with reference to a specific embodiment. For example, only the 2nd matrix standard (|| A || 2) was used. However, other matrix norms, such as the Frobenius matrix norm, could be used. It is understood by those skilled in the art that various changes in form and detail will be made herein without departing from the scope of the invention, which is defined in the following claims.

VP / 3044 / AG

EP 1 985 125 B1