JP5275280B2 - Codebook generation method - Google Patents

Description

  The present invention relates to a code book generation method, and more particularly to a code book generation method in a next generation mobile communication system.

  In a UMTS (Universal Mobile Telecommunications System) network, HSDPA (High Speed Downlink Packet Access) and HSUPA (High Speed Uplink Packet Access) are adopted for the purpose of improving frequency utilization efficiency and data rate. A system based on CDMA (Wideband Code Division Multiple Access) is maximally extracted. For this UMTS network, Long Term Evolution (LTE) is being studied for the purpose of further high data rate and low delay.

  The third generation system can realize a transmission rate of about 2 Mbps at the maximum on the downlink using a fixed band of 5 MHz in general. On the other hand, in the LTE system, a maximum transmission rate of about 300 Mbps on the downlink and about 75 Mbps on the uplink can be realized using a variable band of 1.4 MHz to 20 MHz. In addition, in the UMTS network, a successor system of LTE is also being studied for the purpose of further increasing the bandwidth and speed (for example, LTE Advanced (LTE-A)). For example, in LTE-A, it is planned to extend the maximum system band of LTE specifications, 20 MHz, to about 100 MHz.

  Also, in an LTE system (LTE system), a MIMO (Multi Input Multi Output) system has been proposed as a wireless communication technique for transmitting and receiving data with a plurality of antennas and improving the data rate (frequency utilization efficiency) ( For example, refer nonpatent literature 1). In a MIMO system, a plurality of transmission / reception antennas are prepared in a transmitter / receiver, and different transmission information sequences are transmitted simultaneously from different transmission antennas. On the other hand, on the receiver side, the data rate (frequency utilization efficiency) is increased by separating and detecting simultaneously transmitted information sequences using the fact that different fading fluctuations occur between transmission / reception antennas. Is possible.

  In the LTE system, transmission information sequences transmitted simultaneously from different transmission antennas are all transmitted from the same user, such as single user MIMO (SU-MIMO (Single User MIMO)) transmission, and multi-user MIMO is transmitted from different users. (MU-MIMO (Multiple User MIMO)) transmission is specified. In these SU-MIMO transmission and MU-MIMO transmission, a phase / amplitude control amount (precoding matrix (precoding weight)) to be set in the antenna of the transmitter on the receiver side is associated with this precoding matrix. An optimum PMI is selected from a code book in which a plurality of PMIs (Precoding Matrix Indicators) are determined, and is fed back to the transmitter. On the transmitter side, based on the PMI fed back from the receiver, precoding is performed for each transmission antenna and a transmission information sequence is transmitted.

3GPP TR 25.913 “Requirements for Evolved UTRA and Evolved UTRAN”

  In general, in a codebook suitable for SU-MIMO transmission, a precoding matrix that enables effective signal separation in a mobile station apparatus UE as a receiver when a correlation between antennas of a base station apparatus eNodeB as a transmitter is low. Is preferably determined. On the other hand, in a codebook suitable for MU-MIMO transmission, it is preferable to define a precoding matrix that enables effective signal separation by the mobile station apparatus UE when the correlation between the antennas of the base station apparatus eNodeB is high. For this reason, when data transmission is performed by switching between SU-MIMO transmission and MU-MIMO transmission, the mobile station apparatus UE is effective in both cases where the correlation between the antennas of the base station apparatus eNodeB is high and low. Therefore, it is necessary to determine a precoding matrix that can be used for signal separation.

  The present invention has been made in view of such circumstances, and a code book that defines a precoding matrix that enables effective signal separation in a mobile station apparatus regardless of the level of correlation between antennas in the base station apparatus is provided. It is an object of the present invention to provide a codebook generation method that can be generated.

The code book generation method of the present invention includes a plurality of first code books having a DFT (Discrete Fourier Transform) structure that defines a precoding matrix selected when a correlation between antennas in a base station apparatus is high, and the base A codebook generation method comprising a plurality of second codebooks defining a precoding matrix selected when the correlation between antennas in a station apparatus is low, wherein the first and second codebooks include Calculating the minimum value of the inter-matrix distance (Chordal distance) of the included precoding matrix for each rank, and calculating the minimum value of the inter-matrix distance calculated in advance in the first codebook or in advance. calculating a minimum value of the matrix distance between precoding matrix included in the prior SL second codebook is updated according to each rank, further Repeating the comparison of the minimum values of the inter-matrix distances before and after, and selecting the second code book that maximizes the minimum value of the inter-matrix distances of the precoding matrix included in the entire code book, When calculating the minimum value of the inter-matrix distance, two different precoding matrices included in the first codebook are excluded from the calculation target.

  According to this method, since two different precoding matrices included in the first codebook are excluded from the calculation target when calculating the minimum value of the inter-matrix distance, a part of the first codebook is updated. Therefore, the gain inherently enjoyed by the first codebook having the DFT structure can be obtained, and effective in the mobile station apparatus when the correlation between the antennas in the base station apparatus is high. A precoding matrix capable of signal separation can be selected. On the other hand, since the second codebook that maximizes the minimum value of the inter-matrix distance is selected except for the arithmetic processing between two different precoding matrices included in the first codebook, the characteristics are not similar to each other. Since the precoding matrix can be appropriately selected, when the correlation between antennas in the base station apparatus is low, it is possible to select a precoding matrix that can be effectively separated by the mobile station apparatus. As a result, it is possible to generate a codebook that defines a precoding matrix that can be effectively signal-separated by the mobile station apparatus, regardless of the correlation between the antennas of the base station apparatus.

  According to the present invention, when the minimum value of the inter-matrix distance is calculated, two different precoding matrices included in the first codebook are excluded from the calculation target, so that a part of the first codebook is updated. Therefore, the gain inherently enjoyed by the first codebook having the DFT structure can be obtained, and effective in the mobile station apparatus when the correlation between the antennas in the base station apparatus is high. A precoding matrix capable of signal separation can be selected. On the other hand, since the second codebook that maximizes the minimum value of the inter-matrix distance is selected except for the arithmetic processing between two different precoding matrices included in the first codebook, the characteristics are not similar to each other. Since the precoding matrix can be appropriately selected, when the correlation between antennas in the base station apparatus is low, it is possible to select a precoding matrix that can be effectively separated by the mobile station apparatus. As a result, it is possible to generate a codebook that defines a precoding matrix that can be effectively signal-separated by the mobile station apparatus, regardless of the correlation between the antennas of the base station apparatus.

It is a conceptual diagram for demonstrating the structure of the code book in a LTE system. It is a figure for demonstrating the signal point on the complex plane corresponding to the precoding matrix defined in a DFT codebook. It is a figure which shows an example of the two random code books from which the maximum value in the minimum value of the distance between matrixes changes with ranks.

First, the configuration of the code book in the LTE system will be described. FIG. 1 is a conceptual diagram for explaining the configuration of a code book in the LTE system. FIG. 1 shows a case where the number of PMIs is 16 and the number of ranks (streams) is 8. In FIG. 1, “n” indicates a PMI index, and “dn _{, 1} ” to “dn _{, 8} ” indicate ranks. In the codebook, precoding matrices corresponding to the respective ranks of the respective PMIs are defined (in FIG. 1, the notation of the precoding matrix is omitted for convenience of explanation). In the following, for convenience of explanation, not only the entire code book shown in FIG. 1 but also precoding matrices (that is, eight precoding matrices) associated with each PMI are referred to as codebooks. To do.

  As shown in FIG. 1, codebooks in the LTE system can be divided into two groups from the viewpoint of correlation between antennas in a base station apparatus eNodeB that is a transmitter. That is, in the code book in the LTE system, the correlation between the first code book group that defines the precoding matrix selected when the correlation between the antennas in the base station apparatus eNodeB is high, and the correlation between the antennas in the base station apparatus eNodeB. And a second codebook group that defines a precoding matrix to be selected when it is low.

The codebooks belonging to the first codebook group described above are composed of codebooks having a DFT (discrete Fourier transform) structure (hereinafter referred to as “DFT codebooks”). In this DFT codebook, for example, when a codebook group having a PMI and a rank number of 8 × 8 is taken as an example, the DFT codebook has matrix components shown in (Equation 1).
(Formula 1)

For example, looking at the matrix component of the second column shown in (Equation 1), the first to eighth rows show (1, ω, ω ² , ω ³ , ω ⁴ , ω ⁵ , ω ⁶ , ω ^7. ) Value is defined. When these values are expressed on the complex plane, as shown in FIG. 2, signal points having phases shifted at equal intervals are formed. This means that an equal gain can be obtained in any direction in which a signal arrives when the correlation between antennas in the base station apparatus eNodeB is high. This DFT codebook has the advantage that each matrix component (precoding matrix) can be uniquely determined when the number of DFT codebooks to be defined (for example, eight in the case of FIG. 1) is specified. Have.

  On the other hand, codebooks belonging to the second codebook group do not have a DFT structure, and are configured with codebooks that are determined at random (hereinafter referred to as “random codebooks”). This random code book is calculated based on the calculation process of the inter-matrix distance (Chordal distance) described later in relation to the code book including the DFT code book described above. In the calculation processing of the inter-matrix distance, the calculation is performed so that a precoding matrix that can be effectively separated by the mobile station apparatus UE that is a receiver when the correlation between antennas in the base station apparatus eNodeB is low is determined. Is called.

  Here, the criteria for evaluating the precoding matrix defined in the codebook will be described. As a criterion for evaluating the precoding matrix, a different criterion is adopted depending on the level of correlation between antennas in the base station apparatus eNodeB. An array response is used when the correlation between antennas in the base station apparatus eNodeB is high. On the other hand, when the correlation between the antennas in the base station apparatus eNodeB is low, the inter-matrix distance is used.

（式３）

（式４）

The array response indicates the amplitude of the received signal vector at each antenna element after antenna synthesis in the angular direction θ, and is calculated by, for example, (Equation 2) (Note that (Equation 2) is rank 1). The calculation result of the array response of the precoding matrix is shown). When an array response is used as a reference, a precoding matrix that maximizes the value calculated by (Equation 2) or the like is selected. By selecting the precoding matrix in this way, it is possible to obtain an appropriate gain in the direction in which the signal arrives at the antenna. Here, a (θ) is a steering vector (array response vector) and is represented by (Equation 3). Here, “e” indicates a natural logarithm, and “j” indicates an imaginary number. “Μ” represents the amount of phase rotation, and is determined by the radio wave arrival angle θ, the antenna interval d, and the wavelength λ with respect to the carrier frequency, as shown in (Equation 4).
(Formula 2)

(Formula 3)

(Formula 4)

  A DFT codebook belonging to the first codebook group described above (hereinafter referred to as “DFT codebook group”) has a DFT structure, thereby obtaining a uniform gain in any direction in which a signal arrives. It is possible to define a precoding matrix that can be used.

（式６）

On the other hand, the inter-matrix distance indicates the distance of the precoding matrix to be compared (in other words, the similarity in the direction of the precoding vector corresponding to the precoding matrix), and is calculated by (Equation 5). . When the inter-matrix distance is used as a reference, the minimum value d _min (R) of the inter-matrix distance d _chord calculated by (Equation 5) is calculated by (Equation 6), and this minimum value d _min (R ) Is maximized for each rank. By selecting the code book in this way, precoding matrices whose characteristics are not similar to each other can be determined in the random code book. Here, “R” indicates a rank, “n” and “m” indicate variables corresponding to PMI indexes, and “G” indicates a precoding matrix.
(Formula 5)

(Formula 6)

Such calculation processing of the inter-matrix distance d _chord is performed by the following steps. First, after a predetermined number of DFT codebooks are set, random codebooks belonging to a second codebook group (hereinafter referred to as “random codebook group”) are randomly determined. The matrix distance d _chord all precoding matrix included in the random codebook and DFT codebook is determined, is calculated between all of the same rank in other codebook precoding matrix (formula 5). Then, the minimum value d _min (R) in the calculated inter-matrix distance d _chord is calculated for each rank (Formula 6). Next, the entire random codebook is updated, and inter-matrix distances d _chord of all precoding matrices included in the updated random codebook and DFT codebook are calculated again, and their minimum values d _min (R ) Is calculated for each rank. In this case, when the preceding matrix distance _{d chord} minimum value _{d min} (R) matrix distance subsequent than _{d chord} minimum _{d min} of the (R) is large, the random codebook the updated selection Is done. Such calculation processing is repeatedly performed until the minimum value d _min (R) of the inter-matrix distance d _chord becomes maximum.

Here, the target of the calculation process of the inter-matrix distance d _chord will be described with reference to FIG. As shown in FIG. 1, the calculation process of the inter-matrix distance d _chord is performed between two precoding matrices belonging to the same rank. The two precoding matrices to be subjected to arithmetic processing are (1) two different precoding matrices included in the DFT codebook, and (2) a precoding matrix included in the DFT codebook and a precoding matrix included in the random code. It consists of a coding matrix and (3) two different precoding matrices included in a random code. These arithmetic processes are performed on all precoding matrices included in the entire codebook (DFT codebook and random codebook).

In this case, when the calculation result in the calculation process of (1) is the minimum value d _min (R) of the inter-matrix distance d _chord , one or both DFT codebooks including the corresponding precoding matrix ( In the example shown in FIG. 1, one or both of DMI codebooks of PMI indexes # 2 and # 5) are updated. As described above, in the DFT codebook group, when the number of DFT codebooks to be determined is specified, each matrix component (precoding matrix) is uniquely determined. It is possible to obtain an even gain. For this reason, when a part of the DFT codebook is updated, it is difficult to obtain a uniform gain in any direction in which the signal arrives. As a result, when the correlation between antennas in the base station apparatus eNodeB is high, it may be difficult to effectively perform signal separation in the mobile station apparatus UE. The present inventor can inherently enjoy the DFT codebook by including the two precoding matrices included in the DFT codebook in the calculation target of the minimum value of the inter-matrix distance of the precoding matrix. Focusing on the point that gain cannot be obtained, the present invention has been achieved.

  In the code book generation method according to the present invention, the DFT code book (first code book) selected when the correlation between antennas in the base station apparatus eNodeB is high and the correlation between antennas in the base station apparatus eNodeB is low. And a random codebook (second codebook) selected from the DFT codebook and the minimum value of the inter-matrix distance of the precoding matrix included in the random codebook. For each rank, the minimum value of the inter-matrix distance of the precoding matrix included in the DFT codebook and the random codebook updated according to the calculation result of the minimum value of the inter-matrix distance calculated in advance is calculated for each rank. Calculate and repeat the comparison of the minimum value of the inter-matrix distance before and after the update, and When selecting a random codebook in which the minimum value of the inter-matrix distance of the coding matrix is selected, two different precoding matrices included in the DFT codebook are excluded from the calculation target when calculating the minimum value of the inter-matrix distance. It is what I did.

In the codebook generation method according to the present invention, the a After calculating the matrix distance _{d chord} of all pre-coding matrix by (Equation 5), the minimum value _d'min between these matrices distance _{d chord} (R) Calculated by (Equation 7). Here, as in (Expression 5) and (Expression 6), “n” and “m” indicate variables corresponding to the PMI index, and “N ₁ ” indicates the DFT codebook group included in the DFT codebook group. The numbers are shown (in FIG. 1, the case of N ₁ = 8 is shown).
(Formula 7)

When calculating the minimum value d ′ _min (R) of the inter-matrix distance d _{chord according} to (Equation 7), the precoding matrix included in the DFT codebook is removed from one variable n. At least one of the calculation targets of the minimum value d ′ _min (R) of the distance d _chord includes the precoding matrix included in the random codebook. For this reason, it is possible to avoid the calculation process of the minimum value d ′ _min (R) of the inter-matrix distance d _chord of the precoding matrix between the two precoding matrices included in the DFT codebook. It is possible to prevent a situation in which a part of the DFT codebook is updated and to obtain a gain that can be inherently enjoyed in the DFT codebook. Thereby, when the correlation between the antennas in the base station apparatus eNodeB is high, it is possible to select a precoding matrix that can be effectively separated by the mobile station apparatus UE.

On the other hand, since the minimum value d ′ _min (R) of the inter-matrix distance d _chord is calculated by (Equation 7) excluding the arithmetic processing between two different precoding matrices included in the DFT codebook, Can be appropriately selected. Thereby, when the correlation between the antennas in the base station apparatus eNodeB is low, it is possible to select a precoding matrix that can be effectively separated by the mobile station apparatus UE. As a result, it is possible to generate a codebook that defines a precoding matrix that can be effectively separated by the mobile station apparatus UE regardless of the level of correlation between antennas of the base station apparatus eNodeB.

By the way, when selecting a random codebook including a precoding matrix in which the minimum value d ′ _min (R) of the inter-matrix distance d _chord is maximum, it is not possible to update only the precoding matrix of a specific rank. Need to update the entire random codebook. This is based on the property of nested properties required for codebooks. The nested property is a characteristic that a precoding matrix of a lower rank (for example, rank 2) is composed of a submatrix of a precoding matrix of an upper rank (for example, rank 3). In order to deal with such characteristics, it is not possible to update only the precoding matrix of a specific rank.

However, when updating the entire random codebook, the minimum value d ′ _{min of the} inter-matrix distance d _chord of a certain rank in two random codebooks (here, random codebooks A and B) that are selection candidates. In (R), a larger value is calculated in the random codebook A, whereas in the minimum value d ′ _min (R) of the inter-matrix distance d _chord of other ranks, the random codebook B has a larger value. A calculated situation can occur.

FIG. 3 is a diagram illustrating an example of two random codebooks in which the maximum value of the minimum value d ′ _min (R) of the inter-matrix distance d _chord differs depending on the rank. In FIG. 3, the minimum value d ′ _{min of the} inter-matrix distance d _chord of random code books (hereinafter referred to as “candidate random code books”) that are candidates in the process of selecting random code books with the same PMI index. (R) is shown. FIG. 3 shows a case where codebooks A and B are selected as candidate random codebooks.

As shown in FIG. 3, in the code book A and the code book B, the minimum value d ′ _min (R) of the inter-matrix distance d _chord is the same value in rank 2, rank 4, rank 6 and rank 8. . In ranks 1 and 7, the minimum value d ′ _min (R) of the inter-matrix distance d _chord of the codebook B is larger. On the other hand, in ranks 3 and 5, the minimum value d ′ _min (R) of the inter-matrix distance d _chord of the codebook A is larger.

  In this case, if codebook A is selected, a transmission beam that is easier to separate signals in mobile station apparatus UE than codebook B can be formed when data is transmitted in rank 3 and rank 5, whereas rank 1 and rank 7, a transmission beam that is more difficult to separate in the mobile station apparatus UE than in the code book B is formed. On the other hand, if codebook B is selected, a transmission beam that is easier to separate signals in mobile station apparatus UE than codebook A can be formed when data is transmitted in ranks 1 and 7, whereas ranks 3 and 5 When transmitting data, a transmission beam is formed which is more difficult to separate in the mobile station apparatus UE than in the codebook A.

For this reason, in the codebook generating method according to the present invention, when the maximum value in the minimum value d ′ _min (R) of the inter-matrix distance d _chord exists in different candidate random codebooks, the inter-matrix distance d of each rank _A random codebook is selected based on an adjustment value d ″ _min obtained by weighted averaging the minimum value d ′ _min (R) of _chord . That is, in the code book generating method according to the present invention, as shown in (Equation 8), the minimum value d ′ _min (R) of the inter-matrix distance d _chord of each rank is adjusted by the weighting coefficient α _R , and A random codebook is selected based on an adjustment value d ″ _min obtained by arithmetic averaging of the adjusted values.
(Formula 8)

Here, the weighting factor α _R is determined according to, for example, a rank that is important by the mobile communication system. When data transmission with a low rank is important in a mobile communication system, for example, weight coefficients α ₁ , α ₂ , α ₃ , and α ₄ of rank 1 to rank 4 are 0.3, 0.3, and 0.2, respectively. , 0.2 is set, and 0 is set to the weighting factors α ₅ , α ₆ , α ₇ , α ₈ of ranks 5 to 8. Note that the determination of the weighting factor α _R is not limited to ranks that are important in data transmission, and can be changed as appropriate.

Using the example described above, the adjustment value is calculated from the codebook A shown in FIG. 3 _d'' min (R), as well as the adjustment value is calculated from the code book _B d'' _min (R) is less, respectively It becomes as follows.
Codebook A_d ″ _min = 0.3 * 0.8 + 0.3 * 1.0 + 0.2 * 1.1 + 0.2 * 1.2
= 0.24 + 0.3 + 0.22 + 0.24 = 1.0
Codebook B_d ″ _min = 0.3 * 0.9 + 0.3 * 1.0 + 0.2 * 1.0 + 0.2 * 1.2
= 0.27 + 0.3 + 0.2 + 0.24 = 1.01
In this case, since the adjustment value d ″ _min calculated from the code book B is larger than the adjustment value d ″ _min calculated from the code book A, the code book B is selected.

As described above, according to the code book generating method according to the present invention, when the maximum value in the minimum value d ′ _min (R) of the inter-matrix distance d _chord exists in different candidate random codebooks, the inter-matrix distance of each rank Since the random codebook is selected based on the adjustment value d ″ _min obtained by weighted averaging the minimum value d ′ _min (R) of d _chord , for example, the rank is emphasized in the mobile communication system. Accordingly, since a random codebook can be selected, it is possible to select a precoding matrix that is easily signal-separated by the mobile station apparatus UE when data is transmitted at the selected rank.

In the above description, when the precoding matrix included in the random codebook is selected by the arithmetic processing of the inter-matrix distance d _{chord in} the codebook composed of the DFT codebook and the random codebook, In the code book generation method in which two different precoding matrices included are excluded from the calculation target, a case where a random code book is selected based on the adjustment value d ″ _min is described. However, the method of selecting a random codebook based on an adjustment value obtained by weighted averaging the minimum values of the inter-matrix distances of each rank is applicable to all codebook generation methods for selecting a random codebook. Is possible.

In the calculation process of the inter-matrix distance d _chord , as described above, after the random codebooks belonging to the random codebook group are randomly determined, the inter-matrix distances d _chord of all the precoding matrices are changed to other It is calculated between all the precoding matrices of the corresponding rank in the codebook. Then, the minimum value d _min (R) in the calculated inter-matrix distance d _chord is calculated. Then, the entire random codebook is updated, and the inter-matrix distance d _chord of all precoding matrices after the update is calculated again, and their minimum value d _min (R) is calculated for each rank. In this case, when the preceding matrix distance _{d chord} minimum _{d min} the minimum value of the matrix distance _{d chord} succeeding than (R) _{d min} of (R) is large, the codebook of the updated selected The Such calculation processing is repeatedly performed until the minimum value d _min (R) of the inter-matrix distance d _chord becomes maximum.

That is, in the calculation process of the inter-matrix distance d _chord , a random code book having the minimum minimum value d _min (R) of the inter-matrix distance d _chord is searched while updating the entire random code book. However, in this case, since the inter-matrix distance d _chord of all the precoding matrices after update is an object to be calculated, a random codebook that maximizes the minimum value d _min (R) of the inter-matrix distance d _chord is searched. The amount of computation until it is done becomes enormous.

Therefore, in the codebook generation method according to the present invention, when searching for a random codebook in which the minimum value d _min (R) of the inter-matrix distance d _chord is maximized, the entire random codebook is not updated. Update some of the random codebooks. Specifically, one or both of the random code books for which the minimum value d _min (R) of the inter-matrix distance d _chord is calculated are updated. Note that the random codebook to be updated is selected at random.

In such calculation processing of the inter-matrix distance d _chord , after random codebooks belonging to the random codebook group are randomly determined, all precoding matrices (excluding two precoding matrices included in the DFT codebook) ) Between matrixes d _chord is calculated with all precoding matrices of corresponding ranks in other codebooks. Then, the minimum value d _min (R) in the calculated inter-matrix distance d _chord is calculated. Then, the random codebook including the minimum value d _min (R) in the inter-matrix distance d _chord is updated, and the inter-matrix distances d _chord of all precoding matrices after the update are calculated again, and their minimum A value d _min (R) is calculated for each rank. In this case, when the preceding matrix distance _{d chord} minimum _{d min} the minimum value of the matrix distance _{d chord} succeeding than (R) _{d min} of (R) is large, the codebook of the updated selected The Such calculation processing is repeatedly performed until the minimum value d _min (R) of the inter-matrix distance d _chord becomes maximum.

As described above, according to the code book generating method according to the present invention, when searching for a random code book in which the minimum value d _min (R) of the inter-matrix distance d _chord is maximized, some of the random code books are updated. As a result, the amount of computation required to search for a random codebook in which the minimum value d _min (R) of the inter-matrix distance d _chord is maximized can be greatly reduced as compared with the case of updating the entire random codebook. It becomes possible.

In particular, in the code book generation method according to the present invention, one or both of the random code books for which the minimum value d _min (R) of the inter-matrix distance d _chord is calculated are updated. It is easy to specify a random codebook, and it is possible to efficiently perform a calculation process for searching for a random codebook in which the minimum value d _min (R) of the inter-matrix distance d _chord is maximized.

In the above description, when the precoding matrix included in the random codebook is selected by the arithmetic processing of the inter-matrix distance d _{chord in} the codebook composed of the DFT codebook and the random codebook, In the code book generation method in which two different precoding matrices included are excluded from the calculation target, a case will be described in which one or both of the random code books for which the minimum value d _min (R) of the inter-matrix distance d _chord is calculated are updated. ing. However, the method of updating one or both of the random codebooks for which the minimum value d _min (R) of the inter-matrix distance d _chord is calculated may be applied to all codebook generation methods for selecting a random codebook. Is possible.

  Although the present invention has been described in detail using the above-described embodiments, it is obvious to those skilled in the art that the present invention is not limited to the embodiments described in this specification. The present invention can be implemented as modified and changed modes without departing from the spirit and scope of the present invention defined by the description of the scope of claims. Therefore, the description of the present specification is for illustrative purposes and does not have any limiting meaning to the present invention.

Claims (4)

A plurality of first codebooks having a DFT (discrete Fourier transform) structure defining a precoding matrix selected when the correlation between antennas in the base station apparatus is high, and the correlation between antennas in the base station apparatus is low A codebook generation method comprising a plurality of second codebooks defining precoding matrices to be selected in a case,
Calculating a minimum value of the inter-matrix distance (Chordal distance) of the precoding matrix included in the first and second codebooks for each rank; and the first codebook or the preceding calculation calculating a minimum value of the matrix distance between precoding matrices for each rank contained in the prior SL second codebook is updated according to the calculation result of the minimum value of the matrix between the length of the before and after the update between the matrix Repeating the comparison of the minimum values of the distances, and selecting the second codebook that maximizes the minimum value of the inter-matrix distance of the precoding matrix included in the entire codebook,
A codebook generation method, wherein two different precoding matrices included in the first codebook are excluded from a calculation target when calculating the minimum value of the inter-matrix distance.   The minimum value of the inter-matrix distance is adjusted by a weighting factor for each rank, and the second codebook is selected based on an adjustment value obtained by arithmetically averaging the adjusted values. The code book generation method according to claim 1.   3. The code book according to claim 1, wherein only a part of the plurality of second code books is updated according to a calculation result of a minimum value of the inter-matrix distance calculated in advance. Generation method.   4. The codebook generation method according to claim 3, wherein one or both of the second codebooks used for calculating the minimum value of the inter-matrix distance is updated.

Images