KR102304930B1 - Method of lattice reduction of multiple-input multiple-output communication system - Google Patents

Abstract

A grid reduction method of a multiple input/output wireless communication system is disclosed. In a multiple input multiple output (MIMO) wireless communication system according to the present disclosure, a grid reduction method of a receiving end having a plurality of receiving antennas is a tree-search-based arrangement of first to N-th columns performing QR decomposition on a channel matrix sequentially including a vector (N is an integer greater than or equal to 2), swapping a k-th column vector (k is an integer greater than or equal to 2 and less than or equal to N) included in the channel matrix ) or determining to maintain, and swapping the k-th column vector and the k-1th column vector when it is determined that the k-th column vector is to be swapped in the determining step, wherein the determining comprises: Among the second to N-th column vectors, it may be characterized in that it is performed only on a column vector of a single expansion (SE) stage.

Description

Grid reduction method of multiple input/output communication system

The technical idea of the present disclosure relates to a lattice reduction method, and more particularly, to a lattice reduction method for signal detection in a Multiple-Input Multiple-Output (MIMO) communication system.

The multiple input/output communication system is a method of using multiple antennas in a transmission/reception end to support a high transmission rate within a limited transmission power and frequency resource, and is widely used in an OFDM (Orthogonal Frequency Division Multiplexing)-based wireless communication system. A multiple input/output communication system can perform high-speed data transmission by transmitting and receiving signals using a plurality of antennas at both the transmitting end and the receiving end. Since independent signals are simultaneously transmitted through each antenna of the transmitting end of the multiple input/output communication system, a technique for effectively decoding the independent signals at the receiving end has a great influence on system performance.

According to a maximum likelihood decoding technique as one of the decoding techniques, a received signal is converted into a log-likelihood ratio (LLR) for each coded bit while considering the mutual influence between transmission signals. Here, an accurate LLR can be obtained when all possible combinations of signals are considered, but when the number of simultaneously transmitted signals is large, the number of signal combinations increases exponentially, making it substantially difficult to decode in consideration of all combinations.

As one of the semi-optimal algorithms for reducing the complexity of such a decoding algorithm, there is a successive interference cancellation (SIC) technique as a decoding method with a low computational amount. The SIC sequentially decodes several simultaneously transmitted signals, and removes the decoded signal component from the received signal. Through this, the process of considering a signal combination can be omitted by separating and decoding simultaneously transmitted signals.

A decoding technique in which lattice-reduction (LR) and continuous interference cancellation (SIC) are combined may be considered as a method of increasing the performance of a sub-optimal algorithm such as the continuous interference cancellation (SIC). The lattice reduction method is a method to obtain low complexity and high performance at the same time by applying a suboptimal algorithm such as continuous interference cancellation (SIC) after transforming the channel matrix so that the channel matrix has maximum favorability. The degree of favorability of a specific channel matrix may be determined by orthogonality of the corresponding channel matrix and complexity of a lattice reduction method.

Signal transmission/reception through the multiple input/output communication system using the above-described grid reduction method provides a high data rate, while data transmitted through multiple transmission/reception antennas is exposed to noise and the reception signal is deformed. In addition, as the number of antennas increases, the complexity increases as the number of signal candidates to be considered in signal detection increases exponentially, and accordingly, the degree of favorability decreases, acting as the biggest obstacle to realizing a high-dimensional and high-performance multiple input/output communication system in an actual system. do. As will be described later, exemplary embodiments of the present disclosure provide an algorithm for detecting a transmission signal by efficiently restoring a modified multiple input/output reception signal.

The problem to be solved by the technical idea of the present disclosure is the complexity and consumption of the multiple input/output signal detection algorithm through the reduction of the complexity of the lattice reduction (LR) algorithm used as a preprocessing technique of a plurality of signal detection methods in a multiple input/output communication system An object of the present invention is to provide a grid reduction method for reducing power.

According to an exemplary embodiment of the present disclosure, a method for reducing a grid of a receiving end having a plurality of receiving antennas in a multiple input/output communication system is a tree-search-based first to N-th column vector (N). performing a QR decomposition on a channel matrix sequentially including an integer greater than or equal to 2), swapping or maintaining a k-th column vector (k is an integer greater than or equal to 2 and less than or equal to N) included in the channel matrix and determining the k-th column vector and swapping the k-1th column vector when it is determined that the k-th column vector is to be swapped in the determining step, wherein the determining comprises the steps of: Among the N-th column vectors, it may be characterized in that it is performed only on a column vector of a single expansion (SE) stage.

The lattice reduction method according to an exemplary embodiment of the present disclosure is a lattice reduction method based on a fixed-complexity Lenstra-Lenstra-Lovasz (fcLLL) algorithm of a receiving end having a plurality of receiving antennas in a multiple input/output wireless communication system. , determining swap or maintaining the second to Nth column vectors in a channel matrix sequentially including the first to Nth column vectors (N is an integer of 2 or more), and when the determining step is finished , determining lattice reduction termination or repetition by comparing favorability and reference favorability with respect to the channel matrix, and determining the swap or maintenance is adjacent to each other among the second to N-th column vectors. It may be characterized in that it is sequentially determined whether to swap with respect to the column vectors that have not been performed.

According to the lattice reduction technique according to an exemplary embodiment of the present disclosure, a multiple input/output signal through reduction in complexity of a lattice reduction (LR) algorithm used as a preprocessing technique for a plurality of signal detection methods at the receiving end of a multiple input/output communication system It is possible to reduce the complexity and power consumption of the detection algorithm.

1 is a flowchart illustrating a grating reduction method according to an exemplary embodiment of the present disclosure.
2 is a diagram illustrating a tree-search arrangement structure according to an exemplary embodiment of the present disclosure.
3 is a flowchart illustrating a column search according to an exemplary embodiment of the present disclosure.
4 is a flowchart illustrating a column search according to an exemplary embodiment of the present disclosure.
5 is a flowchart illustrating a grid reduction method according to an exemplary embodiment of the present disclosure.
6 is a flowchart illustrating a grid reduction method according to an exemplary embodiment of the present disclosure.
7 is a flowchart illustrating a grating reduction method according to an exemplary embodiment of the present disclosure.
8 is a diagram illustrating a grating reduction method according to an exemplary embodiment of the present disclosure.
9 is a graph showing a result of evaluating the performance of a lattice reduction method according to an exemplary embodiment of the present disclosure.
10 is a block diagram illustrating a multiple input/output communication system according to an exemplary embodiment of the present disclosure.

1 is a flowchart illustrating a grating reduction method according to an exemplary embodiment of the present disclosure.

According to a maximum likelihood decoding technique as one of the decoding techniques of a multiple input/output communication system, a received signal is converted into a log-likelihood ratio (LLR) for each coded bit while considering the mutual influence between transmission signals. Here, an accurate LLR can be obtained when all possible combinations of signals are considered, but when the number of simultaneously transmitted signals is large, the number of signal combinations increases exponentially, making it substantially difficult to decode in consideration of all combinations.

As one of the semi-optimal algorithms for reducing the complexity of such a decoding algorithm, there is a successive interference cancellation (SIC) technique as a decoding method with a low computational amount. The SIC sequentially decodes several simultaneously transmitted signals, and removes the decoded signal component from the received signal. Through this, the process of considering the signal combination by separating and decoding the simultaneously transmitted signals can be omitted.

A decoding technique in which lattice-reduction (LR) and continuous interference cancellation (SIC) are combined may be considered as a method of increasing the performance of a sub-optimal algorithm such as the continuous interference cancellation (SIC). The lattice reduction method is a method to obtain low complexity and high performance at the same time by applying a suboptimal algorithm such as continuous interference cancellation (SIC) after transforming the channel matrix so that the channel matrix has maximum favorability. The degree of favorability of a specific channel matrix may be determined by orthogonality of the corresponding channel matrix and complexity of a lattice reduction method.

Signal transmission/reception through the multiple input/output communication system using the above-described grid reduction method provides a high data rate, while data transmitted through multiple transmission/reception antennas is exposed to noise and the reception signal is deformed. In addition, as the number of antennas increases, the complexity increases as the number of signal candidates to be considered in signal detection increases exponentially, and accordingly, the degree of favorability decreases, acting as the biggest obstacle to realizing a high-dimensional and high-performance multiple input/output communication system in an actual system. do. As will be described later, exemplary embodiments of the present disclosure provide an algorithm for detecting a transmission signal by efficiently restoring a modified multiple input/output reception signal.

The receiving end of the multiple input/output communication system may perform the lattice reduction method in the process of decoding the received signal matrix into the transmit signal matrix. A vector matrix between input/output signals of a multiple input/output communication system may be expressed as follows [Equation 1].

The reception signal matrix Y is a data matrix of RXD (R is the number of reception antennas, D is the number of data symbols), and the transmission signal matrix S is a data matrix of TXD (T is the number of transmission antennas, D is the number of data symbols) , and the channel matrix H may be a data matrix of RXT. The channel matrix H may be expressed by [Equation 2] as follows.

h ₁ to h _T are column vectors obtained by dividing the channel matrix H into T matrices having a size of RX 1 , and in the present specification, h ₁ to h _T are referred to as first to T-th column vectors, respectively. The first to Tth column vectors are the first to Pth column vectors h ₁ to h _P of the FE (Full Expansion) stage and the P+1 to Tth of the SE (Single Expansion) stage according to the tree-search criterion. It can be divided into column vectors (h _P+1 to h _{T ).} This will be described later in detail with reference to FIG. 2 .

The receiving end of the multiple input/output communication system may perform the lattice reduction method of the present disclosure on the channel matrix (H). The lattice reduction method of the present disclosure includes the steps of swapping the FE stage and the SE stage of the channel matrix H (S100), performing QR decomposition on the channel matrix H (S200), and performing column search for the SE stage It may include a step (S300).

The receiving end of the multiple input/output communication system may swap the FE stage and the SE stage for the channel matrix H (S100). As another embodiment, the receiving end of the multiple input/output communication system may receive a channel matrix in which the FE stage and the SE stage are swapped (not shown). The channel matrix (H') in which the FE stage and the SE stage are swapped can be expressed by [Equation 3].

The receiving end of the multiple input/output communication system may perform QR decomposition on the channel matrix (H') in which the FE stage and the SE stage are swapped (S200). QR decomposition means decomposing a channel matrix (H or H') into a product of an orthogonal matrix (hereinafter, Q vector) and an upper triangular matrix (hereinafter, R vector), and the Gram-Schmidt process (Gram-Schmidt Process or House The House Holder Process may be used.

According to an embodiment of the present disclosure, the receiving end of the multiple input/output communication system may perform QR decomposition using only _{the P+1th to Tth column vectors (h P+1} to h _{T ) included in the SE stage.} Since QR decomposition is performed using only some of the column vectors of the channel matrix (H), QR decomposition with lower complexity can be performed compared to the case of performing QR decomposition using all column vectors of the channel matrix (H). , the lattice reduction method may be more favorable.

The receiving end of the multiple input/output communication system may perform a column search for the SE stage using the R vector obtained in the QR decomposition (S300). According to an embodiment of the present disclosure, the receiving end of the multiple input/output communication system searches for a column using only _{the P+1th to Tth column vectors (h P+1} to h _{T) and the corresponding R vectors included in the SE stage.} can be performed. _{Since only the P+1th to Tth column vectors (h P+1} to h _T ) included in the SE stage are used except for the FE stage, the complexity of the lattice reduction method may be reduced and favorability may be increased. Details of column search will be described later with reference to FIGS. 3 and 4 .

2 is a diagram illustrating a tree-search arrangement structure according to an exemplary embodiment of the present disclosure.

Referring to FIG. 2 , a symbol included in data received by the multiple input/output communication system may be determined according to a Euclidean distance between candidate symbols. Each point in FIG. 2 may represent candidate symbols, and between each point may be quantified according to a Euclidean distance. The receiving end of the multiple input/output communication system may detect a signal having the smallest Euclidean distance in consideration of all candidate symbols.

As a method for detecting a signal having the smallest Euclidean distance, a tree-search arrangement structure may be used. The number of FE stages may be determined based on the number of transmit antennas T, a symbol having a poor channel environment may be disposed on the FE stage, and a symbol with a good channel environment may be disposed on the SE stage. As an example, the number of FE stages ^{may be determined to be T 1/2} −1 (T is the number of transmit antennas) or more. By disposing a symbol having a poor channel environment in the FE stage, Euclidean distances of all candidate symbols are considered, so that an error due to channel distortion can be eliminated. On the other hand, in the SE stage in which a symbol having a good channel environment is disposed, only one symbol having a close Euclidean distance may be considered.

3 is a flowchart illustrating a column search according to an exemplary embodiment of the present disclosure. In detail, FIG. 3 is a flowchart illustrating an example of the step ( S300 ) of performing the column search of FIG. 1 .

1 and 3 , the step of performing column search ( S300 ) includes determining whether to swap or maintain a column vector ( S310 ), and if there is a swap determination, performing a column vector swap ( S320 ) ), changing the determination target column vector in response to the swap or maintenance determination ( S330 ) and determining whether to end the column search ( S340 ).

The receiving end of the multiple input/output communication system may determine to swap or maintain the column vector (S310). The R vector obtained through QR decomposition may be used when determining swap or hold. According to an embodiment of the present disclosure, the receiving end of the multiple input/output communication system determines whether the R vector satisfies the basis reduction conditional expression and the Lobartz conditional expression, and only when both the basis reduction conditional expression and the Lobartz conditional expression are satisfied, the corresponding column It is possible to judge the maintenance of the vector. Details on this will be described later with reference to FIG. 4 .

According to an embodiment of the present disclosure, the receiving end of the multiple input/output communication system may determine swap or maintain only _{for the P+1th to Tth column vectors (h P+1} to h _{T ) included in the SE stage.} The first to Pth column vectors (h ₁ to h _P ) included in the FE stage have to consider all the nearby symbols of the FE stage, so the P+1th to Tth column vectors (h _{P+1) included in the SE stage} to h _T ), the overhead for column search may be large. In addition, since the FE stage considers all candidate symbols, the need for column search may be less than that of the SE stage in terms of affinity. Therefore, the receiving end of the multiple input/output communication system of the present disclosure _{determines swap or maintain only for the P+1th to Tth column vectors (h P+1} to h _T ) included in the SE stage, thereby reducing the overall grid by lowering the complexity. It can increase the level of favorability of the method.

The receiving end of the multiple input/output communication system may perform swap on the column vector determined to be swap ( S320 ). In detail, the receiving end of the multi-input communication system may change the order of the column vector determined to be swap and the column vector before it. By rearranging the column vector through swap, orthogonality of the column vector is increased, and a channel matrix including the column vector can be arranged concisely. Accordingly, the channel matrix H may be arranged according to the affinity characteristic.

The receiving end of the multiple input/output communication system may change the determination target column vector in response to the swap or maintenance determination (S330). In detail, the receiving end of the multiple input/output communication system may change the column vector to be swapped or maintained when there is a swap decision for the corresponding column vector to the column vector of the next row, and swap when there is a retain decision for the column vector Alternatively, the column vector to be maintained may be changed to the column vector of the previous row.

The receiving end of the multiple input/output communication system may determine whether the column search is terminated (S340). When the column vector determined to be swapped or maintained in step S310 is the T-th column vector, or when the end determination value obtained by performing partial SIC only for the SE stage exceeds the reference value, the multiple input/output communication system The receiving end of may terminate the column search. This will be described later with reference to FIG. 5 .

4 is a flowchart illustrating a column search according to an exemplary embodiment of the present disclosure.

3 and 4 , steps S311 and S312 of FIG. 4 may be detailed steps of step S310 of FIG. 3 , and step S321 of FIG. 4 may be a detailed step of step S320 of FIG. 3 , FIG. Steps S331 and S332 of FIG. 3 may be detailed steps of step S330 of FIG. 3 , and steps S341 and S342 of FIG. 4 may be detailed steps of step S340 of FIG. 3 .

The receiving end of the multiple input/output communication system may input P+2 as an initial value of the determination value k of the column search (S301). According to an embodiment of the present disclosure, swap or maintenance can be determined only for the SE stage, and for this purpose, the initial value of the determination value k is the second column vector of the SE stage, the P+2th column vector (hP+2). ) may be set to P+2. As the initial value of the determination value k is set to P+2, it may be determined whether to swap or maintain the P+2th column vector (hP+2) with the P+1th column vector (hP+1). have.

The receiving end of the multiple input/output communication system may determine swapping or maintaining the k-th column vector hK corresponding to the determination value k. In an embodiment of the present disclosure, the receiving end of the multiple input/output communication system _{determines whether the k-th column vector (h K} ) satisfies the basis reduction (Size Reduction) conditional expression (S311), and the k-th column vector (h _K) ) may determine whether or not the Lovasz conditional expression is satisfied (S312). With respect to the R matrix for the channel matrix (H) including the kth column vector (h _K ), the basis reduction condition may be as follows [Equation 4]. In addition, the Lobartz conditional expression may be as shown in [Equation 5] below.

(R(m,n) is the component of m rows and n columns of the R matrix, k is a judgment value, l is a natural number less than k, and δ is a real number between 0.5 and 1 as a Lobartz variable)

When the kth column vector (h _K ) does not satisfy at least one of the basis reduction conditional expression and the Lobartz conditional expression, the receiving end of the multiple input/output communication system performs the kth column vector (h _K ) and the k-1th column vector (h _{K ) -1} ) can be swapped (S321). That is, the receiving end of the multiple input/output communication system may reverse the order of _{the k-th column vector (h K} ) and the k- _{1th column vector (h K-1).} Also, the receiving end of the multiple input/output communication system may input k-1 as the determination value k to change the column vector to be determined into the column vector corresponding to the previous column ( S331 ). That is, the receiving end of the multiple input/output communication system may repeat steps S311 and S312 for determining swap or maintenance for _{the k-1th column vector h K-1 by adjusting the determination value k.}

When the kth column vector (h _K ) satisfies the basis reduction conditional expression and the Lobartz conditional expression, the receiving end of the multiple input/output communication system calculates the kth column vector (h _K ) and the k-1th column vector (h _K-1 ) maintained, and it may be determined whether the determination value k is equal to the number of reception antennas T (S341). When the determination value k is equal to the number of receiving antennas T, the receiving end of the multiple input/output communication system may end the column search.

If the determination value (k) is not equal to the number of reception antennas (T), the receiving end of the multiple input/output communication system receives the end determination value (S _y ) and the reference value obtained through partial SIC for the SE stage. (S _Ref ) can be compared (S342). According to an embodiment of the present disclosure, since the thermal search is performed only for the SE stage, partial SIC for limiting the column vector of the SIC to the SE stage may be performed. As an example, the end determination value S _y may be a Euclidean distance through partial successive interference cancellation (Partial SIC), and the reference value S _Ref may be a predetermined Euclidean distance.

When the end determination value (S _y ) is not smaller than the reference value (S _Ref ), the receiving end of the multiple input/output communication system changes the determination target column vector to a column vector corresponding to the next column. k+1 may be input (S332). That is, the receiving end of the multiple input/output communication system may repeat steps S311 and S312 for determining swap or maintenance for _{the k+1th column vector h K+1 by adjusting the determination value k.} When the end determination value S _y is smaller than the reference value S _Ref , the receiving end of the multiple input/output communication system may end the column search.

5 is a flowchart illustrating a grid reduction method according to an exemplary embodiment of the present disclosure.

Referring to FIG. 5, the receiving end of the multiple input/output communication system sequentially searches columns for second to T-th column vectors at intervals of leaf values (L) in a lattice reduction method using a fixed complexity LLL (fcLLL). can be performed (S400). According to an embodiment of the present disclosure, the leaf value L may be an integer of 2 or more. The receiving end of the multiple input/output communication system of the present disclosure may jump and perform swap or maintenance determination by setting the leaf value to an integer of 2 or more, and the speed and favorability of the lattice reduction method may be improved. The column search of S400 will be described in detail with reference to FIG. 6 .

When the column search for the second to Nth column vectors is completed, the receiving end of the multiple input/output communication system may determine whether the grid reduction method is terminated (S500). According to an embodiment of the present disclosure, the receiving end of the multiple input/output communication system may determine whether to terminate the grid reduction method based on the number of times column search is performed, that is, the number of times step S400 is performed. When the column search is performed a predetermined number of times, the receiving end of the multiple input/output communication system may end the grid reduction method. According to another embodiment of the present disclosure, the receiving end of the multiple input/output communication system may determine whether to terminate the lattice reduction method based on the degree of favorability of the channel matrix. The receiving end of the multiple input/output communication system may terminate the grid reduction method when the degree of favorability of the channel matrix is equal to or greater than the reference degree of favorability. As an example, the Euclidean distance may be used as an index indicating the degree of favorability.

6 is a flowchart illustrating a grid reduction method according to an exemplary embodiment of the present disclosure. In detail, FIG. 6 is a flowchart illustrating an exemplary column search in step S400 of FIG. 5 .

5 and 6 , the receiving end of the multiple input/output communication system may input 2 as an initial value of the determination value k of the column search ( S401 ). As the initial value of the determination value k is set to ₂ , it may be determined whether the second column vector h 2 is swapped or maintained with the first column vector h _{1 .}

The receiving end of the multiple input/output communication system may determine swapping or maintaining the k-th column vector hK corresponding to the determination value k. In an embodiment of the present disclosure, the receiving end of the multiple input/output communication system _{determines whether the k-th column vector (h K} ) satisfies the basis reduction (Size Reduction) conditional expression (S411), and the k-th column vector (h _K) ) may determine whether or not the Lovasz conditional expression is satisfied (S412).

When the kth column vector (h _K ) does not satisfy at least one of the basis reduction conditional expression and the Lobartz conditional expression, the receiving end of the multiple input/output communication system performs the kth column vector (h _K ) and the k-1th column vector (h _{K ) -1} ) can be swapped (S421). That is, the receiving end of the multiple input/output communication system may reverse the order of _{the k-th column vector (h K} ) and the k- _{1th column vector (h K-1).}

Also, the receiving end of the multiple input/output communication system may input k+L to the determination value k in order to change the column vector to be determined (S431). That is, the receiving end of the multiple input/output communication system may repeat steps S411 and S412 for determining swap or maintenance for the _{k+L-th column vector (h k+L) by adjusting the determination value k.} The leaf value L may be an integer of 2 or more.

When the kth column vector (h _K ) satisfies the basis reduction conditional expression and the Lobartz conditional expression, the receiving end of the multiple input/output communication system calculates the kth column vector (h _K ) and the k-1th column vector (h _K-1 ) maintained, and it may be determined whether the determination value k is equal to the number T of the reception antennas (S441). When the determination value k is equal to the number of receiving antennas T, the receiving end of the multiple input/output communication system may end the column search.

When the determination value k is not equal to the number of reception antennas T, the receiving end of the multiple input/output communication system may input k+L to the determination value k to change the column vector to be determined ( S432). That is, the receiving end of the multiple input/output communication system may repeat steps S411 and S412 for determining swap or maintenance for _{the k+L-th column vector (h K+L) by adjusting the determination value k.}

7 is a flowchart illustrating a grating reduction method according to an exemplary embodiment of the present disclosure.

Referring to FIGS. 6 and 7 , the receiving end of the multiple input/output communication system may sequentially perform a column search on the second to T-th column vectors at intervals of the leaf value L ( S400a ). The receiving end of the multiple input/output communication system may determine whether the grid reduction method ends (S500a), and when it is determined that the grid reduction method ends, the grid reduction method may end. Steps S400a and S500a may be the same as or similar to steps S400 and S500 of FIG. 6 , so a description thereof will be omitted.

When the grid reduction method is not terminated (S500a), the receiving end of the multiple input/output communication system may adjust the leaf value L. Thereafter, the column search in step S400a may be performed using the adjusted leaf value (L). According to an embodiment of the present disclosure, the leaf value L may be adjusted to a predetermined leaf value L. According to another exemplary embodiment of the present disclosure, the leaf value L may be adjusted to a different leaf value L by evaluating the degree of favorability for the lattice reduction method. As an embodiment, the leaf value L may be set as the first leaf value when the favorability of the lattice reduction method is lower than the first reference affinity, and when the favorability of the grid reduction method is higher than the first reference affinity, the leaf value ( L) may be set as the second leaf value.

8 is a diagram illustrating a grating reduction method according to an exemplary embodiment of the present disclosure.

5 and 8 , the receiving end of the multiple input/output communication system performs a search on a column vector separated by a leaf value (L) from a previously searched column vector instead of performing a column search on all column vectors. can 8 is an example in which the receiving end of the multiple input/output communication system performs column search based on '2' as the leaf value (L) and column search based on '3' as the leaf value (L). An example of performing the is shown.

Referring to the example in which the leaf value L is '2', the receiving end of the multiple input/output communication system performs column search on the second column vector (h2) and then column searches on the fourth column vector (h ₄ ). can be performed. Thereafter, the receiving end of the multiple input/output communication system may perform a column search while jumping two by two for a column vector. _{When the search for the column vector up to the T-} th column vector (h T ) is finished, the receiving end of the multi-input/output communication system may check whether or not it is terminated (Termination Check). A method of checking whether or not the end is completed will be omitted as described above with reference to FIGS. 5 and 6 . If the predetermined criterion is met, the receiving end of the multiple input/output communication system may end the grid reduction method. If the predetermined criterion is not met, the receiving end of the multiple input/output communication system may return to the beginning and repeat the column search from _{the second column vector h 2 .} In this case, the receiving end of the multiple input/output communication system may add 1 to the number of iterations (iter).

The receiving end of the multiple input/output communication system may maintain or change the leaf value L during repeated column searches. As an example, if the leaf value L was set to '2' in the previous column search, the leaf value L may be set to '3' in the repeated column search.

According to the lattice reduction method of the present disclosure, since a hop search is performed to search a column vector spaced by L from a previously searched column vector, global orthogonality can be increased and the conditional expression can be quickly satisfied. The method execution time can be reduced and the complexity of hardware implementation can be reduced.

9 is a graph showing a result of evaluating the performance of a lattice reduction method according to an exemplary embodiment of the present disclosure.

Referring to FIG. 9 , the horizontal axis of the graph represents the bit energy to noise power ratio (Eb/N0), the vertical axis represents the bit error rate (BER), and each iter of the algorithm means the number of repetitions of the column search. . 'CLLL' may represent the case of a lattice reduction method having an optimal performance that does not limit the maximum number of iterations. That is, 'CLLL' may represent the case of an ideal lattice reduction method. 'fcLLL' indicates a case in which a fixed complexity LLL is performed according to a predetermined number of iterations (iter), and 'proposed' indicates a case in which the iteration number (iter) and leaf value (L) are taken into consideration when performed according to the present disclosure. can represent 9, the graph of CLLL, the graph of fcLLL according to the first iteration number iter1, the graph of fcLLL according to the second iteration number iter2, and the second iteration number iter2 and the first leaf value L1 A graph of the present disclosure according to (Proposed) is shown.

Comparing fcLLL(iter1) and proposed(iter2/L1), the number of iterations (iter) required to obtain a graph close to CLLL is the first number of iterations (iter1) in the case of fcLLL, In this case, it can be confirmed that it is the second iteration number iter2. According to the experimental result, the first iteration number iter1 may be greater than the second iteration number iter2. As an example, the first iteration number iter1 may be '7', and the second iteration number iter2 may be '5'. That is, the lattice reduction method of the present disclosure may have optimal performance with fewer iterations compared to the existing fcLLL algorithm.

In addition, comparing fcLLL(iter2) and proposed(iter2/L1), when the number of iterations (iter) is the same as the second number of iterations (iter2), the case of the present disclosure (proposed) is closer to CLLL compared to fcLLL that can be checked

The lattice reduction method of the present disclosure can obtain a high degree of favorability with less complexity due to a reduction in the number of iterations. As a result, the number of pipelines in hardware implementation can be reduced, so that implementation complexity of hardware can be reduced.

10 is a block diagram illustrating a multiple input/output communication system according to an exemplary embodiment of the present disclosure.

Referring to FIG. 10 , the transmitter may include an encoder 400 , a modulator 402 , a precoder 404 , and a plurality of antennas 408-1 to 408-T. In addition, the receiver may include a plurality of antennas 410 - 1 to 410 -R, a MIMO detector 412 , a decoder 414 , a channel estimator 416 , and a codebook selector 418 .

The encoder 400 may encode transmitted data and output code symbols. For example, the encoder 400 may perform encoding using a Convolutional Code (CC), a Turbo Code (TC), a Convolutional Turbo Code (CTC), or a Low Density Parity Check (LDPC) code. The modulator 402 may generate modulation symbols by modulating the code symbols from the encoder 400 . For example, the modulator 402 may perform modulation using Quadrature Phase Shift Keying (QPSK), Quadrature Amplitude Modulation (16QAM), 32QAM, 64QAM, or the like. An interleaver or the like may be added between the encoder 400 and the modulator 402 .

In the case of a broadband wireless access system, a plurality of antenna signals generated from the precoder 404 are each OFDM (Orthogonal Frequency Division Multiplexing) modulated, and each of the OFDM modulated signals is RF (Radio Frequency) processed through a corresponding antenna. can be sent.

A signal received through the plurality of antennas 410 - 1 to 410 -R may be input to the MIMO detector 412 . In the case of a broadband wireless access system, each of a plurality of received signals may be processed as a baseband signal, and each of the baseband signals may be OFDM-demodulated and provided to the MIMO detector 412 . The MIMO detector 412 decodes the input reception vector according to a predetermined MIMO detection method in consideration of the channel matrix H, and estimates and outputs the transmission vector transmitted by the transmitter. In this case, the MIMO detector 412 may estimate the transmission vector through the lattice reduction method described above with reference to FIGS. 1 to 9 . The decoder 414 decodes the data from the MIMO detector 412 and restores the original information data.

Exemplary embodiments have been disclosed in the drawings and specification as described above. Although the embodiments have been described using specific terms in the present specification, these are used only for the purpose of explaining the technical spirit of the present disclosure and are not used to limit the meaning or the scope of the present disclosure described in the claims. . Therefore, it will be understood by those skilled in the art that various modifications and equivalent other embodiments are possible therefrom. Accordingly, the true technical protection scope of the present disclosure should be defined by the technical spirit of the appended claims.

Claims (10)

In a multiple input multiple output (MIMO) communication system, a method for reducing a grid of a receiving end having a plurality of receiving antennas, the method comprising:
performing a QR decomposition on a channel matrix sequentially including first to N-th column vectors (N is an integer of 2 or more) arranged on a tree-search basis;
determining swap or maintenance for a k-th column vector (k is an integer greater than or equal to 2 and less than or equal to N) included in the channel matrix; and
and swapping the k-th column vector and the k-1th column vector when it is determined that the k-th column vector is to be swapped in the determining step;
The lattice reduction method, characterized in that the determining is performed only with respect to a column vector of a single expansion (SE) stage among the second to Nth column vectors. According to claim 1,
The method further comprising: receiving the channel matrix in which a column vector of a Full Expansion (FE) stage and a column vector of a Single Expansion (SE) stage are swapped. According to claim 1,
The method further comprising: generating the channel matrix by swapping a column vector of a Full Expansion (FE) stage and a column vector of a Single Expansion (SE) stage among the original matrix arranged on a tree-search basis. According to claim 1,
The determining includes determining the k-th column vector as a swap when the k-th column vector does not satisfy a size reduction conditional expression,
For the R matrix for the channel matrix generated as a result of the performing step, the basis reduction condition is
|R(k,l)| ² ≤ 1/2*|R(l,l)| ² (R(m,n) is the component of m rows and n columns of the R matrix, k is the judgment value, and l is a natural number less than k)
A lattice reduction method, characterized in that According to claim 1,
The determining includes determining as a swap with respect to the k-th column vector when the k-th column vector does not satisfy a Lovasz conditional expression,
For the R matrix for the channel matrix generated as a result of the performing step, the Lobartz condition is
δ|R(k-1,k-1)| ² ≤|R(k,k-1)| ² +|R(k,k)| ² (R(m,n) is the component of m rows and n columns of the R matrix, k is the judgment value, and δ is the Lobartz variable, a real number between 0.5 and 1)
A lattice reduction method, characterized in that According to claim 1,
When it is determined that the k-th column vector is swapped in the determining step, after performing the swapping step, determining whether to swap or maintain the k-1th column vector. According to claim 1,
and determining whether to swap or maintain the k+1th column vector after performing the swap if it is determined that the k-th column vector is maintained in the determining step. According to claim 1,
The step of performing the QR decomposition,
A lattice reduction method, characterized in that the QR decomposition is performed on a matrix composed of column vectors of a single expansion (SE) stage among the first to Nth column vectors. According to claim 1,
Whether the lattice reduction is terminated by comparing the result of performing Partial Successive Interference Cancellation (Partial SIC) on the column vector of the SE (Single Expansion) stage among the first to N-th column vectors and a predetermined reference value Lattice reduction method further comprising the step of determining. In a lattice reduction method based on a fixed-complexity Lenstra-Lenstra-Lovasz (fcLLL) algorithm of a receiving end having a plurality of receiving antennas in a multiple input multiple output (MIMO) communication system, the method comprising:
determining swap or maintenance with respect to second to Nth column vectors in a channel matrix sequentially including first to Nth column vectors (N is an integer of 2 or more); and
When the determining step is completed, determining whether the grid reduction is terminated or repeated by comparing favorability and reference favorability for the channel matrix;
The determining of the swap or the maintenance comprises sequentially determining whether to swap the non-adjacent column vectors among the second to N-th column vectors.

Images