CN102724160B - Signal detecting method in high-order modulated multi-input multi-output system - Google Patents
Signal detecting method in high-order modulated multi-input multi-output system Download PDFInfo
- Publication number
- CN102724160B CN102724160B CN201210138399.0A CN201210138399A CN102724160B CN 102724160 B CN102724160 B CN 102724160B CN 201210138399 A CN201210138399 A CN 201210138399A CN 102724160 B CN102724160 B CN 102724160B
- Authority
- CN
- China
- Prior art keywords
- path
- mrow
- group
- paths
- survival
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
- 238000000034 method Methods 0.000 title claims abstract description 88
- 239000013598 vector Substances 0.000 claims abstract description 61
- 239000011159 matrix material Substances 0.000 claims abstract description 60
- 238000001514 detection method Methods 0.000 claims abstract description 36
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 31
- 230000004083 survival effect Effects 0.000 claims description 70
- 230000001174 ascending effect Effects 0.000 claims description 32
- 238000009825 accumulation Methods 0.000 claims description 13
- 238000005457 optimization Methods 0.000 claims description 6
- 238000012216 screening Methods 0.000 claims description 6
- 150000001875 compounds Chemical class 0.000 claims description 2
- 238000012545 processing Methods 0.000 abstract description 3
- 238000005259 measurement Methods 0.000 abstract 1
- 238000004088 simulation Methods 0.000 description 12
- 230000006854 communication Effects 0.000 description 8
- 230000005540 biological transmission Effects 0.000 description 7
- 238000004891 communication Methods 0.000 description 7
- 238000012163 sequencing technique Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 239000004065 semiconductor Substances 0.000 description 3
- 238000004422 calculation algorithm Methods 0.000 description 2
- 238000010187 selection method Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Landscapes
- Radio Transmission System (AREA)
Abstract
The invention discloses a signal detection method in a high-order modulated multi-input multi-output system. The method comprises the following steps of: 1. receiving signals; 2. performing preprocess; 3. detecting a bottom index layer; 4. performing presort; 5. performing path extension; 6. selecting preferred paths; 7. selecting suboptimal paths; 8. determining surviving paths; 9. updating index layers; 10. detecting a top index layer; and 11. outputting obtained detection vectors. According to the method provided in the invention, a sorted QR decomposition is adopted to apply triangularization on a real channel matrix and the presort is adopted to determine a sequence in which child nodes ascend according to branch measurement values; when the suboptimal paths is selected, selection results of the preferred paths are adopted to limit the suboptimal paths in a certain range, and therefore, processing on extended paths beyond the range is omitted. With lower complexity, the method provided in invention can complete the signal detection of the high-order M-QAM modulated MIMO system with detection performance guaranteed.
Description
Technical Field
The invention relates to the technical field of wireless communication, in particular to a signal detection method of a high-order modulation multiple-input multiple-output (MIMO) spatial multiplexing system in the technical field of multiple antennas. The invention can complete the signal detection of the high-order M-QAM modulation MIMO system with lower complexity under the condition of ensuring the detection performance.
Background
The structure of the MIMO communication system is an important means for realizing high-frequency spectrum utilization rate, high-speed rate and high-reliability data transmission of a future mobile communication system. In the MIMO communication process, since multiple channels of transmission data are transmitted in parallel on different antennas, and transmission signals of different transmitting antennas interfere with each other, a more complex detection technique is required to obtain the performance gain caused by the MIMO technique more completely. In recent years, the MIMO system has been researched and found to have a tree structure characteristic, so that it can be detected by a tree search method. The K-best detection in the MIMO system is a tree search method based on breadth first, and the algorithm reduces the complexity by only keeping K expansion paths with the minimum accumulated metric value at each layer in the tree search process. However, with the increase of the modulation order and the number of antennas, the complexity of the conventional K-best detection method is still high, which limits its application to a certain extent.
A patent of "MIMO signal detection method based on breadth-first tree search" applied by the university of electrical science and technology of west ann "(patent application No. 201010577316.9, publication No. CN 102006148A). The patent application mainly provides a MIMO signal detection method based on breadth-first tree search, which utilizes a Schnorr-Euchner enumeration method to sequentially determine a path expansion sequence, an expansion path and path metrics, and sequences the path metrics of the expansion path through a merging and sequencing method to determine a survival path. The patent application has the following disadvantages: the number of new paths expanded by each survivor path is the same as the number of real number constellation points, and under the condition of high-order modulation, more expanded paths need to be processed to complete the detection of signals, so that the detection complexity is very high.
The patent "multiple input multiple output wireless communication data detector" applied by seiko semiconductor integrated technology research and development center limited, su zhou (patent application No. 200910115375.1, publication No. CN 101557281A). The patent application mainly proposes a mimo wireless communication data detector and a mimo wireless communication data detection method. The method introduces a concept of priority into all child nodes of the same father node, mainly inspects the child nodes with high priority, reduces the number of child nodes needing to calculate the branch metric value, and fully utilizes the information contained in the accumulated metric value of the last survivor path in tree search to greatly reduce the calculation of alternative branch paths with low possibility. The patent application has the following disadvantages: in the survivor path sorting unit, it cannot be guaranteed that the selected survivor path is optimal among the extended paths of the corresponding layer, and a part of the extended paths is actually wasted.
Disclosure of Invention
The present invention is directed to overcome the above-mentioned deficiencies in the prior art and to provide a signal detection method in a high-order modulation MIMO system, which can complete signal detection with ideal detection performance and as low complexity as possible in a high-speed MIMO data transmission mode of high-order M-QAM modulation.
The basic idea of the invention is to use the sorted QR decomposition to replace the traditional QR decomposition to realize the triangularization of the real channel matrix; arranging the survivor paths of the same layer according to the ascending order of the accumulated metric values, and pre-sequencing the child nodes of each survivor path according to the ascending order of the branch metric values; dividing the sorted survivor paths of the same layer into two groups, respectively expanding new paths with different numbers for the survivor paths in the two groups, respectively selecting a better path from the expanded paths of the two groups by adopting different selection methods, and then selecting a certain number of optimal paths from the better paths of the two groups as the survivor paths of the index layer.
In order to achieve the above purpose, the invention comprises the following steps:
(1) receiving a signal
1a) Forming a receiving pilot frequency vector and a receiving data vector by signals received by a plurality of antennas of a receiving end;
1b) performing channel estimation according to the received pilot frequency vector to obtain a channel matrix;
(2) pretreatment of
2a) Carrying out real value decomposition on the channel matrix by using a matrix real value decomposition formula to obtain a real number channel matrix;
2b) carrying out real value decomposition on the received data vector by using a vector real value decomposition formula to obtain a real number receiving vector;
2c) determining a real number candidate node set corresponding to a modulation mode by using a real number candidate node set formula according to a D-QAM modulation mode of a transmitted signal;
2d) performing ordered QR decomposition on the real channel matrix obtained in the step 2a) to obtain an upper triangular matrix, an orthogonal matrix and an elementary column exchange matrix;
2e) multiplying the transpose matrix of the orthogonal matrix by the real number receiving vector obtained in the step 2b) to obtain a transformed receiving vector;
2f) initializing a layer index for tracking a layer to be detected to 2Nt, wherein Nt is the number of transmitting antennas.
(3) Bottom most layer detection
3a) Will be provided withUsing the empty path as the survival path of the index layer, and concentrating the real number candidate nodes obtained in step 2c)A candidate node as aboveStarting nodes of the survival paths, wherein D is the number of constellation points modulated by D-QAM;
3b) by applying to the compound described in step 3a)Pre-sorting the initial nodes according to ascending orders of branch metric values to obtain sorting of survival paths corresponding to the initial nodes according to ascending orders of accumulated metric values;
3c) calculating the accumulated metric value of each survivor path in the step 3b) by using an initial metric formula;
3d) and updating the index layer to the adjacent upper layer.
(4) Pre-ordering
In the updated index layer, the survival paths of the adjacent detected layers are taken as the basis, and the survival path corresponds to each survival pathThe sub-nodes are pre-ordered in ascending order of branch metric values.
(5) Path expansion
And (4) sorting the survival paths of the adjacent detected layers, forming the first half of the survival paths into a first survival group, forming the second half of the survival paths into a second survival group, selecting P1 sub nodes with the minimum branch metric value from the sub nodes of each survival path in the first survival group for path expansion by using the sorting of the sub nodes in the step (4), and selecting P2 sub nodes with the minimum branch metric value from the sub nodes of each survival path in the second survival group for path expansion.
(6) Sorting preferred paths
6a) Taking the extension path of each survival path in the first survival group as a path queue, and calculating the accumulated metric value of each extension path in each path queue by using an accumulation method;
6b) every two adjacent path queues form a comparison group, after grouping of all path queues is completed, whether the total number of the comparison groups is 1 or not is judged, if yes, a match-Select-Replace optimal sorting method is used for selecting from only one comparison groupOutputting the paths which are arranged according to the ascending order of the accumulated metric values as better paths, and turning to the step (7); otherwise, go to step 6 c);
6c) in each comparison group, combining two path queues forming the comparison group into a new path queue with the extension paths arranged in ascending order according to the accumulated metric value by utilizing a Compare-Select-Replace optimal sorting method, and then switching to the step 6b) on the basis of the new path queue obtained by each comparison group.
(7) Picking sub-optimal paths
7a) Set a length ofThe sub-optimal path list of any element is not filled in, and the first position of the sub-optimal path list is used as the current index position;
7b) taking the extended path of each survivor path in the second survivor group as a path group; selecting a first extension path as an alternative path of the path group in each path group, calculating an accumulated metric value of the alternative path by using an accumulation method, and if the path group is empty, taking an empty path as the alternative path of the path group;
7c) on the basis of the alternative paths obtained in the step 7b), forming every two adjacent alternative paths into a candidate group, after grouping the alternative paths, storing each candidate group, screening out one alternative path from each candidate group by using an in-group optimization method, and taking the stored candidate group as a first-stage candidate group; the method for selecting the optimal path in the group comprises the steps of counting the number of non-empty alternative paths in a candidate group, analyzing a counting result, comparing the accumulated metric values of the two alternative paths if the number of the non-empty alternative paths in the counting result is 2, and taking the alternative path with a smaller accumulated metric value as the alternative path screened out by the candidate group; if the number of the non-empty alternative paths is 1, taking only one non-empty alternative path as an alternative path screened out by the candidate group; if the number of the non-empty alternative paths is 0, taking an empty path as an alternative path screened out by the candidate group;
7d) repeating the grouping, storing and screening processes in the step 7c) on the basis of the selected alternative paths of each candidate group until the total number of the selected alternative paths is 1, taking only one alternative path as the selected path, and increasing the number of the candidate groups stored each time;
7e) combining the selected path with the path obtained in step (6)Comparing the last path e in the sorted superior paths, judging whether the accumulated metric value of the selected path is greater than or equal to the accumulated metric value of the path e, if so, outputting the path stored in the suboptimum path list as a suboptimum path, and turning to the step (8); otherwise, outputting the selected path to the index position of the suboptimal path list, and updating the index position to the next adjacent position; judging whether the suboptimal path list is filled, if so, outputting the path in the filled suboptimal path list as a suboptimal path, and turning to the step (8), otherwise, turning to the step 7 f);
7f) firstly, updating a path group where the selected path is located, deleting the selected path from a path group q where the selected path is located, secondly, judging whether the updated path group q is empty, if not, taking a first extended path in the updated path group q as an initial updating path, calculating an accumulated metric value of the initial updating path by using an accumulation method, if so, taking an empty path as an initial updating path, and then, sequentially updating all levels of candidate groups stored in the step 7c) and the step 7d) by using a step-by-step replacement method, storing the updating, taking the updating path of the last level as a new selected path, and transferring to the step 7 e); the step-by-step replacement method is that, starting from the first-stage candidate group, in each stage of candidate group, the candidate group where the selected path is located is updated, the selected path in the candidate group is replaced by the updated path of the previous stage, and a candidate path is reselected from the updated candidate group as the updated path of the current stage by the intra-group optimization method in step 7c), and in the first-stage candidate group, the updated path of the starting stage is used as the updated path of the previous stage.
(8) Determining a survival path
Taking the better path obtained in the step (6) as a first path queue, taking the suboptimum path obtained in the step (7) as a second path queue, forming the two path queues into a comparison group, and selecting the comparison group by utilizing a Compare-Select-Replace optimal sorting methodThe optimal path in ascending order of the accumulated metric values is used as the survival path of the index layer.
(9) Updating index layer
And (5) updating the index layer to the adjacent previous layer, checking whether the updated index layer is the 1 st layer, if so, turning to the step (10), otherwise, turning to the step (4).
(10) Topside detection
And (4) and (5) are executed on the uppermost layer 1, the accumulated metric value of the expansion path corresponding to the first child node of each survival path in the two survival groups is calculated by using an accumulation method, an expansion path with the minimum accumulated metric value is selected as an optimal path by using a bubble sorting method, accordingly, an optimal vector with the dimension of 2Nt is obtained, the initial column exchange matrix obtained in the step 2d) is multiplied by the optimal vector, the optimal vector with the dimension of 2Nt is obtained, and complex value combination is carried out on the optimal vector to obtain the detection vector with the dimension of Nt.
(11) And outputting the obtained detection vector.
Compared with the prior art, the invention has the following advantages:
firstly, when the invention selects the suboptimal path, the invention limits the suboptimal path within a certain range by using the selection result of the better path, overcomes the problem that the prior art needs to process more extension paths to complete signal detection during high-order modulation, omits the processing of the extension paths outside the range and further reduces the complexity of signal detection.
Secondly, because the invention selects the better paths meeting certain conditions from the two groups of extension paths of the same layer by adopting different selection methods, and then selects a certain number of optimal paths from the two groups of better paths as the survival paths of the index layer, the defect that the selected survival paths are optimal in the extension paths of the corresponding layer in the prior art can not be overcome, so that a certain number of survival paths obtained by each layer are optimal in the extension paths of the corresponding layer, and the compromise of performance and complexity is better realized.
Drawings
FIG. 1 is a flow chart of the present invention;
FIG. 2 is a simulation effect diagram of the present invention under 64QAM modulation;
fig. 3 is a simulation effect diagram of the present invention under 256QAM modulation.
Detailed Description
The following describes the implementation steps of the present invention in detail with reference to fig. 1.
Step 1, receiving signals
Forming a receiving pilot frequency vector and a receiving data vector by signals received by a plurality of antennas of a receiving end; and performing channel estimation according to the received pilot frequency vector to obtain a channel matrix.
Step 2, pretreatment
And carrying out real value decomposition on the channel matrix by using a matrix real value decomposition formula to obtain a real number channel matrix. The matrix real-valued decomposition formula is:
wherein, H' is a real channel matrix, Re (-) represents the operation of the real part, Im (-) represents the operation of taking the imaginary part, -Im (-) represents the operation of taking the negative of the imaginary part, and H is a channel matrix.
And carrying out real value decomposition on the received data vector by using a vector real value decomposition formula to obtain a real number received vector. The vector real-valued decomposition formula is:
where r' is a real received vector, Re (·) represents the real part operation, Im (·) represents the imaginary part operation, and r is a received data vector.
And determining a real number candidate node set corresponding to the modulation mode by using a real number candidate node set formula according to the D-QAM modulation mode of the transmitted signal. The D-QAM modulation modes are 64QAM and 256QAM, and the real number candidate node set formula is as follows:
wherein Ω is a group consisting ofAnd D is the constellation point number modulated by the D-QAM.
And performing sequenced QR decomposition on the real channel matrix to obtain an upper triangular matrix, an orthogonal matrix and an elementary column exchange matrix. And in the QR decomposition process of the real channel matrix by utilizing a Gram-Schmidt orthogonalization method, rearranging column vectors of the real channel matrix, so that the diagonal elements of the upper triangular matrix show an increasing trend from top to bottom after each orthogonalization, and simultaneously recording the column transformation position of the real channel matrix by using an elementary column exchange matrix.
And multiplying the transposed matrix of the orthogonal matrix by the real number receiving vector to obtain a transformed receiving vector. Initializing a layer index for tracking a layer to be detected to 2Nt, wherein Nt is the number of transmitting antennas.
Step 3, bottom layer detection
Will be provided withTaking the strip empty path as a survival path of the index layer, and concentrating the real number candidate nodesA candidate node as aboveAnd D is the constellation point number of the D-QAM modulation.
By pairsAnd pre-sorting the starting nodes according to ascending orders of the branch metric values to obtain the sorting of survival paths corresponding to the starting nodes according to ascending orders of the accumulated metric values. Because each survival path of the layer only comprises one starting node, and the accumulated metric value of the survival path is equal to the branch metric value of the corresponding starting node, the ascending sorting of the starting nodes according to the branch metric values is the ascending sorting of the corresponding survival paths according to the accumulated metric values. Pre-ordering by Suzhou Zhongke semiconductor integration technology research and development center, IncIn the linear pre-ordering method proposed in the patent "mimo wireless communication data detector" (patent application No. 200910115375.1, publication No. CN101557281a), the pre-ordering steps adopted in the present invention are as follows:
step 1, calculating a comparison parameter according to the following formula:
wherein u is a comparison parameter, y2NtTaking values for the 2Nt line of the transformed received vector, i is the layer index, yiTaking values of the ith row of the transformed received vector, RijIs an upper triangular matrixThe ith row and the jth column of (1) element takes on values,the node corresponding to the j layer on the survivor path corresponding to the child node to be sequenced;
step 2, multiplying the absolute value | d | of the diagonal parameter d by a plurality of positive integers to obtain CxI | d |, wherein the diagonal parameter d is the value of the ith row and the ith column element of the upper triangular matrix, i is the layer index,d is the constellation point number modulated by the D-QAM;
step 3, comparing the absolute value | u | of the parameter u withAnd determining the sorting of the nodes to be sorted according to the ascending order of the branch metric values according to the comparison result.
The comparison parameter u ═ y corresponding to the current index layer2NtIf the transmitted signal is modulated by 64QAM, the ascending order of the branch metric values of the 8 starting nodes can be determined by comparing the | u | with { | d |, 2| d |, 3| d |, 4| d |, 5| d |, 6| d | } values according to the comparison result and referring to the following table.
And calculating the accumulated metric value of each survivor path by using the initial metric formula, and updating the index layer to the adjacent upper layer. The starting metric formula is as follows:
wherein A iskAccumulating metric values, y, for the kth survivor path2NtTaking values for the 2Nt line of the transformed received vector, R2NtValues are taken for the 2Nt row and 2Nt column elements of the upper triangular matrix,for the start node of the kth survivor path, | · computation2Indicating a squaring operation.
Step 4, pre-sequencing
In the updated index layer, the survival paths of the adjacent detected layers are taken as the basis, and the survival path corresponds to each survival pathThe sub-nodes are pre-ordered in ascending order of branch metric values. The pre-sorting adopts a linear pre-sorting method proposed by Suzhou Zhongke semiconductor integrated technology research and development center, Inc. in a patent of 'multiple input multiple output wireless communication data detector' (patent application No. 200910115375.1, publication No. CN101557281A), and the pre-sorting steps adopted in the invention are as follows:
step 1, calculating a comparison parameter according to the following formula:
wherein u is a comparison parameter, y2NtTaking values for the 2Nt line of the transformed received vector, i is the layer index, yiTaking values of the ith row of the transformed received vector, RijThe value is taken for the ith row and the jth column element of the upper triangular matrix,the node corresponding to the j layer on the survivor path corresponding to the child node to be sequenced;
step 2, multiplying the absolute value | d | of the diagonal parameter d by a plurality of positive integers to obtain CxI | d |, wherein the diagonal parameter d is the value of the ith row and the ith column element of the upper triangular matrix, i is the layer index,d is the constellation point number modulated by the D-QAM;
step 3, comparing the absolute value | u | of the parameter u withAnd determining the sorting of the nodes to be sorted according to the ascending order of the branch metric values according to the comparison result.
For example, in the 64QAM modulation scheme, the absolute value | u | of the comparison parameter u corresponding to each survivor path of the adjacent detected layer is compared with the values of { | d |, 2| d |, 3| d |, 4| d |, 5| d |, 6| d | }, and the ascending order of the branch metric value of the 8 child nodes corresponding to each survivor path can be determined according to the comparison result and by referring to the following table.
Step 5, path expansion
And (4) sorting the survival paths of the adjacent detected layers, forming the first half of the survival paths into a first survival group, forming the second half of the survival paths into a second survival group, selecting P1 sub nodes with the minimum branch metric value from the sub nodes of each survival path in the first survival group for path expansion by using the sorting of the sub nodes in the step (4), and selecting P2 sub nodes with the minimum branch metric value from the sub nodes of each survival path in the second survival group for path expansion. The value size relationship between P1 and P2 is that P1 is larger than P2. In the two survival groups, the extension paths of each survival path are arranged according to the arrangement sequence of the corresponding child nodes, and because the arrangement sequence of the extension paths of the same survival path according to the ascending sequence of the accumulated metric values is consistent with the arrangement sequence of the corresponding child nodes according to the ascending sequence of the branch metric values, the extension paths of each survival path are arranged according to the ascending sequence of the accumulated metric values.
Step 6, selecting the better path
6a) And taking the extension path of each survivor path in the first survivor group as a path queue, and calculating the accumulated metric value of each extension path in each path queue by using an accumulation method. The steps of the accumulation method are as follows:
step 1, calculating branch metric values of child nodes corresponding to the extended paths by using the following formula:
wherein,is a child nodeBranch metric value of yiTaking values for the ith row of the transformed received vector, i being the layer index, RijTaking values of ith row and jth column elements of the upper triangular matrix, sjIs a child nodeNode, R, corresponding to the j-th layer on the corresponding survivor pathiValues are taken for the ith row and ith column elements of the upper triangular matrix,to expand the child node corresponding to the path, | · non-woven2Representing a squaring operation;
and 2, adding the branch metric value of the child node and the accumulated metric value of the corresponding survivor path to obtain the accumulated metric value of the extension path corresponding to the child node.
6b) Every two adjacent path queues form a comparison group, after grouping of all path queues is completed, whether the total number of the comparison groups is 1 or not is judged, if yes, a match-Select-Replace optimal sorting method is used for selecting from only one comparison groupOutputting the paths which are arranged according to the ascending order of the accumulated metric values as better paths, and turning to the step (7); otherwise, go to step 6 c). The match-Select-Replace optimal sorting method is a method For synchronously selecting and sorting, which is proposed by Haifang Jian et al in the document of 'A Low Complexity Soft-output QRD-M Algorithm For MIMO-OFDMSystems', and a certain number of optimal paths selected by the method are arranged according to the ascending order of accumulated metric values. The following parts-Select-Replace sorting steps are adopted in the invention:
step 1, setting a path list which is not filled with any element, defining the length of the path list to be equal to the number of paths needing to be selected from a comparison group, and taking the first position of the path list as the current index position;
step 2, comparing the first extended path of the two path queues in the comparison group, taking the path with smaller accumulated metric value as the preferred path to output to the index position of the path list, and updating the index position to the next adjacent position; judging whether the path list is filled, if so, taking the path in the filled path list as output, and finishing the match-Select-Replace optimal sorting; otherwise, turning to the step 3;
step 3, updating a path queue n where the preferred path is located, deleting the preferred path from the path queue n where the preferred path is located, judging whether the updated path queue n is empty, if so, sequentially outputting a certain number of extended paths arranged at the front in another non-empty path queue to the vacant positions of the path list according to the sequence of the extended paths until the path list is filled, then taking the filled path in the path list as output, and finishing the match-Select-Replace optimal sequencing; otherwise, turning to the step 2;
6c) in each comparison group, combining two path queues forming the comparison group into a new path queue with an extended path arranged in ascending order according to an accumulated metric value by using the Compare-Select-Replace optimal sorting method in the step 6b), and then switching to the step 6b on the basis of the new path queue obtained by each comparison group;
step 7, selecting the suboptimal path
7a) Set a length ofThe sub-optimal path list of any element is not filled, and the first position of the sub-optimal path list is used as the current index position.
7b) Taking the extended path of each survivor path in the second survivor group as a path group; selecting a first extension path as an alternative path of the path group in each path group, calculating an accumulated metric value of the alternative path by using the accumulation method in the step 6a), and taking an empty path as the alternative path of the path group if the path group is empty;
7c) on the basis of the alternative paths obtained in the step 7b), forming every two adjacent alternative paths into a candidate group, after grouping the alternative paths, storing each candidate group, screening out one alternative path from each candidate group by using an in-group optimization method, and taking the stored candidate group as a first-stage candidate group; the method for selecting the optimal path in the group comprises the steps of counting the number of non-empty alternative paths in a candidate group, analyzing a counting result, comparing the accumulated metric values of the two alternative paths if the number of the non-empty alternative paths in the counting result is 2, and taking the alternative path with a smaller accumulated metric value as the alternative path screened out by the candidate group; if the number of the non-empty alternative paths is 1, taking only one non-empty alternative path as an alternative path screened out by the candidate group; if the number of the non-empty alternative paths is 0, taking an empty path as an alternative path screened out by the candidate group;
7d) repeating the grouping, storing and screening processes in the step 7c) on the basis of the selected alternative paths of each candidate group until the total number of the selected alternative paths is 1, taking only one alternative path as the selected path, and increasing the number of the candidate groups stored each time;
7e) combining the selected path with the path obtained in step (6)Comparing the last path e in the sorted superior paths, judging whether the accumulated metric value of the selected path is greater than or equal to the accumulated metric value of the path e, if so, outputting the path stored in the suboptimum path list as a suboptimum path, and turning to the step (8); otherwise, outputting the selected path to the index position of the suboptimal path list, and updating the index position to the next adjacent position; judging whether the suboptimal path list is filled, if so, outputting the path in the filled suboptimal path list as a suboptimal path, and turning to the step (8), otherwise, turning to the step 7 f);
7f) firstly, updating the subgroup where the selected path is located, deleting the selected path from the subgroup q where the selected path is located, secondly, judging whether the updated subgroup q is empty, if not, taking the first expanded path in the updated subgroup q as an initial updating path, calculating the accumulated metric value of the initial updating path by using the accumulation method in the step 6a), if so, taking an empty path as the initial updating path, thirdly, sequentially updating each level of candidate groups stored in the step 7c) and the step 7d) by using a step-by-step replacement method, storing the updating path selected by the last level of candidate group as a new selected path, and turning to the step 7 e); the step-by-step replacement method is that, starting from the first-stage candidate group, in each stage of candidate group, the candidate group where the selected path is located is updated, the selected path in the candidate group is replaced by the updated path of the previous stage, an alternative path is reselected from the updated candidate group as the updated path of the current stage by using the intra-group optimization method in the step 7c), and in the first-stage candidate group, the updated path is used as the updated path of the previous stage.
In the process of selecting the suboptimal path, if the selected path is output to the suboptimal path list, the subgroup where the selected path is located needs to be updated, and the selected path is deleted from the subgroup where the selected path is located. Because the selected path selected each time has the smallest accumulated metric value in the expansion paths corresponding to each subgroup, once the accumulated metric value of the selected path is not less than the accumulated metric value of the last path in the better path, the processing of the rest expansion paths in each subgroup is stopped, the detection complexity is further reduced, and meanwhile, the selected suboptimum paths are arranged in an ascending order according to the accumulated metric values.
Step 8, determining survival path
Using the sorted superior path obtained in step 6 as a first path queue, using the sorted suboptimum path obtained in step 7 as a second path queue, forming a comparison group by the two path queues, and selecting from the comparison group by using the match-Select-Replace sorting method described in step 6b)The optimal path in ascending order of the accumulated metric values is used as the survival path of the index layer.
Step 9, updating the index layer
Updating the index layer to the adjacent previous layer, checking whether the updated index layer is the layer 1, if so, turning to the step (10), otherwise, turning to the step (4);
step 10, top layer detection
And (3) executing the step (4) and the step (5) on the uppermost layer 1, calculating the accumulated metric value of the expansion path corresponding to the first child node of each survival path in the two survival groups by using the accumulation method in the step 6a), selecting the expansion path with the minimum accumulated metric value as the optimal path by using a bubble sorting method, accordingly obtaining the optimal vector with the dimension of 2Nt, multiplying the initial column exchange matrix obtained by sorted QR decomposition and the optimal vector to obtain the optimal vector with the dimension of 2Nt, and carrying out complex value combination on the optimal vector to obtain the detection vector with the dimension of Nt.
And step 11, outputting the obtained detection vector.
The effects of the present invention will be further described with reference to fig. 2 and 3.
1. Simulation conditions
The multiple input multiple output system of the invention adopts a 4-transmission 4-reception V-BLAST system, the type of the channel is a complex Gaussian channel, three curves in figure 2 are obtained by adopting a 64QAM modulation mode for the transmission signal, and three curves in figure 3 are obtained by adopting a 256QAM modulation mode for the transmission signal.
2. Emulated content
The simulation performance of the method of the invention and the simulation performance of the existing method are compared under 64QAM and 256QAM modulation modes respectively. Under the 64QAM modulation mode, the number K of survivor paths reserved by each layer is 8, in the simulation, the number P1 of the extension paths of each survivor path in the first survivor group is set to be 3, and the number P2 of the extension paths of each survivor path in the second survivor group is set to be 2. Under the 256QAM modulation mode, the number K of survivor paths reserved by each layer is 16, in the simulation, the number P1 of the extension paths of each survivor path in the first survivor group is set to be 4, and the number P2 of the extension paths of each survivor path in the second survivor group is set to be 2.
3. Simulation result
Fig. 2 is a simulation diagram of the symbol error rate of the system of the present invention under 64QAM modulation, and fig. 3 is a simulation diagram of the symbol error rate of the system of the present invention under 256QAM modulation. In fig. 2 and fig. 3, curve a is the system symbol error rate curve of the conventional K-best method, curve B is the system symbol error rate curve of the present invention, and curve C is the system symbol error rate curve of the sorted K-best method. The sequencing K-best method is characterized in that on the basis of the traditional K-best method, the traditional QR decomposition is replaced by the sequencing QR decomposition, and compared with the traditional K-best method, the performance of the sequencing K-best method is improved to a certain extent, but the complexity is slightly improved. As can be seen from the attached figures 2 and 3, the symbol error rate of the system adopting the method of the invention is very close to that of the sorted K-best method and is obviously lower than that of the traditional K-best method.
Based on the simulation conditions and the simulation contents, the complexity contrast table of the method of the invention and the traditional K-best method is as follows:
as can be seen from the above table, the implementation complexity of the present invention is significantly reduced compared to the conventional K-best method. It should be noted that [ a, b ] in the table indicates that the minimum value of the comparison parameter is a and b, and the complexity comparison result in the table does not take into account the slight increase of the complexity of the present invention due to the introduction of the sorted QR decomposition, which is very small compared to the complexity reduced by the present invention as a whole.
It can be seen from the combination of fig. 2 and fig. 3 and the complexity comparison table that the present invention significantly reduces the implementation complexity while ensuring the detection performance, and is suitable for the MIMO high-speed data transmission mode using high-order M-QAM modulation.
Claims (8)
1. The signal detection method in the high-order modulation multiple input multiple output system comprises the following steps:
(1) receiving a signal
1a) Forming a receiving pilot frequency vector and a receiving data vector by signals received by a plurality of antennas of a receiving end;
1b) performing channel estimation according to the received pilot frequency vector to obtain a channel matrix;
(2) pretreatment of
2a) Carrying out real value decomposition on the channel matrix by using a matrix real value decomposition formula to obtain a real number channel matrix;
2b) carrying out real value decomposition on the received data vector by using a vector real value decomposition formula to obtain a real number receiving vector;
2c) determining a real number candidate node set corresponding to a modulation mode by using a real number candidate node set formula according to a D-QAM modulation mode of a transmitted signal;
the D-QAM modulation mode is 64QAM or 256QAM, and the real number candidate node set formula is as follows:
wherein Ω is a group consisting ofA real number candidate node set of real number nodes, wherein D is the number of constellation points modulated by D-QAM;
2d) performing ordered QR decomposition on the real channel matrix obtained in the step 2a) to obtain an upper triangular matrix, an orthogonal matrix and an elementary column exchange matrix;
2e) multiplying the transpose matrix of the orthogonal matrix by the real number receiving vector obtained in the step 2b) to obtain a transformed receiving vector;
2f) initializing a layer index for tracking a layer to be detected to 2Nt, wherein Nt is the number of transmitting antennas;
(3) bottom most layer detection
3a) Will be provided withUsing the empty path as the survival path of the index layer, and concentrating the real number candidate nodes obtained in step 2c)A real number node as aboveStarting nodes of the survival paths, wherein D is the number of constellation points modulated by D-QAM;
3b) by applying to the compound described in step 3a)Pre-sorting the initial nodes according to ascending orders of branch metric values to obtain sorting of survival paths corresponding to the initial nodes according to ascending orders of accumulated metric values;
3c) calculating the accumulated metric value of each survivor path in the step 3b) by using an initial metric formula;
the starting metric formula is as follows:
wherein A iskAccumulating metric values, y, for the kth survivor path2NtTaking values for the 2Nt line of the transformed received vector, R2NtIs the 2Nt row and 2Nt column element value of the upper triangular matrix,for the start node of the kth survivor path, | · computation2Representing a squaring operation;
3d) updating the index layer to the adjacent upper layer;
(4) pre-ordering
In the updated index layer, the survival paths of the adjacent detected layers are taken as the basis, and the survival path corresponds to each survival pathThe sub-nodes are pre-sorted according to ascending order of branch metric values;
(5) path expansion
According to the survival path sequence of the adjacent detected layer, the first half of the survival paths form a first survival group, the second half of the survival paths form a second survival group, each survival path selects P1 child nodes with the minimum branch metric value from the child nodes in the first survival group for path expansion by utilizing the sequence of the child nodes in the step (4), and each survival path selects P2 child nodes with the minimum branch metric value from the child nodes in the second survival group for path expansion;
(6) sorting preferred paths
6a) Taking the extension path of each survival path in the first survival group as a path queue, and calculating the accumulated metric value of each extension path in each path queue by using an accumulation method;
6b) every two adjacent path queues form a comparison group, after grouping of all path queues is completed, whether the total number of the comparison groups is 1 or not is judged, if yes, a match-Select-Replace optimal sorting method is used for selecting from only one comparison groupOutputting the paths which are arranged according to the ascending order of the accumulated metric values as better paths, and turning to the step (7); otherwise, go to step 6 c);
6c) in each comparison group, combining two path queues forming the comparison group into a new path queue with an extended path arranged in ascending order according to the accumulated metric value by utilizing a Compare-Select-Replace optimal sorting method, and then switching to the step 6b on the basis of the new path queue obtained by each comparison group;
(7) picking sub-optimal paths
7a) Set a length ofThe sub-optimal path list of any element is not filled in, and the first position of the sub-optimal path list is used as the current index position;
7b) taking the extended path of each survivor path in the second survivor group as a path group; selecting a first extension path as an alternative path of the path group in each path group, calculating an accumulated metric value of the alternative path by using an accumulation method, and if the path group is empty, taking an empty path as the alternative path of the path group;
7c) on the basis of the alternative paths obtained in the step 7b), forming every two adjacent alternative paths into a candidate group, after grouping the alternative paths, storing each candidate group, screening out one alternative path from each candidate group by using an in-group optimization method, and taking the stored candidate group as a first-stage candidate group; the method for selecting the optimal path in the group comprises the steps of counting the number of non-empty alternative paths in a candidate group, analyzing a counting result, comparing the accumulated metric values of the two alternative paths if the number of the non-empty alternative paths in the counting result is 2, and taking the alternative path with a smaller accumulated metric value as the alternative path screened out by the candidate group; if the number of the non-empty alternative paths is 1, taking only one non-empty alternative path as an alternative path screened out by the candidate group; if the number of the non-empty alternative paths is 0, taking an empty path as an alternative path screened out by the candidate group;
7d) repeating the grouping, storing and screening processes in the step 7c) on the basis of the selected alternative paths of each candidate group until the total number of the selected alternative paths is 1, taking only one alternative path as the selected path, and increasing the number of the candidate groups stored each time;
7e) combining the selected path with the path obtained in step (6)Last path in ordered superior pathse, comparing, judging whether the accumulated metric value of the selected path is greater than or equal to the accumulated metric value of the path e, if so, outputting the path stored in the suboptimal path list as a suboptimal path, and turning to the step (8); otherwise, outputting the selected path to the index position of the suboptimal path list, and updating the index position to the next adjacent position; judging whether the suboptimal path list is filled, if so, outputting the path in the filled suboptimal path list as a suboptimal path, and turning to the step (8), otherwise, turning to the step 7 f);
7f) firstly, updating a path group where the selected path is located, deleting the selected path from a path group q where the selected path is located, secondly, judging whether the updated path group q is empty, if not, taking a first extended path in the updated path group q as an initial updating path, calculating an accumulated metric value of the initial updating path by using an accumulation method, if so, taking an empty path as an initial updating path, and then, sequentially updating all levels of candidate groups stored in the step 7c) and the step 7d) by using a step-by-step replacement method, storing the updating, taking the updating path of the last level as a new selected path, and transferring to the step 7 e); the step-by-step replacement method is that, starting from the first-stage candidate group, in each stage of candidate group, the candidate group where the selected path is located is updated, the selected path in the candidate group is replaced by the updated path of the previous stage, and a candidate path is reselected from the updated candidate group as the updated path of the current stage by the intra-group optimization method in the step 7c), and in the first-stage candidate group, the updated path is used as the updated path of the previous stage;
(8) determining a survival path
Taking the better path obtained in the step (6) as a first path queue, taking the suboptimum path obtained in the step (7) as a second path queue, forming the two path queues into a comparison group, and selecting the comparison group by utilizing a Compare-Select-Replace optimal sorting methodFortunately, the optimal path arranged by the ascending order of the accumulated metric values is used as the index layerStoring the path;
(9) updating index layer
Updating the index layer to the adjacent previous layer, checking whether the updated index layer is the layer 1, if so, turning to the step (10), otherwise, turning to the step (4);
(10) topside detection
Executing the step (4) and the step (5) on the uppermost layer 1, calculating the accumulated metric value of the expansion path corresponding to the first child node of each survival path in the two survival groups by using an accumulation method, selecting the expansion path with the minimum accumulated metric value as an optimal path by using a bubble sorting method, accordingly obtaining an optimal vector with the dimension of 2Nt, multiplying the initial column exchange matrix obtained in the step 2d) with the optimal vector to obtain an optimal vector with the dimension of 2Nt, and carrying out complex value combination on the optimal vector to obtain a detection vector with the dimension of Nt;
(11) and outputting the obtained detection vector.
2. The method as claimed in claim 1, wherein the matrix real-valued decomposition formula in step 2a) is:
wherein, H' is a real channel matrix, Re (-) represents the operation of the real part, Im (-) represents the operation of taking the imaginary part, -Im (-) represents the operation of taking the negative of the imaginary part, and H is a channel matrix.
3. The method as claimed in claim 1, wherein the vector real-valued decomposition formula in step 2b) is:
where r' is a real received vector, Re (·) represents the real part operation, Im (·) represents the imaginary part operation, and r is a received data vector.
4. The method according to claim 1, wherein the QR decomposition of the ordering in step 2d) is performed in synchronization with the QR decomposition by using an upper triangular matrix diagonal element ordering method, and the column vectors of the real channel matrix are rearranged during QR decomposition of the real channel matrix by using Gram-Schmidt orthogonalization method.
5. The method for detecting signals in a high order modulation multiple input multiple output system according to claim 1, wherein the pre-ordering method in step 3b) and step (4) is as follows:
step 1, calculating a comparison parameter according to the following formula:
wherein u is a comparison parameter, y2NtTaking values for the 2Nt line of the transformed received vector, i is the layer index, yiTaking values of the ith row of the transformed received vector, RijThe value is taken for the ith row and the jth column element of the upper triangular matrix,is to be treatedSorting nodes corresponding to the j layer on the survivor paths corresponding to the child nodes;
step 2, multiplying the absolute value | d | of the diagonal parameter d by a plurality of positive integers to obtain CxI | d |, wherein the diagonal parameter d is the value of the ith row and the ith column element of the upper triangular matrix, i is the layer index,d is the constellation point number modulated by the D-QAM;
step 3, comparing the absolute value | u | of the parameter u withAnd determining the node sorting according to the ascending order of the branch metric values according to the comparison result.
6. The method as claimed in claim 1, wherein the values of P1 and P2 in step (5) are such that P1 is greater than P2.
7. The method for detecting signals in a high order modulation multiple input multiple output system according to claim 1, wherein the accumulating method in step 6a), step 7b), step 7f) and step (10) is as follows:
step 1, calculating branch metric values of child nodes corresponding to the extended paths by using the following formula:
wherein,is a child nodeBranch metric value of yiTaking values for the ith row of the transformed received vector, i being the layer index, RijTaking values of ith row and jth column elements of the upper triangular matrix, sjIs a child nodeNode, R, corresponding to the j-th layer on the corresponding survivor pathiValues are taken for the ith row and ith column elements of the upper triangular matrix,to expand the child node corresponding to the path, | · non-woven2Representing a squaring operation;
and 2, adding the branch metric value of the child node and the accumulated metric value of the corresponding survivor path to obtain the accumulated metric value of the extension path corresponding to the child node.
8. The method as claimed in claim 1, wherein the complex-valued combination of step (10) is an inverse process of real-valued decomposition.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210138399.0A CN102724160B (en) | 2012-05-04 | 2012-05-04 | Signal detecting method in high-order modulated multi-input multi-output system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210138399.0A CN102724160B (en) | 2012-05-04 | 2012-05-04 | Signal detecting method in high-order modulated multi-input multi-output system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102724160A CN102724160A (en) | 2012-10-10 |
| CN102724160B true CN102724160B (en) | 2015-03-04 |
Family
ID=46949825
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210138399.0A Expired - Fee Related CN102724160B (en) | 2012-05-04 | 2012-05-04 | Signal detecting method in high-order modulated multi-input multi-output system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102724160B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105281814B (en) * | 2014-07-23 | 2021-03-26 | 锐迪科（重庆）微电子科技有限公司 | Basic odd-even merging grid unit and survival path selection construction method and device |
| WO2016178896A1 (en) * | 2015-05-01 | 2016-11-10 | Graphiclead LLC | System and method for embedding a two dimensional code in video images |
| CN115001546B (en) * | 2022-05-26 | 2023-09-08 | 西安电子科技大学 | MIMO system signal detection method based on parallel K-Best search |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101557281A (en) * | 2009-05-13 | 2009-10-14 | 苏州中科半导体集成技术研发中心有限公司 | Multiple-input multiple-output wireless communication data detector |
| CN101997657A (en) * | 2010-11-03 | 2011-03-30 | 北京邮电大学 | Detection method for breadth-first sphere decoding in MIMO (multiple input multiple output) system |
| CN102006148A (en) * | 2010-12-07 | 2011-04-06 | 西安电子科技大学 | Multiple-input multiple-output (MIMO) signal detection method based on breadth-first tree search |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR100948259B1 (en) * | 2007-12-13 | 2010-03-18 | 한국전자통신연구원 | Mimo antenna receiving apparatus and receiving method |
| CN101572561B (en) * | 2008-04-30 | 2012-09-05 | 华为技术有限公司 | Method and device for detecting signal of multi-input multi-output system |
-
2012
- 2012-05-04 CN CN201210138399.0A patent/CN102724160B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101557281A (en) * | 2009-05-13 | 2009-10-14 | 苏州中科半导体集成技术研发中心有限公司 | Multiple-input multiple-output wireless communication data detector |
| CN101997657A (en) * | 2010-11-03 | 2011-03-30 | 北京邮电大学 | Detection method for breadth-first sphere decoding in MIMO (multiple input multiple output) system |
| CN102006148A (en) * | 2010-12-07 | 2011-04-06 | 西安电子科技大学 | Multiple-input multiple-output (MIMO) signal detection method based on breadth-first tree search |
Non-Patent Citations (1)
| Title |
|---|
| 一种改进的MIMO-OFDM***信号检测算法;陈帅等;《计算机时代》;20110302(第3期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102724160A (en) | 2012-10-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20170288902A1 (en) | Method and Apparatus for Detecting Data in Wireless Communication Networks | |
| Jaldén et al. | The error probability of the fixed-complexity sphere decoder | |
| JP4854378B2 (en) | Wireless transmission system and wireless transmission method | |
| WO2006101093A1 (en) | Mimo multiplex communication apparatus and signal separation method | |
| KR101671263B1 (en) | Advanced method for decoding in the mimo system and apparatus for implementing thereof | |
| CN103532889B (en) | A kind of parallel storehouse MIMO signal detection method of soft output | |
| CN104580039A (en) | Receiver detection method assisted by lattice reduction algorithm and applied to wireless MIMO system | |
| CN102724160B (en) | Signal detecting method in high-order modulated multi-input multi-output system | |
| EP2928084B1 (en) | Method and apparatus for soft output fixed complexity sphere decoding detection | |
| CN109981151A (en) | Improved Gauss tree approximation message transmission detection algorithm in extensive mimo system | |
| CN103188703A (en) | Survival constellation point choosing method and QRM-maximum likehood detection (QRM-MLD) signal detection method | |
| KR101508700B1 (en) | Apparatus and method for detecting signal in multiple input multiple output wireless communication system | |
| CN101557281B (en) | Multiple-input multiple-output wireless communication data detector | |
| CN101997657B (en) | Detection method for breadth-first sphere decoding in MIMO (multiple input multiple output) system | |
| CN101964667B (en) | High-efficiency multi-antenna detection method for long term evolution scheme | |
| CN102307080B (en) | Method and device for detecting serial block signal in MIMO (multiple-input multiple-output) system | |
| CN105119665B (en) | A kind of MIMO detection method based on lattice reduction | |
| CN109660473B (en) | Spherical decoding detection method and device and computer readable storage medium | |
| Ju et al. | VLSI implementation of an 855 Mbps high performance soft-output K-Best MIMO detector | |
| CN103384180B (en) | Performing space-time decoding method and apparatus for multi-input multi-output system | |
| CN102281091B (en) | Reception method for multi-antenna communication system | |
| Wu et al. | Accelerated sphere decoding for multiple-input multiple-output systems using an adaptive statistical threshold | |
| Li et al. | Ordered sphere decoding detection algorithm for MIMO systems | |
| KR20180100000A (en) | Adaptive signal detection method using MIMO-OFDM system and apparatus thereof | |
| KR101348557B1 (en) | Method for detecting signal using mimo-ofdm system and apparatus thereof |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20150304 Termination date: 20200504 |