CN102088750B - Method and device for clustering propagation paths in multiple input multiple output (MIMO) technology - Google Patents

Abstract

The invention discloses a method and a device for clustering propagation paths in a multiple input multiple output (MIMO) technology. The method comprises the following steps: determining the number of candidate clusters and a corresponding clustering result of each candidate cluster; calculating a corresponding evaluation index of each clustering result; and determining an optimum clustering mode based on the corresponding evaluation index of each clustering result and a joint detection method, and taking the corresponding clustering result of the optimum clustering mode as the final clustering result. By utilizing the proposal provided by the invention, the propagation paths can be effectively clustered.

Description

Propagation path clustering method and device in multiple-input multiple-output technology

Technical Field

The present invention relates to mimo technology, and in particular, to a propagation path clustering method and apparatus in mimo technology.

Background

To achieve higher spectral efficiency, Multiple Input Multiple Output (MIMO) technology has become one of the mainstream candidate technologies for next-generation wireless communication systems. Compared with the traditional Single Input Single Output (SISO) technology, the multiple Input multiple Output (mlmo) technology fully utilizes the degree of freedom of a wireless channel in a spatial domain, a time domain and a frequency domain.

In the implementation process of the existing mimo technology, measurement of a wireless channel needs to be performed in an actual geographic environment, so as to obtain important wireless channel research bases, such as channel impulse response and the like, and then large-scale loss, small-scale fading parameters in space, time and frequency domains, statistical characteristics of the large-scale loss and the small-scale fading parameters and the like are obtained through further analysis based on the channel impulse response and the like; then, statistical characteristics and the like are further reasonably characterized, abstracted and modeled, and a wireless channel model is established based on the statistical characteristics and the like.

In an actual electromagnetic propagation environment, due to the existence of scatterers such as trees, buildings and the like, the propagation direction and amplitude of electromagnetic waves can be changed after the electromagnetic waves are contacted with an emitting body. Propagation paths are commonly used to characterize the propagation of electromagnetic waves. The propagation path may be characterized by a multi-dimensional parameter set, which typically includes: power, time delay, arrival angle, departure angle, and the like, each of which characterizes propagation path characteristics in different dimensions. The propagation paths generated by adjacent scatterers or reflectors with matte surfaces have similar parameters. Therefore, propagation paths with similar parameters can be classified into a set called a cluster, the process of classifying different propagation paths into clusters is called clustering, and the statistical average of the parameters of all propagation paths in each cluster is called a cluster center. In the reasonable representation, abstraction and modeling processes, different propagation paths can be clustered, then intra-cluster parameters and inter-cluster parameters of each cluster are respectively obtained based on each cluster, and subsequent processing is carried out based on the obtained intra-cluster parameters and inter-cluster parameters. Therefore, how to accurately cluster directly affects the objectivity and accuracy of the subsequent wireless model establishment, but no known effective clustering method exists in the prior art.

Disclosure of Invention

In view of the above, the main objective of the present invention is to provide a propagation path clustering method in mimo technology, which can implement effective clustering on propagation paths.

Another object of the present invention is to provide a propagation path clustering device in mimo technology, which can realize effective clustering of propagation paths.

In order to achieve the purpose, the technical scheme of the invention is realized as follows:

a propagation path clustering method in the multiple input multiple output technology comprises the following steps:

determining the number of candidate clusters, and determining a clustering result corresponding to each candidate cluster number;

calculating an evaluation index corresponding to each clustering result;

and determining an optimal clustering mode based on a joint detection method according to the evaluation index corresponding to each clustering result, and taking the clustering result corresponding to the optimal clustering mode as a final clustering result.

The determining the clustering result corresponding to each candidate cluster number includes:

for each candidate cluster number K, the following processing is performed:

A. sequencing all propagation paths to be clustered according to a sequence from small to large, and sequentially dividing all the propagation paths to be clustered into K groups;

B. determining the propagation path with the minimum time delay in each group as an original cluster center, and dividing all the original cluster centers into two sets, wherein one set is an effective cluster center set, and the other set is an undetected cluster center set; the effective cluster center set only comprises an original cluster center of a first group, the original cluster center in the effective cluster center set is called an effective cluster center, and the cluster center in the undetected cluster center set is called an undetected cluster center;

C. for each unchecked cluster center X in the set of unchecked cluster centers, the following is performed:

c1, calculating the distance between an unverified cluster center X and each effective cluster center in the effective cluster center set, comparing each calculated distance with a predetermined minimum distance between clusters, if each calculated distance is greater than or equal to the minimum distance between clusters, classifying the unverified cluster center X into the effective cluster center set, and ending the processing aiming at the unverified cluster center X, otherwise, executing the step C2;

c2, selecting the propagation path with the minimum time delay in the group in which the undetected cluster center X is located except the undetected cluster center X;

c3, calculating the distance between the selected propagation path and each effective cluster center in the effective cluster center set, comparing each calculated distance with the minimum distance between the clusters, if each calculated distance is greater than or equal to the minimum distance between the clusters, classifying the selected propagation path into the effective cluster center set, and ending the processing aiming at the undetected cluster center X, otherwise, executing the step C4;

c4, selecting the propagation path with the minimum time delay from the unselected propagation paths in the group where the unchecked cluster center X is located, and returning to execute the step C3;

D. taking the effective cluster centers in the effective cluster center set which is processed according to the mode shown in the step C as initialization cluster centers;

E. and dividing all propagation paths to be clustered into K clusters based on an inter-path normalized orthogonal distance dividing mode according to the initialized cluster center.

The method further comprises the following steps: if all the propagation diameters in the group where the undetected cluster center X is located in the step C do not meet the condition of being included in the effective cluster center set, respectively determining the minimum value in the distances between each propagation diameter and each effective cluster center in the effective cluster center set, selecting the maximum value in the determined minimum values, and including the propagation diameter corresponding to the maximum value in the effective cluster center set.

The step E comprises the following steps:

e1, calculating the distance between the propagation path of each cluster to be clustered and each initialization cluster center, and classifying the propagation path of each cluster to be clustered into the cluster corresponding to the initialization cluster center with the minimum distance to obtain K clusters;

e2, calculating a new cluster center of each cluster obtained in the step E1: <math> <mrow> <msub> <mover> <mi>c</mi> <mo>&OverBar;</mo> </mover> <mi>k</mi> </msub> <mo>=</mo> <mfrac> <mrow> <msubsup> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <msub> <mi>L</mi> <mi>k</mi> </msub> </msubsup> <mrow> <mo>(</mo> <msub> <mi>P</mi> <mi>l</mi> </msub> <mo>&CenterDot;</mo> <msubsup> <mi>X</mi> <mi>l</mi> <mi>k</mi> </msubsup> <mo>)</mo> </mrow> </mrow> <mrow> <msubsup> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <msub> <mi>L</mi> <mi>k</mi> </msub> </msubsup> <msub> <mi>P</mi> <mi>l</mi> </msub> </mrow> </mfrac> <mo>,</mo> </mrow> </math> wherein, X_l ^kRepresents the parameter set, P, corresponding to the ith propagation path in the kth cluster_lEnergy, L, representing the first propagation path_kRepresenting the number of propagation paths in the kth cluster, wherein the value of K is more than or equal to 1 and less than or equal to K; determining whether the distance between the new cluster center of each cluster and the initialized cluster center of the cluster is smaller than a preset threshold value, if so, ending the processing, otherwise, executing the step E3;

e3, calculating the distance between the propagation path of each cluster to be clustered and each new cluster center, and classifying the propagation path of each cluster to be clustered into the cluster corresponding to the new cluster center with the minimum distance to obtain K clusters;

e4, calculating a new cluster center for each newly divided cluster: <math> <mrow> <msub> <mover> <mi>c</mi> <mo>&OverBar;</mo> </mover> <mi>k</mi> </msub> <mo>=</mo> <mfrac> <mrow> <msubsup> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <msub> <mi>L</mi> <mi>k</mi> </msub> </msubsup> <mrow> <mo>(</mo> <msub> <mi>P</mi> <mi>l</mi> </msub> <mo>&CenterDot;</mo> <msubsup> <mi>X</mi> <mi>l</mi> <mi>k</mi> </msubsup> <mo>)</mo> </mrow> </mrow> <mrow> <msubsup> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <msub> <mi>L</mi> <mi>k</mi> </msub> </msubsup> <msub> <mi>P</mi> <mi>l</mi> </msub> </mrow> </mfrac> <mo>,</mo> </mrow> </math> and E, determining whether the distance between the new cluster center of each cluster calculated this time and the new cluster center of the cluster calculated last time is smaller than a preset threshold value, if so, ending the processing, and if not, returning to the step E3.

The method for determining the minimum distance between the clusters comprises the following steps:

calculating the distance between any two propagation paths in all the propagation paths to be clustered, and finding out the maximum distance in the propagation paths;

and dividing the maximum distance by K to obtain the minimum distance between the clusters.

The calculating of the evaluation index corresponding to each clustering result comprises the following steps: and calculating the CH value and the DB value corresponding to each clustering result.

The calculating the CH value corresponding to each clustering result includes:

computing <math> <mrow> <mover> <mi>c</mi> <mo>&OverBar;</mo> </mover> <mo>=</mo> <mfrac> <mrow> <msubsup> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>L</mi> </msubsup> <mrow> <mo>(</mo> <msub> <mi>P</mi> <mi>l</mi> </msub> <mo>&CenterDot;</mo> <msub> <mi>X</mi> <mi>l</mi> </msub> <mo>)</mo> </mrow> </mrow> <mrow> <msubsup> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>L</mi> </msubsup> <msub> <mi>P</mi> <mi>l</mi> </msub> </mrow> </mfrac> <mo>,</mo> </mrow> </math> Wherein, X_lRepresents the parameter set, P, corresponding to the L propagation path in all the L propagation paths to be clustered_lEnergy representing the l-th propagation path;

computing <math> <mrow> <mi>tr</mi> <mrow> <mo>(</mo> <mi>B</mi> <mo>)</mo> </mrow> <mo>=</mo> <munderover> <mi>&Sigma;</mi> <mrow> <mi>k</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>K</mi> </munderover> <msub> <mi>L</mi> <mi>k</mi> </msub> <mo>&CenterDot;</mo> <mi>MD</mi> <msup> <mrow> <mo>(</mo> <msub> <mi>c</mi> <mi>k</mi> </msub> <mo>,</mo> <mover> <mi>c</mi> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> <mn>2</mn> </msup> <mo>,</mo> </mrow> </math> Wherein L is_kIndicates the number of propagation paths in the kth cluster, MD (c)_kC) represents the distance from the initialized cluster center of the kth cluster to c;

computing <math> <mrow> <mi>tr</mi> <mrow> <mo>(</mo> <mi>W</mi> <mo>)</mo> </mrow> <mo>=</mo> <munderover> <mi>&Sigma;</mi> <mrow> <mi>k</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>K</mi> </munderover> <munderover> <mi>&Sigma;</mi> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> </mrow> <msub> <mi>L</mi> <mi>k</mi> </msub> </munderover> <mi>MD</mi> <msup> <mrow> <mo>(</mo> <msubsup> <mi>X</mi> <mi>j</mi> <mi>k</mi> </msubsup> <mo>,</mo> <msub> <mi>c</mi> <mi>k</mi> </msub> <mo>)</mo> </mrow> <mn>2</mn> </msup> <mo>,</mo> </mrow> </math> Wherein, X_j ^kRepresents the parameter set corresponding to the jth propagation path in the kth cluster, MD (X)_j ^k，c_k) Representing the distance from the jth propagation path in the kth cluster to the initialized cluster center of the kth cluster;

computing $\mathrm{CH}\left(K\right)=\frac{\mathrm{tr}\left(B\right)/(K-1)}{\mathrm{tr}\left(W\right)/(L-K)},$ The CH (k) is a CH value corresponding to each clustering result.

The calculating of the DB value corresponding to each clustering result includes:

computing <math> <mrow> <msub> <mi>S</mi> <mi>k</mi> </msub> <mo>=</mo> <mfrac> <mn>1</mn> <msub> <mi>L</mi> <mi>k</mi> </msub> </mfrac> <munderover> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <msub> <mi>L</mi> <mi>k</mi> </msub> </munderover> <mi>MD</mi> <mrow> <mo>(</mo> <msubsup> <mi>X</mi> <mi>l</mi> <mi>k</mi> </msubsup> <mo>,</mo> <msub> <mi>c</mi> <mi>k</mi> </msub> <mo>)</mo> </mrow> <mo>,</mo> </mrow> </math> Wherein L is_kNumber of propagation paths in the kth cluster, X_l ^kRepresents the parameter set corresponding to the l propagation path in the k cluster, MD (X)_l ^k，c_k) Representing the distance from the l propagation path in the k cluster to the initialized cluster center of the k cluster;

calculating d_ij＝MD(c_i，c_j) Wherein, MD (c)_i，c_j) Indicating the distance between the initialization cluster center of the ith cluster and the initialization cluster center of the jth cluster; the values of K, i and j are all more than or equal to 1 and less than or equal to K;

computing <math> <mrow> <msub> <mi>R</mi> <mi>i</mi> </msub> <mo>=</mo> <munder> <mi>max</mi> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> <mo>,</mo> <mo>.</mo> <mo>.</mo> <mo>.</mo> <mo>,</mo> <mi>K</mi> <mo>,</mo> <mi>j</mi> <mo>&NotEqual;</mo> <mi>i</mi> </mrow> </munder> <mo>{</mo> <mfrac> <mrow> <msub> <mi>S</mi> <mi>i</mi> </msub> <mo>+</mo> <msub> <mi>S</mi> <mi>j</mi> </msub> </mrow> <msub> <mi>d</mi> <mi>ij</mi> </msub> </mfrac> <mo>}</mo> <mo>,</mo> </mrow> </math> Wherein max represents taking the maximum value;

computing <math> <mrow> <mi>DB</mi> <mrow> <mo>(</mo> <mi>K</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfrac> <mn>1</mn> <mi>K</mi> </mfrac> <munderover> <mi>&Sigma;</mi> <mrow> <mi>i</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>K</mi> </munderover> <msub> <mi>R</mi> <mi>i</mi> </msub> <mo>,</mo> </mrow> </math> And DB (K) is the DB value corresponding to each clustering result.

The determining of the optimal clustering mode based on the joint detection method comprises the following steps:

determining the minimum value in the DB value corresponding to each clustering result, and multiplying the minimum value by a preset proportionality constant t to obtain the product of the minimum value and the proportionality constant t;

selecting DB values smaller than or equal to the product from DB values corresponding to each clustering result, and classifying the candidate cluster number corresponding to each selected DB value into a set F;

and determining the maximum value in the CH values corresponding to each candidate cluster number in the set F, and determining the candidate cluster number corresponding to the maximum value as the optimal clustering mode.

Respectively forming each dimension parameter of all L propagation paths to be clustered into a parameter set to obtain M parameter sets, wherein M represents the parameter dimension of the propagation paths;

computing the ith parameter set xⁱWith the jth parameter set x^jCorrelation between them <math> <mrow> <msub> <mi>C</mi> <mi>ij</mi> </msub> <mo>=</mo> <mfrac> <mrow> <munderover> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>L</mi> </munderover> <mrow> <mo>(</mo> <msubsup> <mi>x</mi> <mi>l</mi> <mi>i</mi> </msubsup> <mo>-</mo> <mover> <msup> <mi>x</mi> <mi>i</mi> </msup> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> <mrow> <mo>(</mo> <msubsup> <mi>x</mi> <mi>l</mi> <mi>j</mi> </msubsup> <mo>-</mo> <mover> <msup> <mi>x</mi> <mi>j</mi> </msup> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> </mrow> <mrow> <msqrt> <munderover> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>L</mi> </munderover> <msup> <mrow> <mo>(</mo> <msubsup> <mi>x</mi> <mi>l</mi> <mi>i</mi> </msubsup> <mo>-</mo> <mover> <msup> <mi>x</mi> <mi>i</mi> </msup> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> <mn>2</mn> </msup> </msqrt> <msqrt> <munderover> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>L</mi> </munderover> <msup> <mrow> <mo>(</mo> <msubsup> <mi>x</mi> <mi>l</mi> <mi>j</mi> </msubsup> <mo>-</mo> <mover> <msup> <mi>x</mi> <mi>j</mi> </msup> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> <mn>2</mn> </msup> </msqrt> </mrow> </mfrac> <mo>,</mo> </mrow> </math> Wherein x is_l ⁱA set of representations xⁱThe first sample in (1), xⁱA set of representations xⁱAverage value of (1), x_l ^jA set of representations x^jThe first sample in (1), x^jA set of representations x^jAverage value of (d); the values of i and j are both greater than or equal to 1 and less than or equal to M;

constructing a sample covariance matrix R, wherein the element of the ith row and the jth column of the matrix is C_ijTo obtain

Calculating an inverse matrix R of the matrix R^-1；

Calculating the distance d_i′j′＝(X_i′-X_j′)′R^-1(X_i′-X_j′) Wherein X is_i′Parameter set representing the i ' th propagation path or cluster center of the i ' th cluster, symbol ' representing the transpose operation, X_j′A parameter set representing the jth propagation path or cluster center of the jth cluster; when X is present_i′，X_j′When the parameter sets are propagation paths, d is_i′j′Denotes the distance between the propagation paths, when X_i′，X_j′When the parameter sets are cluster centers, d is_i′j′Denotes the distance between the cluster centers when X_i′，X_j′One is the parameter set of the propagation path, and the other is the parameter set of the cluster center, the d_i′j′The distance between the propagation path and the cluster center is shown.

A propagation path clustering device in the multiple input multiple output technology comprises:

the first determining module is used for determining the number of candidate clusters and determining a clustering result corresponding to each candidate cluster number;

the calculation module is used for calculating the evaluation index corresponding to each clustering result;

and the second determining module is used for determining the optimal clustering mode based on a joint detection method according to the evaluation index corresponding to each clustering result, and taking the clustering result corresponding to the optimal clustering mode as the final clustering result.

The first determining module includes:

a first determining unit for determining the number of candidate clusters;

the second determining unit is configured to determine, for each candidate cluster number K, an initialization cluster center corresponding to each candidate cluster number K, and includes:

A. sequencing all propagation paths to be clustered according to a sequence from small to large, and sequentially dividing all the propagation paths to be clustered into K groups;

B. determining the propagation path with the minimum time delay in each group as an original cluster center, and dividing all the original cluster centers into two sets, wherein one set is an effective cluster center set, and the other set is an undetected cluster center set; the effective cluster center set only comprises an original cluster center of a first group, the original cluster center in the effective cluster center set is called an effective cluster center, and the cluster center in the undetected cluster center set is called an undetected cluster center;

C. for each unchecked cluster center X in the set of unchecked cluster centers, the following is performed:

c1, calculating the distance between an unverified cluster center X and each effective cluster center in the effective cluster center set, comparing each calculated distance with a predetermined minimum distance between clusters, if each calculated distance is greater than or equal to the minimum distance between clusters, classifying the unverified cluster center X into the effective cluster center set, and ending the processing aiming at the unverified cluster center X, otherwise, executing the step C2;

c2, selecting the propagation path with the minimum time delay in the group in which the undetected cluster center X is located except the undetected cluster center X;

c3, calculating the distance between the selected propagation path and each effective cluster center in the effective cluster center set, comparing each calculated distance with the minimum distance between the clusters, if each calculated distance is greater than or equal to the minimum distance between the clusters, classifying the selected propagation path into the effective cluster center set, and ending the processing aiming at the undetected cluster center X, otherwise, executing the step C4;

c4, selecting the propagation path with the minimum time delay from the unselected propagation paths in the group where the unchecked cluster center X is located, and returning to execute the step C3;

D. taking the effective cluster centers in the effective cluster center set which is processed according to the mode shown in the step C as initialization cluster centers;

and the third determining unit is used for dividing all propagation paths to be clustered into K clusters based on the inter-path normalized orthogonal distance dividing mode according to the initialized cluster center corresponding to each candidate cluster number K.

The second determining unit is further configured to, if all propagation diameters in the group where the unchecked cluster center X is located in step C do not satisfy the condition of being included in the effective cluster center set, respectively determine a minimum value of distances between each propagation diameter and each effective cluster center in the effective cluster center set, select a maximum value of the determined minimum values, and include the propagation diameter corresponding to the maximum value in the effective cluster center set.

The third determination unit includes:

the first calculating subunit is used for calculating the distance between the propagation path of each cluster to be clustered and each initialization cluster center, and classifying the propagation path of each cluster to be clustered into the cluster corresponding to the initialization cluster center with the minimum distance to obtain K clusters;

a second calculating subunit, configured to calculate a new cluster center for each cluster: <math> <mrow> <msub> <mover> <mi>c</mi> <mo>&OverBar;</mo> </mover> <mi>k</mi> </msub> <mo>=</mo> <mfrac> <mrow> <msubsup> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <msub> <mi>L</mi> <mi>k</mi> </msub> </msubsup> <mrow> <mo>(</mo> <msub> <mi>P</mi> <mi>l</mi> </msub> <mo>&CenterDot;</mo> <msubsup> <mi>X</mi> <mi>l</mi> <mi>k</mi> </msubsup> <mo>)</mo> </mrow> </mrow> <mrow> <msubsup> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <msub> <mi>L</mi> <mi>k</mi> </msub> </msubsup> <msub> <mi>P</mi> <mi>l</mi> </msub> </mrow> </mfrac> <mo>,</mo> </mrow> </math> wherein, X_l ^kRepresents the parameter set, P, corresponding to the ith propagation path in the kth cluster_lEnergy, L, representing the first propagation path_kRepresenting the number of propagation paths in the kth cluster; the value of K is greater than or equal to 1 and less than or equal to K; determining whether the distance between the new cluster center of each cluster and the initialized cluster center of the cluster is smaller than a preset threshold value, if so, ending the processing, and otherwise, informing a third computing subunit to execute the self function;

the third computing subunit is used for computing the distance between the propagation path to be clustered and each new cluster center, classifying the propagation path to be clustered into the cluster corresponding to the new cluster center with the minimum distance to obtain K clusters, and informing the fourth computing subunit to execute the functions of the fourth computing subunit;

the fourth calculating subunit is configured to calculate a new cluster center of each newly divided cluster: <math> <mrow> <msub> <mover> <mi>c</mi> <mo>&OverBar;</mo> </mover> <mi>k</mi> </msub> <mo>=</mo> <mfrac> <mrow> <msubsup> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <msub> <mi>L</mi> <mi>k</mi> </msub> </msubsup> <mrow> <mo>(</mo> <msub> <mi>P</mi> <mi>l</mi> </msub> <mo>&CenterDot;</mo> <msubsup> <mi>X</mi> <mi>l</mi> <mi>k</mi> </msubsup> <mo>)</mo> </mrow> </mrow> <mrow> <msubsup> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <msub> <mi>L</mi> <mi>k</mi> </msub> </msubsup> <msub> <mi>P</mi> <mi>l</mi> </msub> </mrow> </mfrac> <mo>,</mo> </mrow> </math> and determining whether the distance between the new cluster center of each cluster calculated this time and the new cluster center of the cluster calculated last time is smaller than a preset threshold value, if so, ending the processing, and otherwise, informing the third calculating subunit to execute the self function.

The second determining unit calculates the distance between any two propagation paths in all the propagation paths to be clustered, finds out the maximum distance in the propagation paths, divides the maximum distance by K, and takes the division result as the minimum distance between clusters.

The calculation module comprises:

the first calculating unit is used for calculating a CH value corresponding to each clustering result, and comprises the following steps:

computing <math> <mrow> <mover> <mi>c</mi> <mo>&OverBar;</mo> </mover> <mo>=</mo> <mfrac> <mrow> <msubsup> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>L</mi> </msubsup> <mrow> <mo>(</mo> <msub> <mi>P</mi> <mi>l</mi> </msub> <mo>&CenterDot;</mo> <msub> <mi>X</mi> <mi>l</mi> </msub> <mo>)</mo> </mrow> </mrow> <mrow> <msubsup> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>L</mi> </msubsup> <msub> <mi>P</mi> <mi>l</mi> </msub> </mrow> </mfrac> <mo>,</mo> </mrow> </math> Wherein, X_lRepresents the parameter set, P, corresponding to the L propagation path in all the L propagation paths to be clustered_lEnergy representing the l-th propagation path;

computing <math> <mrow> <mi>tr</mi> <mrow> <mo>(</mo> <mi>B</mi> <mo>)</mo> </mrow> <mo>=</mo> <munderover> <mi>&Sigma;</mi> <mrow> <mi>k</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>K</mi> </munderover> <msub> <mi>L</mi> <mi>k</mi> </msub> <mo>&CenterDot;</mo> <mi>MD</mi> <msup> <mrow> <mo>(</mo> <msub> <mi>c</mi> <mi>k</mi> </msub> <mo>,</mo> <mover> <mi>c</mi> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> <mn>2</mn> </msup> <mo>,</mo> </mrow> </math> Wherein L is_kIndicates the number of propagation paths in the kth cluster, MD (c)_kC) represents the distance from the initialized cluster center of the kth cluster to c;

computing <math> <mrow> <mi>tr</mi> <mrow> <mo>(</mo> <mi>W</mi> <mo>)</mo> </mrow> <mo>=</mo> <munderover> <mi>&Sigma;</mi> <mrow> <mi>k</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>K</mi> </munderover> <munderover> <mi>&Sigma;</mi> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> </mrow> <msub> <mi>L</mi> <mi>k</mi> </msub> </munderover> <mi>MD</mi> <msup> <mrow> <mo>(</mo> <msubsup> <mi>X</mi> <mi>j</mi> <mi>k</mi> </msubsup> <mo>,</mo> <msub> <mi>c</mi> <mi>k</mi> </msub> <mo>)</mo> </mrow> <mn>2</mn> </msup> <mo>,</mo> </mrow> </math> Wherein, X_j ^kRepresents the parameter set corresponding to the jth propagation path in the kth cluster, MD (X)_j ^k，c_k) Representing the distance from the jth propagation path in the kth cluster to the initialized cluster center of the kth cluster;

computing $\mathrm{CH}\left(K\right)=\frac{\mathrm{tr}\left(B\right)/(K-1)}{\mathrm{tr}\left(W\right)/(L-K)},$ The CH (K) is a CH value corresponding to each clustering result;

the second calculating unit is used for calculating a DB value corresponding to each clustering result and comprises the following steps:

computing <math> <mrow> <msub> <mi>S</mi> <mi>k</mi> </msub> <mo>=</mo> <mfrac> <mn>1</mn> <msub> <mi>L</mi> <mi>k</mi> </msub> </mfrac> <munderover> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <msub> <mi>L</mi> <mi>k</mi> </msub> </munderover> <mi>MD</mi> <mrow> <mo>(</mo> <msubsup> <mi>X</mi> <mi>l</mi> <mi>k</mi> </msubsup> <mo>,</mo> <msub> <mi>c</mi> <mi>k</mi> </msub> <mo>)</mo> </mrow> <mo>,</mo> </mrow> </math> Wherein L is_kNumber of propagation paths in the kth cluster, X_l ^kRepresents the parameter set corresponding to the l propagation path in the k cluster, MD (X)_l ^k，c_k) Representing the distance from the l propagation path in the k cluster to the initialized cluster center of the k cluster;

calculating d_ij＝MD(c_i，c_j) Wherein, MD (c)_i，c_j) Indicating the distance between the initialization cluster center of the ith cluster and the initialization cluster center of the jth cluster; the values of K, i and j are all more than or equal to 1 and less than or equal to K;

computing <math> <mrow> <msub> <mi>R</mi> <mi>i</mi> </msub> <mo>=</mo> <munder> <mi>max</mi> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> <mo>,</mo> <mo>.</mo> <mo>.</mo> <mo>.</mo> <mo>,</mo> <mi>K</mi> <mo>,</mo> <mi>j</mi> <mo>&NotEqual;</mo> <mi>i</mi> </mrow> </munder> <mo>{</mo> <mfrac> <mrow> <msub> <mi>S</mi> <mi>i</mi> </msub> <mo>+</mo> <msub> <mi>S</mi> <mi>j</mi> </msub> </mrow> <msub> <mi>d</mi> <mi>ij</mi> </msub> </mfrac> <mo>}</mo> <mo>,</mo> </mrow> </math> Wherein max represents taking the maximum value;

computing <math> <mrow> <mi>DB</mi> <mrow> <mo>(</mo> <mi>K</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfrac> <mn>1</mn> <mi>K</mi> </mfrac> <munderover> <mi>&Sigma;</mi> <mrow> <mi>i</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>K</mi> </munderover> <msub> <mi>R</mi> <mi>i</mi> </msub> <mo>,</mo> </mrow> </math> And DB (K) is the DB value corresponding to each clustering result.

The second determining module includes:

a fourth determining unit, configured to determine a minimum value in the DB values corresponding to each clustering result, and multiply the minimum value by a preset proportionality constant t to obtain a product of the minimum value and the preset proportionality constant t; selecting DB values smaller than or equal to the product from DB values corresponding to each clustering result, and classifying the candidate cluster number corresponding to each selected DB value into a set F;

and a fifth determining unit, configured to determine a maximum value in CH values corresponding to each candidate cluster number in the set F, determine the candidate cluster number corresponding to the maximum value as an optimal clustering mode, and use a clustering result corresponding to the optimal clustering mode as a final clustering result.

Therefore, by adopting the technical scheme of the invention, the number of candidate clusters is determined, the clustering result corresponding to each candidate cluster number is determined, the evaluation index corresponding to each clustering result is calculated, finally, the optimal clustering mode is determined based on the joint detection method according to the evaluation index corresponding to each clustering result, and the clustering result corresponding to the optimal clustering mode is taken as the final clustering result, so that an effective propagation path clustering mode is provided.

Drawings

FIG. 1 is a flow chart of an embodiment of the method of the present invention.

FIG. 2 is a schematic diagram of the structure of the device according to the present invention.

Detailed Description

Aiming at the problems in the prior art, the invention provides a brand-new propagation path clustering scheme in the multiple input multiple output technology, namely, firstly determining the number of candidate clusters and determining the clustering result corresponding to each candidate cluster number; then, calculating an evaluation index corresponding to each clustering result; and finally, determining an optimal clustering mode based on a joint detection method according to the evaluation index corresponding to each clustering result, and taking the clustering result corresponding to the optimal clustering mode as a final clustering result.

In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail below with reference to the accompanying drawings by way of examples.

FIG. 1 is a flow chart of an embodiment of the method of the present invention. As shown in fig. 1, the method comprises the following steps:

step 11: and determining the number of candidate clusters.

The number of candidate clusters mentioned here refers to how many clusters all propagation paths to be clustered can be divided into. The specific value of the number of candidate clusters may be determined empirically, or, in the simplest manner, assuming that the number of propagation paths to be clustered is L, each integer in a closed interval composed of 1 to L may be used as the number of candidate clusters. Assume that the closed interval of the number of candidate clusters in this embodiment is [ K ]_min，K_max]。

Step 12: and determining a clustering result corresponding to each candidate cluster number.

In this step, for [ K_min，K_max]Determining the clustering result corresponding to each candidate cluster number K according to the sequence of the values from small to large, wherein the specific determination mode is as follows:

A. and sequencing all the propagation paths to be clustered according to a sequence from small to large, and dividing all the propagation paths to be clustered after sequencing into K groups.

In this embodiment, all the sorted propagation paths to be clustered can be averagely divided into K groups, and if a special condition occurs, for example, the number of all the propagation paths to be clustered is 10, and the value of K is 3, that is, the two propagation paths cannot be evenly divided, the number of the propagation paths in each group after being grouped can be flexibly set, for example, the first group includes 3 propagation paths, the 2 nd group includes 3 propagation paths, and the 3 rd group includes 4 propagation paths; that is, when the division cannot be performed, it is first ensured that the propagation paths in the previous groups are the same in number, and the insufficient propagation paths or the excessive propagation paths are all put into the last group.

B. Determining the propagation path with the minimum time delay in each group as an original cluster center c_k ⁰K, and dividing all original cluster centers into two sets, one set being a valid cluster center set and one set being an unverified cluster center set; wherein the effective cluster center set only comprises c₁ ⁰', and c is included in the set of unchecked clusters₀ ¹′...c_K ⁰'; the original cluster center in the valid cluster center set is called a valid cluster center, and the cluster center in the undetected cluster center set is called an undetected cluster center.

C. For an unverified cluster center c in a set of unverified cluster centers₂ ⁰', the following treatments were performed:

c1, calculating C₂ ⁰' distance to each of the valid cluster centers in the valid cluster center set (initial state, only c is included in the valid cluster center set)₁ ⁰') and advances each calculated distance to a predetermined minimum inter-cluster distanceComparing rows, if each calculated distance is greater than or equal to the minimum distance between clusters, then c₂ ⁰' to a valid set of clusters, and correspondingly, c₂ ⁰' delete from set of unchecked clusters, end for c₂ ⁰' otherwise, step C2 is performed.

The determination mode of the minimum distance between the clusters is as follows: calculating the distance between any two propagation paths in all the propagation paths to be clustered, and finding out the maximum distance in the propagation paths; and dividing the maximum distance by K to obtain the result, namely the minimum distance between clusters.

C2, selection of C₂ ⁰' division by c in packets₂ ⁰' propagation path with minimum external delay.

The propagation paths in each group are arranged in the order of increasing time delay, so the propagation path selected in this step is the adjacent c₂ ⁰' propagation path.

C3, calculating the distance between the selected propagation path and each effective cluster center in the effective cluster center set, comparing each calculated distance with the minimum distance between clusters, if each calculated distance is greater than or equal to the minimum distance between clusters, classifying the selected propagation path into the effective cluster center set, and ending aiming at C₂ ⁰' otherwise, step C4 is performed.

C4, selection of C₂ ⁰' the propagation path with the smallest time delay among the propagation paths not selected in the packet is located, and the step C3 is returned to.

I.e., until a propagation path is found that satisfies the conditions for inclusion in the active cluster center set.

However, if c₂ ⁰If all the propagation paths in the group do not satisfy the condition, the minimum value of the distances between each propagation path and each effective cluster center in the effective cluster center set can be determined respectively, and the determined minimum value is selectedAnd the maximum value in the minimum values puts the propagation path corresponding to the maximum value into the effective cluster center set.

For example, assume that a total of m valid cluster cores are included in the set of valid cluster cores, and c₂ ⁰The grouping includes n propagation paths, so that m distances can be calculated for each propagation path, the minimum value of the m distances is found out, n minimum values are obtained in total, then the maximum value of the n minimum values is found out, and the propagation path corresponding to the maximum value is classified into the effective cluster center set.

Then, aim at c in turn₃ ⁰′、c₄ ⁰′......c_K ⁰' and performing the step C respectively.

D. Taking the effective cluster center in the effective cluster center set processed according to the mode shown in the step C as an initialization cluster center C₁ ⁰...c_K ⁰。

After the processing in step C, the effective cluster center set includes K effective cluster centers, and in this embodiment, the K effective cluster centers are respectively referred to as an initialization cluster center C₁ ⁰...c_K ⁰And respectively correspond to K groups.

E. According to the obtained initialization cluster center c₁ ⁰...c_K ⁰And dividing all propagation paths to be clustered into K clusters based on the inter-path normalized orthogonal distance dividing mode.

The specific implementation of the inter-path normalized orthogonal distance division mode may include:

e1, calculating the distance between the propagation path of each cluster to be clustered and each initialization cluster center, and classifying the propagation path of each cluster to be clustered into the cluster corresponding to the initialization cluster center with the minimum distance to obtain K clusters;

e2, calculating a new cluster center of each cluster obtained in the step E1: <math> <mrow> <msub> <mover> <mi>c</mi> <mo>&OverBar;</mo> </mover> <mi>k</mi> </msub> <mo>=</mo> <mfrac> <mrow> <msubsup> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <msub> <mi>L</mi> <mi>k</mi> </msub> </msubsup> <mrow> <mo>(</mo> <msub> <mi>P</mi> <mi>l</mi> </msub> <mo>&CenterDot;</mo> <msubsup> <mi>X</mi> <mi>l</mi> <mi>k</mi> </msubsup> <mo>)</mo> </mrow> </mrow> <mrow> <msubsup> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <msub> <mi>L</mi> <mi>k</mi> </msub> </msubsup> <msub> <mi>P</mi> <mi>l</mi> </msub> </mrow> </mfrac> <mo>,</mo> </mrow> </math> wherein, X_l ^kRepresents the parameter set, P, corresponding to the ith propagation path in the kth cluster_lEnergy, L, representing the first propagation path_kRepresenting the number of propagation paths in the kth cluster; k is greater than or equal to 1 and less than or equal to K; determining whether the distance between the new cluster center of each cluster and the initialized cluster center of the cluster is smaller than a preset threshold value, if so, ending the processing, otherwise, executing the step E3; how to obtain each parameter is the prior art;

e3, calculating the distance between the propagation path of each cluster to be clustered and each new cluster center, and classifying the propagation path of each cluster to be clustered into the cluster corresponding to the new cluster center with the minimum distance to obtain K clusters;

e4, calculating a new cluster center for each newly divided cluster: <math> <mrow> <msub> <mover> <mi>c</mi> <mo>&OverBar;</mo> </mover> <mi>k</mi> </msub> <mo>=</mo> <mfrac> <mrow> <msubsup> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <msub> <mi>L</mi> <mi>k</mi> </msub> </msubsup> <mrow> <mo>(</mo> <msub> <mi>P</mi> <mi>l</mi> </msub> <mo>&CenterDot;</mo> <msubsup> <mi>X</mi> <mi>l</mi> <mi>k</mi> </msubsup> <mo>)</mo> </mrow> </mrow> <mrow> <msubsup> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <msub> <mi>L</mi> <mi>k</mi> </msub> </msubsup> <msub> <mi>P</mi> <mi>l</mi> </msub> </mrow> </mfrac> <mo>,</mo> </mrow> </math> and E, determining whether the distance between the new cluster center of each cluster calculated this time and the new cluster center of the cluster calculated last time is smaller than a preset threshold value, if so, ending the processing, and if not, returning to the step E3.

Step 13: and calculating the evaluation index corresponding to each clustering result.

After determining the clustering result corresponding to each candidate cluster number, in this step, calculating a CH value and a DB value corresponding to each clustering result, where the CH value is calculated in the following manner:

1) computing <math> <mrow> <mover> <mi>c</mi> <mo>&OverBar;</mo> </mover> <mo>=</mo> <mfrac> <mrow> <msubsup> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>L</mi> </msubsup> <mrow> <mo>(</mo> <msub> <mi>P</mi> <mi>l</mi> </msub> <mo>&CenterDot;</mo> <msub> <mi>X</mi> <mi>l</mi> </msub> <mo>)</mo> </mrow> </mrow> <mrow> <msubsup> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>L</mi> </msubsup> <msub> <mi>P</mi> <mi>l</mi> </msub> </mrow> </mfrac> <mo>,</mo> </mrow> </math> Wherein, X_lRepresents the parameter set, P, corresponding to the L propagation path in all the L propagation paths to be clustered_lEnergy representing the l-th propagation path;

2) computing <math> <mrow> <mi>tr</mi> <mrow> <mo>(</mo> <mi>B</mi> <mo>)</mo> </mrow> <mo>=</mo> <munderover> <mi>&Sigma;</mi> <mrow> <mi>k</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>K</mi> </munderover> <msub> <mi>L</mi> <mi>k</mi> </msub> <mo>&CenterDot;</mo> <mi>MD</mi> <msup> <mrow> <mo>(</mo> <msub> <mi>c</mi> <mi>k</mi> </msub> <mo>,</mo> <mover> <mi>c</mi> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> <mn>2</mn> </msup> <mo>,</mo> </mrow> </math> Wherein L is_kIndicates the number of propagation paths in the kth cluster, MD (c)_kC) represents the distance from the initialized cluster center of the kth cluster to c; c is commonly referred to as a whole cluster core;

3) computing <math> <mrow> <mi>tr</mi> <mrow> <mo>(</mo> <mi>W</mi> <mo>)</mo> </mrow> <mo>=</mo> <munderover> <mi>&Sigma;</mi> <mrow> <mi>k</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>K</mi> </munderover> <munderover> <mi>&Sigma;</mi> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> </mrow> <msub> <mi>L</mi> <mi>k</mi> </msub> </munderover> <mi>MD</mi> <msup> <mrow> <mo>(</mo> <msubsup> <mi>X</mi> <mi>j</mi> <mi>k</mi> </msubsup> <mo>,</mo> <msub> <mi>c</mi> <mi>k</mi> </msub> <mo>)</mo> </mrow> <mn>2</mn> </msup> <mo>,</mo> </mrow> </math> Wherein, X_j ^kRepresents the parameter set corresponding to the jth propagation path in the kth cluster, MD (X)_j ^k，c_k) Representing the distance from the jth propagation path in the kth cluster to the initialized cluster center of the kth cluster;

4) computing $\mathrm{CH}\left(K\right)=\frac{\mathrm{tr}\left(B\right)/(K-1)}{\mathrm{tr}\left(W\right)/(L-K)},$ CH (K) is the CH value corresponding to each clustering result. The DB values are calculated in the following way:

1) computing <math> <mrow> <msub> <mi>S</mi> <mi>k</mi> </msub> <mo>=</mo> <mfrac> <mn>1</mn> <msub> <mi>L</mi> <mi>k</mi> </msub> </mfrac> <munderover> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <msub> <mi>L</mi> <mi>k</mi> </msub> </munderover> <mi>MD</mi> <mrow> <mo>(</mo> <msubsup> <mi>X</mi> <mi>l</mi> <mi>k</mi> </msubsup> <mo>,</mo> <msub> <mi>c</mi> <mi>k</mi> </msub> <mo>)</mo> </mrow> <mo>,</mo> </mrow> </math> Wherein L is_kNumber of propagation paths in the kth cluster, X_l ^kRepresents the parameter set corresponding to the l propagation path in the k cluster, MD (X)_l ^k，c_k) Representing the distance from the l propagation path in the k cluster to the initialized cluster center of the k cluster;

2) calculating d_ij＝MD(c_i，c_j) Wherein, MD (c)_i，c_j) Indicating the distance between the initialization cluster center of the ith cluster and the initialization cluster center of the jth cluster; k. the values of i and j are both greater than or equal to 1 and less than or equal to K;

3) computing <math> <mrow> <msub> <mi>R</mi> <mi>i</mi> </msub> <mo>=</mo> <munder> <mi>max</mi> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> <mo>,</mo> <mo>.</mo> <mo>.</mo> <mo>.</mo> <mo>,</mo> <mi>K</mi> <mo>,</mo> <mi>j</mi> <mo>&NotEqual;</mo> <mi>i</mi> </mrow> </munder> <mo>{</mo> <mfrac> <mrow> <msub> <mi>S</mi> <mi>i</mi> </msub> <mo>+</mo> <msub> <mi>S</mi> <mi>j</mi> </msub> </mrow> <msub> <mi>d</mi> <mi>ij</mi> </msub> </mfrac> <mo>}</mo> <mo>,</mo> </mrow> </math> Wherein max represents taking the maximum value;

4) computing <math> <mrow> <mi>DB</mi> <mrow> <mo>(</mo> <mi>K</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfrac> <mn>1</mn> <mi>K</mi> </mfrac> <munderover> <mi>&Sigma;</mi> <mrow> <mi>i</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>K</mi> </munderover> <msub> <mi>R</mi> <mi>i</mi> </msub> <mo>,</mo> </mrow> </math> DB (K) is the DB value corresponding to each clustering result.

K in this step is the same as K in step 12, and also indicates the number of candidate clusters.

Step 14: and determining an optimal clustering mode based on a joint detection method according to the evaluation index corresponding to each clustering result, and taking the clustering result corresponding to the optimal clustering mode as a final clustering result.

The specific implementation mode of the combined detection method is as follows:

1) determining the minimum value in the DB value corresponding to each clustering result, multiplying the minimum value by a preset proportionality constant t to obtain the product of the minimum value and the proportionality constant t, wherein the value of t can be determined according to actual needs; and selecting DB values less than or equal to the product from DB values corresponding to each clustering result, and classifying the candidate cluster number corresponding to each selected DB value into a set F.

Will soon satisfy <math> <mrow> <mi>DB</mi> <mrow> <mo>(</mo> <msub> <mi>K</mi> <mi>i</mi> </msub> <mo>)</mo> </mrow> <mo>&le;</mo> <mi>t</mi> <mo>&CenterDot;</mo> <munder> <mi>min</mi> <mi>K</mi> </munder> <mo>{</mo> <mi>DB</mi> <mrow> <mo>(</mo> <mi>K</mi> <mo>)</mo> </mrow> <mo>}</mo> </mrow> </math> The number of candidate clusters of (2) is included in the set F, K_iIndicating different candidate cluster numbers.

2) And determining the maximum value in the CH values corresponding to each candidate cluster number in the set F, and determining the candidate cluster number corresponding to the maximum value as the optimal clustering mode.

Namely calculation <math> <mrow> <msub> <mi>K</mi> <mi>opt</mi> </msub> <mo>=</mo> <munder> <mrow> <mi>arg</mi> <mi>max</mi> </mrow> <mrow> <msub> <mi>K</mi> <mi>i</mi> </msub> <mo>&Element;</mo> <mi>F</mi> </mrow> </munder> <mo>{</mo> <mi>CH</mi> <mrow> <mo>(</mo> <msub> <mi>K</mi> <mi>i</mi> </msub> <mo>)</mo> </mrow> <mo>}</mo> <mo>,</mo> </mrow> </math> WhereinDenotes all CH (K)_i) K corresponding to the maximum value of_iValue of, the K_iThe values must be in the set F.

K determined in step 2)_iThe value is the number of candidate clusters corresponding to the optimal clustering mode, and the clustering result clustered according to the optimal clustering mode is used as the final clustering result.

Thus, the flow shown in the method embodiment of the present invention is completed.

In the above embodiment, the distance calculation between propagation paths, the distance calculation between a propagation path and a cluster center, and the distance calculation between a cluster center and a cluster center are involved in many cases, and in the present embodiment, the following calculation method may be adopted:

1) respectively forming a set by all dimension parameters of L propagation paths to be clustered to obtain M parameter sets, wherein M represents the parameter dimension of the propagation paths;

2) computing the ith parameter set xⁱWith the jth parameter set x^jCorrelation between them <math> <mrow> <msub> <mi>C</mi> <mi>ij</mi> </msub> <mo>=</mo> <mfrac> <mrow> <munderover> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>L</mi> </munderover> <mrow> <mo>(</mo> <msubsup> <mi>x</mi> <mi>l</mi> <mi>i</mi> </msubsup> <mo>-</mo> <mover> <msup> <mi>x</mi> <mi>i</mi> </msup> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> <mrow> <mo>(</mo> <msubsup> <mi>x</mi> <mi>l</mi> <mi>j</mi> </msubsup> <mo>-</mo> <mover> <msup> <mi>x</mi> <mi>j</mi> </msup> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> </mrow> <mrow> <msqrt> <munderover> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>L</mi> </munderover> <msup> <mrow> <mo>(</mo> <msubsup> <mi>x</mi> <mi>l</mi> <mi>i</mi> </msubsup> <mo>-</mo> <mover> <msup> <mi>x</mi> <mi>i</mi> </msup> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> <mn>2</mn> </msup> </msqrt> <msqrt> <munderover> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>L</mi> </munderover> <msup> <mrow> <mo>(</mo> <msubsup> <mi>x</mi> <mi>l</mi> <mi>j</mi> </msubsup> <mo>-</mo> <mover> <msup> <mi>x</mi> <mi>j</mi> </msup> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> <mn>2</mn> </msup> </msqrt> </mrow> </mfrac> <mo>,</mo> </mrow> </math> Wherein x is_l ⁱA set of representations xⁱThe first sample, xⁱA set of representations xⁱAverage value of (1), x_l ^jA set of representations x^jThe first sample in (1), x^jA set of representations x^jAverage value of (d); the values of i and j are both greater than or equal to 1 and less than or equal to M;

3) constructing a sample covariance matrix R, wherein the element of the ith row and the jth column of the matrix is C_ijTo obtain

4) Calculating an inverse matrix R of the matrix R^-1；

5) Calculating the distance d_i′j′＝(X_i′-X_j′)′R^-1(X_i′-X_j′) Wherein X is_i′Parameter set representing the i ' th propagation path or cluster center of the i ' th cluster, symbol ' representing the transpose operation, X_j′A parameter set representing the jth propagation path or cluster center of the jth cluster; when X is present_i′，X_j′When the parameter sets are propagation paths, d is_i′j′Denotes the distance between the propagation paths, when X_i′，X_j′When the parameter sets are cluster centers, d is_i′j′Denotes the distance between the cluster centers when X_i′，X_j′One is a set of propagation path parameters and the other isWhen clustering the parameter sets of the heart, d_i′j′The distance between the propagation path and the cluster center is shown.

Subsequently, the inter-cluster parameters and intra-cluster parameters may also be calculated based on the final clustering result determined in the embodiment shown in fig. 1, so as to provide a basis for the establishment of a subsequent wireless channel model.

The specific way in which the inter-cluster parameters and intra-cluster parameters are calculated is well known in the art, e.g.,

calculating the delay spread parameter in the cluster: the delay spread parameter of the kth cluster is <math> <mrow> <msub> <mi>&sigma;</mi> <mrow> <mi>&tau;</mi> <mo>,</mo> <mi>k</mi> </mrow> </msub> <mo>=</mo> <msqrt> <munderover> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <msub> <mi>L</mi> <mi>k</mi> </msub> </munderover> <msup> <mrow> <mo>(</mo> <msub> <mi>&tau;</mi> <mi>l</mi> </msub> <mo>-</mo> <msub> <mi>&mu;</mi> <mrow> <mi>&tau;</mi> <mo>,</mo> <mi>k</mi> </mrow> </msub> <mo>)</mo> </mrow> <mn>2</mn> </msup> <mo>&CenterDot;</mo> <msub> <mi>P</mi> <mi>l</mi> </msub> <mo>/</mo> <munderover> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <msub> <mi>L</mi> <mi>k</mi> </msub> </munderover> <msub> <mi>P</mi> <mi>l</mi> </msub> </msqrt> <mo>,</mo> </mrow> </math> Wherein L is_kNumber of propagation paths in the kth cluster, τ_lTime of the l-th propagation path in the k-th clusterYan, P_lRepresents the power of the propagation path, mu_τ，kRepresenting the mean time delay of the kth cluster, i.e. <math> <mrow> <msub> <mi>&mu;</mi> <mrow> <mi>&tau;</mi> <mo>,</mo> <mi>k</mi> </mrow> </msub> <mo>=</mo> <msqrt> <munderover> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <msub> <mi>L</mi> <mi>k</mi> </msub> </munderover> <msub> <mi>&tau;</mi> <mi>l</mi> </msub> <mo>&CenterDot;</mo> <msub> <mi>P</mi> <mi>l</mi> </msub> <mo>/</mo> <munderover> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <msub> <mi>L</mi> <mi>k</mi> </msub> </munderover> <msub> <mi>P</mi> <mi>l</mi> </msub> </msqrt> <mo>.</mo> </mrow> </math>

Calculating an intra-cluster angle expansion parameter: the angle spread parameter of the kth cluster is <math> <mrow> <msub> <mi>&sigma;</mi> <mrow> <mi>&theta;</mi> <mo>,</mo> <mi>k</mi> </mrow> </msub> <mo>=</mo> <msqrt> <munderover> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <msub> <mi>L</mi> <mi>k</mi> </msub> </munderover> <msup> <mrow> <mo>|</mo> <msub> <mi>&theta;</mi> <mi>l</mi> </msub> <mo>-</mo> <msub> <mi>&mu;</mi> <mrow> <mi>&theta;</mi> <mo>,</mo> <mi>k</mi> </mrow> </msub> <mo>)</mo> <mo>|</mo> </mrow> <mn>2</mn> </msup> <mo>&CenterDot;</mo> <msub> <mi>P</mi> <mi>l</mi> </msub> <mo>/</mo> <munderover> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <msub> <mi>L</mi> <mi>k</mi> </msub> </munderover> <msub> <mi>P</mi> <mi>l</mi> </msub> </msqrt> <mo>,</mo> </mrow> </math> Wherein σ_θ，kDenotes the angular spread of the kth cluster, θ_lRepresenting the angular value, P, of the l-th propagation path in the k-th cluster_lRepresents the power of the propagation path, mu_θ，kMeans of angle representing the kth cluster <math> <mrow> <msub> <mi>&mu;</mi> <mrow> <mi>&theta;</mi> <mo>,</mo> <mi>k</mi> </mrow> </msub> <mo>=</mo> <msqrt> <munderover> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <msub> <mi>L</mi> <mi>k</mi> </msub> </munderover> <msub> <mi>&theta;</mi> <mi>l</mi> </msub> <mo>&CenterDot;</mo> <msub> <mi>P</mi> <mi>l</mi> </msub> <mo>/</mo> <munderover> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <msub> <mi>L</mi> <mi>k</mi> </msub> </munderover> <msub> <mi>P</mi> <mi>l</mi> </msub> </msqrt> <mo>,</mo> </mrow> </math> |θ_l-μ_θ，kI is: <math> <mrow> <mrow> <mo>|</mo> <msub> <mi>&theta;</mi> <mi>l</mi> </msub> <mo>-</mo> <msub> <mi>&mu;</mi> <mrow> <mi>&theta;</mi> <mo>,</mo> <mi>k</mi> </mrow> </msub> <mo>|</mo> </mrow> <mrow> <mo>=</mo> <mfenced open='{' close=''> <mtable> <mtr> <mtd> <mn>2</mn> <mi>&pi;</mi> <mo>+</mo> <mrow> <mo>(</mo> <msub> <mi>&theta;</mi> <mi>l</mi> </msub> <mo>-</mo> <msub> <mi>&mu;</mi> <mrow> <mi>&theta;</mi> <mo>,</mo> <mi>k</mi> </mrow> </msub> <mo>)</mo> </mrow> <mo>,</mo> </mtd> <mtd> <msub> <mi>&theta;</mi> <mi>l</mi> </msub> <mo>-</mo> <msub> <mi>&mu;</mi> <mrow> <mi>&theta;</mi> <mo>,</mo> <mi>k</mi> </mrow> </msub> <mo>&lt;</mo> <mo>-</mo> <mi>&pi;</mi> </mtd> </mtr> <mtr> <mtd> <msub> <mi>&theta;</mi> <mi>l</mi> </msub> <mo>-</mo> <msub> <mi>&mu;</mi> <mrow> <mi>&theta;</mi> <mo>,</mo> <mi>k</mi> </mrow> </msub> <mo>,</mo> </mtd> <mtd> <mo>|</mo> <msub> <mi>&theta;</mi> <mi>l</mi> </msub> <mo>-</mo> <msub> <mi>&mu;</mi> <mrow> <mi>&theta;</mi> <mo>,</mo> <mi>k</mi> </mrow> </msub> <mo>|</mo> <mo>&le;</mo> <mi>&pi;</mi> </mtd> </mtr> <mtr> <mtd> <mo>-</mo> <mn>2</mn> <mi>&pi;</mi> <mo>+</mo> <mrow> <mo>(</mo> <msub> <mi>&theta;</mi> <mi>l</mi> </msub> <mo>-</mo> <msub> <mi>&mu;</mi> <mrow> <mi>&theta;</mi> <mo>,</mo> <mi>k</mi> </mrow> </msub> <mo>)</mo> </mrow> <mo>,</mo> </mtd> <mtd> <msub> <mi>&theta;</mi> <mi>l</mi> </msub> <mo>-</mo> <msub> <mi>&mu;</mi> <mrow> <mi>&theta;</mi> <mo>,</mo> <mi>k</mi> </mrow> </msub> <mo>></mo> <mi>&pi;</mi> </mtd> </mtr> </mtable> </mfenced> <mo>;</mo> </mrow> </mrow> </math>

the intra-cluster angle expansion parameters may further include a horizontal arrival angle expansion parameter, a horizontal departure angle expansion parameter, a vertical arrival angle expansion parameter, and a vertical departure angle expansion, and the specific calculation methods are the same, and only different angle values need to be brought in the calculation process.

Based on the above method, fig. 2 is a schematic view of a composition structure of an embodiment of the apparatus of the present invention. As shown in fig. 2, includes:

a first determining module 21, configured to determine the number of candidate clusters, and determine a clustering result corresponding to each number of candidate clusters;

the calculating module 22 is configured to calculate an evaluation index corresponding to each clustering result;

and a second determining module 23, configured to determine an optimal clustering mode based on a joint detection method according to the evaluation index corresponding to each clustering result, and use the clustering result corresponding to the optimal clustering mode as a final clustering result.

The first determining module 21 may specifically include:

a first determining unit 211, configured to determine the number of candidate clusters;

the second determining unit 222 is configured to determine, for each candidate cluster number K, an initialization cluster center corresponding to each candidate cluster number K, and includes:

A. sequencing all propagation paths to be clustered according to a sequence from small to large, and sequentially dividing all the propagation paths to be clustered into K groups;

B. determining the propagation path with the minimum time delay in each group as an original cluster center, and dividing all the original cluster centers into two sets, wherein one set is an effective cluster center set, and the other set is an undetected cluster center set; the effective cluster center set only comprises an original cluster center of a first group, the original cluster center in the effective cluster center set is called an effective cluster center, and the cluster center in the undetected cluster center set is called an undetected cluster center;

C. for each unchecked cluster center X in the set of unchecked cluster centers, the following is performed:

c1, calculating the distance between the undetected cluster center X and each effective cluster center in the effective cluster center set, comparing each calculated distance with a predetermined minimum distance between clusters, if each calculated distance is greater than or equal to the minimum distance between clusters, classifying the undetected cluster center X into the effective cluster center set, and finishing the processing aiming at the undetected cluster center X, otherwise, executing the step C2;

c2, selecting the propagation path with the minimum time delay in the group in which the undetected cluster center X is located except the undetected cluster center X;

c3, calculating the distance between the selected propagation path and each effective cluster center in the effective cluster center set, comparing each calculated distance with the minimum distance between clusters, if each calculated distance is greater than or equal to the minimum distance between clusters, classifying the selected propagation path into the effective cluster center set, and ending the processing aiming at the undetected cluster center X, otherwise, executing the step C4;

c4, selecting the propagation path with the minimum time delay from the unselected propagation paths in the group in which the unchecked cluster center X is located, and returning to execute the step C3;

D. taking the effective cluster centers in the effective cluster center set which is processed according to the mode shown in the step C as initialization cluster centers;

a third determining unit 213, configured to divide all propagation paths to be clustered into K clusters based on the inter-path normalized orthogonal distance division manner according to the initialization cluster center corresponding to each candidate cluster number K.

The second determining unit 212 may be further configured to, if all propagation paths in the group where the unchecked cluster center X is located in step C do not satisfy the condition of being included in the effective cluster center set, respectively determine a minimum value of distances between each propagation path and each effective cluster center in the effective cluster center set, select a maximum value of the determined minimum values, and include the propagation path corresponding to the maximum value in the effective cluster center set.

In addition, the third determining unit 213 may further include (not shown in the figure for simplicity):

the first calculating subunit is used for calculating the distance between the propagation path of each cluster to be clustered and each initialization cluster center, and classifying the propagation path of each cluster to be clustered into the cluster corresponding to the initialization cluster center with the minimum distance to obtain K clusters;

a second calculating subunit, configured to calculate a new cluster center for each cluster: <math> <mrow> <msub> <mover> <mi>c</mi> <mo>&OverBar;</mo> </mover> <mi>k</mi> </msub> <mo>=</mo> <mfrac> <mrow> <msubsup> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <msub> <mi>L</mi> <mi>k</mi> </msub> </msubsup> <mrow> <mo>(</mo> <msub> <mi>P</mi> <mi>l</mi> </msub> <mo>&CenterDot;</mo> <msubsup> <mi>X</mi> <mi>l</mi> <mi>k</mi> </msubsup> <mo>)</mo> </mrow> </mrow> <mrow> <msubsup> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <msub> <mi>L</mi> <mi>k</mi> </msub> </msubsup> <msub> <mi>P</mi> <mi>l</mi> </msub> </mrow> </mfrac> <mo>,</mo> </mrow> </math> wherein, X_l ^kRepresents the parameter set, P, corresponding to the ith propagation path in the kth cluster_lEnergy, L, representing the first propagation path_kRepresenting the number of propagation paths in the kth cluster; k is greater than or equal to 1 and less than or equal to K; determining whether the distance between the new cluster center of each cluster and the initialized cluster center of the cluster is smaller than a preset threshold value, if so, ending the processing, and otherwise, informing a third computing subunit to execute the self function;

the third calculating subunit is used for calculating the distance between the propagation path to be clustered and each new cluster center, classifying the propagation path to be clustered into the cluster corresponding to the new cluster center with the minimum distance to obtain K clusters, and informing the fourth calculating subunit to execute the self function;

a fourth calculating subunit, configured to calculate a new cluster center of each newly divided cluster: <math> <mrow> <msub> <mover> <mi>c</mi> <mo>&OverBar;</mo> </mover> <mi>k</mi> </msub> <mo>=</mo> <mfrac> <mrow> <msubsup> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <msub> <mi>L</mi> <mi>k</mi> </msub> </msubsup> <mrow> <mo>(</mo> <msub> <mi>P</mi> <mi>l</mi> </msub> <mo>&CenterDot;</mo> <msubsup> <mi>X</mi> <mi>l</mi> <mi>k</mi> </msubsup> <mo>)</mo> </mrow> </mrow> <mrow> <msubsup> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <msub> <mi>L</mi> <mi>k</mi> </msub> </msubsup> <msub> <mi>P</mi> <mi>l</mi> </msub> </mrow> </mfrac> <mo>,</mo> </mrow> </math> and determining whether the distance between the new cluster center of each cluster calculated this time and the new cluster center of the cluster calculated last time is smaller than a preset threshold value, if so, ending the processing, and otherwise, informing a third calculation subunit to execute the self function.

The second determining unit calculates the distance between any two propagation paths in all the propagation paths to be clustered, finds out the maximum distance in the propagation paths, divides the maximum distance by K, and takes the division result as the minimum distance between clusters.

The calculation module 22 may specifically include:

the first calculating unit 221, configured to calculate a CH value corresponding to each clustering result, includes:

computing <math> <mrow> <mover> <mi>c</mi> <mo>&OverBar;</mo> </mover> <mo>=</mo> <mfrac> <mrow> <msubsup> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>L</mi> </msubsup> <mrow> <mo>(</mo> <msub> <mi>P</mi> <mi>l</mi> </msub> <mo>&CenterDot;</mo> <msub> <mi>X</mi> <mi>l</mi> </msub> <mo>)</mo> </mrow> </mrow> <mrow> <msubsup> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>L</mi> </msubsup> <msub> <mi>P</mi> <mi>l</mi> </msub> </mrow> </mfrac> <mo>,</mo> </mrow> </math> Wherein, X_lRepresents the parameter set, P, corresponding to the L propagation path in all the L propagation paths to be clustered_lEnergy representing the l-th propagation path;

computing <math> <mrow> <mi>tr</mi> <mrow> <mo>(</mo> <mi>B</mi> <mo>)</mo> </mrow> <mo>=</mo> <munderover> <mi>&Sigma;</mi> <mrow> <mi>k</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>K</mi> </munderover> <msub> <mi>L</mi> <mi>k</mi> </msub> <mo>&CenterDot;</mo> <mi>MD</mi> <msup> <mrow> <mo>(</mo> <msub> <mi>c</mi> <mi>k</mi> </msub> <mo>,</mo> <mover> <mi>c</mi> <mo>&OverBar;</mo> </mover> <mo>)</mo> </mrow> <mn>2</mn> </msup> <mo>,</mo> </mrow> </math> Wherein L is_kIndicates the number of propagation paths in the kth cluster, MD (c)_kC) represents the distance from the initialized cluster center of the kth cluster to c;

computing <math> <mrow> <mi>tr</mi> <mrow> <mo>(</mo> <mi>W</mi> <mo>)</mo> </mrow> <mo>=</mo> <munderover> <mi>&Sigma;</mi> <mrow> <mi>k</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>K</mi> </munderover> <munderover> <mi>&Sigma;</mi> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> </mrow> <msub> <mi>L</mi> <mi>k</mi> </msub> </munderover> <mi>MD</mi> <msup> <mrow> <mo>(</mo> <msubsup> <mi>X</mi> <mi>j</mi> <mi>k</mi> </msubsup> <mo>,</mo> <msub> <mi>c</mi> <mi>k</mi> </msub> <mo>)</mo> </mrow> <mn>2</mn> </msup> <mo>,</mo> </mrow> </math> Wherein, X_j ^kRepresents the parameter set corresponding to the jth propagation path in the kth cluster, MD (X)_j ^k，c_k) Representing the distance from the jth propagation path in the kth cluster to the initialization cluster center of the kth cluster;

computing $\mathrm{CH}\left(K\right)=\frac{\mathrm{tr}\left(B\right)/(K-1)}{\mathrm{tr}\left(W\right)/(L-K)},$ CH (K) is the CH value corresponding to each clustering result;

the second calculating unit 222 is configured to calculate a DB value corresponding to each clustering result, and includes:

computing <math> <mrow> <msub> <mi>S</mi> <mi>k</mi> </msub> <mo>=</mo> <mfrac> <mn>1</mn> <msub> <mi>L</mi> <mi>k</mi> </msub> </mfrac> <munderover> <mi>&Sigma;</mi> <mrow> <mi>l</mi> <mo>=</mo> <mn>1</mn> </mrow> <msub> <mi>L</mi> <mi>k</mi> </msub> </munderover> <mi>MD</mi> <mrow> <mo>(</mo> <msubsup> <mi>X</mi> <mi>l</mi> <mi>k</mi> </msubsup> <mo>,</mo> <msub> <mi>c</mi> <mi>k</mi> </msub> <mo>)</mo> </mrow> <mo>,</mo> </mrow> </math> Wherein L is_kNumber of propagation paths in the kth cluster, X_l ^kRepresents the parameter set corresponding to the l propagation path in the k cluster, MD (X)_l ^k，c_k) Representing the distance from the l propagation path in the k cluster to the initialized cluster center of the k cluster;

calculating d_ij＝MD(c_i，c_j) Wherein, MD (c)_i，c_j) Indicating the distance between the initialization cluster center of the ith cluster and the initialization cluster center of the jth cluster; k. the values of i and j are both greater than or equal to 1 and less than or equal to K;

computing <math> <mrow> <msub> <mi>R</mi> <mi>i</mi> </msub> <mo>=</mo> <munder> <mi>max</mi> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> <mo>,</mo> <mo>.</mo> <mo>.</mo> <mo>.</mo> <mo>,</mo> <mi>K</mi> <mo>,</mo> <mi>j</mi> <mo>&NotEqual;</mo> <mi>i</mi> </mrow> </munder> <mo>{</mo> <mfrac> <mrow> <msub> <mi>S</mi> <mi>i</mi> </msub> <mo>+</mo> <msub> <mi>S</mi> <mi>j</mi> </msub> </mrow> <msub> <mi>d</mi> <mi>ij</mi> </msub> </mfrac> <mo>}</mo> <mo>,</mo> </mrow> </math> Wherein max represents taking the maximum value;

computing <math> <mrow> <mi>DB</mi> <mrow> <mo>(</mo> <mi>K</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfrac> <mn>1</mn> <mi>K</mi> </mfrac> <munderover> <mi>&Sigma;</mi> <mrow> <mi>i</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>K</mi> </munderover> <msub> <mi>R</mi> <mi>i</mi> </msub> <mo>,</mo> </mrow> </math> DB (K) is the DB value corresponding to each clustering result.

The second determining module 23 may specifically include:

a fourth determining unit 231, configured to determine a minimum value in the DB values corresponding to each clustering result, and multiply the minimum value by a preset proportionality constant t to obtain a product of the minimum value and the preset proportionality constant t; selecting DB values smaller than or equal to the product from DB values corresponding to each clustering result, and classifying the candidate cluster number corresponding to each selected DB value into a set F;

a fifth determining unit 232, configured to determine a maximum value in CH values corresponding to each candidate cluster number in the set F, determine the candidate cluster number corresponding to the maximum value as an optimal clustering mode, and use a clustering result corresponding to the optimal clustering mode as a final clustering result.

For a specific work flow of the apparatus embodiment shown in fig. 2, please refer to the corresponding description in the method embodiment shown in fig. 1, which is not repeated herein.

In summary, the above description is only a preferred embodiment of the present invention, and is not intended to limit the scope of the present invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims (15)

1. A propagation path clustering method in the multiple input multiple output technology is characterized by comprising the following steps:

determining the number of candidate clusters, and determining a clustering result corresponding to each candidate cluster number;

calculating an evaluation index corresponding to each clustering result;

determining an optimal clustering mode based on a joint detection method according to the evaluation index corresponding to each clustering result, and taking the clustering result corresponding to the optimal clustering mode as a final clustering result;

wherein, the determining the clustering result corresponding to each candidate cluster number comprises:

for each candidate cluster number K, the following processing is performed:

A. sequencing all propagation paths to be clustered according to a sequence from small to large, and sequentially dividing all the propagation paths to be clustered into K groups;

B. determining the propagation path with the minimum time delay in each group as an original cluster center, and dividing all the original cluster centers into two sets, wherein one set is an effective cluster center set, and the other set is an undetected cluster center set; the effective cluster center set only comprises an original cluster center of a first group, the original cluster center in the effective cluster center set is called an effective cluster center, and the cluster center in the undetected cluster center set is called an undetected cluster center;

C. for each unchecked cluster center X in the set of unchecked cluster centers, the following is performed:

c1, calculating the distance between an unverified cluster center X and each effective cluster center in the effective cluster center set, comparing each calculated distance with a predetermined minimum distance between clusters, if each calculated distance is greater than or equal to the minimum distance between clusters, classifying the unverified cluster center X into the effective cluster center set, and ending the processing aiming at the unverified cluster center X, otherwise, executing the step C2;

c2, selecting the propagation path with the minimum time delay in the group in which the undetected cluster center X is located except the undetected cluster center X;

c3, calculating the distance between the selected propagation path and each effective cluster center in the effective cluster center set, comparing each calculated distance with the minimum distance between the clusters, if each calculated distance is greater than or equal to the minimum distance between the clusters, classifying the selected propagation path into the effective cluster center set, and ending the processing aiming at the undetected cluster center X, otherwise, executing the step C4;

c4, selecting the propagation path with the minimum time delay from the unselected propagation paths in the group where the unchecked cluster center X is located, and returning to execute the step C3;

D. taking the effective cluster centers in the effective cluster center set which is processed according to the mode shown in the step C as initialization cluster centers;

E. and dividing all propagation paths to be clustered into K clusters based on an inter-path normalized orthogonal distance dividing mode according to the initialized cluster center.

2. The method of claim 1, further comprising: if all the propagation diameters in the group where the undetected cluster center X is located in the step C do not meet the condition of being included in the effective cluster center set, respectively determining the minimum value in the distances between each propagation diameter and each effective cluster center in the effective cluster center set, selecting the maximum value in the determined minimum values, and including the propagation diameter corresponding to the maximum value in the effective cluster center set.

3. The method of claim 1, wherein step E comprises:

e1, calculating the distance between the propagation path of each cluster to be clustered and each initialization cluster center, and classifying the propagation path of each cluster to be clustered into the cluster corresponding to the initialization cluster center with the minimum distance to obtain K clusters;

e2, calculating a new cluster center of each cluster obtained in the step E1:wherein,represents the parameter set, P, corresponding to the ith propagation path in the kth cluster_lEnergy, L, representing the first propagation path_kRepresenting the number of propagation paths in the kth cluster, wherein the value of K is more than or equal to 1 and less than or equal to K; determining whether the distance between the new cluster center of each cluster and the initialized cluster center of the cluster is smaller than a preset threshold value, if so, ending the processing, otherwise, executing the step E3;

e3, calculating the distance between the propagation path of each cluster to be clustered and each new cluster center, and classifying the propagation path of each cluster to be clustered into the cluster corresponding to the new cluster center with the minimum distance to obtain K clusters;

e4, calculating a new cluster center for each newly divided cluster:and E, determining whether the distance between the new cluster center of each cluster calculated this time and the new cluster center of the cluster calculated last time is smaller than a preset threshold value, if so, ending the processing, and if not, returning to the step E3.

4. The method of claim 1, wherein the minimum distance between clusters is determined by:

calculating the distance between any two propagation paths in all the propagation paths to be clustered, and finding out the maximum distance in the propagation paths;

and dividing the maximum distance by K to obtain the minimum distance between the clusters.

5. The method according to claim 1, wherein the calculating the evaluation index corresponding to each clustering result comprises: and calculating the CH value and the DB value corresponding to each clustering result.

6. The method of claim 5, wherein the calculating the CH value corresponding to each clustering result comprises:

computingWherein, X_lRepresents the parameter set, P, corresponding to the L propagation path in all the L propagation paths to be clustered_lEnergy representing the l-th propagation path;

computingWherein L is_kIndicates the number of propagation paths in the kth cluster,indicating the initialization cluster center of the k-th cluster to theThe distance of (d);

computingWherein,represents the parameter set corresponding to the jth propagation path in the kth cluster,representing the distance from the jth propagation path in the kth cluster to the initialized cluster center of the kth cluster;

computingThe CH (k) is a CH value corresponding to each clustering result.

7. The method of claim 5, wherein the calculating the DB value corresponding to each clustering result comprises:

computingWherein L is_kIndicates the number of propagation paths in the kth cluster,represents the parameter set corresponding to the l propagation path in the k cluster,representing the distance from the l propagation path in the k cluster to the initialized cluster center of the k cluster;

calculating d_ij＝MD(c_i,c_j) Wherein, MD (c)_i,c_j) Indicating the distance between the initialization cluster center of the ith cluster and the initialization cluster center of the jth cluster; the values of K, i and j are all more than or equal to 1 and less than or equal to K;

computingWherein max represents taking the maximum value;

computingAnd DB (K) is the DB value corresponding to each clustering result.

8. The method of claim 5, wherein determining the optimal clustering based on the joint detection method comprises:

determining the minimum value in the DB value corresponding to each clustering result, and multiplying the minimum value by a preset proportionality constant t to obtain the product of the minimum value and the proportionality constant t;

selecting DB values smaller than or equal to the product from DB values corresponding to each clustering result, and classifying the candidate cluster number corresponding to each selected DB value into a set F;

and determining the maximum value in the CH values corresponding to each candidate cluster number in the set F, and determining the candidate cluster number corresponding to the maximum value as the optimal clustering mode.

9. The method according to any one of claims 1 to 8,

respectively forming each dimension parameter of all L propagation paths to be clustered into a parameter set to obtain M parameter sets, wherein M represents the parameter dimension of the propagation paths;

computing the ith parameter set xⁱWith the jth parameter set x^jCorrelation between themWherein,a set of representations xⁱThe number i of the samples in (1),a set of representations xⁱIs determined by the average value of (a) of (b),a set of representations x^jThe number i of the samples in (1),a set of representations x^jAverage value of (d); the values of i and j are both greater than or equal to 1 and less than or equal to M;

constructing a sample covariance matrix R, wherein the element of the ith row and the jth column of the matrix is C_ijTo obtain

Calculating an inverse matrix R of the matrix R^-1；

Calculating the distance d_i'j'＝(X_i'-X_j')'R^-1(X_i'-X_j') Wherein X is_i'Parameter set representing the i ' th propagation path or cluster center of the i ' th cluster, symbol ' representing the transpose operation, X_j'A parameter set representing the jth propagation path or cluster center of the jth cluster; when X is present_i',X_j'When the parameter sets are propagation paths, d is_i'j'Denotes the distance between the propagation paths, when X_i',X_j'When the parameter sets are cluster centers, d is_i'j'Denotes the distance between the cluster centers when X_i',X_j'One is the parameter set of the propagation path, and the other is the parameter set of the cluster center, the d_i'j'The distance between the propagation path and the cluster center is shown.

10. A propagation path clustering device in a multiple-input multiple-output (MIMO) technology, comprising:

the first determining module is used for determining the number of candidate clusters and determining a clustering result corresponding to each candidate cluster number;

the calculation module is used for calculating the evaluation index corresponding to each clustering result;

the second determining module is used for determining an optimal clustering mode based on a joint detection method according to the evaluation index corresponding to each clustering result, and taking the clustering result corresponding to the optimal clustering mode as a final clustering result;

wherein the first determining module comprises:

a first determining unit for determining the number of candidate clusters;

the second determining unit is configured to determine, for each candidate cluster number K, an initialization cluster center corresponding to each candidate cluster number K, and includes:

A. sequencing all propagation paths to be clustered according to a sequence from small to large, and sequentially dividing all the propagation paths to be clustered into K groups;

B. determining the propagation path with the minimum time delay in each group as an original cluster center, and dividing all the original cluster centers into two sets, wherein one set is an effective cluster center set, and the other set is an undetected cluster center set; the effective cluster center set only comprises an original cluster center of a first group, the original cluster center in the effective cluster center set is called an effective cluster center, and the cluster center in the undetected cluster center set is called an undetected cluster center;

C. for each unchecked cluster center X in the set of unchecked cluster centers, the following is performed:

c1, calculating the distance between an unverified cluster center X and each effective cluster center in the effective cluster center set, comparing each calculated distance with a predetermined minimum distance between clusters, if each calculated distance is greater than or equal to the minimum distance between clusters, classifying the unverified cluster center X into the effective cluster center set, and ending the processing aiming at the unverified cluster center X, otherwise, executing the step C2;

c2, selecting the propagation path with the minimum time delay in the group in which the undetected cluster center X is located except the undetected cluster center X;

c3, calculating the distance between the selected propagation path and each effective cluster center in the effective cluster center set, comparing each calculated distance with the minimum distance between the clusters, if each calculated distance is greater than or equal to the minimum distance between the clusters, classifying the selected propagation path into the effective cluster center set, and ending the processing aiming at the undetected cluster center X, otherwise, executing the step C4;

c4, selecting the propagation path with the minimum time delay from the unselected propagation paths in the group where the unchecked cluster center X is located, and returning to execute the step C3;

D. taking the effective cluster centers in the effective cluster center set which is processed according to the mode shown in the step C as initialization cluster centers;

and the third determining unit is used for dividing all propagation paths to be clustered into K clusters based on the inter-path normalized orthogonal distance dividing mode according to the initialized cluster center corresponding to each candidate cluster number K.

11. The apparatus according to claim 10, wherein the second determining unit is further configured to, if all propagation paths in the group where the unchecked cluster center X is located in step C do not satisfy the condition of being included in the valid cluster center set, respectively determine a minimum value of distances between each propagation path and each valid cluster center in the valid cluster center set, select a maximum value of the determined minimum values, and include the propagation path corresponding to the maximum value in the valid cluster center set.

12. The apparatus according to claim 10 or 11, wherein the third determining unit comprises:

the first calculating subunit is used for calculating the distance between the propagation path of each cluster to be clustered and each initialization cluster center, and classifying the propagation path of each cluster to be clustered into the cluster corresponding to the initialization cluster center with the minimum distance to obtain K clusters;

a second calculating subunit, configured to calculate a new cluster center for each cluster:wherein,represents the parameter set, P, corresponding to the ith propagation path in the kth cluster_lEnergy, L, representing the first propagation path_kRepresenting the number of propagation paths in the kth cluster; the value of K is greater than or equal to 1 and less than or equal to K; determining whether the distance between the new cluster center of each cluster and the initialized cluster center of the cluster is smaller than a preset threshold value, if so, ending the processing, and otherwise, informing a third computing subunit to execute the self function;

the third computing subunit is used for computing the distance between the propagation path to be clustered and each new cluster center, classifying the propagation path to be clustered into the cluster corresponding to the new cluster center with the minimum distance to obtain K clusters, and informing the fourth computing subunit to execute the functions of the fourth computing subunit;

the fourth calculating subunit is configured to calculate a new cluster center of each newly divided cluster:and determining whether the distance between the new cluster center of each cluster calculated this time and the new cluster center of the cluster calculated last time is smaller than a preset threshold value, if so, ending the processing, and otherwise, informing the third calculating subunit to execute the self function.

13. The apparatus according to claim 10 or 11, wherein the second determining unit calculates a distance between any every two propagation paths among all the propagation paths to be clustered, and finds a maximum distance therein, and divides the maximum distance by K, taking a result of the division as a minimum distance between clusters.

14. The apparatus of claim 10 or 11, wherein the computing module comprises:

the first calculating unit is used for calculating a CH value corresponding to each clustering result, and comprises the following steps:

computingWherein, X_lRepresents the parameter set, P, corresponding to the L propagation path in all the L propagation paths to be clustered_lEnergy representing the l-th propagation path;

computingWherein L is_kIndicates the number of propagation paths in the kth cluster,indicating the initialization cluster center of the k-th cluster to theThe distance of (d);

computingWherein,represents the parameter set corresponding to the jth propagation path in the kth cluster,representing the distance from the jth propagation path in the kth cluster to the initialized cluster center of the kth cluster;

computingThe CH (K) is a CH value corresponding to each clustering result;

the second calculating unit is used for calculating a DB value corresponding to each clustering result and comprises the following steps:

computingWherein L is_kIndicates the number of propagation paths in the kth cluster,represents the parameter set corresponding to the l propagation path in the k cluster,representing the distance from the l propagation path in the k cluster to the initialized cluster center of the k cluster;

calculating d_ij＝MD(c_i,c_j) Wherein, MD (c)_i,c_j) Indicating the distance between the initialization cluster center of the ith cluster and the initialization cluster center of the jth cluster; the values of K, i and j are all more than or equal to 1 and less than or equal to K;

computingWherein max represents taking the maximum value;

computingAnd DB (K) is the DB value corresponding to each clustering result.

15. The apparatus of claim 14, wherein the second determining module comprises:

a fourth determining unit, configured to determine a minimum value in the DB values corresponding to each clustering result, and multiply the minimum value by a preset proportionality constant t to obtain a product of the minimum value and the preset proportionality constant t; selecting DB values smaller than or equal to the product from DB values corresponding to each clustering result, and classifying the candidate cluster number corresponding to each selected DB value into a set F;

and a fifth determining unit, configured to determine a maximum value in CH values corresponding to each candidate cluster number in the set F, determine the candidate cluster number corresponding to the maximum value as an optimal clustering mode, and use a clustering result corresponding to the optimal clustering mode as a final clustering result.