JP2007300211A - Wireless communication device and time space clustering method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To properly perform time space clustering with comparatively small computational quantity. <P>SOLUTION: A delay time range detection section 1033 detects a correlation time width for each peak of a result of a correlation arithmetic operation as a delay time range, and an arrival angular range detection section 1035 detects a correlation angular width for each peak of a result of the correlation arithmetic operation as an arrival angular range. An initial cluster generating section 1036 generates an initial cluster whose cluster range is a range surrounded by the delay time range and the arrival angular range. A cluster judgment write section 1051 classifies arrival paths to the initial clusters including delay time estimate values and arrival direction estimate values of the arrival paths estimated by a superresolution estimate section 104. Further, correlation path detection sections 1061-1 to 1061-M use signals with high signal correlation coefficients of the arrival paths classified in one and same initial cluster for a new set to perform cluster classification. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a wireless communication apparatus and a spatiotemporal clustering method.

  2. Description of the Related Art In recent years, in a wireless communication system, in order to further increase subscriber capacity and increase transmission speed, space-time signal processing techniques represented by AAA (Adaptive Array Antenna) and MIMO (Multiple Input Multiple Output) have been applied. It is being considered. Here, in order to accurately evaluate these techniques, a spatio-temporal path model that can simultaneously handle the time direction and the arrival direction of radio waves arriving at the base station is essential. In order to propose this model, it is necessary to analyze the spatio-temporal multipath propagation path in the real environment in detail. However, since the spatio-temporal multipath parameters obtained by experiments are specific to each measurement environment, not all statistical parameters are subjected to statistical processing, but elementary waves with similar spatio-temporal propagation characteristics are recognized as clusters. Thus, a method of performing statistical processing is used. FIG. 15 shows a conceptual diagram of a cluster.

  If such a cluster analysis is introduced in a wireless communication apparatus, it is possible to achieve processing such as space-time multiplexing of transmission signals for each block of paths during transmission. Therefore, it is considered that cluster analysis will become an increasingly important technology for wireless communication devices that perform space-time signal processing.

  Conventionally, this cluster analysis is generally performed empirically using a method such as visual observation (see Non-Patent Document 1).

On the other hand, Patent Document 1 discloses a method of performing an incoming wave automatic clustering process so as to be close to visual determination.
Kei Sakaguchi, Junichi Takada "MIMO Propagation Measurement Equipment, Measurement Method, Analysis Method, Modeling", IEICE Transactions, VOL.J88-B, 2005-9. Japanese Patent Laid-Open No. 11-248814

  However, the above-described automatic clustering process accumulates estimated values from several measurement results and performs statistical processing, and there is a problem that it takes time until the results are obtained. Furthermore, since the clustering process is performed only with the arrival direction information without considering the delay time information, there is a problem that it cannot be applied when analyzing the spatio-temporal multipath propagation path.

  The present invention has been made in view of this point, and an object of the present invention is to provide a radio communication apparatus and a spatiotemporal clustering method capable of accurately performing spatiotemporal clustering with a relatively small amount of calculation.

  In order to solve such a problem, the wireless communication apparatus according to the present invention detects the arrival angle range and delay time range of the received signal, and based on the detection result, collects a set of similar arrival angle ranges and delay time ranges. First cluster generation means for generating a plurality of clusters as one cluster, and super-resolution estimation means for estimating the arrival angle and delay time of the received signal with higher accuracy than the detection accuracy of the first cluster generation means; Cluster classification means for classifying to which cluster the received signal belongs to the cluster generated by the first cluster generation means from the highly accurate arrival angle and delay time estimated by the super-resolution estimation means; The structure to do is taken.

  According to this configuration, the delay time and the arrival angle of the received signal are detected using two methods with different resolution accuracy, and a plurality of clusters are obtained from the delay time range and the arrival angle range calculated using the method with low accuracy. The received signal delay time and arrival angle estimated using a method with high accuracy are generated and the received signal belongs to which of the previously generated clusters is classified. As a result, clusters can be generated with a small amount of computation, and cluster classification can be performed accurately.

  According to the present invention, space-time clustering can be performed accurately with a relatively small amount of computation.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

(Embodiment 1)
FIG. 1 shows a main configuration of radio communication apparatus 100 according to the embodiment of the present invention. 1 includes an array antenna 101-1 to 101-L, a receiving unit 102, a first cluster generation unit 103, a super-resolution estimation unit 104, a cluster classification unit 105, and a second The cluster generation unit 106 is provided.

  Array antennas 101-1 to 101 -L receive a signal transmitted from a communication partner (not shown) and output the received signal to receiving unit 102.

  Receiving section 102 performs radio reception processing (down-conversion, A / D conversion, etc.) for each reception signal output from array antennas 101-1 to 101 -L, converts it to a baseband signal, and converts the baseband signal to 1 to the cluster generation unit 103. That is, a plurality of baseband signals corresponding to array antennas 101-1 to 101 -L are output from receiving section 102 to first cluster generating section 103.

  The first cluster generation unit 103 includes a known signal generation unit 1031, a correlation value calculation unit 1032, a delay time range detection unit 1033, an angle spectrum calculation unit 1034, an arrival angle range detection unit 1035, and an initial cluster generation unit 1036, and a rough initial cluster is generated. Hereinafter, each processing unit will be described.

  Known signal generating section 1031 outputs a predetermined known signal to correlation value calculating section 1032.

  Correlation value calculation section 1032 performs correlation calculations between a plurality of baseband signals output from reception section 102 and known signals, and combines them. That is, a correlation calculation result between a baseband signal corresponding to a signal received by a plurality of array antennas 101-1 to 101 -L and a known signal is synthesized. Correlation value calculation section 1032 outputs the combination result of the correlation calculation results to delay time range detection section 1033 and angle spectrum calculation section 1034.

  The delay time range detection unit 1033 detects the correlation time width for each correlation calculation peak output from the correlation value calculation unit 1032 as a delay time range. Here, the correlation time width means an arrival delay time range of signals that arrive at different times via different paths under the influence of multipath. Hereinafter, this will be specifically described with reference to FIG. FIG. 2 is a diagram illustrating a correlation calculation result in the correlation value calculation unit 1032, where the vertical axis represents the correlation value and the horizontal axis represents the delay time. The correlation time width is, for example, a time range surrounded by adjacent local minimum points including a peak, or a time range including a peak and having a correlation calculation result equal to or greater than a predetermined threshold. In other words, the delay time range detection unit 1033 detects the delay time until the same signal as the signal at which the correlation peak is obtained via a different path as the correlation time width, that is, the delay time range. To do. In this way, the delay time range detection unit 1033 detects the delay time range T1 for the peak P1, detects the delay time range T2 for the peak P2, and detects the delay time range TN for the peak PN. The delay time range T1 to TN is output to the initial cluster generation unit 1036.

  The angle spectrum calculation unit 1034 extracts a peak from the correlation calculation result output from the correlation value calculation unit 1032, calculates an angle spectrum for each extracted peak, and outputs the angle spectrum to the arrival angle range detection unit 1035.

  The arrival angle range detection unit 1035 detects the correlation angle width for each peak of the angle spectrum as the arrival angle range. Here, the correlation angle width means an arrival angle range of signals arriving from different directions via different paths under the influence of multipal. Hereinafter, this will be specifically described with reference to FIG. FIG. 3 is a diagram illustrating the angle spectrum calculated by the angle spectrum calculation unit 1034, where the vertical axis indicates the level of the angle spectrum and the horizontal axis indicates the arrival angle. The correlation angle width is, for example, an angle range surrounded by adjacent local minimum points including a peak, an angle range including a peak or higher than a predetermined level, or an angle range within a predetermined range from the peak. In other words, the arrival angle range detection unit 1035 determines the angle range in which the same signal that arrives from the direction in which the angle spectrum reaches a peak and arrives from a different direction via a different path as a correlation angle width, that is, an arrival angle. Detect as a range. 3 shows an example in which R1 and R2 are detected as arrival angle ranges by the arrival angle range detection unit 1035 for the peak P1 shown in FIG.

  The initial cluster generation unit 1036 generates an initial cluster using the delay time range and the arrival angle range. Hereinafter, the generation of the initial cluster will be specifically described with reference to FIG. In FIG. 4, the vertical axis represents the delay time, and the horizontal axis represents the arrival angle. As shown in the figure, the initial cluster generation unit 1036 clusters an area surrounded by T1 and R1 when T1 is detected as a delay time range for the peak P1 and R1 and R2 are detected as arrival angle ranges. An initial cluster 1 (hereinafter also referred to as “cluster 1”) as a range is generated, and an initial cluster 2 (hereinafter also referred to as “cluster 2”) whose area is surrounded by T1 and R2 is generated. Then, an initial cluster M (hereinafter also referred to as “cluster M”) is generated in which the area surrounded by the delay time range detected for each peak and the arrival angle range is a cluster range.

  In this way, the initial cluster generation unit 1036 generates an initial cluster whose cluster range is a range in which the peak is surely included. Therefore, at this time, even if a peak occurs due to the influence of an interference signal such as spurious, an initial cluster having a range including spurious as a cluster range can be generated. The initial cluster created by the influence is deleted.

  Since the correlation value calculation unit 1032 and the angle spectrum calculation unit 1034 calculate the delay time characteristic and the angle spectrum using a method having a lower resolution than the algorithm used in the super-resolution estimation unit 104 described later, the delay time A range having a certain width as the range and the arrival angle range can be detected. That is, in the present embodiment, the correlation time width and the correlation angle width are estimated using a method having a relatively low resolution, and paths that are correlated with each other arrive in the range surrounded by the correlation time width and the correlation angle width. It was considered to be a range, and an initial cluster with this range as the cluster range was generated first. The procedure until the delay time characteristic and angle spectrum are calculated by the correlation value calculation unit 1032 and the angle spectrum calculation unit 1034 described above is the same method as the beamformer method, and therefore an initial cluster is generated with a small amount of calculation. can do.

  On the other hand, the super-resolution estimation unit 104 uses an algorithm having higher measurement accuracy and resolution than the correlation value calculation unit 1032 and the angle spectrum calculation unit 1034, and uses the baseband signal output from the reception unit 102 to determine the delay time and arrival time. Estimate both directions. The estimation algorithm used in the super-resolution estimation unit 104 estimates both delay time and direction of arrival, as represented by MUSIC (MUltiple SIgnal Classification) and ESPRIT (Estimation of Signal. Parameters via Rotational Invariance Techniques). Use an algorithm that can Algorithms such as MUSIC and ESPRIT have higher measurement accuracy and resolution than the beam former method used in the correlation value calculation unit 1032 and the angle spectrum calculation unit 1034. Super-resolution estimation section 104 outputs the estimated delay time estimation value and arrival direction estimation value to cluster classification section 105.

  The cluster classification unit 105 includes a cluster determination writing unit 1051 and a first stage cluster storage unit 1052. Hereinafter, each processing unit will be described.

  The cluster determination writing unit 1051 arrives at one of the initial clusters 1 to M including the delay time estimation value and the arrival direction estimation value from the delay time estimation value and the arrival direction estimation value estimated by the super-resolution estimation unit 104. Classify paths one by one. The cluster classification method will be described later in detail. Cluster determination writing unit 1051 outputs the classified result to first-stage cluster storage unit 1052. As described above, the first-stage clustering is performed.

  The first-stage cluster storage unit 1052 stores the signal sequence of the classified path for each of the initial clusters 1 to M using the result of classification by the cluster determination writing unit 1051, and stores the stored signal sequence as the initial cluster. Output to the second cluster generation unit 106 every time.

  The second cluster generation unit 106 includes correlation path detection units 1061-1 to 1061-M, a second-stage cluster storage unit 1062, and a cluster information storage unit 1063.

  Correlation path detection units 1061-1 to 1061-M calculate signal correlation coefficients between signal sequences of paths for each initial cluster output from first stage cluster storage unit 1052. For example, consider a case where L paths are classified in the initial cluster 1. At this time, the set of all paths is represented as C: {PA1, PA2,..., PAL}, and the path with the maximum power is PA1. The correlation path detection unit 1061-1 calculates a signal correlation coefficient between PA1 and other PA2, ..., PAL. Hereinafter, the calculation method of the signal correlation coefficient will be described using Expressions (1) to (5).

なお、各シンボルは式（２）〜式（５）で表される。

The correlation matrix (covariance matrix) R _xx of the array antenna is expressed by Expression (1).

Each symbol is expressed by Expression (2) to Expression (5).

When the above equation (1) is modified, the following equation (6) is obtained.

From the above equation (6), the signal correlation matrix S is calculated as in equation (7).

  Using equation (1) to equation (7) described above, correlation path detectors 1061-1 to 1061-M calculate signal correlation coefficients between paths classified into the same initial cluster, and signal correlation coefficients. Using the result, the second-stage clustering process is performed. Specifically, the correlation path detection units 1061-1 to 1061-M classify only paths whose signal correlation coefficient with PA1 is equal to or greater than a predetermined threshold into the same cluster as PA1, and perform the classified path information in the second stage. The data is output to the cluster storage unit 1062. Further, the set of paths whose signal correlation coefficient with PA1 is less than a predetermined threshold is C, and the path with the maximum power among them is PA1, and the above procedure is repeated again to perform clustering processing. Then, the above procedure is repeated until there are no paths whose signal correlation coefficient is less than the predetermined threshold, and the incoming paths are classified into detailed clusters. As described above, the second-stage clustering is performed. Correlation path detection units 1061-1 to 1061-M output the classified path information to second-stage cluster storage unit 1062. When the correlation path detection units 1061-1 to 1061-M perform the second-stage clustering process in parallel, the calculation time required for the second-stage clustering process can be significantly reduced.

  Second-stage cluster storage unit 1062 stores path information for each cluster (hereinafter referred to as “detailed cluster”) classified by the second-stage clustering process, and outputs information on all clusters to cluster information storage unit 1063.

  Next, the clustering operation of the wireless communication apparatus 100 configured as described above will be described.

  First, a signal transmitted from a communication partner (not shown) is received via array antennas 101-1 to 101 -L, and the received signal is output to receiving section 102.

  Then, the reception unit 102 performs radio reception processing (A / D conversion, down-conversion, etc.) on the received signal and converts it into a baseband signal, and the baseband signal is a correlation value calculation unit of the first cluster generation unit 103. 1032 and the super-resolution estimation unit 104.

  Then, the correlation value calculation unit 1032 uses the known signal output from the known signal generation unit 1031 to despread the baseband signal, and the correlation calculation result after despreading is obtained from the delay time range detection unit 1033 and the angle spectrum. It is output to the calculation unit 1034. Then, the angle spectrum calculation unit 1034 calculates an angle spectrum from the correlation calculation result, and the angle spectrum is output to the arrival angle range detection unit 1035.

  Then, the delay time range detection unit 1033 detects the delay time range (T1 to T7 in FIG. 2), and the arrival angle range detection unit 1035 detects the arrival angle range (R1 and R2 in FIG. 3). The delay time range and the arrival angle range are output to the initial cluster generation unit 1036.

  Then, the initial cluster generation unit 1036 generates an initial cluster from the delay time range and the arrival angle range. For example, the initial cluster 1 is created in which the range surrounded by the delay time range T1 and the arrival angle range R1 for the peak P1 is the cluster range. Similarly, an initial cluster 2 is created in which the range surrounded by the delay time range T1 and the arrival angle range R2 is the cluster range. FIG. 4 shows how the initial clusters 1 to 10 are generated by the initial cluster generation unit 1036 in this way.

  On the other hand, the super-resolution estimation unit 104 estimates the arrival direction estimation value and the delay time estimation value using the baseband signal after despreading, and the delay time estimation value and the arrival direction estimation value are determined by the cluster classification unit 105. The data is output to the writing unit 1051. As described above, the estimation method used in the super-resolution estimation unit 104 uses a high-precision and high-resolution algorithm as compared with the beamformer method, as represented by MUSIC, ESPRIT, and the like.

  In the cluster determination writing unit 1051, all paths are classified into any one of the initial clusters 1 to M. Hereinafter, a procedure for classifying the L paths into the initial clusters 1 to M will be described with reference to the flowchart of FIG.

  First, it is determined whether the delay time estimation value and the arrival direction estimation value of path i are included in the cluster range of initial cluster j (ST11). If included, path i is stored in cluster j. (ST12). On the other hand, if it is not included, it is determined whether or not the initial cluster M has been checked (ST13). If there is an initial cluster that has not been checked yet, j is incremented (ST14) and ST11 is determined again. . On the other hand, if it is not included in any of the initial clusters 1 to M, the path i is stored in the cluster in which the closest distance path exists (ST15). The closest distance path is, for example, a path having the smallest sum of the square of the difference between the delay time estimation values and the square of the difference between the arrival direction estimation values.

  Then, it is determined whether or not all L paths are stored in any of the initial clusters (ST16). If there is an unstored path, i is incremented, j is set to 1 (ST17), and all paths are stored in one of the initial clusters 1 to M by the above-described procedure. Then, after all paths are stored in any of the initial clusters 1 to M, the initial cluster having no path is deleted (ST18).

  That is, the initial cluster generated by the initial cluster generation unit 1036 is a cluster range that is surrounded by a delay time range or an arrival angle range detected from a peak due to the influence of interference signals such as spurious included in the received signal. There is a possibility that an initial cluster may be included, but it is generated under the influence of spurious etc. by deleting the initial cluster that does not include the delay time estimation value and arrival direction estimation value estimated by the super-resolution estimation unit 104. The initial cluster can be removed. FIG. 6 shows a state of the initial cluster map after deleting the initial cluster having no path in this way. In FIG. 6, black circles indicate paths specified by the delay time estimated value and arrival direction estimated value estimated by the super-resolution estimating unit 104, and a range surrounded by diagonal lines is indicated by the cluster determination writing unit 1051. Indicates the cluster range of the erased initial cluster.

  Then, information regarding the paths classified into the initial clusters by the cluster determination writing unit 1051 is output to the first-stage cluster storage unit 1052, and the first-stage cluster storage unit 1052 stores the paths classified for each initial cluster. A signal sequence is stored.

  In this way, it is possible to perform the first-stage clustering process of the path in consideration of both the delay time and the arrival direction with a relatively simple procedure.

  Further, the correlation path detection unit 1061-1 in the second cluster generation unit 106 uses the signal correlation coefficient between the signal sequences of the paths stored for each initial cluster in the first-stage cluster storage unit 1052, and uses the signal correlation coefficient. A two-stage clustering process is performed. Hereinafter, the second-stage clustering process will be described with reference to the flowchart of FIG. FIG. 7 shows a procedure for classifying the I paths in the initial cluster 1 into more detailed clusters.

  A set of all paths in the initial cluster 1 is expressed as C: {PA1, PA2,..., PAI}, and the path with the maximum power is PA1 (ST21). Then, the signal correlation coefficient between PA1 and the other paths PAi (i = 2,..., I) is calculated using the above equation (7) (ST22). Then, it is determined whether or not the signal correlation coefficient between PA1 and PAi is equal to or greater than a predetermined threshold (ST23). If the signal correlation coefficient is equal to or greater than the predetermined threshold, PAi is stored in the same cluster as PA1. (ST24). On the other hand, if the signal correlation coefficient is less than the predetermined threshold, PAi is stored in set C ′ (ST25). Then, the above-described procedure is repeated until the signal correlation coefficient with PA1 is compared for all PAi in set C (ST26, ST27). Then, it is determined whether the number of paths allocated to the set C ′ that is not stored in the same cluster as PA1 is 0 (ST28), and if the number of paths is not 0 but exists in the set C ′, the set C ′ is set again as set C (ST29), and the above-described procedure is repeated until the determination result in ST28 becomes zero.

  By doing in this way, the initial cluster 1 clustered by the first stage clustering is classified into more detailed clusters 1_1 to 1_N. Information regarding paths classified into detailed clusters is output to the second-stage cluster storage unit 1062 and stored in the second-stage cluster storage unit 1062. Note that, when the second stage clustering is performed in parallel for each of the correlation path detection units 1061-1 to 1061-M, it is possible to significantly reduce the calculation time required for the second stage clustering process.

  Then, all the path information for each cluster classified by the second-stage clustering process is output to the cluster information storage unit 1063.

  As described above, according to the present embodiment, the spatio-temporal multipath cluster processing is performed in two stages. That is, as the first stage clustering, the delay time and the angle spectrum are calculated from the correlation calculation result between the received signal and the known signal by a method with a small calculation amount and a relatively low resolution, and the correlation time width and the correlation angle width for each peak are calculated. From these, an initial cluster having these as cluster ranges is generated, and further, a delay time and an arrival direction are estimated by a method with higher resolution, and the estimated delay time estimation value and arrival direction estimation value are respectively included in the initial clusters. The path was classified. By doing this, a macroscopic cluster is generated from the correlation time width and correlation angle width calculated by the method with relatively low resolution, and the delay time estimation value and arrival direction estimation calculated by the method with relatively high resolution are generated. Since each path is classified into generated clusters using values, each path is clustered with a relatively small amount of computation compared to the case where all the clusters are fractionated by a method with relatively high resolution. Will be able to.

  Further, as the second-stage clustering, signals having high signal correlation coefficients are further classified into the same cluster using signal correlation coefficients between signals classified into the same cluster. As a result, the first-stage clusters described above can be classified into more detailed clusters. Thus, according to the present embodiment, it is possible to accurately perform spatiotemporal clustering with a relatively small amount of computation, and as a result, the efficiency of propagation path analysis can be improved.

(Embodiment 2)
FIG. 8 shows a main configuration of radio communication apparatus 100 according to the second embodiment. In FIG. 8, the same components as those in FIG. 1 are denoted by the same reference numerals as those in FIG. 1, and detailed description thereof will be omitted. 8 employs a configuration in which a super-resolution estimation unit 201 is added to FIG. 1 instead of the super-resolution estimation unit 104.

  The super-resolution estimation unit 201 estimates the delay time and the arrival direction for each delay time range calculated by the delay time range detection unit 1033, and uses the estimated delay time estimation value and arrival direction estimation value as the cluster determination writing unit 1051. Output to. That is, the super-resolution estimation unit 201 estimates the delay time and the arrival direction for each delay time range by limiting the search range for estimating the delay time and the arrival direction to the delay time range including the peak value of the correlation calculation result. .

  Hereinafter, the estimation operation of the super-resolution estimation unit 201 will be described again with reference to FIG. FIG. 2 shows the correlation calculation result calculated by the correlation value calculation unit 1032, the vertical axis indicates the correlation value, and the horizontal axis indicates the delay time. As described above, in the delay time range detection unit 1033, for example, a time range surrounded by adjacent local minimum points including a peak, or a time range including a peak and having a correlation calculation result equal to or greater than a predetermined threshold is a delay time range. Detected as FIG. 2 shows how the delay time ranges T1 to TN are detected.

  Then, for each of these delay time ranges T1 to TN, the super-resolution estimation unit 201 estimates the delay time and the arrival direction. By doing this, the number of samples on the time axis can be reduced and the delay time and the arrival direction can be estimated, so that the amount of calculation can be greatly reduced. In addition, since the delay time ranges T1 to TN always include the peak value as shown in FIG. 2, the delay time and the arrival direction can be reliably estimated while reducing the amount of calculation. Furthermore, by performing the delay time and arrival direction estimation calculation processing in parallel for each delay time range, the calculation time can be shortened and the processing speed can be improved.

  As described above, according to the present embodiment, the delay time and arrival direction estimation search range is limited to the delay time range including the peak value of the correlation calculation result, and the delay time and arrival time are determined for each delay time range. Since the direction is estimated, the amount of calculation can be greatly reduced as compared with the case where the delay time and the arrival direction are estimated by searching a wide range. In addition, when the delay time and arrival direction estimation calculation processing is performed in parallel for each delay time range, the calculation time can be further reduced.

(Embodiment 3)
FIG. 9 shows a main configuration of radio communication apparatus 100 according to the third embodiment. In FIG. 9, the same reference numerals as those in FIG. 1 are assigned to components common to those in FIG. 1, and detailed description thereof is omitted. 9 employs a configuration in which a super-resolution estimation unit 301 is provided instead of the super-resolution estimation unit 104 and a search range setting unit 302 and an initial value setting unit 303 are added to FIG.

  Super-resolution estimation section 301 uses a SAGE (Space Alternating Generalized EM) algorithm for estimation of delay time and direction of arrival. The SAGE algorithm is not a null scanning method such as MUSIC or ESPRIT but an algorithm using maximum likelihood estimation. In the present embodiment, the super-resolution estimation unit 301 is limited to the case where the SAGE algorithm is used. First, the SAGE algorithm will be described.

In parameter estimation based on maximum likelihood estimation, when the set of observed received signals is Y and the set of parameters to be estimated is H, H that maximizes the log likelihood function f (Y, H) is obtained. However, parameter estimation based on maximum likelihood estimation has a problem that the amount of computation becomes enormous with the number of parameters M. To solve this problem, “APDempster, NMLaird, and DBRubin,“ Maximum likelihood from incomplete data via the EM algorithm ”J. Royal Statist. Soc., Ser. B, vol.39, no. 1, pp. 1 -38, 1977. "discloses an EM (Expectation Maximization) algorithm that reduces the amount of computation and approximately realizes maximum likelihood estimation. The EM algorithm is a sequential algorithm composed of two steps represented by Equation (8) and Equation (9).

Here, Q (H, H ^[i] ) is a conditional log likelihood function, H ^[i] is the i-th estimate of H, and E [] is a conditional average scan. By repeating the two steps shown in the above equations (8) and (9), H ^[i] converges asymptotically to the estimated value of the maximum likelihood estimation.

  However, the EM algorithm has a problem that the convergence speed is slow. The algorithm that increases the convergence speed is the SAGE algorithm. The SAGE algorithm sequentially repeats an operation of fixing only some parameters while updating other parameters, and sequentially realizes parameter estimation based on maximum likelihood estimation. “JAFessler and AOHero,“ Space-alternationg generalized expectation-maximization algorithm ”IEEE Trans. Signal Processing, vol. 42, pp.2664-2677, Oct. 1994.” and “Bernard H. Fleury et al.,” Channel Parameter Estimation In SAGE Algorithm “IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, VOL. 17, NO.3, MARCH 1999” Has been.

In the SAGE algorithms disclosed therein, H is divided into N subsets, the nth subset is H _n , H _n ^[i] _where n ≠ 1 is fixed, and H _n ^[i] is Update with the EM algorithm, repeat until n = N, and perform the same update as i → i + 1.

  However, since the above-described SAGE algorithm is a local algorithm, it has been pointed out that the initial value may converge to a completely different value, and the setting of the initial value is an important issue. Further, there is a problem that the estimation time becomes enormous as the search range to be estimated becomes wider. The wireless communication apparatus according to the present embodiment has added a configuration for improving these problems when the SAGE algorithm is used in super-resolution estimation section 301, that is, search range setting section 302 and initial value setting section 303. Take the configuration. Hereinafter, the search range setting unit 302 and the initial value setting unit 303 will be described.

  The search range setting unit 302 uses the cluster information output from the first-stage cluster storage unit 1052, and the search range used by the SAGE algorithm when the next super-resolution estimation unit 301 estimates the delay time and the arrival direction. And the search range is output to the super-resolution estimation unit 301. By limiting the search range, it is not necessary to search the entire range when estimating the next delay time and direction of arrival, and the estimation time can be shortened. The method for determining the search range will be described in detail later.

  The initial value setting unit 303 determines initial values used when the super-resolution estimation unit 301 estimates the delay time and the arrival direction. Specifically, after the detailed cluster is determined by the second-stage cluster storage unit 1062, the average value or the centroid value of the delay time estimated value and the arrival direction estimated value of the path in each detailed cluster is calculated as the initial value. And output to the super-resolution estimation unit 301. In the SAGE algorithm, as described above, setting an initial value is an important issue, but an average value or a centroid value calculated using an actual arrival path delay time estimation value or arrival direction estimation value is used as an initial value. Thus, the problem that the estimation result converges to a different value can be avoided.

  Hereinafter, the operation of the clustering process of the wireless communication apparatus 100 configured as described above will be described.

  First, a signal transmitted from a communication partner (not shown) is received via the array antennas 101-1 to 101-L, and the received signal is output to the reception unit 102. The reception unit 102 performs wireless reception processing (A / D). Conversion, down-conversion, etc.) and converted into a baseband signal, and the baseband signal is output to correlation value calculation section 1032 and super-resolution estimation section 301. Then, the correlation value calculation unit 1032 uses the known signal output from the known signal generation unit 1031 to despread the baseband signal, and the correlation calculation result after despreading is obtained from the delay time range detection unit 1033 and the angle spectrum. It is output to the calculation unit 1034. In the angle spectrum calculation unit 1034, the angle spectrum is calculated using the correlation calculation result after despreading, and is output to the arrival angle range detection unit 1035.

  Then, the delay time range is detected from the correlation calculation result after diffusion by the delay time range detection unit 1033, the arrival angle range is detected by the arrival angle range detection unit 1035, and the delay time range and the arrival angle range are the initial clusters. The data is output to the generation unit 1036.

  Then, the initial cluster generation unit 1036 uses the delay time range and the arrival angle range to generate an initial cluster, and information about the initial cluster is output to the cluster determination writing unit 1051.

  On the other hand, information on the delay time estimated value and the arrival direction estimated value of each path estimated by the super-resolution estimating unit 301 is output to the cluster determination writing unit 1051.

  Then, the cluster determination writing unit 1051 uses the initial cluster and the information on the arrival path delay time estimation value and the arrival direction estimation value to classify the arrival path into the initial cluster according to the flowchart shown in FIG. Is done. Then, information regarding paths classified into the initial clusters is output to the first-stage cluster storage unit 1052, and information regarding paths classified for each initial cluster is stored by the first-stage cluster storage unit 1052. In this way, the first stage clustering process is performed. The first-stage clustering process is the same as that in the first embodiment or the second embodiment.

  Then, when the search range setting unit 302 uses each cluster information output from the first-stage cluster storage unit 1052 to estimate the delay time and the arrival direction in the next super-resolution estimation unit 301, the SAGE algorithm The search range to be used is determined. Hereinafter, setting of the search range of the search range setting unit 302 will be described with reference to FIG. FIG. 10 is a diagram showing estimation results of delay times and arrival directions of the initial cluster and the respective paths, and black circles are paths specified by the delay time estimation values and arrival direction estimation values estimated by the super-resolution estimation unit 301. Indicates.

  First, the search range setting unit 302 detects the minimum value and the maximum value of the delay time estimated value and the arrival direction estimated value of the path classified into each cluster output from the first-stage cluster storage unit 1052. A predetermined margin is provided for the minimum value and the maximum value of the delay time estimated value and the arrival direction estimated value, and an area surrounded by the delay time range and the arrival angle range after the margin setting is set as a search range. For example, the search range is set by providing a margin of +5 nsec from the maximum delay time estimated value, −5 nsec from the minimum value, +5 deg from the maximum value of the arrival direction estimated value, and −5 deg from the minimum value. FIG. 10 shows an example of the search range. In the figure, a range surrounded by a dotted line indicates a search range. Information regarding the search range is output to the super-resolution estimation unit 301.

  In this way, by limiting the search range used in the next SAGE algorithm using the delay time estimated value and the arrival direction estimated value of the arrival path, there is no need to estimate the delay time and the arrival direction over a wide range. The calculation amount by the algorithm can be greatly reduced.

  However, when the path delay time estimation value and the arrival direction estimation value fluctuate greatly, there may be a case where the path is not included in the set search range, so every predetermined number of times (for example, 1 in 10 times). The delay time and the arrival direction may be estimated by searching a wide range without limiting the search range.

  Meanwhile, the correlation path detection units 1061-1 to 1061-M use the signal correlation coefficient between the signal sequences of the paths stored for each initial cluster in the first stage cluster storage unit 1052 to perform the second stage clustering process. Done. All the path information for each cluster classified by the second stage clustering process is output to the cluster information storage unit 1063 and the initial value setting unit 303 via the second stage cluster storage unit 1062.

  Then, the initial value used by the SAGE algorithm is set in the super-resolution estimation unit 301 by using the path information classified into the detailed clusters after the second-stage clustering process by the initial value setting unit 303.

  As an initial value, an average value or a centroid value of a delay time estimated value and an arrival direction estimated value of a path in each cluster is set. FIG. 11 is an example in the case where the average value of the delay time estimated value and the arrival direction estimated value of the paths 1 to 3 classified into the cluster 1_1 is calculated as an initial value to be used next time by the SAGE algorithm. By doing this, the delay time estimation value and the arrival direction estimation value of the actual arrival path are reflected in the initial value, and the rate of convergence to an incorrect value is reduced compared to the case where the initial value is arbitrarily set. As a result, the measurement accuracy can be improved. In addition, since the convergence time is shortened compared to the case where the initial value is arbitrarily set, the search time can be shortened as a result.

  As described above, according to the present embodiment, when using the SAGE algorithm in the super-resolution estimation unit 301, the delay time estimated value and the arrival direction estimated value of the path classified into the initial cluster by the first-stage clustering are obtained. The search range is set by using the delay time estimation value and the arrival direction estimation value of the path classified into the detailed cluster by the second stage clustering, so that the search time is shortened, In addition, the direction of arrival can be estimated with high accuracy. As a result, the clustering processing time can be shortened, and the accuracy of the delay time estimation value and the arrival direction estimation value estimated by the super-resolution estimation unit 301 is improved, so that the spatio-temporal clustering can be performed accurately. it can.

(Embodiment 4)
FIG. 12 shows a main configuration of radio communication apparatus 100 according to the fourth embodiment. In FIG. 12, the same components as those in FIG. 1 are denoted by the same reference numerals as those in FIG. 1, and detailed description thereof is omitted. 12 employs a configuration in which an angle spread / center angle calculation unit 401, a transmission beam setting unit 402, a modulation unit 403, and an array transmission unit 404 are added to FIG.

  The angle spread / center angle calculation unit 401 calculates the angle spread of the arrival path classified into the detailed clusters after the second-stage clustering process for each detailed cluster. In addition, the angle spread / center angle calculation unit 401 is configured to calculate the arrival direction estimated value of the arrival paths classified into the detailed clusters, the center of gravity value, or the arrival direction of the maximum power path among the arrival paths classified into the detailed clusters. The estimated value is calculated for each detailed cluster as the center angle. The angle spread / center angle calculation unit 401 outputs the calculated angle spread and center angle for each detailed cluster to the transmission beam setting unit 402.

  Transmit beam setting section 402 determines directivity by setting the transmit beam width and the like based on the angle spread and the center angle, and outputs information related to directivity to array transmitter 404. FIG. 13 is a diagram illustrating an example of directivity determined by the transmission beam setting unit 402. FIG. 13 shows that the incoming paths are classified into three clusters of cluster 1_1 to cluster 1_3 in the present embodiment, and the angle spread / center angle calculation unit 401 shows the angle spread and center angle of cluster 1_1 to cluster 1_3 as shown in FIG. It is an example when the values as shown in FIG.

  Modulating section 403 converts the transmission data into a modulated signal and outputs it to array transmitting section 404.

  Array transmission section 404 controls array antennas 101-1 to 101-L using information on directivity, and controls the transmission direction of the modulation signal output from modulation section 403. By doing so, it is possible to reliably transmit the modulation signal in the arrival direction.

  As described above, according to the present embodiment, the angle spread and the center angle are calculated for each detailed cluster after the second-stage clustering process, and array antennas 101-1 to 101-1 are used by using the angle spread and the center angle. The directivity of 101-L was controlled to transmit the modulated signal in the direction of arrival. By doing so, the directivity of the array antenna can be controlled in the direction in which the incoming path is actually transmitted, and communication can be performed reliably.

  In the above description, the case where the angle spread and the center angle are calculated using the arrival direction estimation value of the path classified into the detailed cluster has been described. However, from the arrival direction estimation value of the path classified into the initial cluster. The angle spread and the center angle may be calculated. Also in this case, it is possible to control the array antennas 101-1 to 101 -L to transmit the modulated signal in the arrival direction, and not the result after the second stage clustering process, but the first stage clustering process. Since the array antennas 101-1 to 101 -L are controlled using the later results, the amount of calculation is reduced compared to the case where the angle spread and the center angle are calculated using the estimated arrival direction values of the paths classified into the detailed clusters. Can be reduced.

  The wireless communication apparatus according to the first aspect of the present invention detects an arrival angle range and a delay time range of a received signal, and collects a collection of similar arrival angle ranges and delay time ranges into one cluster based on the detection result. First cluster generation means for generating a plurality of clusters, super-resolution estimation means for estimating the arrival angle and delay time of the received signal with higher accuracy than the detection accuracy of the first cluster generation means, A cluster classification unit that classifies which cluster the received signal belongs to which of the clusters generated by the first cluster generation unit from the highly accurate arrival angle and delay time estimated by the super-resolution estimation unit; Take.

  According to this configuration, the delay time and the arrival angle of the received signal are detected using two methods with different resolution accuracy, and a plurality of clusters are obtained from the delay time range and the arrival angle range calculated using the method with low accuracy. The received signal delay time and arrival angle estimated using a method with high accuracy are generated and the received signal belongs to which of the previously generated clusters is classified. As a result, clusters can be generated with a small amount of computation, and cluster classification can be performed accurately.

  A wireless communication apparatus according to a second aspect of the present invention calculates a correlation value between the received signals for each cluster classified by the cluster classification means in the first aspect, and based on the calculation result, The configuration further includes second cluster generation means for generating a new cluster.

  According to this configuration, the received signals are further clustered into new clusters based on the correlation values of the received signals classified into the same cluster, so spatio-temporal multipath propagation that takes both delay time and arrival direction into consideration. The clustering of roads can be performed accurately.

  The wireless communication apparatus according to a third aspect of the present invention is the wireless communication apparatus according to the first aspect, wherein the first cluster generation means includes a correlation value calculation means for calculating a correlation value between the received signal and the known signal, A delay time detection means for detecting a delay time range based on a correlation value; a means for calculating an angle spectrum using the correlation value; an arrival angle range detection means for detecting an arrival angle range based on the angle spectrum; The structure which comprises is taken.

  According to this configuration, a plurality of clusters having a cluster range that is surrounded by the delay time range and the arrival angle range are generated, and the arrival angle and delay time estimated with high accuracy are compared with the cluster range of these clusters. Thus, the received signals can be classified into the generated clusters, and the received signals that are roughly similar with a relatively simple configuration can be classified into one cluster.

  The wireless communication apparatus according to a fourth aspect of the present invention is the wireless communication apparatus according to the second aspect, wherein the second cluster generation means calculates a signal correlation value between the received signals classified into the same cluster by the cluster classification means. Signal correlation calculating means for calculating, and adopting a configuration in which a collection of received signals having a signal correlation value equal to or greater than a predetermined threshold is used as one new cluster.

  According to this configuration, since a more similar received signal is generated as one new cluster according to the signal correlation value of the received signals belonging to the roughly classified cluster, the signal correlation value is a predetermined value. Only received signals that are equal to or higher than the threshold and strongly correlated with each other can be classified into the same cluster, and can be more accurately clustered.

  The wireless communication apparatus according to a fifth aspect of the present invention is the wireless communication apparatus according to the fourth aspect, wherein the second cluster generation means includes a signal correlation value between the received signals classified into the same cluster by the cluster classification means. Is calculated in parallel for each cluster.

  According to this configuration, since the signal correlation values of the received signals classified into the same cluster are calculated in parallel for each cluster, the calculation time of the spatiotemporal clustering can be shortened.

  The radio communication apparatus according to a sixth aspect of the present invention employs a configuration in which, in the first aspect, the super-resolution estimation means calculates the arrival angle and the delay time for each delay time range.

  According to this configuration, the time range for calculating the angle of arrival and the delay time with high accuracy is reduced to the delay time range, the number of samples required for the calculation can be reduced, and the calculation amount of the spatio-temporal clustering can be greatly reduced. it can. In addition, since computation can be performed in parallel for each delay time range, the computation time for spatiotemporal clustering can be greatly reduced.

  The wireless communication apparatus according to a seventh aspect of the present invention is the wireless communication apparatus according to the first aspect, wherein the super-resolution estimation unit calculates the arrival angle and the delay time using a MUSIC, ESPRIT, or SAGE algorithm. Take.

  According to this configuration, the arrival angle and the delay time can be calculated with high accuracy.

  The wireless communication apparatus according to an eighth aspect of the present invention is the wireless communication apparatus according to the first aspect, wherein the super-resolution estimation unit calculates the angle of arrival and the delay time using a SAGE algorithm. A configuration further includes search range determining means for determining a search range used by the SAGE algorithm from the arrival angle and the delay time of the received signals classified into the same cluster by the cluster classification means.

  According to this configuration, the search range of the SAGE algorithm is narrowed down from the arrival angle and delay time of the received signals classified into the same cluster, and the arrival angle and delay time are estimated, so the estimation time is shortened. Will be able to.

  The wireless communication apparatus according to a ninth aspect of the present invention is the wireless communication apparatus according to the first aspect, wherein the super-resolution estimation means calculates the angle of arrival and the delay time using a SAGE algorithm. The configuration further includes initial value determining means for determining an initial value used by the SAGE algorithm from the arrival angle and the delay time of the received signal classified into a new cluster by the second cluster generating means.

  According to this configuration, the arrival angle and the delay time are estimated using the SAGE algorithm using the initial value calculated from the arrival angle and the delay time of the received signal classified into the new cluster, which is completely different. The rate of convergence to the value decreases, and the arrival angle and delay time can be accurately estimated.

  A wireless communication apparatus according to a tenth aspect of the present invention, in the first aspect, obtains an angle spread and a center arrival angle from a plurality of array antennas and received signals classified into the same cluster by the cluster classification means. An angle spread calculating means for calculating and a control means for controlling the directivity and transmission beam width of the array antenna from the angle spread and the center arrival angle are employed.

  According to this configuration, the directivity and transmission beam width of the array antenna can be roughly controlled with a relatively small amount of calculation.

  A wireless communication apparatus according to an eleventh aspect of the present invention is the wireless communication apparatus according to the first aspect, wherein a plurality of array antennas and receptions classified into the same cluster in a new cluster generated by the second cluster generation means are provided. An angle spread calculating means for calculating an angle spread and a center arrival angle from the signal, and a control means for controlling the directivity and transmission beam width of the array antenna from the angle spread and the center arrival angle. Take.

  According to this configuration, the directivity and transmission beam width of the array antenna can be controlled more accurately than in the radio communication apparatus according to the eighth aspect.

  The spatio-temporal clustering method according to the twelfth aspect of the present invention detects an arrival angle range and a delay time range of a received signal and, based on the detection result, collects similar arrival angle ranges and delay time ranges into one As a cluster, a first cluster generation step of generating a plurality of clusters, a super-resolution estimation step of estimating the arrival angle and delay time of the received signal with higher accuracy than the detection accuracy of the first cluster generation step, A cluster classification step of classifying which cluster generated by the first cluster generation step the received signal belongs to from the highly accurate arrival angle and delay time estimated in the super-resolution estimation step I made it.

  According to this method, the delay time and the arrival angle of the received signal are detected using two methods having different resolution accuracy, and a plurality of clusters are obtained from the delay time range and the arrival angle range calculated using a method with low accuracy. The received signal delay time and arrival angle estimated using a method with high accuracy are generated and the received signal belongs to which of the previously generated clusters is classified. As a result, clusters can be generated with a small amount of computation, and cluster classification can be performed accurately.

  INDUSTRIAL APPLICABILITY The present invention can accurately perform spatiotemporal clustering with a relatively small amount of computation, and is useful for a wireless communication apparatus, a spatiotemporal clustering method, and the like.

The block diagram which shows the principal part structure of the radio | wireless communication apparatus which concerns on Embodiment 1 of this invention. Diagram for explaining detection of delay time range Diagram for explaining detection of arrival angle range Illustration for explaining the initial cluster Flow chart for explaining the first-stage clustering procedure The figure which shows the initial cluster after the clustering processing of the 1st step Flow chart for explaining the clustering procedure of the second stage The block diagram which shows the principal part structure of the radio | wireless communication apparatus which concerns on Embodiment 2 of this invention. The block diagram which shows the principal part structure of the radio | wireless communication apparatus which concerns on Embodiment 3 of this invention. The figure which shows the range which performs direction of arrival estimation concerning Embodiment 3 The figure which shows the initial value of the DOA estimation which concerns on Embodiment 3. The block diagram which shows the principal part structure of the radio | wireless communication apparatus which concerns on Embodiment 4 of this invention. Diagram showing an example of array antenna directivity Diagram showing an example of center angle and angle spread Diagram showing cluster concept

Explanation of symbols

101-1 to 101-L array antenna 102 receiving unit 103 first cluster generation unit 1031 known signal generation unit 1032 correlation value calculation unit 1033 delay time range detection unit 1034 angle spectrum calculation unit 1035 arrival angle range detection unit 1036 initial cluster generation Unit 104, 201, 301 super-resolution estimation unit 105 cluster classification unit 1051 cluster determination writing unit 1052 first stage cluster storage unit 106 second cluster generation unit 1061-1 to 1061-M correlation path detection unit 1062 second stage cluster Storage unit 1063 Cluster information storage unit 302 Search range setting unit 303 Initial value setting unit 401 Angle spread / center angle calculation unit 402 Transmit beam setting unit 403 Modulation unit 404 Array transmission unit

Claims (12)

First cluster generation means for detecting an arrival angle range and a delay time range of a received signal, and generating a plurality of clusters based on a result of the detection, with a collection of similar arrival angle ranges and delay time ranges as one cluster When,
Super-resolution estimation means for estimating the arrival angle and delay time of the received signal with higher accuracy than the detection accuracy of the first cluster generation means;
Cluster classification means for classifying to which cluster the received signal belongs to the cluster generated by the first cluster generation means from the highly accurate arrival angle and delay time estimated by the super-resolution estimation means;
A wireless communication apparatus comprising: The apparatus further comprises second cluster generation means for calculating a correlation value between the received signals for each cluster classified by the cluster classification means and generating a new cluster based on the calculation result. The wireless communication device described. The first cluster generation means includes correlation value calculation means for calculating a correlation value between the received signal and the known signal;
A delay time detecting means for detecting a delay time range based on the correlation value;
Means for calculating an angular spectrum using the correlation value;
The wireless communication apparatus according to claim 1, further comprising: an arrival angle range detecting unit that detects an arrival angle range based on the angle spectrum. The second cluster generation means comprises signal correlation calculation means for calculating a signal correlation value between received signals classified into the same cluster by the cluster classification means,
The wireless communication apparatus according to claim 2, wherein a collection of received signals having a signal correlation value equal to or greater than a predetermined threshold is defined as one new cluster. The wireless communication apparatus according to claim 4, wherein the second cluster generation unit calculates a signal correlation value between received signals classified into the same cluster by the cluster classification unit in parallel for each cluster. The radio communication apparatus according to claim 1, wherein the super-resolution estimation unit calculates the arrival angle and the delay time for each delay time range. The wireless communication apparatus according to claim 1, wherein the super-resolution estimation unit calculates the arrival angle and the delay time using a MUSIC, ESPRIT, or SAGE algorithm. The super-resolution estimation means calculates the arrival angle and the delay time using a SAGE algorithm,
The wireless communication apparatus further includes search range determining means for determining a search range used by the SAGE algorithm from the arrival angle and the delay time of the received signals classified into the same cluster by the cluster classification means. The wireless communication device according to 1. The super-resolution estimation means calculates the arrival angle and the delay time using a SAGE algorithm,
The wireless communication apparatus further includes initial value determining means for determining an initial value used by the SAGE algorithm from the arrival angle and the delay time of the received signal classified into a new cluster by the second cluster generating means. The wireless communication device according to claim 1. Multiple array antennas,
Angle spread calculation means for calculating an angle spread and a center arrival angle from the received signals classified into the same cluster by the cluster classification means;
The wireless communication apparatus according to claim 1, further comprising: a control unit that controls directivity and transmission beam width of the array antenna from the angle spread and the center arrival angle. Multiple array antennas,
Angle spread calculation means for calculating an angle spread and a center arrival angle from received signals classified into the same cluster in the new cluster generated by the second cluster generation means;
The wireless communication apparatus according to claim 1, further comprising: a control unit that controls directivity and transmission beam width of the array antenna from the angle spread and the center arrival angle. A first cluster generation step of detecting an arrival angle range and a delay time range of a received signal, and generating a plurality of clusters based on the detection result, with a collection of similar arrival angle ranges and delay time ranges as one cluster When,
A super-resolution estimation step of estimating the arrival angle and delay time of the received signal with higher accuracy than the detection accuracy of the first cluster generation step;
A cluster classification step of classifying to which cluster the received signal belongs to which is generated by the first cluster generation step from the highly accurate arrival angle and delay time estimated in the super-resolution estimation step;
A spatiotemporal clustering method comprising:

Images