JP6331841B2 - Direction of arrival estimation device - Google Patents

Description

  The present invention relates to an arrival direction estimation device that estimates the arrival direction of waves such as radio waves, light waves, and sound waves.

  There is a VESPA (Virtual ESPRIT Algorithm), which is a super-resolution angle measurement algorithm using higher-order statistics, as an algorithm for separating signals with multiple wave interference and estimating the direction of arrival. VESPA receives a signal from a target as a received signal, creates a fourth-order cumulant using a reception vector corresponding to the array antenna, and uses ESPRIT (Estimation of Signal Parameters) shown in Non-Patent Document 1 from the fourth-order cumulant. This is an algorithm for extracting information on the direction of arrival in accordance with an algorithm similar to that of Via Rotational Invariance Techniques). The VESPA can be applied to an array of an arbitrary arrangement, and does not require an element antenna pattern except for a two-element phase pattern called a guiding sensor, and does not cause an arrival direction estimation error due to these measurement errors. Signal Classification) and other advantages not found in conventional ESPRIT. However, for the operation, the arrival wave needs to be uncorrelated as in Non-Patent Document 2.

Roy, R. and Kailath, T .: “Esprit-Estimation of Signal Parameters ia Rotational Invariance Techniques”, IEEE Trans. Acoustics, Speech and Signal Processing, vol. 37, no. 7, JULY 1989. Jerry M. Mendel and Mithat C. Dogan: “Application of Cumulants to Array Processing Part IV: Direction Finding in Coherent Signals Case”, IEEE Trans. On Signal Processing, Vol. 45, no. 9, SEPTEMBER 1997.

  However, in the conventional direction-of-arrival estimation device, when estimating the direction of arrival at the time of interference of a plurality of waves, conventionally, it has been necessary to take a long observation time and sufficiently reduce the correlation of signals. In addition, it has not been studied in detail how much observation time should be taken. Therefore, in order to separate multiple waves and perform highly accurate direction-of-arrival estimation, it takes a long observation time. was there.

  The present invention has been made to solve the above-described problems, and enables arrival direction estimation in a shorter time than the prior art, and further clarifies the standard of how long the observation time should be set. It is an object of the present invention to obtain an arrival direction estimation device that can perform the above.

  An arrival direction estimation apparatus according to the present invention includes an array antenna that receives a signal with an unknown arrival direction, a matrix creation unit that creates a correlation matrix or cumulant matrix of the received signal received by the array antenna, and the matrix creation unit Using the correlation matrix or the cumulant matrix created by the step of calculating a phase rotation matrix including information representing the phase rotation caused by the difference in the observation position of the signal whose direction of arrival is unknown as a diagonal component, A phase analysis unit that determines a data observation time based on a time change of a diagonal component, and an arrival direction estimation unit that performs an arrival direction estimation using a received signal corresponding to the data observation time determined by the phase analysis unit, It is characterized by having.

  According to the present invention, it is possible to calculate an observation time in which arrival direction estimation is possible even when signals are correlated, and it is possible to perform arrival direction estimation in a shorter time than in the prior art.

The block diagram which shows the arrival direction estimation apparatus by Embodiment 1 of this invention. The block diagram of the phase evaluation part 7 in Embodiment 1 of this invention. The figure which shows the graph which calculates | requires the candidate of suitable data observation time from the phase component in each observation time in Embodiment 1 of this invention. The phase difference between the real array and the virtual array plotted on the complex plane in the first embodiment of the present invention. The block diagram which shows the arrival direction estimation apparatus by Embodiment 2 of this invention. The block diagram of the signal singular value evaluation part 10 in Embodiment 2 of this invention. The block diagram which shows the arrival direction estimation apparatus by Embodiment 3 of this invention. The block diagram which shows the arrival direction estimation apparatus by Embodiment 4 of this invention. The block diagram of the signal eigenvalue evaluation part 12 in Embodiment 4 of this invention.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
Embodiment 1 FIG.
FIG. 1 is a block diagram showing an arrival direction estimation apparatus according to Embodiment 1 of the present invention. FIG. 1 shows a direction-of-arrival estimation apparatus 1A to which VESPA (Virtual ESPRIT Algorithm), which is a super-resolution angle measurement algorithm, is applied. The invention described in Embodiment 1 uses a fourth-order cumulant that is a fourth-order statistic. Any algorithm other than VESPA may be used as long as it is an algorithm that performs direction-of-arrival estimation.

  In FIG. 1, a radiation source 2 is a signal source that emits a signal whose direction of arrival θ is unknown, or a reflector that reflects a signal emitted from another radiation source. The receiving antennas 3-1 to 3-M are element antennas constituting an array antenna, and receive a signal whose arrival direction θ is unknown.

  The observation time is set automatically or manually by the observation time setting unit 4, and the reception units 5-1 to 5 -M receive various signals with respect to the RF signals that are reception signals of the reception antennas 3-1 to 3 -M. By performing processing (for example, signal amplification processing, band pass filter processing, frequency conversion processing, A / D conversion processing, etc.), a baseband complex signal that is a digital signal is obtained, and the baseband complex signal is cumulated. It is a receiver that outputs to the matrix creation unit 6. When the digital signal obtained by the receiving units 5-1 to 5-M is an IF (Intermediate Frequency) real signal, the baseband complex signal is obtained by performing Hilbert transform or digital quadrature detection on the IF real signal. You may make it the structure which obtains. The receiving units 5-1 to 5-M output all signals received up to the observation time set by the observation time setting unit 4 to the cumulant matrix creation unit 6 that is a matrix creation unit.

  The cumulant matrix creation unit 6 creates a cumulant matrix from the output baseband complex signal, and outputs the created cumulant matrix to the phase analysis unit 7. The arrival direction estimation unit 8 performs arrival direction estimation using the phase component received from the phase analysis unit 7 and outputs the arrival direction estimation result to the direction measurement result output unit 9.

  FIG. 2 shows the components of the phase analysis unit 7. The phase analysis unit 7 includes a phase rotation matrix derivation unit 71, a phase calculation unit 72, a phase evaluation unit 73, and a data observation time derivation unit 74. In each figure, the same numerals indicate the same or corresponding parts.

  Below, operation | movement of the arrival direction estimation apparatus 1A in this Embodiment 1 is demonstrated. When the receiving antennas 3-1 to 3-M receive an RF signal whose arrival direction θ is unknown, the observation time setting unit 4 extracts a signal in a time range based on the observation time set automatically or manually. The observation time setting unit 4 outputs signals extracted from the RF signals received by the reception antennas 3-1 to 3-M to the reception units 5-1 to 5-M, respectively.

  When receiving units 5-1 to 5-M receive signals from observation time setting unit 4, various types of signal processing (for example, signal amplification processing, bandpass filter processing, frequency conversion processing, A / D conversion processing, etc.) are performed. The baseband complex signal which is a digital signal is output to the cumulant matrix creation unit 6.

When receiving the baseband complex signal from the receiving units 5-1 to 5-M, the cumulant matrix creating unit 6 creates a cumulant matrix from the baseband complex signal.

式（１）、（２）において、ｍ（ｍ＝１、２、・・・、Ｍ）、ｎ（ｎ＝１、２、・・・、Ｍ）はガイディングセンサと呼ばれる受信アンテナ３−１〜３−Ｍに含まれるアンテナの
素子番号(Ｍは素子数)であり、r_m(t)、r_ｎ(t)はガイディングセンサの受信する時間信号である。また、ｒ(t)は各素子の受信する時間信号を成分に持つベクトル、^*は複素共役を、^Hは複素共役転置を表し、ここでcum(＊)は式（４）で定義されるキュムラント演算である。

ここで、Ｅ[＊]は＊のアンサンブル平均もしくは時間平均を表す。式（４）を用いると、式（１）、式（２）の４次キュムラントは式（５）、式（６）のように４次モーメントと２次モーメントとの差の形に展開することができる。

Here, the cumulant matrix is composed of fourth-order higher-order cumulants. For example, the fourth-order cumulant matrix can be expressed as equations (1) and (2), and the fourth-order cumulant matrix can be expressed as equation (3).

In equations (1) and (2), m (m = 1, 2,..., M) and n (n = 1, 2,..., M) are reception antennas 3-1 called guiding sensors. an antenna element number included in the to 3-M (M is the number of _{elements), r m (t),} r n (t) is a time signal received by the guiding sensor. R (t) is a vector having the time signal received by each element as a component, ^* is a complex conjugate, ^H is a complex conjugate transpose, and cum (*) is a cumulant defined by equation (4). It is an operation.

Here, E [*] represents an ensemble average or a time average of *. Using Equation (4), the fourth-order cumulants in Equation (1) and Equation (2) should be expanded into the difference between the fourth-order moment and the second-order moment as in Equation (5) and Equation (6). Can do.

When the cumulant matrix creation unit 6 creates the cumulant matrix of Expression (3), the cumulant matrix creation unit 6 outputs the created cumulant matrix to the phase analysis unit 7.

  When the phase rotation matrix derivation unit 71 in the phase analysis unit 7 receives the cumulant matrix from the cumulant matrix creation unit 6, the phase rotation matrix is subjected to singular value decomposition and the TLS method (Total-Least-Square) from the cumulant matrix. Is derived and output to the phase calculator 72. Here, instead of the TLS method, the LS method (Least-Square) may be used as the processing after the singular value decomposition.

ここで、Σは対角行列、Ｖはユニタリ行列である。
次に式（８）のように、行列U₁を縦に分割した行列Ｅｘ、Ｅｙを得る。

ここで，Exは図1における，実際に存在する実アレー３−１〜３−Ｍに対応する信号部分空間行列，Eyは４次キュムラントにより作成した仮想的なアレーに対応する信号部分空間行列である． Details of the process up to the derivation of the phase rotation matrix are shown below. First, a quaternary cumulant matrix created from two quaternary cumulant matrices is subjected to singular value decomposition as shown in Equation (7), and a matrix U _{1 in} which signal subspace vectors are aggregated and a matrix in which noise subspace vectors are aggregated. Divide into U ₂ .

Here, Σ is a diagonal matrix, and V is a unitary matrix.
Next, as shown in Expression (8), matrices Ex and Ey obtained by vertically dividing the matrix U ₁ are obtained.

Here, Ex is the signal subspace matrix corresponding to the actual arrays 3-1 to 3-M that actually exist in FIG. 1, and Ey is the signal subspace matrix corresponding to the virtual array created by the fourth-order cumulant. is there.

Here, the virtual array is determined by the fourth-order cumulant matrix of Equation (6). If the position vector between the guiding sensors is d and the wave number vector of the incoming wave is k, Equation (6) can be expressed as Equation (9).

  Equation (9) corresponds to a fourth-order cumulant matrix (equation (6)) of a virtual array at a position obtained by moving the real arrays 3-1 to 3-M by the position vector between the guiding sensors. A virtual array at a position moved by the position vector between the guiding sensors is hereinafter referred to as a virtual array. VESPA is essentially an algorithm for determining the direction of arrival by determining the phase difference between a real array and a virtual array.

ここで、ｋ₁,…,ｋ_Pは波数ベクトル、dはガイディングセンサ間の位置ベクトルであり、Ｐは到来波の波数である。 Here, if the steering matrix in the real subarray is A ₁ , and the one corresponding to the virtual subarray is A ₂ , Ex and Ey are related by the regular matrix T by Equations (10) and (11). Φ is a diagonal matrix that gives a phase rotation to each column of A ₁ , which is called a phase rotation matrix, and can be expressed as Equation (12).

Here, k ₁ ,..., K _P are wave number vectors, d is a position vector between guiding sensors, and P is the wave number of an incoming wave.

In order to obtain the phase rotation matrix (formula (12)), formula transformation is performed as follows using formula (10) and formula (11).

Equation (14) is nothing but an eigenvalue expansion of Ψ. That is, if the matrix Ψ can be obtained, the phase rotation matrix Φ can be obtained. The method for obtaining this matrix Ψ uses the TLS method or LS method described above. Both are methods for obtaining Ψ satisfying Equation (13) using the least square method.

から導出される。 The phase rotation matrix Φ can be obtained by the above method. Here, when the P-th diagonal element of Φ is ν _P , k _P = | k _P | (sin θ, cos θ), d = (1,0), and the wavelength of the P-th incoming wave is λ _P , the direction of arrival Is

Is derived from

When the phase calculation unit 72 receives the phase rotation matrix from the phase rotation matrix deriving unit 71, it obtains the phase component of the diagonal term of the phase rotation matrix at each observation time and outputs it to the phase evaluation unit 73.

The phase evaluation unit 73 obtains a direct current component of a phase that periodically changes with respect to the observation time, and obtains a direct current component of the obtained phase value and a phase of each diagonal component that varies with time as shown in FIG. The intersection of the graph of the time change of the value is obtained, the time at the intersection is set as a suitable data observation time candidate, and the time is output to the data observation time deriving unit 74. The DC component can be extracted by applying a low-pass filter, extracting a peak using a histogram, using an average value, or any other method for extracting a DC component. In addition, when finding suitable data observation time candidates, each diagonal component of the phase rotation matrix is plotted on the complex plane, and the convergence point of the orbit is set as a true value, and the true value is drawn from the origin as shown in FIG. The time to reach the intersection of the straight line and the trajectory may be used. The reason will be explained in the next paragraph.

Here, Φ ₁₁ and Φ ₂₂ in FIG. 4 represent the (1, 1) and (2, 2) components of the phase rotation matrix. T is the observation time, and _{Tρ = 0} represents the time when the correlation of the incoming waves is uncorrelated. Φ ₁₁ , Φ ₂₂ represent the phase difference between the real array and the virtual array made in signal processing, and the angle between Φ ₁₁ , Φ _{22 and the} real axis or the imaginary axis, that is, arg (Φ ₁₁ ), arg (Φ ₂₂ ) has information on the angle of arrival of the incoming wave as shown in equation (14). Here, arg (*) indicates that the phase angle of * is calculated. Φ ₁₁ and Φ ₂₂ converge to true values when the correlation of the incoming waves is uncorrelated, that is, when the observation time is T _{ρ = 0} . Further, as the observation time increases from T = 0, Φ ₁₁ and Φ ₂₂ move on the spiral curve in FIG. 4 toward the true value. When T = T _{ρ = 0} , the arrival direction of the incoming wave can be estimated, but at the intersection of the straight line connecting the origin and the true value and the trajectory drawn by Φ ₁₁ , Φ ₂₂ , arg (Φ ₁₁ ), arg Since (Φ ₂₂ ) is equal, it is understood that there is a time in which the direction of arrival can be estimated in a time shorter than T _{ρ = 0} as shown in FIG. As shown in FIG. 4, as the observation time increases, the characteristic that the values of Φ ₁₁ and Φ ₂₂ move toward the true value while drawing a spiral curve is a characteristic newly found by the present embodiment. . By setting an appropriate observation time using this characteristic, in the first embodiment, the arrival angle is estimated with the same degree of accuracy as when the observation time is T _{ρ = 0} with an observation time shorter than that of the prior art. It becomes possible. When the data observation time derivation unit 74 receives a suitable data observation time candidate from the phase evaluation unit 73, the observation time setting unit sets the minimum value among the suitable data observation time candidates as the suitable data observation time. 4 is output.

The observation time setting unit 4 updates the signal observation time to the appropriate data observation time output from the data observation time deriving unit 74, and thereafter receives the incoming signal with the updated observation time. In addition, with respect to the output of the observation time setting unit 4, the receiving units 5-1 to 5-M, the cumulant matrix creating unit 6, the phase rotation matrix deriving unit 71, and the phase calculating unit 72 perform the same processing as described above. Do. When the arrival direction estimation unit 8 receives the phase component of the diagonal term of the phase rotation matrix from the phase calculation unit 72, the arrival direction estimation unit 8 performs arrival direction estimation according to the equation (15), and the arrival direction estimation result is sent to the direction measurement result output unit 9. Output.

As is apparent from the above, according to the first embodiment, the observation time setting unit 4 and the phase analysis unit 7 for evaluating the temporal change of the phase difference between the signals received by the real array and the virtual array are provided. Thus, it is possible to calculate a time shorter than the time when the signal correlation becomes uncorrelated for the first time, and to update the time as the observation time, to obtain a clear reference for determining the observation time, and to be shorter than before It is possible to estimate the direction of arrival in time.

As described above, the arrival direction estimation apparatus according to the first embodiment includes the reception antennas 3-1 to 3-M that are array antennas that receive signals with unknown arrival directions, and the cumulant of the reception signals received by the array antennas. Using the cumulant matrix creating unit 6 which is a matrix creating unit for creating a matrix and the cumulant matrix created by the cumulant matrix creating unit 6, information representing the phase rotation caused by the difference in the observation position of the signal whose direction of arrival is unknown. The phase rotation matrix included in the diagonal component is calculated, and the data observation time is determined based on the time change of the diagonal component of the phase rotation matrix, and this corresponds to the data observation time determined by the phase analysis unit 7 And an arrival direction estimation unit 8 that performs arrival direction estimation using the received signal. With this configuration, it is possible to estimate the direction of arrival in a shorter time than the prior art.

In the arrival direction estimation apparatus according to the first embodiment, the cumulant matrix is between the reception signals at the reception antennas 3-1 to 3-M, which are array antennas, and the two antennas that are included in the array antenna and are called guiding sensors. It is a matrix obtained by cumulant calculation with respect to the received signal at the virtual array antenna at the position where the array antenna is moved by the difference between the positions. By using this cumulant matrix, it is possible to estimate the direction of arrival with high accuracy assuming a virtual array.

In addition, in the arrival direction estimation apparatus according to Embodiment 1, a DC component is obtained for the phase component of the diagonal term of the phase rotation matrix, and the obtained DC component and the phase component of the diagonal term of the phase rotation matrix that changes with time. The time when the two coincide is defined as the data observation time. By determining the data observation time in this way, it is possible to set the data observation time that can reduce the influence of the temporal fluctuation of the phase component, and to perform the arrival direction estimation with high accuracy.

Moreover, in the arrival direction estimation apparatus of Embodiment 1, the phase analysis unit 7 obtains the trajectory and convergence point of the diagonal component of the phase rotation matrix for a plurality of observation times, and the trajectory is the same as the convergence point on the complex plane. The characteristic is that the observation time in the direction is the data observation time. By determining the data observation time in this way, a phase having the same phase component as the convergence point can be acquired in a short observation time, and accurate arrival direction estimation can be performed in a shorter time than in the prior art. . In the present embodiment, the cumulant matrix is created by a fourth-order cumulant, but the cumulant matrix may be a fourth-order or higher-order cumulant.

Embodiment 2. FIG.
In the first embodiment, the arrival direction estimation apparatus has the observation time setting unit 4 and the phase analysis unit 7, whereas the second embodiment has a signal singular value evaluation unit in addition thereto.

In the first embodiment, when the SN ratio (Signal-Noise Ratio: signal-to-noise ratio) is low, the singular value decomposition of the cumulant matrix cannot be performed accurately due to the influence of noise. ₁₁ and Φ ₂₂ orbits vary, and as a result, the intersection may not be accurately obtained. In the second embodiment, an arrival direction estimation device 1B that can accurately obtain a suitable data observation time even when the SN ratio is low is disclosed.

5 is a block diagram showing an arrival direction estimating apparatus 1B according to Embodiment 2 of the present invention. In the figure, the same reference numerals as those in FIG. In the second embodiment, a signal singular value evaluation unit 10 is added. FIG. 6 is a configuration diagram of the signal singular value evaluation unit 10. The signal singular value evaluation unit 10 includes a cumulant matrix singular value decomposition unit 101, a difference value determination unit 102, and a singular value decomposition limit time calculation unit 103.

  Below, operation | movement of the arrival direction estimation apparatus 1B in Embodiment 2 is demonstrated. Note that the operation of each unit other than the signal singular value evaluation unit 10 is the same as that of the arrival direction estimation apparatus 1B in the first embodiment, and details of the same operation are omitted.

  The singular value decomposition unit 101 of the cumulant matrix in the signal singular value evaluation unit 10 performs singular value decomposition on the cumulant matrix output from the cumulant matrix creation unit 6 and outputs the calculated singular value to the difference value determination unit 102.

  The difference value determination unit 102 can take a ratio of the largest and second largest singular values output from the singular value decomposition unit 101 of the cumulant matrix, calculate a time difference value thereof, and singular value decomposition of the result The result is output to the limit time calculation unit 103. Here, the time difference value of the ratio between the largest and second largest singular value was used, but when the number of incoming waves is L, any combination that can be considered from the largest singular value to the Lth singular value Or a time difference value of each of the singular values from the maximum singular value to the L-th singular value may be used.

The singular value decomposable limit time calculation unit 103 compares the difference value of the ratio of the singular values output from the difference value determination unit 102 with a certain threshold value, and when the difference value becomes smaller than the threshold value a plurality of times continuously, The minimum time among them is defined as a limit time necessary for accurately performing singular value decomposition, and is output to the data observation time deriving unit 74 of the phase analyzing unit 7. If the observation is performed in a time shorter than this time, the singular value is not stable at the time of singular value decomposition of the cumulant matrix. As a result, the phase differences Φ ₁₁ and Φ ₂₂ between the real array and the virtual array vary, and the phase analysis unit 7 Therefore, a suitable data observation time cannot be obtained. The threshold level can be set using simulations or measured values, or can be changed according to the SNR.

ここで、σ１（ｔ）は時刻ｔにおけるキュムラント行列の最大特異値、σ２（ｔ）は時刻ｔにおけるキュムラント行列の２番目に大きい特異値である。Δｔは隣り合う時間間隔である。限界時間算出部９３は特異値の比Q(t)及びその所定時間での差分ΔQ(t)を位相分析部７に出力する。 Here, the ratio Q (t) between the largest singular value and the next largest singular value is the ratio of this singular value (16).
The difference of Q (t) at a predetermined time can be expressed as in Expression (17).

Here, σ1 (t) is the maximum singular value of the cumulant matrix at time t, and σ2 (t) is the second largest singular value of the cumulant matrix at time t. Δt is an adjacent time interval. The limit time calculation unit 93 outputs the singular value ratio Q (t) and the difference ΔQ (t) at the predetermined time to the phase analysis unit 7.

  The data observation time deriving unit 74 in the phase analysis unit 7 outputs the time output from the limit time calculation unit 103 capable of singular value decomposition and the suitable data observation time candidate output from the phase evaluation unit 73. The observation time deriving unit 74 compares and selects the smallest value that is larger than the limit time calculation unit 103 capable of singular value decomposition from the candidates for suitable data observation times.

As apparent from the above, according to the second embodiment, the signal singular value evaluation unit 10 that calculates the time difference value of the ratio of the singular values of the cumulant matrix is added to the configuration of the first embodiment, and the phase analysis unit 7 By setting the lower limit value of the observation time using the time difference value, it becomes possible to obtain the minimum data observation time necessary for the arrival direction estimation according to the situation at that time even if the SNR changes. .

As described above, the arrival direction estimation apparatus according to the second embodiment can perform singular value decomposition by comparing the time variation of the ratio of the singular values of the cumulant matrix created by the cumulant matrix creating unit 6 with a predetermined threshold. A signal singular value evaluation unit 10 for calculating a critical time limit, and the phase analysis unit 7 calculates a data observation time candidate based on a temporal change in the diagonal component of the phase rotation matrix, The data observation time is determined by comparison with the limit time that can be decomposed by the singular value calculated by the signal singular value evaluation unit 10. With this configuration, even if the SNR changes, it is possible to obtain the minimum data observation time required for arrival direction estimation according to the situation at that time.

Further, the ratio of the singular values of the cumulant matrix calculated by the signal singular value evaluation unit 10 is a ratio of the maximum singular value of the cumulant matrix to the second largest singular value. With this configuration, it is possible to appropriately perform phase analysis.

Embodiment 3 FIG.
While the cumulant matrix creation unit 6 is handled in the first embodiment, the arrival direction estimation device including the signal correlation matrix creation unit 11 is disclosed in the third embodiment. The third embodiment is based on the premise that the direction of arrival is estimated by an ESPRIT (Estimation of Signal Parameters Via Rotational Invariance Techniques) algorithm.

When the receiving antennas 3-1 to 3 -M can be divided into a linear array or two similar arrays, a signal correlation matrix creating unit 11 that is a matrix creating unit can be arranged instead of the cumulant matrix creating unit 6. FIG. 7 is a block diagram showing an arrival direction estimation apparatus 1C according to Embodiment 1 of the present invention. In the figure, the same reference numerals as those in FIG.

  The signal correlation matrix creation unit 11 creates a signal correlation matrix from the signals output by the reception units 5-1 to 5 -M, and outputs the result to the phase analysis unit 7. Here, the correlation matrix of this signal is a second-order statistic, and in this respect, the signal correlation matrix creation unit 11 is different from the cumulant matrix creation unit 6 that outputs a fourth-order or higher-order cumulant matrix. The part may be the same as in the first embodiment. This is because the ESPRIT algorithm differs from VESPA only in that the correlation matrix of the signal is created from the observed signal vector, and other parts are the same as VESPA.

  Therefore, the processing in the phase analysis unit 7 in the first embodiment can be applied to the ESPRIT algorithm.

  As described above, the arrival direction estimation apparatus according to Embodiment 3 correlates reception antennas 3-1 to 3-M, which are array antennas that receive signals with unknown arrival directions, and reception signals received by the array antennas. A signal correlation matrix creation unit 11 which is a matrix creation unit for creating a matrix and a correlation matrix created by the signal correlation matrix creation unit 11 represent phase rotation caused by a difference in observation position of a signal whose arrival direction is unknown. A phase rotation matrix including information in a diagonal component and calculating a data observation time based on a time change of the diagonal component of the phase rotation matrix; a data observation time determined by the phase analysis unit 7 And a direction-of-arrival estimation unit 8 that performs direction-of-arrival estimation using a received signal corresponding to the above. With this configuration, it is possible to estimate the direction of arrival in a shorter time than in the prior art even when using the ESPRIT algorithm.

Embodiment 4 FIG.
The fourth embodiment is based on the premise that the arrival direction is estimated by the ESPRIT algorithm as in the third embodiment. In the second embodiment, the arrival direction estimation apparatus includes the cumulant matrix creation unit 6 and the signal singular value evaluation unit 10, whereas in the fourth embodiment, the arrival direction estimation apparatus includes the signal correlation matrix creation unit 11 and the signal eigenvalue evaluation. A configuration including the unit 12 is handled.

In the second embodiment, when the receiving antennas 3-1 to 3-M can be divided into a linear array or two similar arrays, the signal correlation matrix creating unit 11 is replaced with the signal singular value evaluating unit 10 instead of the cumulant matrix creating unit 6. Instead, the signal eigenvalue evaluation unit 12 can be arranged instead. FIG. 8 is a block diagram showing an arrival direction estimation apparatus 1D according to Embodiment 4 of the present invention. In the figure, the same reference numerals as those in FIG. A block diagram of the signal eigenvalue evaluation unit 12 is shown in FIG. The signal eigenvalue evaluation unit 12 includes a signal correlation matrix eigenvalue decomposition unit 121, a difference value determination unit 122, and a limit time calculation unit 123 capable of eigenvalue decomposition.

  The signal correlation matrix eigenvalue decomposition unit 121 performs eigenvalue decomposition on the signal correlation matrix, which is the correlation matrix output from the signal correlation matrix creation unit 11, and outputs the result to the difference value determination unit 122.

  The difference value determination unit 122 calculates the time difference value of the eigenvalue output from the eigenvalue decomposition unit 121 of the signal correlation matrix and the second largest one, calculates the time difference value, and calculates the time limit for eigenvalue decomposition. Output to the unit 123.

  The eigenvalue decomposable limit time calculation unit 123 compares the difference value of the ratio of the eigenvalues output from the difference value determination unit 122 with a certain threshold value, and when the difference value is continuously smaller than the threshold value a plurality of times, The minimum time is defined as a limit time that allows accurate eigenvalue decomposition, and is output to the data observation time deriving unit 74 of the phase analysis unit 7.

As described above, the arrival direction estimation apparatus according to the fourth embodiment compares the singular value ratio of the correlation matrix created by the signal correlation matrix creation unit 11 with the predetermined threshold value, thereby comparing the singular value decomposition. A signal eigenvalue evaluation unit 12 for calculating a possible limit time is provided, and the phase analysis unit 7 calculates a data observation time candidate based on the temporal change of the diagonal component of the phase rotation matrix, and the data observation time candidate and The data observation time is determined by comparison with a limit time that can be resolved by the eigenvalue calculated by the signal eigenvalue evaluation unit 12. With this configuration, even when the ESPRIT algorithm is used, even if the SNR changes, the minimum data observation time required for arrival direction estimation can be obtained according to the situation at that time.

Further, the ratio of the eigenvalues of the correlation matrix calculated by the signal eigenvalue evaluation unit 12 is a ratio of the largest singular value of the correlation matrix to the second largest eigenvalue. With this configuration, it is possible to appropriately perform phase analysis.

In Embodiments 1 to 4, each form has been described. However, when these are handled comprehensively, the inventions described in Embodiments 1 to 4 are arranged such that signals with unknown arrival directions are arrayed. A matrix including information on the arrival direction of an unknown signal using the reception signals received by the reception antennas 3-1 to 3-M and the reception antennas 3-1 to 3-M. An arrival direction estimation device that calculates a singular value or an eigenvalue and estimates the arrival direction of a signal whose arrival direction is unknown using the calculated singular value or eigenvalue, and the received signal used for calculating the singular value or eigenvalue The observation time is shorter than the observation time of the received signal required to converge the singular value or eigenvalue phase component, and the calculated singular value or eigenvalue phase component converges to the singular value or eigenvalue phase component. It is characterized in that the same extent as. With this configuration, it is possible to estimate the direction of arrival with the same degree of accuracy as the convergence time in a shorter time than the observation time required for the convergence of the singular or eigenvalue phase component. Direction estimation can be performed.

1A, 1B, 1C, 1D: Arrival direction estimation device, 2: Radiation source, 3-1 to 3-M: Reception antenna, 4: Observation time setting unit, 5-1 to 5-M: Reception unit, 6: Cumulant Matrix creation unit, 7: phase analysis unit, 8: arrival direction estimation unit, 9: direction measurement result output unit, 10: signal singular value evaluation unit, 11: signal correlation matrix creation unit, 12: signal eigenvalue evaluation unit, 71: Phase rotation matrix derivation unit, 72: phase calculation unit, 73: phase evaluation unit, 74: data observation time derivation unit, 101: singular value decomposition unit of cumulant matrix, 102: difference value determination unit, 103: singular value decomposition is possible Limit time calculation unit 121: Eigenvalue decomposition unit of signal correlation matrix 122: Difference value determination unit 123: Limit time calculation unit capable of eigenvalue decomposition

Claims (11)

An array antenna that receives a signal with an unknown direction of arrival;
A matrix creation unit that creates a cumulant matrix of received signals received by the array antenna, and a phase rotation caused by a difference in the observation position of the signal whose arrival direction is unknown, using the cumulant matrix created by the matrix creation unit Calculating a phase rotation matrix including information representing the diagonal component, and determining a data observation time based on a time change of the diagonal component of the phase rotation matrix; and
A direction-of-arrival estimation unit that performs direction-of-arrival estimation using a received signal corresponding to the data observation time determined by the phase analysis unit;
An arrival direction estimating apparatus comprising: The cumulant matrix includes a cumulant calculation for a received signal at the array antenna, and reception at a virtual array antenna at a position where the array antenna is moved by a difference in position between two antennas included in the array antenna. The arrival direction estimation apparatus according to claim 1, wherein the arrival direction estimation apparatus is a matrix obtained by cumulant calculation on a signal. The phase analysis unit obtains a direct current component with respect to the phase component of the diagonal term of the phase rotation matrix, and the time when the obtained direct current component coincides with the time-varying phase component of the diagonal term corresponds to the data observation. The arrival direction estimation apparatus according to claim 1 or 2, wherein time is used. The phase analysis unit obtains a trajectory and a convergence point of a diagonal component of the phase rotation matrix for a plurality of observation times, and an observation time in which the trajectory is in the same direction as the convergence point on a complex plane is set as the data observation time. The direction of arrival estimation apparatus according to claim 1 or 2, wherein A signal singular value evaluation unit that calculates a limit time that can be decomposed by a singular value by comparing a time variation of a ratio of singular values of the cumulant matrix created by the matrix creation unit with a predetermined threshold, and the phase The analysis unit calculates a data observation time candidate based on the temporal change of the diagonal component of the phase rotation matrix, and the singular value decomposition limit calculated by the data observation time candidate and the signal singular value evaluation unit The arrival direction estimation apparatus according to any one of claims 1 to 4, wherein the data observation time is determined by comparison with time. 6. The direction of arrival according to claim 5, wherein the ratio of the singular values of the cumulant matrix calculated by the signal singular value evaluation unit is a ratio of the largest singular value of the cumulant matrix to the second largest singular value. Estimating device. An array antenna that receives a signal with an unknown direction of arrival;
A matrix creating unit that creates a correlation matrix of received signals received by the array antenna;
Using the correlation matrix created by the matrix creation unit, calculating a phase rotation matrix including information representing phase rotation caused by a difference in observation position of the signal whose arrival direction is unknown in a diagonal component, and the phase rotation matrix A phase analysis unit for determining a data observation time based on a time change of a diagonal component of the direction, an arrival direction estimation unit for performing an arrival direction estimation using a received signal corresponding to the data observation time determined by the phase analysis unit,
An arrival direction estimating apparatus comprising: The phase analysis unit obtains a direct current component with respect to the phase component of the diagonal term of the phase rotation matrix, and the time when the obtained direct current component coincides with the time-varying phase component of the diagonal term corresponds to the data observation. The arrival direction estimation apparatus according to claim 7, wherein time is used. The phase analysis unit obtains a trajectory and a convergence point of a diagonal component of the phase rotation matrix for a plurality of observation times, and an observation time in which the trajectory is in the same direction as the convergence point on a complex plane is set as the data observation time. The arrival direction estimation apparatus according to claim 7 or 8, wherein A signal eigenvalue evaluation unit that calculates a time limit of eigenvalue decomposition by comparing the time change of the ratio of eigenvalues of the correlation matrix created by the matrix creation unit with a predetermined threshold,
The phase analysis unit calculates a data observation time candidate based on a temporal change of a diagonal component of the phase rotation matrix, and the eigenvalue decomposition limit calculated by the data observation time candidate and the signal eigenvalue evaluation unit The arrival direction estimation apparatus according to any one of claims 7 to 9, wherein the data observation time is determined by comparison with time. The direction-of-arrival estimation apparatus according to claim 10, wherein a ratio of eigenvalues of the correlation matrix calculated by the signal eigenvalue evaluation unit is a ratio of a maximum eigenvalue of the correlation matrix to a second largest eigenvalue.

Images