An iterative direction finding method for a multi-channel acoustic acquisition array based on local SRP

Abstract

The present invention discloses an iterative direction finding method for a multi-channel acoustic acquisition array based on local SRP, which includes: synchronously sampling signals in the environment through a multi-channel acoustic sensor array to obtain N signals; where N is the number of channels of the multi-channel acoustic sensor array; randomly selecting K channels from the multi-channel acoustic sensor array to construct a reference value of the SRP distribution; accumulating the reference values of the SRP distribution of the selected K channels to obtain the SRP distribution in the observation space; according to the SRP distribution, performing SRP distribution quality evaluation to determine whether to continue iteration and introduce new channel information; and outputting a direction finding estimate after meeting the quality requirements. The present invention can achieve better direction finding performance in different scenarios.

Description

Technical Field

[0001] The present invention relates to the technical field of direction finding, and particularly to an iterative direction finding method for a multi-channel acoustic acquisition array based on local SRP. Background Art

[0002] Array sensor signal processing has important value in modern acoustic acquisition applications. By synchronously collecting signals through multiple sensor channels, the purpose of enhanced signal acquisition is achieved. In the field of acoustic detection, the most common application scenario of array acquisition is source direction finding. By processing the signals of two or more synchronous sensor channels, attributes such as the time delay between signal channels and the signal direction are obtained. Multi-channel signal processing methods represented by classical algorithms such as BEAMFORMING, CAPON, MUSIC, and ESPRIT are currently applied to various scenarios and have achieved good application effects.

[0003] The above traditional signal processing methods involve simultaneous processing and calculation of multi-channel data in the solution, including matrix calculation, eigenvalue decomposition, etc. When the number of channels is large and the sampling frame duration is long, the computing power requirements for such processing increase rapidly. Moreover, for multi-channel sensors, there is redundant information between each channel. It is difficult for traditional methods to flexibly select the number of channels for dynamic adjustment, usually requiring recalculation, which reduces efficiency. Summary of the Invention

[0004] In view of this, the present invention provides an iterative direction finding method for a multi-channel acoustic acquisition array based on local SRP, which decouples the coupled multi-channel processing in the traditional method into more flexible and superimposable processing units through time delay calculation and steering response power method, and uses local array channel data and asymptotic incremental iterative processing for estimation, realizing the simplification and efficiency improvement of complex and large data volume problems, and providing support for its operability.

[0005] The present invention discloses an iterative direction finding method for a multi-channel acoustic acquisition array based on local SRP, which includes the following steps:

[0006] Step 1: A multi-channel acoustic sensor array synchronously samples signals in the environment to obtain N signals; where N is the number of channels of the multi-channel acoustic sensor array;

[0007] Step 2: Randomly select K channels from the multi-channel acoustic sensor array to construct a reference value of the SRP distribution;

[0008] Step 3: Accumulate the reference value of the SRP distribution to obtain the SRP distribution in the observation space;

[0009] Step 4: According to the SRP distribution, perform SRP distribution quality evaluation and output a direction finding estimate.

[0010] Further, step 2 includes:

[0011] Randomly select K channels from the multi-channel acoustic sensor array, and calculate the reference value R(1, k, θ) of the SRP distribution, where k ∈ 2,..., N. The reference value R(1, k, θ) is the distribution of the steering response power function calculated based on the signals of channel 1 and channel k in the search space, and τ is the time delay of the channel.

[0012] Further, the reference value of the SRP distribution adopts a generalized cross-correlation function R 1，k (t(θ)) or probability density distribution P(t(θ), τ 1，k , a), where t(θ) is the time delay generated in the current sensor channel when the signal is incident from the θ direction.

[0013] Further, when using the generalized cross-correlation function R 1，k (t(θ)) as the reference value of the SRP distribution, R(1, k, θ) = R 1，k (t(θ)), where R 1，k (τ) is the generalized cross-correlation function:

[0014] R 1，k (τ) = ∫ψ(f)G 1，k (f)e j2πfτ df

[0015] where ψ(f) is the weighting function of the generalized cross-correlation function, G 1，k (f) is the cross-power spectrum function of the signals of channel 1 and channel k, and f is the frequency;

[0016] The generalized cross-correlation function is weighted and smoothed for optimization before use to improve performance:

[0017] R′ 1，k (τ) = ∫R 1，k (t)w(τ - t)dt

[0018] where w(t) is the weighting window function.

[0019] Further, when using the probability density distribution P(t(θ), τ 1，k , a), based on the time delay τ 1，k already obtained between the signals of channel 1 and channel k

[0020] R(1, k, θ) = P(t(θ), τ 1，k , a)

[0021] Among them, the P function can adopt the distribution function of normal distribution or T distribution, a is the parameter required for probability density distribution. When P adopts normal distribution, τ 1，k is the mean value, and a is the variance.

[0022] Furthermore, step 3 includes:

[0023]

[0024] Among them, D SRP (θ) is the SRP distribution.

[0025] Furthermore, step 4 includes:

[0026] Step 41: If the quality of the SRP distribution quality assessment meets the preset conditions, then output the direction finding estimation;

[0027] Step 42: If the quality of the SRP distribution quality assessment does not meet the preset conditions, then perform the iterative process degradation assessment; if the iterative process degradation assessment does not meet the condition for continuing iteration, then the valuation fails. Otherwise, introduce new sensor channel data, or exclude bad channel data, and calculate the new D SRP (θ), and continue to execute step 41.

[0028] Furthermore, when adding a new channel, calculate the reference value R(1, k, θ) of the SRP distribution of the corresponding channel and superimpose it on the original SRP distribution D SRP (θ); when excluding abnormal channel data, subtract the reference value of the SRP distribution of the corresponding channel from the original SRP distribution D SRP (θ), and continue to execute step 41.

[0029] Furthermore, normalize the data values of the obtained SRP distribution, and divide the data values of the SRP distribution into three categories according to the effective value threshold and the peak value threshold:

[0030] The first category: The values less than the effective value threshold h 1 are defined as invalid values, which are composed of non-intersecting SRP distribution values, errors, and noises, and do not participate in subsequent calculations;

[0031] The second category: The values greater than or equal to the effective value threshold h 1 are defined as valid values, and their proportion is expressed as L 1 ;

[0032] The third category: Among the valid values, the values greater than or equal to the peak value threshold h 2 are defined as peaks, and the peak proportion is expressed as L 2 ; The part greater than the effective threshold h 1 and less than the peak value threshold h 2 is the flat value; The quality of the SRP distribution is defined as:

[0033]

[0034] Among them, The item represents the ratio of the peak value to the effective value, represents the sum of all values greater than the threshold h 1 value, The item represents the ratio of the area under the curve in the effective area. P is a multi-objective penalty term, and the number of peaks found is used as the P value. k 1 , k 2 , k 3 is the weight of the three items, and its value is between [0, 1], which is determined and adjusted according to requirements. Among them,

[0035] k 1 The larger the value, the more inclined to give a higher evaluation to the SRP distribution with sharp peaks;

[0036] k 2 The larger the value, the more inclined to give a higher evaluation to the SRP distribution with obvious prominent peaks;

[0037] k 3 The larger the value, the more inclined to give a higher evaluation to the SRP distribution with fewer peaks; in the potential multi-objective scenario, k 3 should be set to 0.

[0038] Furthermore, when the quality value Q of the SRP distribution is greater than the set threshold Q T , it is considered that the result estimation meets the preset conditions, and the direction finding estimation is directly output; the quality threshold Q T should gradually decrease as the iteration increases and the number of introduced channels increases.

[0039] Due to the adoption of the above technical solutions, the present invention has the following advantages:

[0040] 1. The present invention decouples the coupled multi-channel processing in the traditional method into more flexible superposable processing units through time delay calculation and steering response power method, and uses local array channel data and asymptotic incremental iterative processing for estimation, realizing the simplification and efficiency improvement of complex and large data volume problems, and providing support for its operability.

[0041] 2. The present invention can achieve good direction finding performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments recorded in the embodiments of the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings.

[0043] Figure 1(a) is a schematic diagram of the ideal SRP distribution according to an embodiment of the present invention;

[0044] Figure 1(b) is a schematic diagram of the proportion of different amplitude results of the ideal SRP distribution according to an embodiment of the present invention;

[0045] Figure 2(a) is a schematic diagram of the non-ideal SRP distribution according to an embodiment of the present invention;

[0046] Figure 2(b) is a schematic diagram of the proportion of different amplitude results of the non-ideal SRP distribution according to an embodiment of the present invention;

[0047] Figure 3 is a schematic diagram of the division of the SRP quality determination region according to an embodiment of the present invention;

[0048] Figure 4 is a schematic diagram of the flow of an iterative direction finding method for a multi-channel sound acquisition array based on local SRP according to an embodiment of the present invention;

[0049] Figure 5(a) is a schematic diagram of the SRP distribution corresponding to 5 channels in the initial round according to an embodiment of the present invention;

[0050] Figure 5(b) is a schematic diagram of the SRP distribution of 7 channels in the 3rd round according to an embodiment of the present invention;

[0051] Figure 5(c) is a schematic diagram of the SRP distribution of 10 channels in the 6th round according to an embodiment of the present invention;

[0052] Figure 5(d) is a schematic diagram of the SRP distribution of 13 channels in the 9th round according to an embodiment of the present invention;

[0053] Figure 5(e) is a schematic diagram of the SRP distribution of 16 channels in the 12th round according to an embodiment of the present invention;

[0054] Figure 5(f) is a schematic diagram of the SRP distribution of 22 channels in the 18th round according to an embodiment of the present invention. Detailed implementation manners

[0055] The present invention will be further described in conjunction with the accompanying drawings and embodiments. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art shall fall within the scope of protection of the embodiments of the present invention.

[0056] See Figure 4 , the present invention provides an embodiment of an iterative direction finding method for a multi-channel sound acquisition array based on local SRP, which includes the following steps:

[0057] Step 1: The multi-channel acoustic sensor array performs synchronous multi-channel sampling on the signals in the environment to obtain N signals; where N is the number of channels of the multi-channel acoustic sensor array;

[0058] Step 2: Randomly select K channels from the multi-channel acoustic sensor array to construct the reference value of the SRP (Steering Response Power) distribution;

[0059] Step 3: Accumulate the reference values of the SRP distribution to obtain the SRP distribution in the observation space;

[0060] Step 4: According to the SRP distribution, perform SRP distribution quality evaluation and output the direction-finding estimation.

[0061] In this embodiment, Step 2 includes:

[0062] Randomly select K channels from the multi-channel acoustic sensor array, and calculate the reference value R(1, k, θ) of the SRP distribution, k ∈ 2,..., N. The reference value R(1, k, θ) is the distribution of the steering response power function calculated based on the signals of channel 1 and channel k in the search space, and τ is the time delay of the channel.

[0063] In this embodiment, the reference value of the SRP distribution adopts the generalized cross-correlation function R 1，k (t(θ)) or the probability density distribution P(t(θ), τ 1，k , a), where t(θ) is the time delay generated when the signal is incident from the θ direction corresponding to the current sensor channel.

[0064] In this embodiment, when using the generalized cross-correlation function R 1，k (t(θ)) as the reference value of the SRP distribution, R(1, k, θ) = R 1，k (t(θ)), where R 1，k (τ) is the generalized cross-correlation function:

[0065] R 1，k (τ) = ∫ψ(f)G 1，k (f)e j2πfτ df

[0066] where ψ(f) is the weighting function of the generalized cross-correlation function, G 1，k (f) is the cross-power spectrum function of the signals of channel 1 and channel k, and f is the frequency;

[0067] In practice, due to the influence of various factors such as noise and estimation error, it is impossible to achieve such excellent performance results, and problems such as unclear peaks or "defocusing" may occur. Therefore, some research focuses on optimizing the robustness of the SRP method. That is, the generalized cross-correlation function is optimized by weighted smoothing and then used to improve the performance:

[0068] R′ 1，k R(τ) = ∫R(t)w(τ - t)dt 1，k (t)w(τ - t)dt

[0069] Among them, w(t) is a weighted window function, and constants, Gaussian functions, etc. can be used as the weighting function.

[0070] In this embodiment, when using the probability density distribution P(t(θ), τ 1，k , a), based on the time delay τ between the signals of channel 1 and channel k that has been obtained 1，k , construct the reference value of the SRP distribution:

[0071] R(1, k, θ) = P(t(θ), τ 1，k , a)

[0072] Among them, the P function can use the distribution function of the normal distribution or the T distribution, and a is the parameter required for the probability density distribution. When P uses the normal distribution, τ 1，k is the mean value and a is the variance.

[0073] In this embodiment, step 3 includes:

[0074]

[0075] Among them, D SRP (θ) is the SRP distribution.

[0076] In this embodiment, step 4 includes:

[0077] Step 41: If the quality of the SRP distribution quality assessment meets the preset conditions, output the direction finding estimate;

[0078] Step 42: If the quality of the SRP distribution quality assessment does not meet the preset conditions, perform an iterative process degradation assessment; if the iterative process degradation assessment does not meet the condition for continuing iteration, the valuation fails. Otherwise, introduce new sensor channel data, or exclude bad channel data, and calculate the new D SRP (θ), and continue to execute step 41.

[0079] The main feature of the degradation of the iterative process is that the quality of the result distribution value no longer improves with iteration. Its essence is that there are errors in the system's observations, the quality is too poor, or the system's information is already close to redundancy. In this method, we use a combination of quality information increment assessment and a hard index of iteration number limit for control.

[0080] The incremental evaluation of quality information can be determined by the fact that the distribution quality does not increase or even decreases in successive iterations. In this case, it is advisable to terminate the iteration, output the current estimation result, and issue a warning of inefficiency. Additionally, for systems with a large number of channels, a hard threshold limit for the number of channels is introduced.

[0081] In this embodiment, when adding a new channel, the reference value R(1, k, θ) of the SRP distribution of the corresponding channel is calculated and superimposed on the original SRP distribution D SRP (θ); when excluding abnormal channel data, the reference value of the SRP distribution of the corresponding channel is subtracted from the original SRP distribution D SRP (θ), and step 41 is continued.

[0082] The ideal single-target SRP result shows a maximum only in the correct target direction, with smaller values at other positions. The ideal scenario in a multi-target situation is similar, but there will be multiple target direction peaks. As shown in the single-target direction finding SRP results of an 8-channel array with almost no measurement error in Figures 1(a) and 1(b). The non-ideal SRP results are shown in Figures 2(a) and 2(b).

[0083] Considering the above situation, in this embodiment, refer to Figure 3 , the data values of the obtained SRP distribution are normalized, and according to the effective value threshold and peak threshold, the data values of the SRP distribution are divided into three categories:

[0084] The first category: Values less than the effective value threshold h 1 are defined as invalid values, which are composed of non-overlapping SRP distribution values, errors, and noises, and do not participate in subsequent calculations;

[0085] The second category: Values greater than or equal to the effective value threshold h 1 are defined as effective values, and their proportion is represented as L 1 ;

[0086] The third category: Among the effective values, values greater than or equal to the peak threshold h 2 are defined as peaks, and the peak proportion is represented as L 2 ; The part greater than the effective threshold h 1 and less than the peak threshold h 2 is the flat value; The quality of the SRP distribution is defined as:

[0087]

[0088] Among them, represents the proportion of the peak in the effective value, represents the sum of all values greater than the threshold h 1 , The item represents the proportion of the area under the curve in the effective region, P is the multi-objective penalty term, and the number of peaks found is used as the value of P, k 1 , k 2 , k 3 is the weight of the three items, with values between [0, 1], determined and adjusted according to requirements. Among them,

[0089] k 1 The larger the value of k, the more inclined to give a higher evaluation to the SRP distribution with sharp peaks;

[0090] k 2 The larger the value of k, the more inclined to give a higher evaluation to the SRP distribution with obvious prominent peaks;

[0091] k 3 The larger the value of k, the more inclined to give a higher evaluation to the SRP distribution with fewer peaks; in the potential multi-objective scenario, k 3 should be set to 0.

[0092] In this embodiment, when the quality value Q of the SRP distribution is greater than the set threshold Q T , it is considered that the result estimation meets the preset conditions, and the direction finding estimation is directly output; the quality threshold Q T should gradually decrease as the iteration increases and the number of introduced channels increases.

[0093] For the sake of easy understanding, the present invention gives a more specific embodiment:

[0094] Now, the method in the present invention will be further elaborated with reference to an example. Here, a 64-channel planar acoustic array with a central position deployed at the origin (0, 0, 0), an aperture of about 0.18 meters, and the positive direction facing the positive Z-axis is simulated as the signal acquisition device. Time delay introduction calculation is used, and the normal distribution is used as the reference value R(1, k, θ).

[0095] Step 1: The acoustic sensor array sampling stage. Here, a target is set at (30, 30, 30). Relative to the origin, the corresponding azimuth angle and elevation angle are 45° and 35.3° respectively. Since the array aperture is small relative to the target distance, it can be approximately regarded as a plane wave incident.

[0096] Step 2: Initially randomly select 5 channels k 1 , k 2 , k 3 , k 4 , k 5 , and calculate the time delay of each of these 5 channels relative to channel 1 through the spatial position and the speed of sound To simulate the estimation error in practice, a zero-mean random error e is added to each time delay here, and 5% of the maximum time delay measurement result is used as the standard deviation of this error in the simulation.

[0097] The probability density distribution function of the normal distribution is used as the reference value for constructing the SRP distribution:

[0098]

[0099] Here, t(θ) is the theoretical time difference for the plane wave incident in the θ direction to reach the reference node 1 and the selected node k n which is calculated through the geometric relationship based on the positions of the array sensors. The search space of θ is set to azimuth -90° to 90°, elevation -90° to 90°, and resolution 1° in this simulation. The standard deviation σ is set to 3% of the maximum theoretical time delay difference. The maximum theoretical time delay difference here refers to the distance between the two farthest sensors on the array divided by the speed of sound.

[0100] Step 3: Accumulate the SRP results that need to be introduced into the calculation to obtain the SRP distribution D SRP (θ) in the observation space.

[0101]

[0102] Step 4: Quality assessment step. Here, 0.35 and 0.7 are used as the thresholds for the invalid value and the peak value after normalization. In this single-target scenario, the case of multiple peaks is not considered, and the following weight assignment and formula are used:

[0103]

[0104] The iterative effect of the algorithm is shown in Figure 5.

[0105] Step 5: Evaluate the degree of deterioration of the iteration. In this iteration, a hard threshold of 50% of the number of channels is set, but the distribution quality reaches the threshold requirement before reaching this hard threshold. As can be seen from Figures 5(a) to 5(e) it, the distribution quality is in a steady improvement process during the iteration. The quality threshold of the system is initially set to 0.98, and the threshold is reduced by 1% after each round of iteration. Finally, the threshold is reached after introducing 22 channels.

[0106] Step 6: Introduce new information. After each iteration, a new channel k n+1 is randomly selected, and after calculating the time delay τ 1，n+1 , the reference value R(1, k, θ) of the SRP of the new channel is calculated separately and superimposed on the original SRP result.

[0107] The final algorithm reached the threshold after introducing 22 channels, and the peak result is shown in Figure 5(f) as (45°, 36°), and the result is correct. It can be seen that through the method described in this patent, good direction-finding performance is achieved without the need for high-dimensional matrix calculations and global channel data synchronization processing.

[0108] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that: modifications or equivalent replacements can still be made to the specific embodiments of the present invention, and any modifications or equivalent replacements that do not depart from the spirit and scope of the present invention should be covered by the protection scope of the claims of the present invention.

Claims

1. An iterative direction finding method for a multi-channel acoustic acquisition array based on local SRP, characterized in that, it includes the following steps: Step 1: The multi-channel acoustic sensor array performs synchronous multi-channel sampling on the signals in the environment to obtain N signals; where N is the number of channels of the multi-channel acoustic sensor array; Step 2: Randomly select K channels from the multi-channel acoustic sensor array to construct a reference value of the SRP distribution; Step 3: Accumulate the reference values of the SRP distribution to obtain the SRP distribution in the observation space; Step 4: According to the SRP distribution, perform SRP distribution quality evaluation and output a direction finding estimate; The said Step 4 includes: Step 41: If the quality of the SRP distribution quality evaluation meets the preset conditions, output the direction finding estimate; Step 42: If the quality of the SRP distribution quality assessment does not meet the preset conditions, perform an iterative process degradation assessment; if the iterative process degradation assessment does not meet the conditions for continuing the iteration, the valuation fails, otherwise, introduce new sensor channel data, or exclude bad channel data, and calculate the new D SRP (θ), and continue to execute Step 41; When adding a new channel, calculate the reference value R(1,k,θ) of the SRP distribution of the corresponding channel and superimpose it on the original SRP distribution D SRP (θ); when excluding abnormal channel data, subtract the reference value of the SRP distribution of the corresponding channel from the original SRP distribution D SRP (θ), and continue to execute step 41; Normalize the data values of the obtained SRP distribution, and divide the data values of the SRP distribution into three categories according to the effective value threshold and the peak value threshold: Category 1: Less than the valid value threshold h 1 The value is defined as an invalid value, which consists of non-intersecting SRP distribution values, errors, and noises, and does not participate in subsequent calculations; The second category: greater than or equal to the effective value threshold h 1 is defined as the effective value, and its proportion is expressed as L 1 ; Category 3: Among the valid values, those greater than or equal to the peak threshold h 2 are defined as peaks, and the peak ratio is expressed as L 2 ; those greater than the effective threshold h 1 and less than the peak threshold h 2 are gentle values; the quality of the SRP distribution is defined as: Among them, The item represents the ratio of the peak value to the effective value, represents the sum of all values greater than the threshold h 1 value, The item represents the ratio of the area under the curve in the effective region. P is a multi-objective penalty term, and the number of peaks found is used as the P value. k 1 , k 2 , k 3 are the weights of the three items, and the values are between [0, 1], which are determined and adjusted according to requirements. Among them, k 1 The larger the value, the more inclined to give a higher evaluation to the SRP distribution with a sharp peak; k 2 The larger the value, the more inclined to give a higher evaluation to the SRP distribution with an obvious prominent peak; k 3 The larger the value of, the more inclined to give a higher evaluation to the SRP distribution with fewer peaks; in the potential multi-objective scenario, k 3 should be set to 0.

2. The method according to claim 1, characterized in that, the said Step 2 includes: Randomly select K channels from the multi-channel acoustic sensor array, calculate the reference value R(1,k,θ) of the SRP distribution, k ∈ 2,…,N, and the reference value R(1,k,θ) is the distribution of the steering response power function calculated based on the signals of channel 1 and channel k in the search space, and τ is the time delay of the channel.

3. The method according to claim 2, characterized in that, The reference value of the SRP distribution adopts the generalized cross-correlation function R 1，k (t(θ)) or the probability density distribution P(t(θ), τ 1，k , a), where t(θ) is the time delay generated in the current sensor channel when the signal is incident from the θ direction; τ 1，k is the time delay between the signals of channel 1 and channel k, and a is the parameter required for the probability density distribution.

4. The method according to claim 3, characterized in that, When using the generalized cross-correlation function R 1，k (t(θ)) as the reference value of the SRP distribution, R(1,k,θ) = R 1,k (t(θ)), where R 1,k (τ) is the generalized cross-correlation function: R 1,k (τ) = ∫ψ(f)G 1,k (f)e j2πfτ df where ψ(f) is the weighting function of the generalized cross-correlation function, and G 1,k (f) is the cross-power spectral function of the signals of channel 1 and channel k, and f is the frequency; Perform weighted smoothing type optimization on the generalized cross-correlation function and then use it to improve the performance: R′ 1,k (τ) = ∫R 1,k (t) w (τ - t)dt where w(t) is a weighted window function.

5. The method according to claim 3, characterized in that, When using the probability density distribution P(t(θ), τ 1，k , a), based on the time delay τ 1，k already obtained between the signals of channel 1 and channel k, construct the reference value of the SRP distribution: R(1, k, θ) = P(t(θ), τ 1，k , a) Among them, the P function can adopt the distribution function of the normal distribution or the T distribution, a is the parameter required for the probability density distribution. When P adopts the normal distribution, τ 1，k is the mean value, and a is the variance.

6. The method according to claim 2, characterized in that, the said Step 3 includes: where D SRP (θ) is the SRP distribution.

7. The method according to claim 1, characterized in that, When the quality value Q of the SRP distribution is greater than the set threshold Q T then it is considered that the result estimation meets the preset conditions, and the direction finding estimation is directly output; the quality threshold Q T should gradually decrease as the iteration increases and the number of introduced channels increases.

Images