WO2012122744A1

WO2012122744A1 - Method for extracting parameters of multi-dimensional channel

Info

Publication number: WO2012122744A1
Application number: PCT/CN2011/075055
Authority: WO
Inventors: 张建华; 黄程祥; 张平; 刘宝玲
Original assignee: 北京邮电大学
Priority date: 2011-03-15
Filing date: 2011-05-31
Publication date: 2012-09-20
Also published as: CN102098705A; CN102098705B

Abstract

A method for extracting parameters of multi-dimensional channel is provided in the present invention, and the method includes the following steps: S1, calculating ranges of local areas and the number of measuring snapshots contained in the range of each local area according to parameter settings of a measuring device and the information related to the measurement environment in the process of the channel measurement; S2, determining the position of a current measuring snapshot in the range of the local area where it is located according to the calculated results in S1; S3, initiating each path channel parameter of the current measuring snapshot by selecting a serial interference elimination method or by a method based on the correlations of the measuring snapshots according to the determined results in S2; S4, performing iteration operation to each path channel parameter by employing a search strategy with priority, and updating the channel parameters according to the principles that the likelihood function of each path channel parameter is maximized and monotone non-decreasing in the process of the iteration. With the application of the present invention, the efficiency of a multi-dimensional channel parameter extracting process is improved, while the accuracy of the extracted result is ensured.

Description

Multidimensional channel parameter extraction method

Technical field

The present invention relates to wireless communication technologies, and in particular, to a multi-dimensional channel parameter extraction method.

Background technique

With the development of wireless communication technology, in order to effectively carry a large amount of multimedia services, the requirements for the transmission rate and spectrum efficiency of the wireless communication system are getting higher and higher. Multiple input multiple output (Multi-Input Multi-Output, MIMO The technology is a key technology for a new generation of wireless communication systems proposed to improve the transmission rate and spectrum efficiency of wireless communication systems. This technology utilizes signals transmitted and received over multiple antennas to suppress channel fading, and can double the capacity and spectrum utilization of wireless communication systems without increasing bandwidth. But because The performance of MIMO technology is also subject to the space-time fading characteristics of the channel, and the wideband channel will exhibit new frequency selection characteristics. Therefore, a suitable wideband MIMO channel model is established to study broadband MIMO. The spatial, temporal and frequency three-dimensional fading characteristics of the channel are the premise and key to maximize the advantages of MIMO technology.

In the modeling process, it is necessary to measure the broadband wireless channel in the actual geographical propagation environment, obtain a large number of channel impulse responses, and then extract characteristic parameters that characterize the channel characteristics, including the propagation delay of the multipath signal, and leave. Angle, angle of arrival, Doppler shift, polarization complex amplitude coefficient, etc., and then model the channel propagation characteristics of the environment according to the extracted characteristic parameters or correct the existing model, and finally obtain a relatively complete broadband The MIMO channel model provides a reference for transmission technology, resource management and network planning in wireless communication systems.

It can be seen that channel parameter extraction is a key link in the channel modeling process. In order to extract the channel characteristic parameters from the measured channel impulse response, a high-precision, multi-dimensional joint extraction method is needed. At present, the application is more widely used is the spatially alternating generalized expectation maximization method. As an extension of the maximum likelihood estimation method, the method decomposes a multi-dimensional estimation problem into multiple one-dimensional estimation problems, and can jointly estimate various parameters of the multipath signal, and is not limited to the structure of the transmitting and receiving antenna array. .

Figure 1 is a schematic flow chart of an existing channel parameter estimation method. Referring to Figure 1, the process includes:

Step 101: Set a measurement snapshot sequence number k , and start channel parameter extraction on the kth measurement snapshot.

Step 102: Initialize channel parameters of all L- path signals by using a serial interference cancellation method to obtain initial values of channel parameters of all L- path signals.

In step 103, the initial value of the iteration round n is set to 1, and the first round of iteration is started.

Step 104 : Initialize the sequence number l of the multipath signal to 1, and start extracting the channel parameters of the first path signal.

Step 105 calculates a first signal path l conditional expectation.

Step 106, the search path of the first signal l maximizing the likelihood function of channel parameter value, to give the l-path channel parameter signal after the n-th iteration extraction result. In this step 106, the search order so that the likelihood function maximizes propagation delay of the l-path signals, Doppler shift, away from the horizontal angle, vertical angle away from the horizontal angle of arrival, angle of arrival and polarization multiplexing vertical amplitude coefficients For other parameters, when searching for a parameter, other parameters remain fixed and equal to the value after the last update.

Step 107: Determine whether the sequence number l of the multipath signal is equal to the maximum multipath number L. If yes, the nth round of iteration ends, and the channel parameter extraction result of all L- path signals after the n-th iteration is obtained; otherwise, the sequence number l is incremented by 1, and the process returns to step 105.

Step 108, after the n-th wheel is determined iterative channel parameter extracting all L multipath signals whether the result of the result of the same n -1 iterations. If yes, the result is converged. The result is the final channel parameter extraction result of all the L- path signals of the kth measurement snapshot; otherwise, the result is not converged, n is incremented by 1, and the process returns to step 104.

Step 109: Determine whether the sequence number k of the measurement snapshot is equal to the maximum number of snapshots K. If so, the entire channel parameter extraction process ends; otherwise, k is incremented by 1, and the process returns to step 101.

It can be seen from the above process that the existing channel parameter extraction methods mainly have the following two problems:

1 The correlation between measurement snapshots is ignored, and the channel parameters of adjacent measurement snapshots are similar. In the existing method, when channel parameters are extracted for each measurement snapshot, the channel parameter values are initialized independently using a serial interference cancellation method. The serial interference cancellation method is less efficient and takes a long time in the actual execution process, so that the initialization process accounts for the execution time of the entire channel parameter extraction process. 70% or so.

2 It is neglected that multipath signals are usually propagated in the form of clusters, and the channel parameters of the path signals in the same cluster are similar. In the existing method, for each path signal, when searching for the channel parameter value that maximizes the likelihood function, for each channel parameter, a global search is needed in the value field. Such a search strategy is less efficient.

technical problem

The technical problem to be solved by the present invention is how to provide a multi-dimensional channel parameter extraction scheme to improve the efficiency of channel parameter initialization in the channel parameter extraction process, and at the same time improve the efficiency of searching channel parameter values that maximize the likelihood function.

Technical solution

To solve the above technical problem, the present invention provides a multi-dimensional channel parameter extraction method, the method comprising the steps of:

S1 Calculating a range of the local area and a number of measurement snapshots included in the range of each local area according to the parameter setting of the measuring device in the channel measurement process and the related information of the measurement environment; the local area refers to dividing a continuous measurement route Multiple areas

S2, determining, according to the calculation result of step S1, a position of the current measurement snapshot within a range of the local area in which it is located;

S3, according to step S2 The determination result is selected by using a serial interference cancellation method or a method for measuring correlation between snapshots to initialize channel parameters of the current measurement snapshot;

S4 The priority search algorithm is used to iteratively calculate the channel parameters of each path, and the channel parameters are updated in the iterative operation according to the principle that the likelihood function of each path channel parameter is maximized and monotonously reduced.

The measurement device includes a mobile station, and the parameter setting of the measurement device and related information of the measurement environment include: a center carrier frequency f _{c of the} measurement signal, a channel sampling rate f _{s ,} and a mobile station motion speed v . The method for calculating the range d of the local area in step S1 is: d=20 λ =10c/ ( π f _c ), where c represents the speed of light. The method for calculating the number of measurement snapshots included in the range d of each local area is:

The symbol └ · ┘ indicates rounding down. The method for determining the position of the current measurement snapshot in the range of the local area in the step S2 is: calculating ( kK ₀ ) mod Q, where k is the sequence number of the current measurement snapshot, and K ₀ is the initial value of the sequence number of the measurement snapshot. If the result of the above calculation is 0, the current measurement snapshot is the first measurement snapshot in the range of the local area in which it is located; if not, the current measurement snapshot is not the first in the range of the local area in which it is located. Measurement snapshots. If it is determined that the current measurement snapshot is the first measurement snapshot in the range of the local area in which the current measurement snapshot is located, the serial interference cancellation method is used to initialize the path parameters of each path; otherwise, an initialization method based on measuring the correlation between snapshots is adopted. The initialization method based on measuring the correlation between snapshots is: θ _l ⁽⁰⁾ ( k )= θ _l ^{( R )} ( k -1), l =1,..., L where θ _l ⁽⁰⁾ ( k ) indicates the initial value of the channel parameter of the l- th path signal in the kth measurement snapshot, and θ _l ^{( R )} ( k -1) represents the final channel parameter extraction result of the l -th path signal in the k -th measurement snapshot, L is The maximum number of multipaths. Wherein the prioritized search strategy is:

Wherein [theta] _l ⁽ⁿ⁾ represents a channel parameter value of l-path signal after the n-th iteration, the _U-l for the first search field, z (θ _l) is the likelihood function of l-path signal, the first search field is defined as The union of the intervals with a certain multiple of the search step radius, centered on the corresponding channel parameter values of the other L -1 path signals. Wherein, the priority search field is calculated as follows:

Where Δ represents the search step size and σ is a positive integer representing the weight of the search step size. Wherein, updating the channel parameter according to the principle that the likelihood function of each channel parameter is monotonous or not decreases means: whether the judgment formula z (θ _l ^{( n )} ) ≥ z (θ _l ^{( n -1)} ) is established, and if so, channel parameters will be the l-path signal updates the value of [theta] _l ^(n), otherwise continue supplemented focus the search channel parameter priority search area, where [theta] _l ⁽ⁿ⁾ represents the channel parameters of l-path signal after the n-th iteration The value, z ( θ _l ) is the likelihood function of the l- th path signal.

Beneficial effect

The invention utilizes a method for measuring the path parameters of the measurement snapshot based on the method for measuring the correlation between snapshots, so that for most measurement snapshots, the calculation process of the initial values of the channel parameters in the channel parameter extraction process is actually omitted, therefore, The efficiency of channel parameter initialization in the channel parameter extraction process is improved; the iterative operation is performed on each channel parameter by using a prioritized search strategy, which improves the efficiency of searching for channel parameter values that maximize the likelihood function.

DRAWINGS

FIG. 1 is a schematic flowchart of an existing channel parameter extraction method;

2 is a flow chart of a method of the present invention;

3 is a flowchart of a multi-dimensional channel parameter extraction method according to an embodiment of the present invention;

FIG. 4 is a specific flowchart of performing multi-dimensional channel parameter extraction according to an embodiment of the present invention.

Embodiments of the invention

The specific embodiments of the present invention will be further described in detail below with reference to the drawings and embodiments. The following examples are intended to illustrate the invention but are not intended to limit the scope of the invention. image 3 A flowchart of a multi-dimensional channel parameter extraction method according to an embodiment of the present invention. Referring to FIG. 3 and referring to steps S1 to S4 of FIG. 2, the process of the method includes:

Step 201, setting a sequence number k of the measurement snapshot;

Step 202: Initialize channel parameters of the L- path signal by using a serial interference cancellation method or an initialization method based on measuring correlation between snapshots, and obtain initial values of channel parameters of all L- path signals, where L is a positive integer. In step 202, the concept of a local area is defined: dividing a continuous measurement route into multiple local areas. For the first measurement snapshots of each local area, the interference cancellation method used to initialize all the parameters of the L channel signal path; snapshots for other measurements, initialization method for initializing the correlation between measured based on the snapshot, i.e. The initial channel parameter extraction result of the previous measurement snapshot is directly used as the initial value of the channel parameter of all the L- path signals in the measurement snapshot. Thus, for most measurement snapshots, the calculation process of the initial values of the channel parameters in the channel parameter extraction process is actually omitted, and the efficiency of channel parameter initialization is improved.

In step 203, the initial value of the iteration round n is set to 1, and the first round of iteration is prepared.

Step 204 : Initialize the sequence number l of the multipath signal to 1, and prepare to start extracting the channel parameters of the first path signal.

Step 205 calculates a first signal path l conditional expectation.

Step 206, a search using a search strategy with priority so maximizing the likelihood function of the channel parameters of the multipath signals l, l to obtain the channel parameters of the multipath signals in the n-th iteration extraction result. In step 206, the search order of the first signal path l maximizing the likelihood function of propagation delay, Doppler shift, away from the horizontal angle, vertical angle away from the horizontal angle of arrival, angle of arrival and polarization multiplexing vertical amplitude coefficients Such parameters, when searching for a certain parameter, keep the other parameters fixed and fixed to the last updated value. And when searching for each parameter, a priority search field is defined for the parameter, and the priority search field is defined as a certain multiple of the search step (preset value) with the corresponding parameter value of the other L -1 path signal as the center. The union of the intervals of the radius. The search is first performed in the priority search domain to find out whether there is a parameter value that maximizes the likelihood function and monotonically decreases, and if so, updates the corresponding parameter to the parameter value; otherwise, the search in the non-first search domain is satisfied. The parameter value of the above condition (maximizing the likelihood function and monotonously decreasing).

Step 207: Determine whether the sequence number l of the multipath signal is equal to the maximum multipath number L. If yes, the nth round of iteration ends, and the channel parameter extraction result of all L- path signals after the n-th iteration is obtained; otherwise, l is incremented by 1, and the process returns to step 205.

Step 208, after the n-th wheel is determined iterative channel parameter extracting all L multipath signals whether the result of the result of the same n -1 iterations. If yes, the result is converged. The result is the final channel parameter extraction result of all the L- path signals of the kth measurement snapshot; otherwise, the result does not converge, n is incremented by 1, and the process returns to step 204.

Step 209: Determine whether the sequence number k of the measurement snapshot is equal to the preset maximum number of snapshots K. If so, the entire channel parameter extraction process ends; otherwise, k is incremented by 1, and the process returns to step 201. FIG. 4 is a specific flowchart of performing multi-dimensional channel parameter extraction according to an embodiment of the present invention. Referring to Figure 4, the process includes:

Step 301: Calculate a range of the local area and a number of measurement snapshots included in each local area according to the parameter setting of the measurement device and the related information of the measurement environment during the channel measurement process (this step is a preparation step performed before step 201). The measuring device is a tool for measuring channel parameters, and mainly includes a transmitting unit for transmitting data, a receiving unit (that is, a mobile station) for receiving data, a data storage unit for storing data, and a transmitting and receiving antenna. For a certain measurement scenario, the center carrier frequency of the transmitted and received signals during measurement is f _c , and the corresponding carrier wavelength is λ = c / ( 2 π f _c ), where c is the speed of light. According to experience, the range d of the local area is set to 20 times the carrier wavelength, that is, d=20 λ =10c/ ( π f _c ). If the moving speed of the mobile station is v in this scene and the sampling frequency of the channel is f _s , then The number of measurement snapshots Q contained in each local area can be calculated as

Among them, the symbol └ · ┘ means rounding down.

Step 302: Set a sequence number k of the measurement snapshot. (This step corresponds to step 201)

Step 303: Determine, according to the calculation result, a position of the current measurement snapshot in the local area where the measurement snapshot is located, that is, determine whether the kth measurement snapshot is the first measurement snapshot in the local area in which it is located. If yes, go to step 304; if no, go to step 305. The judgment method is as follows: the starting sequence number of the measurement snapshot is K ₀ , K ₀ > 0 and is an integer, and the calculation is: ( k - K ₀ ) mod Q where mod represents the remainder. If the result of the above calculation is 0, the kth measurement snapshot is the first measurement snapshot in the local area in which it is located. If not equal to 0, the kth measurement snapshot is not the first in the local area in which it is located. Measure the snapshot.

Step 304: If the kth measurement snapshot is the first measurement snapshot in the local area in which it is located, the channel interference parameter is used to initialize the channel parameters of all the L path signals included in the local area to obtain all the L path signals. The initial value of the channel parameter. In step 304, the serial interference cancellation initialization method is used for the first measurement snapshot in each local area, which can avoid or mitigate the error accumulation caused by the initialization method based on measuring the correlation between snapshots.

Step 305: If the kth measurement snapshot is not the first measurement snapshot in the local area, the result of the final channel parameter extraction result of the k -1th measurement snapshot is directly used as the L path in the kth measurement snapshot. The initial value of the channel parameter of the signal. Let θ _l ^{( R )} ( k )={A _l ^{( R )} , τ _l ^{( R )} , _{^{Ф 1, l (R),}} θ 1, l (R), Φ 2, l (R), θ 2, l (R), f d, l (R)} denotes the k-th measurement snapshots of l Diameter The final channel parameter extraction result of the signal, the superscript R is an abbreviation of result, indicating the extraction result, where A _l , τ _l , Φ _{1 , l} , θ _{1, l} , Φ _{2, l} , θ _{2, l} , f _d,l respectively represent the polarization complex amplitude matrix of the l- path signal, propagation delay, horizontal exit angle, vertical exit angle, horizontal arrival angle, vertical angle of arrival and Doppler Frequency shift, let θ _l ⁽⁰⁾ ( k )={A _l ⁽⁰⁾ , τ _l ⁽⁰⁾ , Φ _{1, l} ⁽⁰⁾ , θ _{1, l} ⁽⁰⁾ , Φ _{2, l} ⁽⁰⁾ , θ _{2, l} ⁽⁰⁾ , f _d,l ⁽⁰⁾ } represents the initial value of the channel parameter of the l- th path signal in the kth measurement snapshot, and the superscript 0 represents the initial value, then The steps can be expressed as θ _l ⁽⁰⁾ ( k ) = θ _l ^{( R )} ( k -1), l =1, ..., L where L is the maximum multipath number. The above steps 303~305 correspond to step 202.

In step 306, the initial value of the iteration round n is set to 1, and the first round of iteration is started. This step corresponds to step 203.

Step 307 : Initialize the sequence number l of the multipath signal to 1, and prepare to start extracting the channel parameters of the first path signal. This step corresponds to step 204.

Step 308, calculating the firstl The condition of the path signal is expected. This step corresponds to step 205. At step 308 B l Conditional expectation of the path signalx _l (t) is calculated as

among them y (tSaidk Measuring the channel impulse response measured by the snapshot, ands _l(t)=exp{j2Πf _d,l t }C₂( Φ _{2, l},θ _{2, l})A_lC₁ ^T( Φ _{1, l},θ _{1, l})δ (t-τ _l) among them

Indicates the response of the transmit antenna array at the corresponding angle,

Representing the response of the receiving antenna array at a corresponding angle,δ (t) means unit impulse function, symbol ( · )^T Represents the transpose of a matrix.

Step 309, determining the firstl The preferred search field for each channel parameter in the path signal. In step 309, the firstl Priority search field for each channel parameter in the path signal U_l Defined as otherL -1 The corresponding channel parameter value of the path signal is the center, and the union of the intervals with a certain multiple of the search step radius is

Where θ_l ^{( n )} (k)={A_l ^{( n )},τ _l ^{( n )} , Φ _{1, l} ^{( n )} , θ_{1, l} ^{( n )} , Φ _{2, l} ^{( n )} , θ _{2, l} ^{( n )} ,f _d,l ^{( n )}} indicates the firstn After the iterationk Number of measurement snapshotsl Channel parameter value of the path signal, Δ represents the search step size, σ is a positive integer representing the weighting of the search step, and ' U ' to the right of the equal sign indicates the union. In the following steps, for the sake of convenience, without causing confusion, θ_l ^{( n )} (k) abbreviated as θ_l ^{( n )} . Correspondingly, the priority search fields of propagation delay, Doppler shift, horizontal exit angle, vertical exit angle, horizontal arrival angle, and vertical arrival angle are respectively

among them,

Search steps for propagation delay, horizontal exit angle, vertical exit angle, horizontal arrival angle, vertical angle of arrival, and Doppler shift, respectively

The weighting coefficients of the respective step lengths are respectively indicated, and the coefficients are flexibly adjusted according to the proportion of the search step size in the entire value range. E.g σ _τ Generally take 1 , when the search step of the angle domain is 2 °,

Generally take 2 .

Step 310, to find the first signal path l maximizes the likelihood function of propagation delay values first search field. Likelihood function of path signal l

Expressed as

among them

Which symbol

Represents the Kronecker product, ( · ) ^H represents the conjugate transpose of the matrix,

Where the symbol ( . ) ^* denotes a conjugate, X _l ( τ _l , f _{d, l} ) is a matrix of N × M dimensions, and N and M are the total number of receiving and transmitting antennas, respectively. The elements in X _l ( τ _l , f _d,l ) are: X _l,n,m ( τ _l , f _d,l )=exp(- j 2 πf _d,l t _n,m ) · δ ( t - τ _l ) x _l ( tt _n,m ) , where t _n,m represents the time at which the mth transmit antenna is transmitted and the nth receive antenna is received. After obtaining the specific expression of the likelihood function, in this step, the process of finding the propagation delay value that maximizes the likelihood function in the priority search domain can be expressed as

Step 311: Determine whether the likelihood function is monotonous or not. The propagation delay value that was found in the previous step to maximize the likelihood function must satisfy the condition that the likelihood function monotonically decreases, and it is possible that it is the optimal value in the entire delay value range. The condition that the likelihood function monotonically decreases in this step is expressed as z ( Φ _1, _l ^{( n -1)} , θ _1, _l ^{( n- 1)} , Φ _2, _l ^{( n- 1)} , θ _2, _l ^{( n- 1)} , τ _l ^{( n )} , f _d,l ^{( n -1)} ; x _l ) ≥ z ( Φ _1, _l ^{( n -1)} , θ _1, _l ^{( n- 1)} , Φ _2, _l ^{( n- 1)} , θ _2, _l ^{( n- 1)} , τ _l ^{( n- 1)} , f _d,l ^{( n -1)} ; x _l ) If the conditions shown in the above formula are satisfied, then The τ _l ^{( n )} obtained in step 310 is the propagation delay value of the l- th path signal after the n-th iteration, skipping step 312 and directly performing step 313; if not, the τ _l ^{( n )} obtained in step 310 It is not possible that the entire delay is optimal in the value range, and step 312 is performed.

Step 312, to find the first signal path l maximizes the likelihood function of the non-preferential propagation delay value search field. The search process in this step is expressed as

among them

Express

Complementary. Since the likelihood function is inevitably monotonous, if this step is performed, the τ _l ^{( n )} obtained in this step is necessarily the optimal in the entire delay value range.

Step 313, in the first search field to find the first signal path l maximizing the likelihood function of Doppler shift values. The process of finding the Doppler shift value that maximizes the likelihood function in the priority search domain in this step can be expressed as

In step 314, it is determined whether the likelihood function is monotonous or not. The Doppler shift value that was found in the previous step to maximize the likelihood function must satisfy the condition that the likelihood function monotonically does not decrease, and it is possible that it is the optimal value in the entire Doppler shift range. The condition that the likelihood function monotonically decreases in this step is expressed as z ( Φ _1, _l ^{( n -1)} , θ _1, _l ^{( n- 1)} , Φ _2, _l ^{( n- 1)} , θ _2, _l ^{( n- 1)} , τ _l ^{( n )} , f _d,l ^{( n )} ; x _l ) ≥ z ( Φ _1, _l ^{( n -1)} , θ _1, _l ^{( n- 1)} , Φ _2, _l ^{( n- 1)} , θ _2, _l ^{( n- 1)} , τ _l ^{( n )} , f _d,l ^{( n -1)} ; x _l ) If the condition shown in the above formula is satisfied, step 313 The obtained f _d,l ^{( n )} is the Doppler shift value of the l- th path signal after the n-th iteration, skipping step 315 and directly performing step 316; if not, the f _d obtained in step 313 _{, l} ^{( n ) It is} impossible to be optimal in the entire Doppler shift value range, and step 315 is performed.

Step 315, the non-first search domain to find the first signal path l maximizing the likelihood function of Doppler frequency shift values. The search process in this step is expressed as

among them

Express

Complementary. Since the likelihood function is inevitably monotonous, if this step is performed, the f _d,l ^{( n )} obtained in this step is necessarily the optimal in the entire Doppler shift value range.

Step 316, in the first search field to find the first l-path signal level maximizes the likelihood function value departure angle. The process of finding the horizontal leaving angle value that maximizes the likelihood function in the priority search domain in this step can be expressed as

Step 317: Determine whether the likelihood function is monotonous or not. The horizontal leaving angle value that is found in the previous step to maximize the likelihood function must satisfy the condition that the likelihood function monotonically does not decrease, and it is possible that the horizontal value of the entire horizontal leaving angle is the optimal value. The condition that the likelihood function monotonously decreases in this step is expressed as z ( Φ _1, _l ^{( n )} ,θ _1, _l ^{( n- 1)} , Φ _2, _l ^{( n- 1)} ,θ _2, _l ^{( n- 1)} ,τ _l ^{( n )} ,f _d,l ^{( n )};x _l) ≥z ( Φ _1, _l ^{( n -1)} ,θ _1, _l ^{( n- 1)} , Φ _2, _l ^{( n- 1)} ,θ _2, _l ^{( n- 1)} ,τ _l ^{( n )} ,f _d,l ^{( n )};x _lIf the condition shown in the above formula is satisfied, then step 316 is obtained.Φ _{1 ,l} ^{( n )} As the firstn After the iterationl The horizontal value of the path signal leaves the angle value. Skip step 318 and directly perform step 319. If not, then step 316 Φ _{1 ,l} ^{( n )} It is not possible to optimize the entire horizontal departure angle range, and step 318 is performed.

Step 318, looking for the first in the non-first search domainl The likelihood function of the path signal maximizes the horizontal departure angle value. The search process in this step is expressed as

among them

Express

Complementary. Since the likelihood function is inevitably monotonous, if this step is performed, the step is obtained. Φ _{1 ,l} ^{( n )} It must be the best in the entire horizontal departure angle range.

Step 319, the first search field to find the first signal path l maximizing the likelihood function value of the angle away from vertical. The process of finding the vertical departure angle value that maximizes the likelihood function in the priority search domain in this step can be expressed as

In step 320, it is determined whether the likelihood function is monotonous or not. The value of the vertical departure angle obtained by the previous step to maximize the likelihood function must satisfy the condition that the likelihood function monotonically does not decrease, and it is possible that it is the optimal value in the entire vertical departure angle range. The condition that the likelihood function monotonically decreases in this step is expressed as z ( Φ _1, _l ^{( n )} , θ _1, _l ^{( n )} , Φ _2, _l ^{( n- 1)} , θ _2, _l ^{( n- 1)} , τ _l ^{( n )} , f _d,l ^{( n )} ; x _l ) ≥ z ( Φ _1, _l ^{( n )} , θ _1, _l ^{( n- 1)} , Φ _2, _l ^{( n- 1)} , θ _2, _l ^{( n- 1)} , τ _l ^{( n )} , f _d,l ^{( n )} ; x _l ) If the condition shown in the above formula is satisfied, then step 319 is obtained. θ _{1 , l} ^{( n )} is the vertical departure angle value of the l- th path signal after the n-th iteration, skipping step 321 and directly performing step 322; if not, step 319 obtains θ _{1 , l} ^{( n )} It is impossible to optimize the entire vertical departure angle range, and step 321 is performed.

Step 321, the non-first search domain to find the first signal path l maximizes the likelihood function of the angle away from vertical value. The search process in this step is expressed as

among them

Express

Complementary. Since the likelihood function is inevitably monotonous, if this step is performed, the θ _{1, l} ^{( n )} obtained in this step is necessarily the best in the entire vertical departure angle range.

Step 322, in the first search field to find the first l-path signal level maximizes the likelihood function value of the angle of arrival. The process of finding the horizontal arrival angle value that maximizes the likelihood function in the priority search domain in this step can be expressed as

Step 323: Determine whether the likelihood function is monotonous or not. The horizontal arrival angle value found in the previous step to maximize the likelihood function must satisfy the condition that the likelihood function monotonically does not decrease, and it is possible that it is the optimal value in the entire horizontal arrival angle range. The condition that the likelihood function monotonously decreases in this step is expressed as z (Φ _1, _l ^{( n )} ,θ _1, _l ^{( n )} , Φ _2, _l ^{( n )} ,θ _2, _l ^{( n- 1)} ,τ _l ^{( n )} ,f _d,l ^{( n )};x _l) ≥z (Φ _1, _l ^{( n )} ,θ _1, _l ^{( n )} , Φ _2, _l ^{( n- 1)} ,θ _2, _l ^{( n- 1)} ,τ _l ^{( n )} ,f _d,l ^{( n )};x _lIf the condition shown in the above formula is satisfied, then step 322 is obtained. Φ _{2, l} ^{( n )} As the firstn After the iterationl The horizontal value of the path signal reaches the angle value. Skip step 324 and directly perform step 325. If not, then step 322Φ _{2, l} ^{( n )} It is not possible to optimize the entire horizontal arrival angle range, and step 324 is performed.

Step 324, looking for the first in the non-first search domainl The horizontal signal of the diameter signal maximizes the angle of arrival. The search process in this step is expressed as

among them

Express

Complementary. Since the likelihood function is inevitably monotonous, if this step is performed, the step is obtained. Φ _{2, l} ^{( n )} It must be the optimal in the entire horizontal angle of arrival.

Step 325, in the first search field to find the first signal path l maximizing the likelihood function of the vertical angle of arrival values. The process of finding the vertical angle of arrival value that maximizes the likelihood function in the priority search domain in this step can be expressed as

Step 326: Determine whether the likelihood function is monotonous or not. The vertical angle of arrival value found in the previous step to maximize the likelihood function must satisfy the condition that the likelihood function monotonically does not decrease, and it is possible that it is the optimal value in the entire vertical angle of arrival. The condition that the likelihood function monotonously decreases in this step is expressed as z ( Φ _1, _l ^{( n )} ,θ _1, _l ^{( n )} ,Φ _2, _l ^{( n )} ,θ _2, _l ^{( n )} ,τ _l ^{( n )} ,f _d,l ^{( n )};x _l) ≥z ( Φ _1, _l ^{( n )} ,θ _1, _l ^{( n )} , Φ _2, _l ^{( n )} ,θ _2, _l ^{( n- 1)} ,τ _l ^{( n )} ,f _d,l ^{( n )};x _lIf the condition shown in the above formula is satisfied, then step 325 is obtained.θ _{2, l} ^{( n )} As the firstn After the iterationl The vertical arrival angle value of the path signal is skipped to step 327, and step 328 is directly executed. If not, then step 325θ _{2 ,l} ^{( n )} It is impossible to be optimal in the entire vertical angle of arrival range, and step 327 is performed.

Step 327, the non-first search domain to find the first signal path l maximizing the likelihood function of the vertical angle of arrival values. The search process in this step is expressed as

among them

Express

Complementary. Since the likelihood function is inevitably monotonous, if this step is performed, the θ _{2, l} ^{( n )} obtained in this step is necessarily the optimal in the entire vertical angle of arrival.

Step 328 calculates a first polarization multiplexing l-path signal amplitude coefficients. In this step, the polarization complex amplitude coefficient is calculated as

Where the symbol ( · ) ^-1 represents the matrix inversion. This step directly calculates the polarization complex amplitude coefficient, so there is no need to calculate its priority search field in step 308. The above steps 309~328 correspond to step 206.

Step 329: Determine whether the sequence number l of the multipath signal is equal to the maximum multipath number L. If yes, the nth round of iteration ends, and the channel parameter extraction result of all the L- path signals after the n-th iteration is obtained; otherwise, l is incremented by 1, and the process returns to step 308. This step corresponds to step 207.

Step 330: Determine whether the channel parameter extraction result of all the L- path signals after the n-th iteration has converged, and the determination condition is θ _l ^{( n )} = θ _l ^{( n -1)} , l =1, ..., L if the equation If satisfied, the result converges, θ _l ^{( n ) is} the final channel parameter extraction result of all L- path signals in the kth measurement snapshot; otherwise, the result does not converge, n is incremented by 1, and the process returns to step 307. This step corresponds to step 208.

Step 331: Determine whether the sequence number k of the measurement snapshot is equal to the maximum number of snapshots K. If so, the entire channel parameter extraction process ends; otherwise, k is incremented by 1, and the process returns to step 302. This step corresponds to step 209. The above embodiments are merely illustrative of the present invention and are not intended to be limiting of the invention, and various modifications and changes can be made without departing from the spirit and scope of the invention. Equivalent technical solutions are also within the scope of the invention, and the scope of the invention is defined by the claims.

Claims

A multi-dimensional channel parameter extraction method, characterized in that the method comprises the steps of: S1 Calculating a range of the local area and a number of measurement snapshots included in the range of each local area according to the parameter setting of the measuring device in the channel measurement process and the related information of the measurement environment; the local area refers to dividing a continuous measurement route Multiple areas S2, determining, according to the calculation result of step S1, the position of the current measurement snapshot in the range of the local area in which it is located; S3, according to step S2 The determination result is selected by using a serial interference cancellation method or a method for measuring correlation between snapshots to initialize channel parameters of the current measurement snapshot; S4 The priority search algorithm is used to iteratively calculate the channel parameters of each path, and the channel parameters are updated in the iterative operation according to the principle that the likelihood function of each path channel parameter is maximized and monotonously reduced.
The method according to claim 1, wherein the measuring device comprises a mobile station, and the parameter setting of the measuring device and the related information of the measuring environment comprise: a central carrier frequency f c of the measurement signal, a channel sampling rate f s and the moving speed of the mobile station v .
The method according to claim 2, wherein the method of calculating the range d of the local region in step S1 is: d = 20 λ = 10c / ( π f c ), where c represents the speed of light.
The method according to claim 3, wherein said calculating the number of measured snapshots in the range d of each local area is:
The symbol └·┘ indicates rounding down.
The method according to claim 1, wherein the method for determining the position of the current measurement snapshot in the range of the local area in which it is located in step S2 is: calculating ( kK 0 ) mod Q, where k is the sequence number of the current measurement snapshot , K 0 is the initial value of the sequence number of the measurement snapshot. If the calculation result of the above formula is 0, the current measurement snapshot is the first measurement snapshot in the range of the local area in which it is located; if not, the current measurement snapshot is not The first measurement snapshot in the range of its local area.
The method according to claim 1 or 5, wherein if it is determined that the current measurement snapshot is the first in the range of the local area in which it is located For the measurement snapshot, the serial interference cancellation method is used to initialize the parameters of each channel; otherwise, an initialization method based on measuring the correlation between snapshots is adopted.
The method according to claim 6, wherein said initializing method based on measuring correlation between snapshots is: θ l (0) ( k ) = θ l ( R ) ( k -1), l =1, ..., L where θ l (0) ( k ) represents the initial value of the channel parameter of the l- th path signal in the kth measurement snapshot, θ l ( R ) ( k -1) represents the lth in the k -1 measurement snapshot The final channel parameter extraction result of the path signal, L is the maximum multipath number.
The method of claim 7 wherein said prioritized search strategy is:
Wherein [theta] l (n) represents a channel parameter value of l-path signal after the n-th iteration, the U-l for the first search field, z (θ l) is the likelihood function of l-path signal, the first search field is defined as The union of the intervals with a certain multiple of the search step radius, centered on the corresponding channel parameter values of the other L -1 path signals.
The method of claim 8 wherein said prioritized search field is calculated as follows:
Where Δ represents the search step size and σ is a positive integer representing the weight of the search step size.
The method according to any one of claims 1 to 9, wherein the updating the channel parameter according to the principle that the likelihood function of each of the path parameters is monotonous or not decreases means: judging the equation z ( θ l ( n ) ) ≥ Whether z ( θ l ( n -1) ) holds, if yes, updates the channel parameter value of the l- path signal to θ l ( n ) , otherwise continues to search for channel parameters in the complementary set of the priority search domain, where θ l (n) denotes the n-th wheel iterative channel parameter value of l-path signal, z (θ l) is the likelihood function of l-path signal.