CN100373832C - Code filtering self-adaptive wave beam formation method - Google Patents

Abstract

The present invention discloses a method for forming code filtering self-adaption beams, which utilizes pre-multibeam to carry out a time delay search. Array datum before and after despreaded can be calculated in autocorrelation mode, then, two matrixes mutually subtract so as to obtain generalized eigenvector corresponding to a maximum feature value of a difference matrix, and the generalized eigenvector is used as a weight value of the time slot. The beam of the weight value which is found by step 2 forms despreading datum so as to conflate a plurality of paths and despread the conflation of the datum. The method avoids the inversion of a large matrix so as to greatly reduce the operational amount, the autocorrelation matrixs can be found by the datum before and after despreaded and mutually subtract so as to obtain the generalized eigenvector corresponding to the maximal generalized feature value, and thus, the output of the maximum signal-to-noise ratio can be ensured. The formation of self-adaption beams can be realized by the present invention, and thus, the formation of a system is simplified so as to conveniently realize a hardware, research development period can be greatly shortened, and accordingly, the performance of the smart antennas base station processing of a broadband cdma system can be greatly enhanced.

Description

Sign indicating number filtering adaptive beam formation method

Technical field

The present invention relates to a kind of Wideband Code Division Multiple Access (WCDMA) communication system, more specifically refer to a kind of sign indicating number filtering adaptive beam formation method of Wideband Code Division Multiple Access (WCDMA) communication system intelligent antenna base station.

Background technology

Wideband Code Division Multiple Access (WCDMA) WCDMA is a third generation mechanics of communication, compare with two generations because of it better confidentiality is arranged, higher performances such as transmission rate, become the trend of communication system development.And intelligent antenna technology is the key technology of WCDMA system, it help to anti-interference, increase power system capacity and increase the sub-district area coverage.Most crucial technology is exactly an adaptive beam formation technology in the intelligent antenna technology, and smart antenna depends primarily on the performance of adaptive beam formation method to the improvement degree of communication system.Proposed various adaptive beam formation method at present, as permanent modeling method, but this method is a kind of total blindness's method, does not utilize the pilot bit of WCDMA agreement defined, is a kind of waste of resource; Also have some wave beam methods of formationing, as ask dimension to receive solution and cascade training method, all relate to the computing that large-scale matrix multiplies each other and inverts in implementation procedure, complexity and operand are excessive, are not easy to the hardware realization.

Summary of the invention

The shortcoming that exists at above-mentioned traditional wave beam formation method,

One of purpose of the present invention is to propose a kind of improved sign indicating number filtering adaptive beam formation method, only relates to simple addition and multiplication in method, has avoided matrix inversion and large-scale matrix to multiply each other, and has reduced complexity, has reduced operand;

Two of goal of the invention is to utilize the poor matrix that receives data array despreading front and back autocorrelation matrix to obtain the pairing generalized eigenvector of maximum generalized eigenvalue, as adaptive weight, thereby improves one's methods processing accuracy, improves arithmetic speed;

Three of goal of the invention is to generate several switched-beams earlier, beam signal is carried out delay time search, solved the known theory hypothesis of previous method supposition time delay, make method have practicality more, earlier beam signal is carried out delay time search, after utilize adaptive approach to handle, can more effective antagonism channel disturbance, method is become possess the adaptive beam formation method of practical value.

To achieve these goals, the present invention adopts following technical scheme,

Sign indicating number filtering adaptive beam formation method of the present invention may further comprise the steps:

Step 1 utilizes switched-beam to carry out delay time search, specifically comprises: utilize quick attached upright leaf transformation to generate a plurality of fixing narrow beams, wave beam forms and receives data array, to a plurality of beam signal delay time search, obtains the time delay value of every paths;

Array data auto-correlation computation before and after the step 2, despreading, two matrixes subtract each other then, obtain the pairing generalized eigenvector of eigenvalue of maximum of poor matrix, as the weights of current time slots;

Step 3, the weights wave beam that step 2 is obtained forms despread data;

Step 4 merges merging with despread data with mulitpath.

Between described step 1 and step 2, also further may further comprise the steps:

A1), all time slots of judging current frame data whether processing finish, and the despread data if processing finishes with each time slot merges, as the despreading output of this frame; Do not finish if handle, then,

A2), judge whether multipath finishes, if finish, then the despread data complex gain with mulitpath carries out the RAKE merging, if finish, then enters step 2.

Described step 2 specifically may further comprise the steps:

B1), the data without any processing of array received being asked auto-correlation computation, is that unit asks on average with the time slot, obtains

R＝E[x(t)x ^H(t)],?R _xx＝(SlotNum-1)/SlotNum*R _xx+1/SlotNum*R.

The same despreading data afterwards of handling obtain

$R=E\left[y_1\left(t\right)y_1^H\left(t\right)\right]$ And R _Yy=(SlotNum-1)/SlotNum*R _Yy+ 1/SlotNum*R;

B2), the autocorrelation matrix before and after the despreading is subtracted each other, obtain R=R _Yy-R _Xx

B3), matrix R is carried out generalized eigenvalue decompose, obtain the pairing generalized eigenvector a of maximum generalized eigenvalue, obtain the reception weight vector w=a in this this path of time slot;

B4), form the despread data of this time slot with the w wave beam.

In the described step 4,

Mulitpath merged be meant that utilizing complex gain to carry out RAKE to the despread data of mulitpath merges;

Merging to despread data is that the result of merging exports as the despreading of this frame with the despread data merging of each time slot.

Because the present invention has adopted above technical scheme, compare with traditional method, the inventive method has avoided large-scale matrix to invert, greatly reduce operand, utilize the despreading front and back to ask autocorrelation matrix and subtract each other and obtain the pairing generalized eigenvector of maximum generalized eigenvalue, guarantee the output maximum signal to noise ratio; Adopt method of the present invention to realize that adaptive beam forms, can make system constitute the realization of simplifying, be easier to hardware more, the R﹠D cycle also shortens greatly, and the performance that the intelligent antenna of broadband CDMA system base station is handled improves greatly.

Description of drawings

The schematic flow sheet of Fig. 1 for realizing that the inventive method adopted.

The structural principle block diagram of Fig. 2 for realizing that the inventive method adopted.

Fig. 3～Fig. 5 simulation result schematic diagram of the present invention.

Embodiment

The core concept of wave beam formation method of the present invention is to carry out delay time search in the wave beam territory earlier, obtain time delay information more accurately, obtain the autocorrelation matrix of the array baseband signal despreading front and back of time delay alignment, calculate the pairing generalized eigenvector of maximum generalized eigenvalue of their poor matrix, as adaptive weight, utilize these weights to carry out wave beam and form.

According to above-mentioned thought, the wave beam of the present invention side of formation concrete steps are described below,

Utilize switched-beam to carry out delay time search:

At first, suppose that the signal of array received is designated as X, utilize fast fourier transform to generate a plurality of fixing narrow beams earlier, wave beam forms and receives data array, to a plurality of beam signal delay time search, obtains the time delay value of every paths, here we get four paths, and time delay is designated as τ _i, i=1,2,3,4.The present invention utilizes switched-beam to carry out delay time search, has guaranteed the precision of delay time search.

Array data auto-correlation computation before and after the despreading, two matrixes subtract each other then, obtain the pairing generalized eigenvector of eigenvalue of maximum of poor matrix, as the weights of this time slot:

The description of following process is an example with single user's individual paths, and multi-user's mulitpath by that analogy.

Following relevant parameter explanation: R represents the result of auto-correlation computation; E represents the auto-correlation computation of 2 matrixes; The signal data that x (t) representative antennas receives; H represents conjugate matrices; R _XxThe mean value that comprises all data autocorrelation value of current time slots before the expression current time slots; SlotNum represents the sequence of time slots number of current frame data; R _YyExpression comprises the mean value of data autocorrelation value behind all descramblings of this time slot before the current time slots.

At first, the data of each time slot of receiving data array are asked auto-correlation computation, R=E[x (t) x ^H(t)], suppose that current time is a SlotNum time slot, so to comprising that all autocorrelation value of this time slot ask average R before this time slot _Xx=(SlotNum-1)/SlotNum*R _Xx+ 1/SlotNum*R

Simultaneously, with the data delay alignment of the data 211 SlotNum time slot of array received, descrambler 222 by despreader 223 despreadings, obtains the symbolic vector matrix y of this time slot to the Q circuit-switched data ₁, it is carried out auto-correlation computation $R=E\left[y_1\left(t\right)y_1^H\left(t\right)\right],$ Computing is carried out 224, asks on average then to obtain

R _yy＝(SlotNum-1)/SlotNum*R _yy+1/SlotNum*R

The result of autocorrelation matrix adds up in 15 time slots of every frame data, thereby makes the data available length of method reach a frame, and this has guaranteed the handling property of method more.

Again two autocorrelation matrixes are subtracted each other R=R _Yy-R _Xx, matrix R is carried out characteristic value decomposition, find out the pairing generalized eigenvector a of maximum generalized eigenvalue, note w=a is as the wave beam formation weights of this time slot.The present invention improves the back and compares with former method, has avoided large-scale matrix to invert, and greatly reduces operand and implementation complexity, but the method performance does not reduce.And utilize the method for asking autocorrelation matrix before and after the despreading, and obtain the pairing generalized eigenvector of maximum generalized eigenvalue, guarantee to make system have preferable performance by the output maximum signal to noise ratio.

Wave beam forms:

The weights wave beam that the last step was obtained forms despread data, and then wave beam forms and is output as w ^*y ₁

If mulitpath is arranged, mulitpath is carried out RAKE merge.The multipath merged data of each time slot is continuous in proper order, be exactly the reception data of this frame.

Please consult shown in Figure 1ly again, this figure illustrated for the flow process that realizes that said method of the present invention adopted, at this Cheng Tuzhong,

The first step 102 generates a plurality of fixing narrow beams, to the beam data delay time search, selects maximum wave beam, obtains the information in the path delay of time of array data;

In second step 103, all time slots of judging these frame data whether processing finish, and handle if finish then forward 110 to;

The 3rd step 104, judge whether multipath finishes, go on foot 109 processing if finish then forward the 8th to;

The 4th step 105, the data without any processing of array received are asked auto-correlation computation, be that unit asks on average with the time slot, obtain

R＝E[x(t)x ^H(t)],?R _xx＝(SlotNum-1)/SlotNum*R _xx+1/SlotNum*R.

The same despreading data afterwards of handling obtain

$R=E\left[y_1\left(t\right)y_1^H\left(t\right)\right]$ And R _Yy=(SlotNum-1)/SlotNum*R _Yy+ 1/SlotNum*R

The 5th step 106, the autocorrelation matrix before and after the despreading is subtracted each other, obtain R=R _Yy-R _Xx

The 6th step 107, matrix R is carried out generalized eigenvalue decompose, obtain the pairing generalized eigenvector a of maximum generalized eigenvalue, obtain the reception weight vector w=a in this this path of time slot;

The 7th goes on foot 108, forms the despread data of this time slot with the w wave beam;

In the 8th step 109, utilize complex gain to carry out RAKE all multi path despreading data of this time slot and merge;

The 9th step 110 was linked in sequence the RAKE result of each time slot, obtained the reception of this frame.

Please continue to consult shown in Figure 2, this figure illustrated to be the structural principle of realizing that the inventive method adopted, and this structure comprises three modules, that is, delay time search module 21, processing module 22, wave beam forms module 23.

Described delay time search module 21 further comprises K (K=1,2...M) group of received data array 211, switched-beam processing and delay time search module 212,

Described processing module 22 further comprises the preceding correlator 221 of despreading, descrambler 222, and despreader 223, autocorrelator 224 after the despreading, find the solution generalized eigenvector computing module 225;

Described wave beam forms module 23 and further comprises multiplier 231, adder 232;

Described delay time search module 21 at first utilizes switched-beam delay time search module 212 to generate a plurality of fixing narrow beams, utilize these a plurality of fixing narrow beams to carry out wave beam respectively and form reception data array 211, obtain a plurality of beam signals, they are delivered to the delay time search device, carry out delay time search, finding the time delay of maximum energy value place wave beam, promptly is the time delay information of desired user; Processing module 22 does not at first have the time delay alignment to carry out correlator 221 computings to the reception data array of each time slot, it is average with the time slot to be that unit asks then, simultaneously the reception data array time delay of this time slot is alignd, send into descrambler 222, the Q circuit-switched data that descrambling comes out is sent into despreader 223, next the data that despreading is come out are asked autocorrelator 224 computings, the same with correlator 221 the auto-correlation result is asked average, at last the auto-correlation average result of correlator 221 and autocorrelator 224 outputs is input to generalized eigenvector computing module 225, obtains the pairing generalized eigenvector of maximum generalized eigenvalue; The generalized eigenvector that the last step was tried to achieve is designated as the weights that wave beam forms, and the despread data that forms 23 pairs of these time slots of module at wave beam is carried out wave beam formation.

Above process merges as long as mulitpath carries out RAKE at a paths.

And multipath and multi-user are the stacks of unique user individual paths.

The signal of array received becomes baseband signal through radio-frequency front-end and baseband-converted, carries out self-adaptive processing in base band.Baseband signal is because the influence of space channel can produce changes of properties such as time delay expansion and angle spread, if directly carry out delay time search to receiving data array, the precision of time delay can not guarantee.For this reason, generate a plurality of fixing narrow beams earlier, utilize this several wave beam receiving array data respectively, obtain a plurality of beam signals, it is carried out delay time search, and the wave beam of energy value maximum is exactly a desired user place wave beam, and corresponding time delay value is exactly this user's a time delay; With the data auto-correlation computation of array received, and the result of each time slot will add up and ask average, makes the variation of the variation reflection channel of weights like this, and the correlation of front and back time slot is bigger; To receive the data that data array 211, descrambler 222, despreader 223 obtain simultaneously and carry out auto-correlation computation, the result also requires on average; The autocorrelation matrix of correlator 221 before 211 despreadings of reception data array with autocorrelator 224 subtracted each other, obtain the pairing generalized eigenvector of maximum generalized eigenvalue of poor matrix, as the reception weight vector in this this path of time slot; This weight vector is delivered to wave beam formation module 23 carry out wave beam formation, output.

Fig. 3 has illustrated to adopt the emulation beamformer output of wave beam formation method of the present invention, among this figure, array number is 6, number of users is 2, Doppler frequency shift is 20HZ, angle spread is 0 degree, and each user has 1 paths, contains 5 the sub-multipaths that can not differentiate in every distinguishable multipath, noise power is 0.1, maximum delay expands to 10 chips, and the power of signal and interference is 1, supposes that sense is 0 degree, interference radiating way is 30 degree, the ripple of expression desired user reaches discovery (as follows) among the figure, and the weights of simulation result can form maximum the sensing at direction of arrival, are 34.0561dB through the emulation output signal-to-noise ratio.

Among Fig. 4, array number is 6, and number of users is 10, Doppler frequency shift is 20HZ, angle spread is 5 degree, and each user has 3 paths, and noise power is 0.1, maximum delay expands to 10 chips, contain 10 the sub-multipaths that can not differentiate in every distinguishable multipath, the power of signal and interference is 1, and 3 paths directions of each signal and interference produce at random, the weights of simulation result can form maximum the sensing at direction of arrival, and output signal-to-noise ratio is 13.4559dB.

Among Fig. 5, array number is that 6,15 users' Doppler frequency shift all is 20HZ, and angle spread is 5 degree, each user has 3 paths, noise power is 20, and maximum delay expands to 10 chips, contains 10 the sub-multipaths that can not differentiate in every distinguishable multipath, the power of signal and interference is 1,3 paths directions of each signal and interference are for producing at random, and the weights of simulation result can form maximum the sensing at direction of arrival, and output signal-to-noise ratio is 8.4957dB.

Code filtering adaptive beam formation method operand of the present invention is little, complexity is low, performance is excellent, have more practicality, and the method has the following advantages:

The first, the method is not utilized reference signal, has just saved the hardware configuration that produces reference signal yet, implements simpler.

The second, the method can be applied under the various channel circumstances, and several channel circumstances that suggestion is used such as ITU such as macrocell, Microcells can both be exported higher signal to noise ratio and the lower bit error rate.

The 3rd, the method can form at the direction of arrival of desired user the maximum sensing of wave beam, effectively receives the expectation subscriber signal, improves the performance of desired signal processing, can also follow the tracks of user's direction of arrival, adaptive adjusting weight vector.

Claims (4)

1. sign indicating number filtering adaptive beam formation method is characterized in that this method may further comprise the steps:

Step 1 utilizes switched-beam to carry out delay time search, specifically comprises: utilize fast fourier transform to generate a plurality of fixing narrow beams, wave beam forms and receives data array, to a plurality of beam signal delay time search, obtains the time delay value of every paths; Array data auto-correlation computation before and after the step 2, despreading, two matrixes subtract each other then, obtain the pairing generalized eigenvector of eigenvalue of maximum of poor matrix, as the weights of current time slots;

Step 3, the weights wave beam that step 2 is obtained forms despread data;

Step 4 merges merging with despread data with mulitpath.

2. sign indicating number filtering adaptive beam formation method as claimed in claim 1 is characterized in that, between described step 1 and step 2, also further may further comprise the steps:

A1), all time slots of judging current frame data whether processing finish, and the despread data if processing finishes with each time slot merges, as the despreading output of this frame; Do not finish if handle, then,

A2), judge whether multipath finishes, if finish, then the despread data complex gain with mulitpath carries out the RAKE merging, if finish, then enters step 2.

3. sign indicating number filtering adaptive beam formation method as claimed in claim 1 is characterized in that described step 2 specifically may further comprise the steps:

B1), the data without any processing of array received being asked auto-correlation computation, is that unit asks on average with the time slot, obtains

R＝E[x(t)x ^H(t)]，R _xx＝(SlotNum-1)/SlotNum*R _xx+1/SlotNum*R，

The same despreading data afterwards of handling obtain

$R=E\left[y_1\left(t\right)y_1^H\left(t\right)\right]$ And R _Yy=(SlotNum-1)/SlotNum*R _Yy+ 1/SlotNum*R;

B2), the autocorrelation matrix before and after the despreading is subtracted each other, obtain R=R _Yy-R _Xx

B3), matrix R is carried out generalized eigenvalue decompose, obtain the pairing generalized eigenvector a of maximum generalized eigenvalue, obtain the reception weight vector w=a in this this path of time slot;

B4), form the despread data of this time slot with the w wave beam;

Wherein, R represents the result of auto-correlation computation; E represents the auto-correlation computation of 2 matrixes; The signal data that x (t) representative antennas receives; H represents conjugate matrices; Rxx represents that current time slots comprises the mean value of all data autocorrelation value of current time slots before; SlotNum represents the sequence of time slots number of current frame data; Ryy represents to comprise before the current time slots mean value of data autocorrelation value behind all descramblings of this time slot.

4. sign indicating number filtering adaptive beam formation method as claimed in claim 1 is characterized in that, in the described step 4,

Mulitpath merged be meant that utilizing complex gain to carry out RAKE to the despread data of mulitpath merges;

Merging to despread data is that the result of merging exports as the despreading of this frame with the despread data merging of each time slot.

Images