CN101867402A - MIMO system and application method thereof for adaptive antenna selection - Google Patents

Abstract

The invention discloses an MIMO system and an application method thereof for adaptive antenna selection. The MIMO system comprises a transmitter, a receiver and a wireless channel; the process of the antenna selection which is realized in the MIMO system adopts an iterative feedback structure; a subset is selected quickly from a large amount of antenna sets by using the characteristic of 60 GHz channel without accurate channel estimation, so that a subchannel corresponding to an antenna subarray is optimal in the sense of a defined target function; and high speed data transmission is performed by using the selected optimal antenna subarray. The invention provides the rapid and efficient application method for the MIMO system along with low cost, and solves the technical problems of high complexity and low efficiency of the traditional antenna selection method in the large-scale the MIMO system. An experiment proves that: the adopted method can obtain good performance only through several iterations and low computational complexity; and the system can better support the application of the large-scale the antenna array.

Description

Mimo system and application process thereof that a kind of adaptive antenna is selected

Technical field:

The invention belongs to the communications field, relate to a kind of mimo system and application process thereof, especially a kind of adaptive antenna system of selection and signal processing method thereof based on random optimization.

Background technology:

The transmitting terminal of multiple-input and multiple-output (MIMO) system or receiving terminal or both have a plurality of antennas usually to constitute aerial array.Utilize spatial reuse or transmit diversity, mimo system can obtain the gain of channel capacity, improves the transmission performance of system.Therefore, mimo system (or multiaerial system) becomes the first-selected physical layer architecture of following high-speed communication system.The 60GHz communication system, this is considered to provide the system up to number Gbps transmission rate to adopt multi-antenna array as receiving and transmitting front end and not at all surprising.Be devoted to also will to use in a motion of standardized 802.15 working groups of 60GHz fairly large aerial array (the dual-mode antenna array respectively has 32 element antennas) and wave beam forming (beamforming) key technology as physical layer.

Though large-scale aerial array is being employed progressively, the needed independent radio frequency link number of communication system transmit-receive end is but because the restriction of factors such as cost and power consumption can not increase arbitrarily thereupon.Sub-array antenna selects technology to select the sub-array antenna of one group of optimum from all combination of antennas, and it is connected with existing radio frequency link.Because this technology has made full use of larger aerial array with fewer purpose independence radio frequency link, can select to close some radio frequency link or only use fewer purpose antenna to receive and dispatch and can obtain full array gain and receive much concern with lower power consumption and cost, thereby become the key technology that the multiaerial system that reduces system complexity and power consumption is optimized.Undoubtedly, this meaning in the extensive antenna array system of 60GHz is more outstanding.In the existing mimo system, the Antenna Selection Algorithem that is realized supposes that all the wireless channel between the receiver/transmitter is known or can obtains by channel estimating, and the target function (capacity or performance of BER) by optimization sub-array antenna correspondence finally obtains optimum sub-array antenna on this basis.Yet, in the 60GHz communication system of using extensive aerial array, consider that channel matrix is unknown, and dimension being bigger, the estimated value that obtains huge channel matrix accurately is difficult.In addition, more existing iterative algorithms hour can use in number of antennas, but when the antenna number further increases, because the search volume is huge, the algorithmic statement performance that realizes is very poor, the efficient of it line options is comparatively low, makes that the efficient of mimo system aerial array adaptive optimization is lower.

Summary of the invention:

At in view of above-mentioned existing sub-array antenna system of selection mimo system at 60GHz, at a high speed, deficiency under extensive many antenna applications background in the system, particularly the optimized Algorithm of sub-array antenna relies on the precise channels estimated result otherwise is difficult to obtain preferable performance, the present invention is intended to disclose mimo system and the application process thereof that a kind of adaptive antenna is selected, the structure of iterative feedback that this mimo system had adopted, utilize the characteristics of 60GHz channel, comprise the scalar output of noise rather than need not under the condition that precise channels estimates according to receiving terminal, utilize the characteristics of 60GHz channel, from the bigger antenna set of number, choose the antenna subset of an optimum fast, it is connected the back adopts the method for beam forming to carry out high-speed data transmission with radio frequency link, make that the subchannel of selected sub-array antenna correspondence is optimum under defined target function meaning.

The technical problem to be solved in the present invention provides the extensive MIMO antenna array system under a kind of 60GHz of being operated in channel condition and the quick self-adapted optimization method and the signal processing method of sub-array antenna thereof, solve existing mimo system lower problem of antenna optimized choice efficient under above-mentioned application background, thereby effectively reduce the power consumption and the cost of mimo system, improve quick day line options problem of the robustness of system.The present invention includes transmitter, receiver and wireless channel model.

Described transmitter (by signal flow to) comprise information source module, the baseband processing module of making a start (comprise coding, interweave, scrambling, modulation etc.), the beam forming device (multiplier) of making a start, a plurality of parallel radio frequency link of making a start (comprising upconverter, low noise amplifier, linear filter etc.), make a start aerial array and adapter thereof, sub-array antenna select to make a start submodule.The described sub-array antenna submodule of making a start comprises antenna index manager and on-off controller.

Described receiver (by signal flow to) comprise receiving antenna array and adapter thereof, receiving end radio frequency link (comprising frequency converter, filter etc.), receiving end beam forming device (multiplier), receiving end baseband processing module (comprising demodulation, descrambling, deinterleaving, decoding etc.), receiving end mixer (adder), the stay of two nights and sub-array antenna and select the receiving end submodule.Described sub-array antenna receiving end submodule comprises the antenna index manager and on-off controller module, target function estimation logic and iteration are upgraded control module etc.

Described wireless channel is the 60GHz wireless channel, it is characterized by (the line of sight that generally has stronger sighting distance, LOS) component, Mathematical Modeling is represented with the random matrix H that the ranks number equals dual-mode antenna usually, its element is obeys the stochastic variable that Rice distributes.In addition, between receiver/transmitter, there are a low rate, reliable based on feedback link, are used for instructing the renewal selection result of making a start.

Described (receipts) end beam forming device utilizes multiple antenna that signal projection is transmitted in the subspace that is made of the beam forming weights, has reduced the interference between multiple signals, has obtained array (beam forming) gain.

At described (receipts) letter machine place, has a radio frequency link module (comprising frequency mixer, low noise amplifier, filter etc.) at least.

Described receiving end mixer is exported signal that each reception antenna received after according to the weighted sum of receiving end beam forming weights.

Described mimo system has N _tIndividual transmitting antenna and N _rIndividual reception antenna, N here _tAnd N _rAll bigger.N is arranged respectively in transmitter and receiver _tAnd n _rIndividual available radio frequency link, and n _t≤ N _t, n _r≤ N _t

Described mimo system replaces is operated in two different processes, and promptly process one, and dual-mode antenna subarray selection course has N _tIndividual transmitting antenna and N _rIndividual reception antenna, N here _tAnd N _rAll bigger, as 32.In addition, n is arranged respectively in transmitter and receiver _tAnd n _rIndividual available radio frequency link (n _t≤ N _t, n _r≤ N _r).The function that described sub-array antenna chooser module is finished be from Select a kind of combination in the individual possible combination of antennas, make that the pairing wireless subchannel of this sub-array antenna is optimum under the meaning of defined target function.Process two, data transmission procedure, data message, are launched through selected bay in radio frequency link and the process one and the machine that is received receives after the beam forming device is handled by baseband processing module again, finish a data transfer.

In order to solve the problems of the technologies described above, the signal processing method that the present invention also provides the adaptive antenna subarray in the described mimo system to select said method comprising the steps of:

(1) system enters sub-array antenna selection course (being process one), and the antenna index manager generates one group of unduplicated antenna index sequence subset ω at random ⁽⁰⁾With this antenna subset as initial current sub-array antenna ω=ω ⁽⁰⁾Selected sub-array antenna

And the probability Estimation vector in the initialization iteration update controller

(2) sampling, estimation and iterative process: each iterative process is decomposed into n _t+ n _rIndividual sub-iterative process; In k sub-iteration, the antenna index manager substitutes k element in the current antenna subset with a new antenna index at random, and obtaining one only has k the new antenna subset that element is different with current antenna subset

Transmitting training sequence (as complete 1 sequence) and by two antenna subset ω of estimation in the target function estimation logic module ⁽ⁿ⁾With

Pairing target function φ (ω ⁽ⁿ⁾) and

(3) adaptive-filtering process.At first compare φ (ω ⁽ⁿ⁾) and

Size, and the greater is decided to be the current sub-array antenna of next iteration, promptly

${\mathrm{\ω}}^{(n+1)}=\left\{\begin{array}{cc}{\mathrm{\ω}}^{\left(n\right)},& \mathrm{\φ}\left({\mathrm{\ω}}^{\left(n\right)}\right)>\mathrm{\φ}\left({\widetilde{\mathrm{\ω}}}_k^{\left(n\right)}\right)\\ {\widetilde{\mathrm{\ω}}}_k^{\left(n\right)},& \mathrm{\φ}\left({\mathrm{\ω}}^{\left(n\right)}\right)\≤\mathrm{\φ}\left({\widetilde{\mathrm{\ω}}}_k^{\left(n\right)}\right)\end{array}\right.,$

Secondly, check whether it has record in probability Estimation vector π, do not append a record π=[π (ω for it if do not write down then need ⁽ⁿ⁺¹⁾, 0) ^T]; Then adaptive-filtering being done in the record in the current probability Estimation vector handles:

π ⁽ⁿ⁺¹⁾＝[1-μ(n+1)]π ⁽ⁿ⁾

π ⁽ⁿ⁺¹⁾(ω)＝π ⁽ⁿ⁺¹⁾(ω)+μ(n+1)

In the formula, π ⁽ⁿ⁺¹⁾Expression π ⁽ⁿ⁾Probability vector after upgrading once; π ⁽ⁿ⁾(ω) record of expression antenna subset ω in probability vector; μ (n)=1/n is the forgetting factor of adaptive process, increases with iterations and diminishes the intensity that reflection is once upgraded;

(4) select and the iteration renewal process: last in each iterative process, according to the probability Estimation vector after upgrading, the antenna subset of therefrom selecting a probability Estimation value maximum selects antenna subset as this iteration, promptly This result will notify the antenna index manager of making a start to upgrade the current antenna index information that is connected with radio frequency link by based on feedback link by the iteration update controller.In the same way, receiving end antenna index manager obtains this information and executing index upgrade by the inner passage.

(5) sub-iteration continues, up to k＞n _t+ n _r

(6) iterative process finishes if the condition of convergence satisfies, otherwise will change next iterative process over to.

Suppose that channel matrix is H, the optimization aim function definition of certain antenna subset is H ^HThe dominant eigenvalue of H, i.e. λ ₁(H ^HQ), it is equivalent to the main singular value of channel matrix, () ^HBe conjugate transpose.Described step (2) is meant:

(a) be antenna subset ω estimation target function: information source input training sequence and the bay that passes through among the antenna subset ω are launched, through corresponding wireless channel H _ω, after the weighting of receiving end beam forming weight coefficient can obtain the output signal corresponding to k transmitting antenna after merging

1≤k≤n _t, h wherein _kBe the wireless channel of k transmitting antenna to receiving antenna array; V (k) is an additive white Gaussian noise;

(b) in order to eliminate The noise, independently repeat step a) M time, obtain y ^(m)(k), 1≤k≤n _t, 1≤m≤M; And further do following smoothing processing

$\mathrm{\β}\left(k\right)=\frac{1}{M}\left\{\right[y^{\left(1\right)}{\left(k\right)}^H{y}^{\left(2\right)}\left(k\right)+y^{\left(2\right)}{\left(k\right)}^H{y}^{\left(3\right)}\left(k\right)+...+y^{\left(M\right)}{\left(k\right)}^H{y}^{\left(1\right)}\left(k\right)]$

$+\underset{l=1,l\≠k}{\overset{n_t}{\mathrm{\Σ}}}\left|\underset{m=1}{\overset{M}{\mathrm{\Σ}}}{y}^{\left(m\right)}{\left(k\right)}^H{y}^{\left(m\right)}\left(l\right)\right|\}$

Obtain the target function estimated value of antenna subset ω by following formula

$B=\frac{1}{\sqrt{n_t}}\max \{\mathrm{\β}\left(1\right),\mathrm{\β}\left(2\right),...,\mathrm{\β}\left(n_t\right)\}\≈{\mathrm{\λ}}_1\left(H_{\mathrm{\ω}}^H{H}_{\mathrm{\ω}}\right)$

The realization of aforesaid target function estimation logic need not precise channels to be estimated, has reduced the complexity of system to a great extent and has realized cost.

The antenna selection course of described system obtains the sub-array antenna of one group of optimum when finishing, it is connected with radio frequency link constitutes a mimo system of optimizing.Under the hypothesis of quasi-static channel opens, system can enter process two, promptly utilizes beam forming technology technology to carry out high speed data transfer.If channel changes, system can enter process two once more and select optimum sub-array antenna for transfer of data.

Description of drawings:

Fig. 1 is the mimo system structure chart based on beam forming method of the present invention;

Fig. 2 is the flow chart that the present invention is based on sequential stochastic approximation optimized Algorithm;

Fig. 3 is the performed program flow diagram of target function estimation logic of the present invention;

Fig. 4 is an average behavior curve chart of the present invention.

Wherein: be respectively 1 for information source; 2 is the transmitting terminal Base-Band Processing; 3 is the transmitting terminal multiplier; 4 is transmitting terminal beam forming device; 5 is transmitting terminal parallel radio frequency link; 6 is transmitting terminal radio frequency link and antenna a period of time adapter; 7 is the transmitting terminal aerial array; 8 is the transmitting terminal Antenna Selection Module; 8.1 be transmitting terminal antenna index manager; 8.2 be the transmitting terminal on-off controller; 9 is the receiving terminal Antenna Selection Module; 9.1 be receiving terminal antenna index manager; 9.2 be that receiving terminal on-off controller 9.3 is target function estimation logic; 9.4 be the iteration control device; 10 is the receiving terminal aerial array; 11 is receiving terminal radio frequency link and aerial array adapter; 12 is receiving terminal parallel radio frequency link; 13 is receiving terminal beam forming device; 14 is the receiving terminal multiplier; 15 is adder; 16 is the receiving terminal baseband processing module; 17 are the stay of two nights.

Embodiment:

Below in conjunction with accompanying drawing the present invention is done and to describe in further detail:

Referring to Fig. 1,2,3,4, the stochastic approximation optimized Algorithm that MIMO Antenna Selection Module disclosed in this invention adopts sequential scheme is carried out fast optimized choice and then is effectively reduced complexity and the cost that system realizes the sending and receiving end sub-array antenna of the extensive antenna array system of 60GHz.In order to maximize the beam forming gain of mimo system, related optimization aim function definition is the subchannel matrix master singular value of aerial array correspondence in the described method.The target function of related day line options is not that the result by channel estimating calculates and obtains, but utilize the 60GHz channel to have strong these characteristics of sighting distance component, estimate that by the scalar output signal that receives the lower bound of the lid that disk right margin of channel matrix effectively obtains the estimation of target function.

Select the implementation method of module to be elaborated below in conjunction with diagram to adaptive MIMO sub-array antenna of the present invention.

Fig. 1 is mimo system and the adaptively selected structural representation of sub-array antenna based on the beam forming technology of the present invention.

With single user is example, and transmitting-receiving two-end has N respectively _r=10 and N _tThe aerial array of=32 array elements.n _rAnd n _tBe respectively the radio frequency link number that transmitting-receiving two-end can be used, and n _t≤ N _t, n _r≤ N _rSub-array antenna selects module will select n from 32 transmitting antennas exactly _tIndividual and from 10 reception antennas, select n _rIndividual composition sub-array antenna is used for transfer of data, makes this be combined in and has the optimal target functional value in all combinations.For the convenience of explaining, be example explanation, i.e. n only with the adaptively selected of transmitting antenna _t=10; n _r=10=N _rShould be understood that, if only consider reception antenna or consider that simultaneously the adaptively selected of dual-mode antenna is extending naturally of described method, also within the protection range of this invention.

On the whole, be positioned at of the instruction of the antenna chooser module of transmitter and receiver according to selection algorithm, the training sequence of information source being imported by antenna index manager and on-off controller is connected to selected transmitting antenna array element emission, and feed antenna chooser module once more behind the back output invariant signal and estimate and judge back output iteration renewal control information by receive merging by reception antenna behind the wireless channel, next according to this information, the antenna index manager of transmitting-receiving two-end is made accordingly more new element.Finish (program circuit such as Fig. 2 are shown in Figure 3) up to the sub-array antenna selection course.Specifically, as described below:

1) system enters the sub-array antenna selection course, and it is unit matrix that the beam forming weight coefficient of making a start is set

(that is, other elements are zero except that the diagonal entry non-zero) is provided with receiving end beam forming weight coefficient and is

(wherein

Square formation for element complete 1); The antenna index manager generates one group of unduplicated antenna index sequence subset at random,

a _i∈ 1,2 ..., N _tAnd a _i≠ a _j, i ≠ j.With this antenna subset as the initial sub-array antenna that selects

And the vector of the probability Estimation in initialization iteration update controller π=(ω ⁽⁰⁾, 1) ^T

2) enter iterative process.Each iterative process is decomposed into n _tIndividual sub-iterative process.In k sub-iteration, selected k element in the antenna subset to substitute with current with a new antenna index at random, promptly obtaining one only has k the element new antenna subset different with selecting antenna subset

Launch complete 1 training sequence and utilize following steps calculate each antenna subset ( With

) corresponding target function:

C) general, suppose be antenna subset ω estimation target function this moment.Information source is imported complete 1 training sequence signal also by the emission of the bay among the antenna subset ω, through wireless channel H _ω, after the weighting of receiving end beam forming weight coefficient can obtain the output signal corresponding to k transmitting antenna after merging

1≤k≤n _t, h wherein _kBe the wireless channel of k transmitting antenna to receiving antenna array; V (k) is an additive white Gaussian noise.

D) in order to eliminate The noise, independently repeat step a) M time, obtain y ^(m)(k), 1≤k≤n _t, 1≤m≤M.And further, obtain

$\mathrm{\β}\left(k\right)=\frac{1}{M}\left\{\right[y^{\left(1\right)}{\left(k\right)}^H{y}^{\left(2\right)}\left(k\right)+y^{\left(2\right)}{\left(k\right)}^H{y}^{\left(3\right)}\left(k\right)+...+y^{\left(M\right)}{\left(k\right)}^H{y}^{\left(1\right)}\left(k\right)]$

$+\underset{l=1,l\≠k}{\overset{n_t}{\mathrm{\Σ}}}\left|\underset{m=1}{\overset{M}{\mathrm{\Σ}}}{y}^{\left(m\right)}{\left(k\right)}^H{y}^{\left(m\right)}\left(l\right)\right|\}$

Estimator with target function

$B=\frac{1}{\sqrt{n}t}\max \{\mathrm{\β}\left(1\right),\mathrm{\β}\left(2\right),...,\mathrm{\β}\left(n_t\right)\}$

Can prove that estimator B is the estimator of the corresponding target function of sub-array antenna during the strong and M in the direct-view path of channel → ∞.It should be noted that owing in each sub-iteration, estimate two antenna subsets of target function and have only an antenna element difference, therefore in estimation process, only need on the same antenna element, get final product by training sequence of emission.

3) in each sub-iteration of step 2, obtained the target function of two antenna subsets With

Then to do the work of two aspects.Check on the one hand that bigger antenna subset of target function existing record whether in the probability Estimation vector, if record then need to append a record not for it; On the other hand, the adaptive-filtering operation done in every record in the current probability Estimation vector:

π ⁽ⁿ⁺¹⁾＝[1-μ(n+1)]π ⁽ⁿ⁾

π ⁽ⁿ⁺¹⁾(ω)＝π ⁽ⁿ⁺¹⁾(ω)+μ(n+1)

In the formula, π ⁽ⁿ⁾The probability vector of representing the n time iteration; π ⁽ⁿ⁾(ω) expression antenna subset ω is in probability vector

Record; μ (n)=1/n is the forgetting factor of adaptive process, the intensity that expression is once upgraded.

Select and the iteration renewal process.Last in each iterative process according to upgrading probability Estimation vector later, therefrom selected the antenna selection result of the antenna subset of a probability Estimation value maximum as this iteration.Finish if the condition of convergence satisfies, otherwise will change next iterative process over to.

Simulation result:

In the realization of above-mentioned example, use following parameter setting:

■ number of transmit antennas: N _t=32; Transmitting terminal radio frequency link number: n _t=10

■ reception antenna number: N _r=10; Receiving terminal radio frequency link number: n _r=10

The ■ wireless channel model Rician K factor: K=10dB;

■ receiving terminal average signal-to-noise ratio: 10dB

The ■ target function is estimated level and smooth number of times: M=10

Carry out 100 independently Computer Simulations according to the said system parameter, and obtain the average behavior curve of system works as shown in Figure 4.For relatively, also provided the performance of traditional iterative algorithm.Because the number of antennas of instance system is more, can't use the method for exhaustion to obtain theoretical performance, so in all search volumes, 1000 sky line options of selection at random, and find out and wherein have result best and the poorest performance, and compare with the method that is proposed as theoretical performance.

As can be seen, the estimation of sequential random optimization algorithm of the present invention next time of right margin of your disk of lid, its value is slightly larger than real target function, but overall trend is consistent.Can prove that along with the increase of M and K, this gap will be dwindled.The iterative algorithm that the algorithmic statement performance that proposes is more traditional is good, just since the latter efficiency of algorithm is not high under the huge situation in search volume causes, and SEQUENTIAL ALGORITHM of the present invention with each random search be limited to last " near ", help the maintenance of good results.In addition, the method applied in the present invention only needs the iteration of minority, and performance has been better than in selecting at random for 1000 times best, and this has also embodied the method that is adopted and has had higher efficient and progressive optimality.

Above content is to further describing that the present invention did in conjunction with concrete preferred implementation; can not assert that the specific embodiment of the present invention only limits to this; for the general technical staff of the technical field of the invention; without departing from the inventive concept of the premise; can also make some simple deduction or replace, all should be considered as belonging to the present invention and determine scope of patent protection by claims of being submitted to.

Claims (3)

1. a mimo system of selecting based on adaptive antenna is characterized in that: comprise transmitter, receiver and wireless channel; Described transmitter comprises information source module, the baseband processing module of making a start, the beam forming device of making a start, a plurality of parallel radio frequency link of making a start, the aerial array of making a start, the aerial array adapter of making a start, sub-array antenna select to make a start submodule; Described receiver comprises receiving antenna array, receiving antenna array adapter, receiving end radio frequency link, receiving end beam forming device, receiving end baseband processing module, receiving end mixer, the stay of two nights, sub-array antenna selection receiving end submodule; The described sub-array antenna submodule of making a start comprises antenna index manager and on-off controller; Described sub-array antenna receiving end submodule comprises antenna index manager, on-off controller module, target function estimation logic module and iteration and upgrades control module; Described wireless channel is the 60GHz wireless channel.

2. a kind of according to claim 1 application process of the mimo system of selecting based on adaptive antenna is characterized in that:

(1) alternation of described system is at sub-array antenna selection course and data transmission procedure; Described sub-array antenna selection course is meant that sub-array antenna selects module to select the subarray of an optimum to be used for transfer of data in aerial array; Described data transmission procedure is meant and utilizes selected optimal antenna subarray to carry out high speed data transfer in conjunction with the beam forming technology;

(2) system enters the sub-array antenna selection course, and the antenna index manager generates one group of unduplicated antenna index sequence subset ω at random ⁽⁰⁾With this antenna subset as initial current sub-array antenna ω=ω ⁽⁰⁾Selected sub-array antenna

And the probability Estimation vector in the initialization iteration update controller

(3) sampling, estimation and iterative process: each iterative process is decomposed into n _t+ n _rIndividual sub-iterative process; In k sub-iteration, the antenna index manager substitutes k element in the current antenna subset with a new antenna index at random, and obtaining one only has k the element new antenna subset different with selecting antenna subset Transmitting training sequence and by two antenna subset ω of estimation in the target function estimation logic module ⁽ⁿ⁾

Pairing target function φ (ω ⁽ⁿ⁾) and

(4) adaptive-filtering process: at first compare φ (ω ⁽ⁿ⁾) and

Size, and the greater is decided to be the current sub-array antenna of next iteration, promptly

${\mathrm{\ω}}^{(n+1)}=\left\{\begin{array}{cc}{\mathrm{\ω}}^{\left(n\right)},& \mathrm{\φ}\left({\mathrm{\ω}}^{\left(n\right)}\right)>\mathrm{\φ}\left({\widetilde{\mathrm{\ω}}}_k^{\left(n\right)}\right)\\ {\widetilde{\mathrm{\ω}}}_k^{\left(n\right)},& \mathrm{\φ}\left({\mathrm{\ω}}^{\left(n\right)}\right)\≤\mathrm{\φ}\left({\widetilde{\mathrm{\ω}}}_k^{\left(n\right)}\right)\end{array}\right.---\left(1\right)$

Secondly, check whether it has record in probability Estimation vector π, do not append a record π=[π (ω for it if do not write down then need ⁽ⁿ⁺¹⁾, 0) ^T]; Then adaptive-filtering being done in the record in the current probability Estimation vector handles:

$\begin{array}{c}{\mathrm{\π}}^{(n+1)}=[1-\mathrm{\μ}(n+1)]{\mathrm{\π}}^{\left(n\right)}\\ {\mathrm{\π}}^{(n+1)}\left(\mathrm{\ω}\right)={\mathrm{\π}}^{(n+1)}\left(\mathrm{\ω}\right)+\mathrm{\μ}(n+1)\end{array}---\left(2\right)$

In the formula, π ⁽ⁿ⁺¹⁾Expression π ⁽ⁿ⁾Probability vector after upgrading once; π ⁽ⁿ⁾(ω) record of expression antenna subset ω in probability vector; μ (n)=1/n is the forgetting factor of adaptive process, the intensity that reflection is once upgraded;

(5) select and the iteration renewal process: last in each iterative process, according to the probability Estimation vector after upgrading, the antenna subset of therefrom selecting a probability Estimation value maximum selects antenna subset as this iteration, promptly

This result will notify the antenna index manager of making a start to upgrade the current antenna index information that is connected with radio frequency link by based on feedback link by the iteration update controller; In the same way, receiving end antenna index manager obtains this information and executing index upgrade by the inner passage;

(6) sub-iteration continues, up to k＞n _t+ n _r

(7) iterative process continues to satisfy or surpass maximum iteration time up to the condition of convergence; If the Antenna Selection Algorithem convergence then enters data transmission procedure.

3. as the application process of a kind of mimo system based on beam forming method as described in the claim 2, it is characterized in that described step (3) is meant:

A) be antenna subset ω estimation target function: information source is imported complete 1 training sequence and is launched by the bay among the antenna subset ω, through corresponding wireless channel H _ω, after the weighting of receiving end beam forming weight coefficient can obtain the output signal corresponding to k transmitting antenna after merging

1≤k≤n _t, h wherein _kBe the wireless channel of k transmitting antenna to receiving antenna array; V (k) is an additive white Gaussian noise;

B) in order to eliminate The noise, independently repeat step a) M time, obtain y ^(m)(k), 1≤k≤n _t, 1≤m≤M; And further do following smoothing processing

$\mathrm{\β}\left(k\right)=\frac{1}{M}\left\{\right[y^{\left(1\right)}{\left(k\right)}^H{y}^{\left(2\right)}\left(k\right)+y^{\left(2\right)}{\left(k\right)}^H{y}^{\left(3\right)}\left(k\right)+...+y^{\left(M\right)}{\left(k\right)}^H{y}^{\left(1\right)}\left(k\right)]$

$+\underset{l=1,l\≠k}{\overset{n_t}{\mathrm{\Σ}}}\left|\underset{m=1}{\overset{M}{\mathrm{\Σ}}}{y}^{\left(m\right)}{\left(k\right)}^H{y}^{\left(m\right)}\left(l\right)\right|\}---\left(3\right)$

Obtain the target function estimated value of antenna subset ω by following formula

$B=\frac{1}{\sqrt{n_t}}\max \{\mathrm{\β}\left(1\right),\mathrm{\β}\left(2\right),...,\mathrm{\β}\left(n_t\right)\}\≈{\mathrm{\λ}}_1\left(H_{\mathrm{\ω}}^H{H}_{\mathrm{\ω}}\right)---\left(4\right)$

Images