CN104219188A - Method for searching double-end time-domain wave beams by aid of compressed sensing - Google Patents

Abstract

The invention belongs to the technical field of wireless communication, and particularly relates to a method for searching the optimal wave beam vectors by the aid of compressed sensing. The method is applied to multi-antenna orthogonal frequency division multiplexing (OFDM) communication systems. The method is applied to searching the optimal wave beam vectors by the aid of compressed sensing double-end time-domain wave beams in the multi-antenna OFDM communication systems. The method has the advantages that wave beam searching is converted into compressed sensing by the aid of the sparsity of departure angles and arrival angles, the compressed sensing is combined with the symmetry of communication channels, and accordingly the optimal emission and receiving wave beam vectors can be determined by the aid of the repeated iteration method.

Description

Utilize the both-end time-domain wave beam searching method of compressed sensing

Technical field

The invention belongs to wireless communication technology field, be specifically related to employing compressed sensing in duplicating multi-antenna orthogonal frequency division (Orthogonal Frequency Division Multiplexing, OFDM) communication system to search for the method for optimal beam vector.

Background technology

UWB system and 60GHz system are mainly used in short distance high-speed transfer, have wide range of applications, and comprise Wireless Personal Network (WPAN, Wireless Personal Area Network), WirelessHD multimedia interface, imaging of medical, trailer-mounted radar etc.In order to adapt to the needs of the aspect such as High Data Rate and high power system capacity, UWB system and 60GHz system often utilize multi-antenna multi-carrier-wave technology for transmitting data.

Multi-antenna technology comprises multiple-input and multiple-output (Multiple Input Multiple Output, MIMO), multiple input single output (Multiple Input Single Output, and single input and multi-output (Single Input Multiple Output, SIMO) MISO).Beam forming technique based on array antenna utilizes the directivity of signal transmission to improve signal to noise ratio (Signal to Noise Ratio, SNR), suppresses interference, improves systematic function.

Array antenna have impact on the correlation of channel space in the distribution situation in space, beam forming technique in smart antenna make use of this correlation and processes signal, the radiation beam producing high directivity in the desired direction strengthens useful signal, zero lobe direction is aimed at interference source and is reached inhibitory action, improves signal to noise ratio thus and increases transmission range.Receive/make a start application antenna array beam be shaped there is following advantage: first, reduce the requirement to power amplifier.If during transmitting terminal use individual antenna, very high to PA gain requirement.If transmitting terminal uses aerial array to send signal, before each bay, increase a power amplifier, like this by using the PA of multiple lower-wattage gain just can meet transmitting power requirement.Secondly, antenna array beam is shaped and is convenient to directional transmissions.In the constant situation of transmitting power, equivalence increases the power of receiver Received signal strength, effectively can also reduce multi-path delay spread simultaneously.The baseband design of transceiver can be simplified like this, reduce the resolution index of analog-digital converter.Finally, antenna array system dynamically adjusts the direction of wave beam, obtains maximum power and reduce the power in other directions to make desired orientation.Not only improve signal-to-jamming ratio like this, also improve the capacity of system, expand system communication coverage, reduce transmitting power requirement.

OFDM is the one of multi-carrier modulation.Its main thought is: channel is divided into some orthogonal sub-channels, high-speed data signal is converted to parallel low speed sub data flow, is modulated to and transmits on each of the sub-channels.Orthogonal signalling by adopting correlation technique to separate at receiving terminal, can reduce between subchannel and mutually disturbing ISI like this.Signal bandwidth on every sub-channels is less than the coherence bandwidth of channel, and the flatness of can regarding as therefore on every sub-channels declines, thus can eliminate intersymbol interference.And due to the bandwidth of every sub-channels be only the sub-fraction of former channel width, channel equalization becomes relatively easy.Beam forming based on OFDM needs before transmitting terminal antenna, do inverse fast Fourier transform conversion, and receiving terminal does fast Fourier transform conversion and carrys out demodulation.

Beam switchover is a kind of beam search rule, and it all pre-sets wave beam and controls vector code book at transmitter and receiver two ends, only need therefrom to choose during use.Therefore, switching-beam is formed also referred to as the beam forming based on code book, uses switched antenna array, and before transmission packet, transmitter repeatedly will send and carry the information that different beams controls vector.

Based on the beam forming technique of channel condition information, transmitter and receiver can find an optimum beam forming control vector.Its method detailed can reference: Yoon S, Jeon T, Lee W.Hybrid beam-forming and beam-switching for OFDM based wireless personal area networks [J] .Selected Areas in Communications, IEEE Journal on, 2009,27 (8): 1425-1432. physical layer (PHY) solutions can provide optimum systematic function, beam forming operation is often considered to carry out in physical layer, but obtain complete channel condition information will very high time cost and expense.Beam forming technique based on code book contributes to reducing complexity and expense, and code book both can design according to base band signal process completely, also can realize in conjunction with key-course (MAC).

Search strategy during beam search is vital, efficient beam search strategy effectively can reduce search time, suppose that transmitting terminal has N number of launching beam vector, M received beam vector, then need at most N × M search, 802.15.3c have employed the codebook structure of two-stage in: a fan-shaped code book and a wave beam code book, each column vector of wave beam code book represents a wave beam, each beam pattern represents an accurate direction, each sector is the set of several wave beam, represent wider direction in space, add up and cover whole space in all sectors.Search procedure is also divided into two benches: the first stage is finding optimum sector according to signal to noise ratio, second stage finds optimum wave beam in the sector of optimum.Its method detailed can reference: Wang J, Lan Z, Pyo C W, et al.Beam codebook based beamforming protocol for multi-Gbps millimeter-wave WPAN systems [J] .Selected Areas in Communications, IEEE Journal on, 2009,27 (8): 1390-1399..

Beam search strategy stage by stage significantly can lower searching times, but when aerial array is very large, the searching times of needs remains huge.Therefore, study one fast and effectively beam search algorithm be one and have novelty and important practical usage and the challenging task of tool.

Summary of the invention

The invention provides a kind of a kind of both-end frequency domain wave beam of compressed sensing that utilizes in multiple antennas ofdm communication system to search for the method for optimal beam vector.The method utilizes the openness problem by beam search of the angle of departure, the angle of arrival to be converted into the problem of compressed sensing, uses different transmitting and receiving vectors by transmitting terminal and receiving terminal, determines separately optimum transmitting/receiving beam vector by receiving terminal.

The object of the invention is to realize as follows:

S1, make the dual-mode antenna number of equipment 1 be Nt, the wave beam number in the code book of described equipment 1 is Ct, and described equipment 1 uses omnidirectional antenna to launch to equipment 2, and launching beam vector is ${\stackrel{\→}{w}}_1=\left[\begin{array}{c}1\\ 0\\ .\\ .\\ .\\ 0\end{array}\right],$ Described launching beam length is Nt, described equipment 1 use OFDM technology frequency domain transmission sequence for [1,1 ..., 1], the sequence length of described transmitting is N

Make the dual-mode antenna number of equipment 2 be Nr, the wave beam number in the code book of described equipment 2 is Cr, and m the sub-carrier signal vector that the antenna of described equipment 2 receives is the receiving terminal of described equipment 2 uses P _rindividual reception vector carrys out Received signal strength, and any one receives vector the vector of to be all length be Nr, value Stochastic choice from set [1, i ,-1 ,-i] of each element in described reception vector, forms a calculation matrix described calculation matrix Φ _reach provisional capital correspondence once receives, and sub-carrier signal is measuring-signal vector is wherein, the noise vector of to be length be Nr, H _mbe the exponent number of m frequency be Nr × Nt channel matrix, the element representation that in described channel matrix, xth row y arranges from transmitting terminal y root antenna to receiving terminal xth root antenna frequency domain channel impulse response, wherein, m=1,2 ..., N, y=1,2, ..., Nt, x=1,2 ..., Nr, d=1,2 ... P _r, i is imaginary unit, noise vector, in the corresponding measured value of each element, () ^ttranspose of a matrix computing, P _rfor being greater than the integer of 1, N, Nt, Nr, Ct and Cr are the integer being greater than 1;

S2, according to S1, build dictionary matrix be an angle in each row corresponding [-90 °, 90 °] of D, D, signal described in S1 can launch under D, and be sparse, expansion coefficient is plural number, be expansion coefficient under D;

S3, use the orthogonal matching pursuit algorithm of N number of task (Orthogonal Matching Pursuit, OMP) to combine sub-carrier signal described in all S1 jointly recover all be specially:

S31、 ${\stackrel{\→}{y}}_{m,r}={\mathrm{\Φ}}_r{\stackrel{\→}{g}}_{m,r}={\mathrm{\Φ}}_{r}D{\stackrel{\→}{\mathrm{\ω}}}_{m,r}+{\stackrel{\→}{n}}_{m,r}^{\′}=V_r{\stackrel{\→}{\mathrm{\ω}}}_{m,r}+{\stackrel{\→}{n}}_{m,r}^{\′},$ V _r=Φ _rd, described in can at V _rlower expansion, be at V _runder expansion coefficient;

S32, from V described in S31 _rin find row make maximum, structural matrix calculate all at V _cunder expansion coefficient $[{\stackrel{\→}{\mathrm{\ω}}}_{1,r,c},{\stackrel{\→}{\mathrm{\ω}}}_{2,r,c},...,{\stackrel{\→}{\mathrm{\ω}}}_{N,r,c}]={\left({V_c}^H{V}_c\right)}^{-1}{V_c}^H[{\stackrel{\→}{y}}_{1,r},{\stackrel{\→}{y}}_{2,r},...,{\stackrel{\→}{y}}_{N,r}],$ Represent the surplus of current recovery extent $e_r=[{\stackrel{\→}{y}}_{1,r},{\stackrel{\→}{y}}_{2,r},...,{\stackrel{\→}{y}}_{N,r}]-V_c[{\stackrel{\→}{\mathrm{\ω}}}_{1,r,c},{\stackrel{\→}{\mathrm{\ω}}}_{2,r.c}...,{\stackrel{\→}{\mathrm{\ω}}}_{N,r,c}],$ Wherein, () ^-1the inversion operation of matrix, () ^hthe conjugate transpose operation of matrix, || represent the amplitude of getting plural number, || || ₂represent two norm computings of vector;

S33, from V _rin find make maximum, wherein matrix e _rin m row, will be added to V described in S32 _cin obtain upgrade after calculate v in the updated _cunder expansion coefficient;

S34, circulation S34 to S33, until e _rf norm be less than the a times of F norm time stop, the coefficient linear combination in conjunction with the column vector in dictionary matrix D described in S2 and correspondence position recovers wherein, a is threshold value, 0<a<1, and a is real number;

S4, recover all frequencies be denoted as the beam vector that one applicable is found from code book make spectrum efficiency maximum, ${\stackrel{\&RightArrow;}{c}}_1=\underset{\stackrel{\&RightArrow;}{c}}{\arg }\max \left(\frac{1}{N}\underset{m=1}{\overset{N}{\mathrm{\&Sigma;}}}{\log }_2(1+{\mathrm{\&gamma;}}_{r,m})\right),$ Wherein ${\mathrm{\&gamma;}}_{r,m}=\frac{{\left|{\stackrel{\&RightArrow;}{c}}^T{\widehat{\stackrel{\&RightArrow;}{g}}}_{r,m}\right|}^2}{N_r{N}_t{\mathrm{\&sigma;}}^2},$ σ ²the power of noise, the complex vector of to be length be Nr;

S5, the OMP of N number of task is used to combine measuring-signal described in all S1 jointly recover all specific as follows:

S51、 ${\stackrel{\&RightArrow;}{y}}_{m,t}={\mathrm{\&Phi;}}_t{\stackrel{\&RightArrow;}{g}}_{m,t}={\mathrm{\&Phi;}}_{t}D{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{m,t}+{\stackrel{\&RightArrow;}{n}}_{m,t}^{\&prime;}=V_t{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{m,t}+{\stackrel{\&RightArrow;}{n}}_{m,t}^{\&prime;},$ Described can at V _tlower expansion, be at V _tunder expansion coefficient, wherein, V _t=Φ _td;

S52, from V described in S51 _tin find row make maximum, structural matrix calculate all at V _tunder expansion coefficient $[{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{1,r,c},{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{2,r,c},...,{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{N,r,c}]={\left({V_c}^H{V}_c\right)}^{-1}{V_c}^H[{\stackrel{\&RightArrow;}{y}}_{1,r},{\stackrel{\&RightArrow;}{y}}_{2,r},...,{\stackrel{\&RightArrow;}{y}}_{N,r}],$ Represent the surplus of current recovery extent $e_r=[{\stackrel{\&RightArrow;}{y}}_{1,r},{\stackrel{\&RightArrow;}{y}}_{2,r},...,{\stackrel{\&RightArrow;}{y}}_{N,r}]-V_c[{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{1,r,c},{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{2,r.c}...,{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{N,r,c}],$ Wherein, () ^-1the inversion operation of matrix, () ^hthe conjugate transpose operation of matrix, || represent the amplitude of getting plural number, || || ₂represent two norm computings of vector;

S53, from V _rin find make maximum, wherein matrix e _rin m row, will be added to V described in S32 _cin obtain upgrade after calculate v in the updated _cunder expansion coefficient;

S54, circulation S51 to S53, until e _rf norm be less than the a times of F norm time stop, the coefficient linear combination in conjunction with the column vector in dictionary matrix D described in S2 and correspondence position recovers be designated as an optimal beam vector is found from code book make spectrum efficiency maximum, wherein σ ²the power of noise, wherein, a is threshold value to the complex vector of Nr that to be length be, 0<a<1, and a is real number, σ ²the power of noise, the complex vector of to be length be Nr;

The receiving terminal of S6, equipment 2 uses P _tindividual reception vector carrys out Received signal strength, the same with the step of S1.3, and the signal in conjunction with all frequencies uses multitask orthogonal matching pursuit algorithm, and measuring-signal is ${\stackrel{\&RightArrow;}{y}}_{m,t}={\mathrm{\&Phi;}}_t{\stackrel{\&RightArrow;}{g}}_{m,t}={\mathrm{\&Phi;}}_{t}D{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{m,t}+{\stackrel{\&RightArrow;}{n}}_{m,t}^{\&prime;}=V_t{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{m,t}+{\stackrel{\&RightArrow;}{n}}_{m,t}^{\&prime;},$ Wherein V _t=Φ _td, ${\mathrm{\&Phi;}}_t={\left[\begin{array}{ccc}{\stackrel{\&RightArrow;}{\mathrm{\&phi;}}}_{t,1}& ...& {\stackrel{\&RightArrow;}{\mathrm{\&phi;}}}_{t,P_t}\end{array}\right]}^T$ Be calculation matrix, every a line receives the corresponding one-shot measurement of vector, and any one receives vector (d=1,2 ..., P _r) vector of to be all length be Nt, value Stochastic choice from set [1, i ,-1 ,-i] of each element, be at V _tunder expansion coefficient, duplicate measurements P _tsecondary, according to recover be designated as when all frequencies (m=1,2 ..., N) all recover after, find from code book one optimal make spectrum efficiency maximum, wherein p _tfor being greater than the integer of 1; the complex vector of to be length be Nt;

S7, equipment 1 with launch to equipment 2 as launching beam vector, equipment 2 finds optimum reception vector by the step repeating S1-S6

S8, through iterating, for equipment 1 and equipment 2, when the beam vector to find for adjacent twice, namely with iteration ends time identical, and will finally find with as the beam vector of equipment 1 and equipment 2 optimum.

Further, the correspondence in dictionary matrix D described in arbitrarily angled q, S2 is classified as $\left[\begin{array}{c}1\\ e^{\mathrm{i\&pi;}\sin \left(\mathrm{\&theta;}\right)}\\ e^{i2\mathrm{\&pi;}\sin \left(\mathrm{\&theta;}\right)}\\ .\\ .\\ .\\ e^{i(\mathrm{Nt}-1)\mathrm{\&pi;}\sin \left(\mathrm{\&theta;}\right)}\end{array}\right].$

Further, a=0.05 described in S54.

The invention has the beneficial effects as follows: beam search desired times is relevant with total number of path, search complexity can not increase along with number of antennas.The scope of application of the present invention is extremely wide, can be used for all slow fading sighting distances or non line of sight channel.

Accompanying drawing explanation

Fig. 1 is the structure chart that the present invention utilizes the both-end time-domain wave beam searching algorithm of compressed sensing.

Fig. 2 be the present invention for 802.11.ad channel rays search for probability of success performance chart.

Embodiment

Below in conjunction with embodiment and accompanying drawing, describe technical scheme of the present invention in detail.

As shown in Figure 1, the present invention utilizes the structure chart of the both-end time-domain wave beam searching algorithm of compressed sensing.Whole process completes at frequency domain, and equipment 1 is launched to equipment 2 with omnidirectional antenna, and equipment 2 repeats to receive P _rsecondary, at every turn with different reception vectors, equipment 2 is according to P _rindividual measured value uses compressed sensing to reduce to original received signal, and the processing procedure about signal is all complete at frequency domain, finds an optimum reception vector to make spectrum efficiency maximum according to the primary signal restored from code book.Due to symmetry, optimum reception vector is exactly optimum transmitting vector, and subsequently, equipment 2 receives vector using the optimum found and launches to equipment 1 as transmitting vector, and same, equipment 1 repeats to receive P _tsecondary, the reception vector that each use is different, equipment 1 uses compressed sensing to signals revivification, finds optimum reception vector to make spectrum efficiency maximum according to the signal restored from code book.Such process is carried out repeatedly, and the optimum that each equipment goes out with last computation after calculating this suboptimal reception vector receives vector and contrasts, and the optimum reception vector phase Simultaneous Iteration gone out when twice adjacent calculation just can stop.

S1, make the dual-mode antenna number of equipment 1 be Nt, the wave beam number in the code book of described equipment 1 is Ct, and described equipment 1 uses omnidirectional antenna to launch to equipment 2, and launching beam vector is ${\stackrel{\&RightArrow;}{w}}_1=\left[\begin{array}{c}1\\ 0\\ .\\ .\\ .\\ 0\end{array}\right],$ Described launching beam length is Nt, described equipment 1 use OFDM technology frequency domain transmission sequence for [1,1 ..., 1], the sequence length of described transmitting is N

Make the dual-mode antenna number of equipment 2 be Nr, the wave beam number in the code book of described equipment 2 is Cr, and m the sub-carrier signal vector that the antenna of described equipment 2 receives is the receiving terminal of described equipment 2 uses P _rindividual reception vector carrys out Received signal strength, and any one receives vector the vector of to be all length be Nr, value Stochastic choice from set [1, i ,-1 ,-i] of each element in described reception vector, forms a calculation matrix described calculation matrix Φ _reach provisional capital correspondence once receives, and sub-carrier signal is measuring-signal vector is wherein, the noise vector of to be length be Nr, H _mbe the exponent number of m frequency be Nr × Nt channel matrix, the element representation that in described channel matrix, xth row y arranges from transmitting terminal y root antenna to receiving terminal xth root antenna frequency domain channel impulse response, wherein, m=1,2 ..., N, y=1,2, ..., Nt, x=1,2 ..., Nr, d=1,2 ... P _r, i is imaginary unit, noise vector, in the corresponding measured value of each element, () ^ttranspose of a matrix computing, P _rfor being greater than the integer of 1, N, Nt, Nr, Ct and Cr are the integer being greater than 1;

S2, according to S1, build dictionary matrix be an angle in each row corresponding [-90 °, 90 °] of D, D, signal described in S1 can launch under D, and be sparse, expansion coefficient is plural number, be expansion coefficient under D;

S3, use the orthogonal matching pursuit algorithm of N number of task (Orthogonal Matching Pursuit, OMP) to combine sub-carrier signal described in all S1 jointly recover all be specially:

S31、 ${\stackrel{\&RightArrow;}{y}}_{m,r}={\mathrm{\&Phi;}}_r{\stackrel{\&RightArrow;}{g}}_{m,r}={\mathrm{\&Phi;}}_{r}D{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{m,r}+{\stackrel{\&RightArrow;}{n}}_{m,r}^{\&prime;}=V_r{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{m,r}+{\stackrel{\&RightArrow;}{n}}_{m,r}^{\&prime;},$ V _r=Φ _rd, described in can at V _rlower expansion, be at V _runder expansion coefficient;

S32, from V described in S31 _rin find row make maximum, structural matrix calculate all at V _cunder expansion coefficient $[{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{1,r,c},{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{2,r,c},...,{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{N,r,c}]={\left({V_c}^H{V}_c\right)}^{-1}{V_c}^H[{\stackrel{\&RightArrow;}{y}}_{1,r},{\stackrel{\&RightArrow;}{y}}_{2,r},...,{\stackrel{\&RightArrow;}{y}}_{N,r}],$ Represent the surplus of current recovery extent $e_r=[{\stackrel{\&RightArrow;}{y}}_{1,r},{\stackrel{\&RightArrow;}{y}}_{2,r},...,{\stackrel{\&RightArrow;}{y}}_{N,r}]-V_c[{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{1,r,c},{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{2,r.c}...,{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{N,r,c}],$ Wherein, () ^-1the inversion operation of matrix, () ^hthe conjugate transpose operation of matrix, || represent the amplitude of getting plural number, || || ₂represent two norm computings of vector;

S33, from V _rin find make maximum, wherein matrix e _rin m row, will be added to V described in S32 _cin obtain upgrade after calculate expansion coefficient under Vc in the updated;

S34, circulation S34 to S33, until e _rf norm be less than the a times of F norm time stop, the coefficient linear combination in conjunction with the column vector in dictionary matrix D described in S2 and correspondence position recovers wherein, a is threshold value, 0<a<1, and a is real number;

S4, recover all frequencies be denoted as the beam vector that one applicable is found from code book make spectrum efficiency maximum, ${\stackrel{\&RightArrow;}{c}}_1=\underset{\stackrel{\&RightArrow;}{c}}{\arg }\max \left(\frac{1}{N}\underset{m=1}{\overset{N}{\mathrm{\&Sigma;}}}{\log }_2(1+{\mathrm{\&gamma;}}_{r,m})\right),$ Wherein ${\mathrm{\&gamma;}}_{r,m}=\frac{{\left|{\stackrel{\&RightArrow;}{c}}^T{\widehat{\stackrel{\&RightArrow;}{g}}}_{r,m}\right|}^2}{N_r{N}_t{\mathrm{\&sigma;}}^2},$ σ ²the power of noise, the complex vector of to be length be Nr;

S5, the OMP of N number of task is used to combine measuring-signal described in all S1 jointly recover all specific as follows:

S51、 ${\stackrel{\&RightArrow;}{y}}_{m,t}={\mathrm{\&Phi;}}_t{\stackrel{\&RightArrow;}{g}}_{m,t}={\mathrm{\&Phi;}}_{t}D{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{m,t}+{\stackrel{\&RightArrow;}{n}}_{m,t}^{\&prime;}=V_t{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{m,t}+{\stackrel{\&RightArrow;}{n}}_{m,t}^{\&prime;},$ Described can at V _tlower expansion, be at V _tunder expansion coefficient, wherein, V _t=Φ _td;

S52, from V described in S51 _tin find row make maximum, structural matrix calculate all at V _tunder expansion coefficient $[{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{1,r,c},{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{2,r,c},...,{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{N,r,c}]={\left({V_c}^H{V}_c\right)}^{-1}{V_c}^H[{\stackrel{\&RightArrow;}{y}}_{1,r},{\stackrel{\&RightArrow;}{y}}_{2,r},...,{\stackrel{\&RightArrow;}{y}}_{N,r}],$ Represent the surplus of current recovery extent $e_r=[{\stackrel{\&RightArrow;}{y}}_{1,r},{\stackrel{\&RightArrow;}{y}}_{2,r},...,{\stackrel{\&RightArrow;}{y}}_{N,r}]-V_c[{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{1,r,c},{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{2,r.c}...,{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{N,r,c}],$ Wherein, () ^-1the inversion operation of matrix, () ^hthe conjugate transpose operation of matrix, || represent the amplitude of getting plural number, || || ₂represent two norm computings of vector;

S53, from V _rin find make maximum, wherein matrix e _rin m row, will be added to V described in S32 _cin obtain upgrade after calculate v in the updated _cunder expansion coefficient;

S54, circulation S51 to S53, until e _rf norm be less than the a times of F norm time stop, the coefficient linear combination in conjunction with the column vector in dictionary matrix D described in S2 and correspondence position recovers be designated as an optimal beam vector is found from code book make spectrum efficiency maximum, wherein σ ²the power of noise, wherein, a is threshold value to the complex vector of Nr that to be length be, 0<a<1, and a is real number, σ ²the power of noise, the complex vector of to be length be Nr;

The receiving terminal of S6, equipment 2 uses P _tindividual reception vector carrys out Received signal strength, the same with the step of S1.3, and the signal in conjunction with all frequencies uses multitask orthogonal matching pursuit algorithm, and measuring-signal is ${\stackrel{\&RightArrow;}{y}}_{m,t}={\mathrm{\&Phi;}}_t{\stackrel{\&RightArrow;}{g}}_{m,t}={\mathrm{\&Phi;}}_{t}D{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{m,t}+{\stackrel{\&RightArrow;}{n}}_{m,t}^{\&prime;}=V_t{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{m,t}+{\stackrel{\&RightArrow;}{n}}_{m,t}^{\&prime;},$ Wherein V _t=Φ _td, ${\mathrm{\&Phi;}}_t={\left[\begin{array}{ccc}{\stackrel{\&RightArrow;}{\mathrm{\&phi;}}}_{t,1}& ...& {\stackrel{\&RightArrow;}{\mathrm{\&phi;}}}_{t,P_t}\end{array}\right]}^T$ Be calculation matrix, every a line receives the corresponding one-shot measurement of vector, and any one receives vector (d=1,2 ..., P _r) vector of to be all length be Nt, value Stochastic choice from set [1, i ,-1 ,-i] of each element, be at V _tunder expansion coefficient, duplicate measurements P _tsecondary, according to recover be designated as when all frequencies (m=1,2 ..., N) all recover after, find from code book one optimal make spectrum efficiency maximum, wherein p _tfor being greater than the integer of 1; the complex vector of to be length be Nt;

S7, equipment 1 with launch to equipment 2 as launching beam vector, equipment 2 finds optimum reception vector by the step repeating S1-S6

S8, through iterating, for equipment 1 and equipment 2, when the beam vector to find for adjacent twice, namely with iteration ends time identical, and will finally find with as the beam vector of equipment 1 and equipment 2 optimum.

Embodiment 1,

Total number of sub-carriers is 512, and sample frequency is 1GHz, and equipment 1 and equipment 2 have 20 antennas, and the wave beam number in code book is 40, is an interval with 5 degree during structure dictionary, cM4 is non-line-of-sight channel, has many multipaths.

As shown in Figure 2, the search of 802.11.ad channel rays probability of success performance chart, in Fig. 2, abscissa abscissa is the duplicate measurements number of times of each equipment when at every turn receiving, and be under the condition of 0dB in signal to noise ratio, each point emulates 1000 times.

Can find out that the probability of success increases along with the increase of pendulous frequency according to Fig. 2.

Claims (3)

1. utilize the both-end time-domain wave beam searching method of compressed sensing, it is characterized in that, comprise the steps:

S1, make the dual-mode antenna number of equipment 1 be Nt, the wave beam number in the code book of described equipment 1 is Ct, and described equipment 1 uses omnidirectional antenna to launch to equipment 2, and launching beam vector is ${\stackrel{\&RightArrow;}{w}}_1=\left[\begin{array}{c}1\\ 0\\ .\\ .\\ .\\ 0\end{array}\right],$ Described launching beam length is Nt, described equipment 1 use OFDM technology frequency domain transmission sequence for [1,1 ..., 1], the sequence length of described transmitting is N

Make the dual-mode antenna number of equipment 2 be Nr, the wave beam number in the code book of described equipment 2 is Cr, and m the sub-carrier signal vector that the antenna of described equipment 2 receives is the receiving terminal of described equipment 2 uses P _rindividual reception vector carrys out Received signal strength, and any one receives vector the vector of to be all length be Nr, value Stochastic choice from set [1, i ,-1 ,-i] of each element in described reception vector, forms a calculation matrix described calculation matrix Φ _reach provisional capital correspondence once receives, and sub-carrier signal is measuring-signal vector is wherein, the noise vector of to be length be Nr, H _mbe the exponent number of m frequency be Nr × Nt channel matrix, the element representation that in described channel matrix, xth row y arranges from transmitting terminal y root antenna to receiving terminal xth root antenna frequency domain channel impulse response, wherein, m=1,2 ..., N, y=1,2, ..., Nt, x=1,2 ..., Nr, d=1,2 ... P _r, i is imaginary unit, noise vector, in the corresponding measured value of each element, () ^ttranspose of a matrix computing, P _rfor being greater than the integer of 1, N, Nt, Nr, Ct and Cr are the integer being greater than 1;

S2, according to S1, build dictionary matrix be an angle in each row corresponding [-90 °, 90 °] of D, D, signal described in S1 can launch under D, and be sparse, expansion coefficient is plural number, be expansion coefficient under D;

S3, use the orthogonal matching pursuit algorithm of N number of task (Orthogonal Matching Pursuit, OMP) to combine sub-carrier signal described in all S1 jointly recover all be specially:

S31、 ${\stackrel{\&RightArrow;}{y}}_{m,r}={\mathrm{\&Phi;}}_r{\stackrel{\&RightArrow;}{g}}_{m,r}={\mathrm{\&Phi;}}_{r}D{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{m,r}+{\stackrel{\&RightArrow;}{n}}_{m,r}^{\&prime;}=V_r{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{m,r}+{\stackrel{\&RightArrow;}{n}}_{m,r}^{\&prime;},$ V _r=Φ _rd, described in can at V _rlower expansion, be at V _runder expansion coefficient;

S32, from V described in S31 _rin find row make maximum, structural matrix calculate all at V _cunder expansion coefficient $[{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{1,r,c},{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{2,r,c},...,{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{N,r,c}]={\left({V_c}^H{V}_c\right)}^{-1}{V_c}^H[{\stackrel{\&RightArrow;}{y}}_{1,r},{\stackrel{\&RightArrow;}{y}}_{2,r},...,{\stackrel{\&RightArrow;}{y}}_{N,r}],$ Represent the surplus of current recovery extent $e_r=[{\stackrel{\&RightArrow;}{y}}_{1,r},{\stackrel{\&RightArrow;}{y}}_{2,r},...,{\stackrel{\&RightArrow;}{y}}_{N,r}]-V_c[{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{1,r,c},{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{2,r.c}...,{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{N,r,c}],$ Wherein, () ^-1the inversion operation of matrix, () ^hthe conjugate transpose operation of matrix, || represent the amplitude of getting plural number, || || ₂represent two norm computings of vector;

S33, from V _rin find make maximum, wherein matrix e _rin m row, will be added to V described in S32 _cin obtain upgrade after calculate v in the updated _cunder expansion coefficient;

S34, circulation S34 to S33, until e _rf norm be less than the a times of F norm time stop, the coefficient linear combination in conjunction with the column vector in dictionary matrix D described in S2 and correspondence position recovers wherein, a is threshold value, 0<a<1, and a is real number;

S4, recover all frequencies be denoted as the beam vector that one applicable is found from code book make spectrum efficiency maximum, ${\stackrel{\&RightArrow;}{c}}_1=\underset{\stackrel{\&RightArrow;}{c}}{\arg }\max \left(\frac{1}{N}\underset{m=1}{\overset{N}{\mathrm{\&Sigma;}}}{\log }_2(1+{\mathrm{\&gamma;}}_{r,m})\right),$ Wherein ${\mathrm{\&gamma;}}_{r,m}=\frac{{\left|{\stackrel{\&RightArrow;}{c}}^T{\widehat{\stackrel{\&RightArrow;}{g}}}_{r,m}\right|}^2}{N_r{N}_t{\mathrm{\&sigma;}}^2},$ σ ²the power of noise, the complex vector of to be length be Nr;

S5, the OMP of N number of task is used to combine measuring-signal described in all S1 jointly recover all specific as follows:

S51、 ${\stackrel{\&RightArrow;}{y}}_{m,t}={\mathrm{\&Phi;}}_t{\stackrel{\&RightArrow;}{g}}_{m,t}={\mathrm{\&Phi;}}_{t}D{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{m,t}+{\stackrel{\&RightArrow;}{n}}_{m,t}^{\&prime;}=V_t{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{m,t}+{\stackrel{\&RightArrow;}{n}}_{m,t}^{\&prime;},$ Described can at V _tlower expansion, be at V _tunder expansion coefficient, wherein, V _t=Φ _td;

S52, from V described in S51 _tin find row make maximum, structural matrix calculate all at V _tunder expansion coefficient $[{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{1,r,c},{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{2,r,c},...,{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{N,r,c}]={\left({V_c}^H{V}_c\right)}^{-1}{V_c}^H[{\stackrel{\&RightArrow;}{y}}_{1,r},{\stackrel{\&RightArrow;}{y}}_{2,r},...,{\stackrel{\&RightArrow;}{y}}_{N,r}],$ Represent the surplus of current recovery extent $e_r=[{\stackrel{\&RightArrow;}{y}}_{1,r},{\stackrel{\&RightArrow;}{y}}_{2,r},...,{\stackrel{\&RightArrow;}{y}}_{N,r}]-V_c[{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{1,r,c},{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{2,r.c}...,{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{N,r,c}],$ Wherein, () ^-1the inversion operation of matrix, () ^hthe conjugate transpose operation of matrix, || represent the amplitude of getting plural number, || || ₂represent two norm computings of vector;

S53, from V _rin find make maximum, wherein matrix e _rin m row, will be added to V described in S32 _cin obtain upgrade after calculate v in the updated _cunder expansion coefficient;

S54, circulation S51 to S53, until e _rf norm be less than the a times of F norm time stop, the coefficient linear combination in conjunction with the column vector in dictionary matrix D described in S2 and correspondence position recovers be designated as an optimal beam vector is found from code book make spectrum efficiency maximum, wherein σ ²the power of noise, wherein, a is threshold value to the complex vector of Nr that to be length be, 0<a<1, and a is real number, σ ²the power of noise, the complex vector of to be length be Nr;

The receiving terminal of S6, equipment 2 uses P _tindividual reception vector carrys out Received signal strength, the same with the step of S1.3, and the signal in conjunction with all frequencies uses multitask orthogonal matching pursuit algorithm, and measuring-signal is ${\stackrel{\&RightArrow;}{y}}_{m,t}={\mathrm{\&Phi;}}_t{\stackrel{\&RightArrow;}{g}}_{m,t}={\mathrm{\&Phi;}}_{t}D{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{m,t}+{\stackrel{\&RightArrow;}{n}}_{m,t}^{\&prime;}=V_t{\stackrel{\&RightArrow;}{\mathrm{\&omega;}}}_{m,t}+{\stackrel{\&RightArrow;}{n}}_{m,t}^{\&prime;},$ Wherein V _t=Φ _td, ${\mathrm{\&Phi;}}_t={\left[\begin{array}{ccc}{\stackrel{\&RightArrow;}{\mathrm{\&phi;}}}_{t,1}& ...& {\stackrel{\&RightArrow;}{\mathrm{\&phi;}}}_{t,P_t}\end{array}\right]}^T$ Be calculation matrix, every a line receives the corresponding one-shot measurement of vector, and any one receives vector (d=1,2 ..., P _r) vector of to be all length be Nt, value Stochastic choice from set [1, i ,-1 ,-i] of each element, be at V _tunder expansion coefficient, duplicate measurements P _tsecondary, according to recover be designated as when all frequencies (m=1,2 ..., N) all recover after, find from code book one optimal make spectrum efficiency maximum, wherein p _tfor being greater than the integer of 1; the complex vector of to be length be Nt;

S7, equipment 1 with launch to equipment 2 as launching beam vector, equipment 2 finds optimum reception vector by the step repeating S1-S6

S8, through iterating, for equipment 1 and equipment 2, when the beam vector to find for adjacent twice, namely with iteration ends time identical, and will finally find with as the beam vector of equipment 1 and equipment 2 optimum.

2. utilize the both-end time-domain wave beam searching method of compressed sensing according to claim 1, it is characterized in that: the correspondence in dictionary matrix D described in arbitrarily angled q, S2 is classified as $\left[\begin{array}{c}1\\ e^{\mathrm{i\&pi;}\sin \left(\mathrm{\&theta;}\right)}\\ e^{i2\mathrm{\&pi;}\sin \left(\mathrm{\&theta;}\right)}\\ .\\ .\\ .\\ e^{i(\mathrm{Nt}-1)\mathrm{\&pi;}\sin \left(\mathrm{\&theta;}\right)}\end{array}\right].$

3. utilize the both-end time-domain wave beam searching method of compressed sensing according to claim 1, it is characterized in that: a=0.05 described in S54.