CN103259586B - A kind of multi-hop cooperating relay beam-forming method based on genetic algorithm - Google Patents

Abstract

Jump the problem of the poor effect that the optimization of many junction waves beam shaping faces for existing three, the present invention proposes a kind of method realizing two groups of trunk group beamforming weight vectors combined optimizations based on genetic algorithm.The method makes every effort under via node gross power meets the prerequisite of certain constraint, the received signal to noise ratio of destination node be maximized.This optimization problem is multidirectional amount, be difficult to be realized by common method, present invention finds the inner link between two vectors, original problem is changed into the problem only comprising first group of trunk group beamforming weight vectors, and then use genetic algorithm to try to achieve the two groups of beamforming weight vectors that can be considered global optimum.

Description

A kind of multi-hop cooperating relay beam-forming method based on genetic algorithm

Technical field

The invention discloses a kind of beam forming technique of jumping many junction networks based on three of genetic algorithm, belong to signal transacting, wireless communication technology field.

Background technology

The application of wireless communication technology presents the situation of explosive growth in recent years, people to the communication quality of radio communication and the requirement of Consumer's Experience also more and more higher.Under the background that frequency spectrum resource is limited, multiple-input and multiple-output (MIMO) technology is arisen at the historic moment, and it can improve power system capacity and the availability of frequency spectrum effectively.In recent years, comprise the diversity techniques such as space, time, frequency, coding to be widely studied.But MIMO technology needs to settle multiple antennas, the constraint of acceptor sum power in terminal, uses MIMO technology very difficult at portable equipment.

For effectively overcoming the above problems, a kind of technology being called collaboration communication is more and more subject to people's attention.In fact, because wireless channel has the characteristic of broadcast, we can be considered as the node that geographical position is contiguous the set of spaced antenna, and therefore, these nodes closed in radio communication can by working in coordination signal transmission.For signal sending node, a cooperative node can regard a via node as.Cooperative communication technology can form virtual MIMO link between communication node.The communication technology based on user collaboration is exactly " collaboration communication ".In collaboration communication, user as via node, can be formed the multiple transmission link from signal sending node to destination node, can realize the diversity gain identical with mimo system.Multiple cooperative communication strategy is studied widely, and such as amplification forwarding, decoding forwarding, coding forward, select relaying etc.In above cooperative communication strategy, due to amplification forwarding have realize simple, the advantage that time delay is less and being valued by the people.The collaboration communication model of the present invention's research also will adopt the mode of amplification forwarding.

The junction waves beam shaping model of double bounce is mainly tended in previous research, and in the middle of the radio communication of reality, three relay-models of jumping also are more common, and relative to the junction waves beam shaping model of double bounce, the three network model coverages of jumping are wider.Jump in junction network three, comprise Liang Ge via node group, the corresponding relaying beamforming weight vectors of each group.In order to improve communication quality, meet the condition of certain power constraint at via node under, the received signal to noise ratio maximizing destination node is one and significantly works.The problems referred to above were approximately Positive semidefinite problem by Zeng You researcher in the past, and then try to achieve closely excellent beamforming weight vectors with convex optimization tool, but this method is easily absorbed in local optimum, effect is not ideal enough.

Summary of the invention

The object of the present invention is to provide a kind of optimized beam-forming method of jumping many junction networks for three based on genetic algorithm, meet the prerequisite of certain constraint at via node power under, combined optimization is carried out to two groups of via node beamforming weight vectors, the received signal to noise ratio of destination node is maximized.In this method, based on to the research of first group of via node by relation between beamforming weight vectors w and second group of via node beamforming weight vectors v, by one of them vector with another vector representation out, then completely use genetic algorithm realize, can globally optimal solution be realized.

Technical solution of the present invention is as follows:

A kind of relay cooperative beam-forming method based on genetic algorithm, the method jumps multi-relay cooperation communication system based on three, this system is made up of a source node, two groups of via node groups and a destination node, and all nodes in network are all only equipped with single antenna, due to the impact of channel fading, all directly cannot set up communication link between source node and destination node, between source node and second group of via node group, between first group of via node group and destination node, sending node needs set up the communication between destination node by two via node groups, the number participating in the via node of collaboration communication is known, and the channel parameter of whole network model can be obtained by channel estimating, in communication process, source node is to first group of relay node broadcasts signal, noise on the first channel aliasing of jumping, first group of via node beamforming weight vectors w is weighted to the received signal, then the mode of amplification forwarding is used through the second hop channel to second group of relay node broadcasts, after second group of via node receives signal, be weighted with another beamforming weight vectors v to received signal, then signal is sent in the mode of amplification forwarding to destination node, this signal is by after noise on the 3rd channel aliasing of jumping, receive at destination node place,

Under the node that continues in each group all meets the prerequisite of certain power constraint, combined optimization is carried out to two groups of via node beamforming weight vectors, thus improve the received signal to noise ratio of destination node, in concrete enforcement, based on the relation of first group of via node between beamforming weight vectors w and second group of via node beamforming weight vectors v, second group of via node beamforming weight vectors v, first group of via node beamforming weight vectors w is showed, multivariable optimization problem is converted into the unitary variant optimization problem only comprising a beamforming weight vectors, then genetic algorithm is used to realize completely, concrete steps are as follows:

Step one, set first group and second group of via node power constraint separately, by channel estimating, the parameter obtaining first, second and third hop channel is respectively f, H, g;

The population of the beamforming weight vectors w of step 2, initialization first group of via node, in the feasible zone that this population must be determined at the power constraint by first group of via node, for the w meeting arbitrarily first group of via node power constraint, optimized second group of via node beamforming weight vectors v w shows, calculate the target function value corresponding to each w in current population, i.e. the received signal to noise ratio of destination node;

Step 3, according to above-mentioned a series of target function value, use genetic algorithm, generate the new population of w, calculate the target function value corresponding to each w in this population, and select wherein maximum target function value;

Step 4, judge whether current maximum target function value meets iterated conditional, if meet, repeat step 3, if do not meet, carry out step 5;

The w making target function maximum in step 5, output population, and calculate corresponding optimized v, be two required beamforming weight vectors.

Preferably, the pass between first group of via node beamforming weight vectors w and second group of via node beamforming weight vectors v is wherein P _rrepresent the power constraint of second group of via node, be respectively the noise power that via node place is jumped in the first jumping and second, for the purpose of the noise power D of Nodes _rfor diagonal matrix, ${\left[D_R\right]}_{k,k}=\underset{m=1}{\overset{M}{\mathrm{\Σ}}}{\left|h_{k,m}\right|}^2({\left|f_m\right|}^2{P}_s+{\mathrm{\σ}}_T^2){\left|w_m\right|}^2+{\mathrm{\σ}}_R^2,k=\mathrm{1,2},...,K,$ K represents first group of via node number, and M represents second group of via node number.

Preferably, former optimization problem can be expressed as the unitary variant optimization problem only comprising a beamforming weight vectors w, for: $\underset{w}{\max }\frac{w^H{R}_{w}w}{w^H{Q}_{w}w+{\mathrm{\σ}}_R^2{\left|\right|v^{H}g\left|\right|}^2+{\mathrm{\σ}}_T^2}$

Make

Wherein for diagonal matrix, , R _w=P _sf ^hh ^hg ^hvv ^hgHF, F=diag (f), G=diag (g), Q _w=H ^hg ^hvv ^hgH, $v=\sqrt{P_R}{D}_R^{-1}{(D_R^{-1/2}{Q}_v{D}_R^{-1/2}+{\mathrm{\σ}}_D^{2}I)}^{-1}{D}_R^{-1/2}\mathrm{GHw},$ I representation unit matrix, p _srepresent source node transmitted power.

Accompanying drawing explanation

Fig. 1: system model figure of the present invention;

Fig. 2: the workflow diagram of this method;

Fig. 3: simulation result figure.

Embodiment

Jump the problem of the poor effect that the optimization of many junction waves beam shaping faces for existing three, the present invention proposes a kind of combined optimization method realizing two groups of trunk group beamforming weight vectors based on genetic algorithm.This combined optimization problem is multidirectional amount, be difficult to be realized by common method, present invention finds the inner link between two vectors, original problem is changed into the problem of unidirectional amount, and then use genetic algorithm to try to achieve the beamforming weight vectors that can be considered global optimum.

Below in conjunction with drawings and Examples, the present invention is further detailed, but is not limited thereto example.

Consider that one is jumped multi-relay cooperation communication system based on three of amplification forwarding mechanism, as shown in Figure 1, this system is by a source node (S), and two groups of via node groups and a destination node (D) form, and all nodes in network all assemble single antenna.Assuming that between source node and destination node, between source node and second group of via node group, all cannot set up direct communication link between trunk group and destination node for first group.Therefore, source node needs to set up the communication with destination node by Liang Ge via node group.Transmitted power is P _s, the number participating in the via node of collaboration communication is known, and wherein first group of via node group comprises M=4 or M=6 via node, and second group comprises K=4 or K=6 via node, remembers that the first hop channel parameter is f=[f ₁, f ₂..., f _m] ^t, the second hop channel parameter matrix is H, and the 3rd hop channel parameter is g=[g ₁, g ₂..., g _k] ^t; First jumping via node beamforming weight vectors is w=[w ₁, w ₂..., w _m] ^t, the second jumping via node beamforming weight vectors is v=[v ₁, v ₂..., v _k] ^t; All noises are stationary white Gaussian noise, and the first noise power of jumping via node place is second noise power of jumping via node place is the noise power at destination node place is we use the complex weighting vector of these two via node groups of genetic Algorithm Design, and under making via node power meet the prerequisite of certain constraint, destination node place received signal to noise ratio maximizes.

As Fig. 2, the method step is as follows:

Step one, setting first are jumped and the second power constraint of jumping is respectively P _tand P _r, utilize channel estimating, obtain f, H, g,

Step 2, initialization t=0, t represents iterations, and P (t) represents the population of t for beamforming weight vectors w, and the population of the beamforming weight vectors w of initialization first group of via node is P (0), s represents the feasible zone of w, this feasible zone is determined by the power constraint of first group of via node, for the w meeting arbitrarily first group of via node power constraint, optimized second group of via node beamforming weight vectors v can show with w, and then former optimization problem can be expressed as the problem only comprising a beamforming weight vectors w:

$\underset{w}{\max }\frac{w^H{R}_{w}w}{w^H{Q}_{w}w+{\mathrm{\σ}}_R^2{\left|\right|v^{H}g\left|\right|}^2+{\mathrm{\σ}}_D^2}$

Make

Wherein for diagonal matrix, , R _w=P _sf ^hh ^hg ^hvv ^hgHF, F=diag (f), G=diag (g), Q _w=H ^hg ^hvv ^hgH, $v=\sqrt{P_R}{D}_R^{-1}{(D_R^{-1/2}{Q}_v{D}_R^{-1/2}+{\mathrm{\σ}}_D^{2}I)}^{-1}{D}_R^{-1/2}\mathrm{GHw},$ I representation unit matrix, D _ralso be diagonal matrix, ${\left[D_R\right]}_{k,k}=\underset{m=1}{\overset{M}{\mathrm{\Σ}}}{\left|h_{k,m}\right|}^2({\left|f_m\right|}^2{P}_s+{\mathrm{\σ}}_T^2){\left|w_m\right|}^2+{\mathrm{\σ}}_R^2,k=\mathrm{1,2},...,K,$ calculate the target function value (received signal to noise ratio of destination node) in current population P (0) corresponding to each w, wherein maximum target function value is expressed as m (0);

Step 3, the value of t is updated to a series of target function values of t+1 corresponding to P (t-1), use genetic algorithm, generate population P (t) that w is new, calculate the target function value corresponding to each w in this population, and select wherein maximum target function value m (t).

Step 4, judge whether current maximum target function value meets iterated conditional make ε=0.0001, if meet, repeat step 3, if do not meet, carry out step 5;

The w making target function value maximum in step 5, output population, and export the optimized of correspondence $v=\sqrt{P_R}{D}_R^{-1}{(D_R^{-1/2}{Q}_v{D}_R^{-1/2}+{\mathrm{\σ}}_D^{2}I)}^{-1}{D}_R^{-1/2}\mathrm{GHw},$ Be two required beamforming weight vectors.Shown in accompanying drawing 3, the signal to noise ratio that accepts that method of the present invention realizes than additive method is significantly improved, and because genetic algorithm is not easy to be absorbed in local optimum, therefore the method can be considered global optimum.

As shown in Figure 3, compared with existing method, method of the present invention is under via node meets the prerequisite of certain power constraint, and attainable reception Signal to Interference plus Noise Ratio is larger, and communication quality is better.

Claims (2)

1. the relay cooperative beam-forming method based on genetic algorithm, the method jumps multi-relay cooperation communication system based on three, this system is by a source node, two groups of via node groups and a destination node composition, and all nodes in network are all only equipped with single antenna, due to the impact of channel fading, between source node and destination node, between source node and second group of via node group, all directly communication link cannot be set up between first group of via node group and destination node, sending node needs set up the communication between destination node by two via node groups, the number participating in the via node of collaboration communication is known, and the channel parameter of whole network model can be obtained by channel estimating, in communication process, source node is to first group of relay node broadcasts signal, noise on the first channel aliasing of jumping, first group of via node beamforming weight vectors w is weighted to the received signal, then the mode of amplification forwarding is used through the second hop channel to second group of relay node broadcasts, after second group of via node receives signal, be weighted with another beamforming weight vectors v to received signal, then signal is sent in the mode of amplification forwarding to destination node, this signal is by after noise on the 3rd channel aliasing of jumping, receive at destination node place,

Under the node that continues in each group all meets the prerequisite of certain power constraint, combined optimization is carried out to two groups of via node beamforming weight vectors, thus improve the received signal to noise ratio of destination node, in concrete enforcement, based on the relation of first group of via node between beamforming weight vectors w and second group of via node beamforming weight vectors v, second group of via node beamforming weight vectors v, first group of via node beamforming weight vectors w is showed, multivariable optimization problem is converted into the unitary variant optimization problem only comprising a beamforming weight vectors, then genetic algorithm is used to realize completely, concrete steps are as follows:

Step one, set first group and second group of via node power constraint separately, by channel estimating, the parameter obtaining first, second and third hop channel is respectively f, H, g;

The population of the beamforming weight vectors w of step 2, initialization first group of via node, in the feasible zone that this population must be determined at the power constraint by first group of via node, for the w meeting arbitrarily first group of via node power constraint, optimized second group of via node beamforming weight vectors v w shows, namely wherein P _rrepresent the power constraint of second group of via node, be respectively the noise power that via node place is jumped in the first jumping and second, for the purpose of the noise power D of Nodes _rfor diagonal matrix, ${\[D_R\]}_{k,k}=\underset{m=1}{\overset{M}{\Σ}}\left|h_{k,m}{|}^2\left(\right|f_m{|}^2{P}_s+{\mathrm{\σ}}_T^2)\right|w_m{|}^2+{\mathrm{\σ}}_R^2,k=1,2,...,K,$ M represents first group of via node number, and K represents second group of via node number, I representation unit matrix, G=diag (g), P _srepresent source node transmitted power, w _mfor m element in vectorial w, w=[w ₁, w ₂..., w _m] ^tbe the first jumping via node beamforming weight vectors, h _k,mthe element of the row k m row of the second hop channel parameter matrix H that expression is, calculates the target function value corresponding to each w in current population, i.e. the received signal to noise ratio of destination node;

Step 3, according to above-mentioned a series of target function value, use genetic algorithm, generate the new population of w, calculate the target function value corresponding to each w in this population, and select wherein maximum target function value;

Step 4, judge whether current maximum target function value meets iterated conditional, if meet, repeat step 3, if do not meet, carry out step 5;

The w making target function maximum in step 5, output population, and calculate corresponding optimized v, be two required beamforming weight vectors.

2., as claimed in claim 1 based on the relay cooperative beam-forming method of genetic algorithm, former optimization problem can be expressed as the unitary variant optimization problem only comprising a beamforming weight vectors w, for:

$\underset{w}{max}\frac{w^H{R}_{w}w}{w^H{Q}_{w}w+{\mathrm{\σ}}_R^2\left|\right|v^{H}g|{|}^2+{\mathrm{\σ}}_T^2}$

Make $w^H{D}_{T}w\≤P_T,w^H{D}_{R}w\≤P_R-{\mathrm{\σ}}_R^2{v}^{H}v$

Wherein w ^hd _tw represents the transmission gross power of first group of trunk group, D _rfor diagonal matrix, r _w=P _sf ^hh ^hg ^hvv ^hgHF, F=diag (f), G=diag (g), Q _w=H ^hg ^hvv ^hgH, i representation unit matrix, p _srepresent source node transmitted power, f _mbe m element in the first hop channel parameter f, v _kbe the kth element in the second jumping via node beamforming weight vectors v, P _tand P _rbe respectively the first jumping and the second power constraint of jumping.