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

Abstract

Aiming at the problem that the existing three-hop multi-relay beamforming optimization is not effective, the present invention proposes a method based on a genetic algorithm to realize the joint optimization of two groups of relay group beamforming weight vectors. This method strives to maximize the receiving signal-to-noise ratio of the destination node under the premise that the total power of the relay node satisfies certain constraints. This optimization problem is multi-vector, which is difficult to realize by common methods. The present invention discovers the intrinsic connection between two vectors, and transforms the original problem into a problem that only includes the first group of relay group beamforming weight vectors. Then the genetic algorithm is used to obtain two sets of beamforming weight vectors which can be regarded as the global optimum.

Description

A Multi-hop Cooperative Relay Beamforming Method Based on Genetic Algorithm

technical field

The invention discloses a beamforming technology of a three-hop multi-relay network based on a genetic algorithm, and belongs to the technical fields of signal processing and wireless communication.

Background technique

The application of wireless communication technology has shown an explosive growth trend in recent years, and people have higher and higher requirements on the communication quality and user experience of wireless communication. In the context of limited spectrum resources, multiple-input multiple-output (MIMO) technology emerges as the times require, which can effectively improve system capacity and spectrum utilization. In recent years, diversity techniques including space, time, frequency, coding, etc. have been extensively studied. However, MIMO technology needs to install multiple antennas in the terminal, and it is very difficult to use MIMO technology in portable devices due to the constraints of size and power.

In order to effectively solve the above problems, a technology called cooperative communication has been paid more and more attention by people. In fact, due to the broadcasting characteristics of wireless channels, we can regard geographically adjacent nodes as a collection of distributed antennas. Therefore, these adjacent nodes in wireless communication can transmit signals by cooperating with each other. For the signaling node, a coordinating node can be regarded as a relay node. Cooperative communication technology can form a virtual MIMO link between communication nodes. The communication technology based on user cooperation is "cooperative communication". In cooperative communication, the user can act as a relay node to form multiple transmission links from the signal sending node to the destination node, which can achieve the same diversity gain as the MIMO system. A variety of cooperative communication strategies have been extensively studied, such as amplification and forwarding, decoding and forwarding, encoding and forwarding, selective relay and so on. Among the above cooperative communication strategies, amplification and forwarding has been paid attention to because of its advantages of simple implementation and small delay. The cooperative communication model studied in the present invention will also adopt the way of amplification and forwarding.

Previous research mainly tends to the two-hop relay beamforming model, and in actual wireless communication, the three-hop relay model is also relatively common. Compared with the two-hop relay beamforming model, the three-hop network model Coverage is wider. In the three-hop relay network, there are two relay node groups, and each group corresponds to a relay beamforming weight vector. In order to improve the communication quality, it is very meaningful to maximize the receiving signal-to-noise ratio of the destination node under the condition that the relay node satisfies certain power constraints. In the past, researchers have approximated the above problem as a semi-positive definite problem, and then used convex optimization tools to obtain a nearly optimal beamforming weight vector. However, this method is prone to fall into local optimum, and the effect is not ideal.

Contents of the invention

The purpose of the present invention is to provide a genetic algorithm-based optimized beamforming method for a three-hop multi-relay network. Under the premise that the relay node power satisfies a certain constraint, two groups of relay node beamforming weight vectors are Joint optimization to maximize the receiving signal-to-noise ratio of the destination node. In this method, based on the research on the relationship between the beamforming weight vector w used by the first group of relay nodes and the beamforming weight vector v of the second group of relay nodes, one of the vectors is represented by another vector, and then fully used The implementation of genetic algorithm can realize the global optimal solution.

Technical solution of the present invention is as follows:

A relay cooperative beamforming method based on genetic algorithm, which is based on a three-hop multi-relay cooperative communication system, which consists of a source node, two groups of relay node groups and a target node, and all nodes in the network They are only equipped with a single antenna. Due to the influence of channel fading, it is impossible to directly establish a network between the source node and the target node, between the source node and the second group of relay node groups, and between the first group of relay node groups and the target node. In the communication link, the sending node needs to establish communication with the target node through two relay node groups; the number of relay nodes participating in cooperative communication is known, and the channel parameters of the entire network model can be estimated through channel estimation. Obtained; in the communication process, the source node broadcasts the signal to the first group of relay nodes, and the noise is aliased on the channel of the first hop, and the first group of relay nodes weights the received signal with the beamforming weight vector w, Then use the amplification and forwarding method to broadcast to the second group of relay nodes through the second hop channel. After receiving the signal, the second group of relay nodes weights the received signal with another beamforming weight vector v, and then uses the amplification and forwarding way to send a signal to the destination node, the signal is received at the destination node after the signal is aliased with noise on the channel of the third hop;

On the premise that each group of relay nodes satisfies a certain power constraint, the beamforming weight vectors of the two groups of relay nodes are jointly optimized to improve the receiving signal-to-noise ratio of the destination node. In the specific implementation, based on the first group of relay nodes Using the relationship between the beamforming weight vector w and the beamforming weight vector v of the second group of relay nodes, the beamforming weight vector v of the second group of relay nodes is expressed by the beamforming weight vector w of the first group of relay nodes , transforming the multivariate optimization problem into a single variable optimization problem containing only one beamforming weight vector, and then completely using the genetic algorithm to realize it. The specific steps are as follows:

Step 1. Set the respective power constraints of the first group and the second group of relay nodes, and obtain the parameters of the first, second and third hop channels as f, H and g respectively through channel estimation;

Step 2. Initialize the population of the beamforming weight vector w of the first group of relay nodes. The population must be within the feasible region determined by the power constraints of the first group of relay nodes. w, the optimized beamforming weight vector v of the second group of relay nodes is represented by w, and the objective function value corresponding to each w in the current population is calculated, that is, the receiving signal-to-noise ratio of the destination node;

Step 3. According to the above-mentioned series of objective function values, use the genetic algorithm to generate a new population of w, calculate the objective function value corresponding to each w in the population, and select the largest objective function value;

Step 4. Judging whether the current maximum objective function value satisfies the iteration condition, if so, repeat step 3, and if not, go to step 5;

Step 5: Output w that maximizes the objective function in the population, and calculate the corresponding optimized v, which is the two beamforming weight vectors sought.

Preferably, the relationship between the beamforming weight vector w used by the first group of relay nodes and the beamforming weight vector v of the second group of relay nodes is: where P _R represents the power constraints of the second set of relay nodes, are the noise power at the relay node of the first hop and the second hop, respectively, is the noise power _DR at the destination node is a diagonal matrix, ${\left[{\mathrm{D.}}_R\right]}_{k,k}=\underset{m=1}{\overset{m}{\mathrm{\Σ}}}{\left|h_{k,m}\right|}^2({\left|f_m\right|}^2{P}_{\mathrm{the\; s}}+{\mathrm{\σ}}_T^2){\left|w_m\right|}^2+{\mathrm{\σ}}_R^2,k=\mathrm{1,2},...,K,$ K represents the number of relay nodes in the first group, and M represents the number of relay nodes in the second group.

Preferably, the original optimization problem can be expressed as a single variable optimization problem containing only one beamforming weight vector w, as: $\underset{w}{\max }\frac{w^h{R}_{w}w}{w^h{Q}_{w}w+{\mathrm{\σ}}_R^2{\left|\right|v^{h}g\left|\right|}^2+{\mathrm{\σ}}_T^2}$

make

in is a diagonal matrix, , R _w =P _s F ^H H ^H G ^H vv ^H GHF, F = diag(f), G = diag(g), Q _w =H ^H G ^H vv ^H GH, $v=\sqrt{P_R}{\mathrm{D.}}_R^{-1}{({\mathrm{D.}}_R^{-1/2}{Q}_v{\mathrm{D.}}_R^{-1/2}+{\mathrm{\σ}}_{\mathrm{D.}}^{2}I)}^{-1}{\mathrm{D.}}_R^{-1/2}\mathrm{wxya},$ I represents the identity matrix, _PS represents the transmit power of the source node.

Description of drawings

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

Fig. 2: the work flowchart of this method;

Figure 3: Simulation result graph.

detailed description

Aiming at the problem of poor effect of existing three-hop multi-relay beamforming optimization, the present invention proposes a joint optimization method based on genetic algorithm to realize beamforming weight vectors of two groups of relays. The joint optimization problem is multi-vector, which is difficult to realize by common methods. The present invention discovers the intrinsic relationship between two vectors, converts the original problem into a single-vector problem, and then uses the genetic algorithm to obtain a global Optimal beamforming weight vector.

The present invention will be further described below in conjunction with the drawings and embodiments, but not limited to this example.

Consider a three-hop multi-relay cooperative communication system based on the amplification and forwarding mechanism, as shown in Figure 1, the system consists of a source node (S), two groups of relay node groups and a destination node (D), and All nodes in the network are equipped with a single antenna. It is assumed that no direct communication link can be established between the source node and the target node, between the source node and the second relay node group, and between the first relay group and the target node. Therefore, the source node needs to establish communication with the target node through two groups of relay nodes. The transmission power is P _S , and the number of relay nodes participating in cooperative communication is known, wherein the first group of relay node groups includes M=4 or M=6 relay nodes, and the second group includes K=4 or K=6 relay nodes, the channel parameter of the first hop is f=[f ₁ ,f ₂ ,...,f _M ] ^T , the channel parameter matrix of the second hop is H, and the channel parameter of the third hop is g= [g ₁ ,g ₂ ,...,g _K ] ^T ; the beamforming weight vector of the first hop relay node is w=[w ₁ ,w ₂ ,...,w _M ] ^T , the second hop relay node The node beamforming weight vector is v=[v ₁ ,v ₂ ,...,v _K ] ^T ; all noises are stationary Gaussian white noise, and the noise power at the first hop relay node is The noise power at the second hop relay node is The noise power at the destination node is We use the genetic algorithm to design the complex weight vectors of the two relay node groups, so that the receiving signal-to-noise ratio at the destination node can be maximized under the premise that the relay node power meets certain constraints.

As shown in Figure 2, the method steps are as follows:

Step 1. Set the power constraints of the first hop and the second hop as PT and _PR respectively, and use channel estimation to obtain f, _H , g,

Step 2, initialize t=0, t represents the number of iterations, P(t) represents the population of the t-th generation beamforming weight vector w, and initialize the population of the beamforming weight vector w of the first group of relay nodes as P(0), , S represents the feasible region of w, which is determined by the power constraints of the first group of relay nodes, for any w that satisfies the power constraints of the first group of relay nodes, the optimized beamforming weight vector of the second group of relay nodes v can be expressed by w, and then the original optimization problem can be expressed as a problem involving only one beamforming weight vector w:

$\underset{\mathrm{<span\; class="notranslate">\; w</span>}}{\mathrm{<span\; class="notranslate">\; max</span>}}\frac{{\mathrm{<span\; class="notranslate">\; w</span>}}^{\mathrm{<span\; class="notranslate">\; h</span>}}{\mathrm{<span\; class="notranslate">\; R</span>}}_{\mathrm{<span\; class="notranslate">\; w</span>}}\mathrm{<span\; class="notranslate">\; w</span>}}{{\mathrm{<span\; class="notranslate">\; w</span>}}^{\mathrm{<span\; class="notranslate">\; h</span>}}{\mathrm{<span\; class="notranslate">\; Q</span>}}_{\mathrm{<span\; class="notranslate">\; w</span>}}\mathrm{<span\; class="notranslate">\; w</span>}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; R</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}{<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; v</span>}}^{\mathrm{<span\; class="notranslate">\; h</span>}}\mathrm{<span\; class="notranslate">\; g</span>}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; D.</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}$

make

in is a diagonal matrix, , R _w =P _s F ^H H ^H G ^H vv ^H GHF, F = diag(f), G = diag(g), Q _w =H ^H G ^H vv ^H GH, $v=\sqrt{P_R}{\mathrm{D.}}_R^{-1}{({\mathrm{D.}}_R^{-1/2}{Q}_v{\mathrm{D.}}_R^{-1/2}+{\mathrm{\&sigma;}}_{\mathrm{D.}}^{2}I)}^{-1}{\mathrm{D.}}_R^{-1/2}\mathrm{wxya},$ I represents the identity matrix, D _R is also a diagonal matrix, ${\left[{\mathrm{D.}}_R\right]}_{k,k}=\underset{m=1}{\overset{m}{\mathrm{\&Sigma;}}}{\left|h_{k,m}\right|}^2({\left|f_m\right|}^2{P}_{\mathrm{the\; s}}+{\mathrm{\&sigma;}}_T^2){\left|w_m\right|}^2+{\mathrm{\&sigma;}}_R^2,k=\mathrm{1,2},...,K,$ Calculate the objective function value corresponding to each w in the current population P(0) (the receiving signal-to-noise ratio of the destination node), where the maximum objective function value is expressed as m(0);

Step 3. Update the value of t to t+1. According to a series of objective function values corresponding to P(t-1), use the genetic algorithm to generate a new population P(t) of w, and calculate the value of each w in the population. Corresponding objective function values, and select the largest objective function value m(t).

Step 4. Determine whether the current maximum objective function value satisfies the iteration condition Let ε=0.0001, if it is satisfied, repeat step three, if not, go to step five;

Step 5. Output the w that maximizes the objective function value in the population, and output the corresponding optimized $v=\sqrt{P_R}{\mathrm{D.}}_R^{-1}{({\mathrm{D.}}_R^{-1/2}{Q}_v{\mathrm{D.}}_R^{-1/2}+{\mathrm{\&sigma;}}_{\mathrm{D.}}^{2}I)}^{-1}{\mathrm{D.}}_R^{-1/2}\mathrm{wxya},$ That is, the two beamforming weight vectors sought. As shown in accompanying drawing 3, the method of the present invention has significantly improved acceptance signal-to-noise ratio than other methods, because the genetic algorithm is not easy to fall into the local optimum, so the method can be regarded as the global optimum.

As shown in FIG. 3 , compared with the existing methods, the method of the present invention can achieve a larger received signal-to-interference-noise ratio and better communication quality under the premise that the relay node satisfies a certain power constraint.

Claims (2)

1. A relay cooperative beamforming method based on genetic algorithm, the method is based on a three-hop multi-relay cooperative communication system, the system consists of a source node, two groups of relay node groups and a target node, and the network All nodes are only equipped with a single antenna. Due to the influence of channel fading, there is no connection between the source node and the target node, between the source node and the second group of relay nodes, and between the first group of relay node groups and the target node. To establish a communication link directly, the sending node needs to establish communication with the target node through two relay node groups; the number of relay nodes participating in cooperative communication is known, and the channel parameters of the entire network model can be obtained through the channel In the communication process, the source node broadcasts the signal to the first group of relay nodes, and the noise is aliased on the channel of the first hop, and the first group of relay nodes uses the beamforming weight vector w to process the received signal weighted, and then broadcast to the second group of relay nodes through the second hop channel in the way of amplification and forwarding. After receiving the signal, the second group of relay nodes weights the received signal with another beamforming weight vector v, and then amplifies The forwarding method sends a signal to the destination node, and the signal is received at the destination node after the noise is aliased on the channel of the third hop; On the premise that each group of relay nodes satisfies a certain power constraint, the beamforming weight vectors of the two groups of relay nodes are jointly optimized to improve the receiving signal-to-noise ratio of the destination node. In the specific implementation, based on the first group of relay nodes Using the relationship between the beamforming weight vector w and the beamforming weight vector v of the second group of relay nodes, the beamforming weight vector v of the second group of relay nodes is expressed by the beamforming weight vector w of the first group of relay nodes , transforming the multivariate optimization problem into a single variable optimization problem containing only one beamforming weight vector, and then completely using the genetic algorithm to realize it. The specific steps are as follows: Step 1. Set the respective power constraints of the first group and the second group of relay nodes, and obtain the parameters of the first, second and third hop channels as f, H and g respectively through channel estimation; Step 2. Initialize the population of the beamforming weight vector w of the first group of relay nodes. The population must be within the feasible region determined by the power constraints of the first group of relay nodes. w, the optimized beamforming weight vector v of the second group of relay nodes is represented by w, that is where P _R represents the power constraints of the second set of relay nodes, are the noise power at the relay node of the first hop and the second hop, respectively, is the noise power _DR at the destination node is a diagonal matrix, ${\&lsqb;{\mathrm{D.}}_R\&rsqb;}_{k,k}=\underset{m=1}{\overset{m}{\&Sigma;}}\left|h_{k,m}{|}^2\left(\right|f_m{|}^2{P}_{\mathrm{the\; s}}+{\mathrm{\&sigma;}}_T^2)\right|w_m{|}^2+{\mathrm{\&sigma;}}_R^2,k=1,2,...,K,$ M represents the number of relay nodes in the first group, K represents the number of relay nodes in the second group, I represents the unit matrix, G=diag(g), _PS represents the source node transmission power, w _m is the mth in the vector w element, w=[w ₁ ,w ₂ ,...,w _M ] ^T is the beamforming weight vector of the first-hop relay node, h _k,m represent the k-th row of the second-hop channel parameter matrix H The elements in the m column calculate the objective function value corresponding to each w in the current population, that is, the receiving signal-to-noise ratio of the destination node; Step 3. According to the above-mentioned series of objective function values, use the genetic algorithm to generate a new population of w, calculate the objective function value corresponding to each w in the population, and select the largest objective function value; Step 4. Judging whether the current maximum objective function value satisfies the iteration condition, if so, repeat step 3, and if not, go to step 5; Step 5: Output w that maximizes the objective function in the population, and calculate the corresponding optimized v, which is the two beamforming weight vectors sought. 2. The relay cooperative beamforming method based on genetic algorithm as claimed in claim 1, the original optimization problem can be expressed as a single variable optimization problem comprising only one beamforming weight vector w, which is: $\underset{\mathrm{<span\; class="notranslate">\; w</span>}}{\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; x</span>}}\frac{{\mathrm{<span\; class="notranslate">\; w</span>}}^{\mathrm{<span\; class="notranslate">\; h</span>}}{\mathrm{<span\; class="notranslate">\; R</span>}}_{\mathrm{<span\; class="notranslate">\; w</span>}}\mathrm{<span\; class="notranslate">\; w</span>}}{{\mathrm{<span\; class="notranslate">\; w</span>}}^{\mathrm{<span\; class="notranslate">\; h</span>}}{\mathrm{<span\; class="notranslate">\; Q</span>}}_{\mathrm{<span\; class="notranslate">\; w</span>}}\mathrm{<span\; class="notranslate">\; w</span>}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; R</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; v</span>}}^{\mathrm{<span\; class="notranslate">\; h</span>}}\mathrm{<span\; class="notranslate">\; g</span>}<span\; class="notranslate">\; |</span>{<span\; class="notranslate">\; |</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; T</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}$ make $w^h{\mathrm{D.}}_{T}w\&le;P_T,w^h{\mathrm{D.}}_{R}w\&le;P_R-{\mathrm{\&sigma;}}_R^2{v}^{h}v$ Where w ^H D _T w represents the total transmission power of the first group of trunk groups, _DR is a diagonal array, R _w = P _s F ^H H ^H G ^H vv ^H GHF, F = diag(f), G = diag(g), Q _w = H ^H G ^H vv ^H GH, I represents the identity matrix, P _S represents the transmit power of the source node, f _m is the mth element in the channel parameter f of the first hop, v _k is the kth element in the beamforming weight vector v of the relay node in the second hop, P _T and P _R are the power constraints of the first hop and the second hop, respectively.