CN104853400A - Link selection method based on multi-user buffer auxiliary relay system - Google Patents

Abstract

The invention belongs to the technical field of wireless communication, and especially relates to a link selection optimized method which maximizes throughput in a multi-user buffer auxiliary relay system. The invention provides a link selection method based on a multi-user buffer auxiliary relay system. The method uses a convex optimization theory, through broadening discrete restriction and using a KKT (Karush-Kuhn-Tucker) condition, an optimal link selection method is obtained. Compared with a conventional relay selection method, the method can obtain larger throughput. The method carries on research aimed at multi-user scenes. The method has better generality, and for a multi-user scene whose number of nodes and links is large, the method can obtain superior performance.

Description

A kind of link selecting method cushioning alternative routing system based on multi-user

Technical field

The invention belongs to wireless communication technology field, particularly relate to multi-user and cushion throughput-maximized link selection optimization method in alternative routing system.

Background technology

Relaying technique has become an important technology in current and future mobile communication system.Traditional half-duplex relay transmission technology is all receive a rear time slot based on the previous time slot in relaying place to send out the pattern hocketed like this.But traditional relay transmission pattern is comparatively large by the impact of channel conditions, namely when the channel conditions of relaying sending and receiving two links wherein any link is poor, the such constant transmissions pattern in relaying place all obtains ideal performance by making system be difficult to.

Buffering alternative routing technology mainly contains two kinds of application now according to the difference of via node quantity.The first application is when system exists multiple via node, buffering on each via node place is equipped with, like this when adopting single relay selection, make system always can in the relay reception information of the information source-repeated link of previous Slot selection optimum, the relaying of the relaying-stay of two nights link of a rear Slot selection optimum sends information, namely cushions auxiliary relay selection technology.The second application is when system only has single via node, owing to introducing buffering, relaying can be allowed while sending and receiving information to have more freedom.Can according to the quality of the different link channel conditions of same time slot, and certain selection algorithm selects link transmission information adaptively.

For the relay selection algorithm of buffering alternative routing system, applying maximum is that the unidirectional decoding of tradition forwards max-min relay selection mode in (Decode and Forward, DF) relaying and max-max relay selection mode.Max-min relay selection mode is mainly used in the middle of non-cushioned DF relay system, and information source sends a packet to relaying at a time slot, then this relaying and then at next time slot by Packet Generation to the stay of two nights.Selected relaying has end-to-end channel quality between best information source-stay of two nights.For max-max algorithm, relaying is equipped with buffering, and thus relaying is at the best information source-repeated link of a Slot selection, and information source sends a packet to selected relaying; Relaying-the stay of two nights link best at next Slot selection carries out transfer of data.Max-min and max-max relay selection algorithm can apply in the middle of the multiple source-multidestination junction network in the present invention.

The link selecting method of current buffering alternative routing system is also few, and is nearly all the situation for the stay of two nights of single information source list, i.e. double bounce 3 relay-models.Relaying is according to the quality of link circuit condition, and this time slot of unrestricted choice is received, or sends out, and system adopts optimum link selection strategy to ensure that throughput of system reaches maximization.In multi-user system, situation is very complicated, because multi-user's relay system has multiple information source and multiple stay of two nights, number of links is numerous, adaptive link selection scheme has no idea directly to obtain, thus at present not for the link selection optimization method of multi-user's relay system.

Summary of the invention

The present invention proposes a kind of link selecting method cushioning alternative routing system based on multi-user, the method utilizes convex optimum theory, by relaxing discrete restriction and utilizing KKT (Karush-Kuhn-Tucker) condition to obtain optimum link selection mode, the maximization of throughput of system can be realized.

Cushion a link selecting method for alternative routing system based on multi-user, concrete steps are as follows:

When S1, t=0, initialization data information, information source and relaying obtain channel condition information (Channel StateInformation, CSI), are specially: according to with obtain with wherein, represent at the i time slot, from a kth information source to the SNR of relaying, represent from the instant SNR being relayed to a kth stay of two nights, for described expectation, represent that i-th time slot is from a kth information source to the average signal-to-noise ratio of relaying, for described expectation, represent that i-th time slot is from the average SNR being relayed to a kth stay of two nights, for described probability density function, for described probability density function, wherein, j represents jth information source, and i represents i-th time slot, k ∈ 1 ..., N}, N represent that system has N number of information source, and N number of stay of two nights and one are equipped with the relaying of N number of buffering, and N is natural number;

S2, according to restrictive condition calculate optimum gate limit value λ _k, specific as follows:

S21、 $E\left\{D_j\left(i\right)C_{\mathrm{SR}}^{\left(j\right)}\left(i\right)\right\}={\∫}_0^{\∞}\left[\underset{k=1,k\≠j}{\overset{N}{\mathrm{\Π}}}{\∫}_0^{H_{\mathrm{SR}}^{\left(k\right)}}f\left(r_S^{\left(k\right)}\right){\mathrm{dr}}_S^{\left(k\right)}\right]\left[\underset{k=1}{\overset{N}{\mathrm{\Π}}}{\∫}_0^{H_{\mathrm{RD}}^{\left(k\right)}}f\left(r_R^{\left(k\right)}\right){\mathrm{dr}}_S^{\left(k\right)}\right],$ Wherein, ${\log }_2(1+r_S^{\left(j\right)})f\left(r_S^{\left(j\right)}\right){\mathrm{dr}}_S^{\left(j\right)}$

$H_{\mathrm{SR}}^{\left(k\right)}={(1+r_S^{\left(j\right)})}^{\frac{{\mathrm{\λ}}_j}{{\mathrm{\λ}}_k}}-1,H_{\mathrm{RD}}^{\left(k\right)}={(1+r_S^{\left(j\right)})}^{-\frac{{\mathrm{\λ}}_j}{1+{\mathrm{\λ}}_k}}-1,$ j∈{1,...,N}；

S22、 $E\left\{D_j\left(i\right)C_{\mathrm{RD}}^{(j-N)}\left(i\right)\right\}={\∫}_0^{\∞}\left[\underset{k=1,k\≠j-N}{\overset{N}{\mathrm{\Π}}}{\∫}_0^{H_{\mathrm{RD}}^{\left(k\right)}}f\left(r_R^{\left(k\right)}\right){\mathrm{dr}}_S^{\left(k\right)}\right]\left[\underset{k=1}{\overset{N}{\mathrm{\Π}}}{\∫}_0^{H_{\mathrm{SR}}^{\left(k\right)}}f\left(r_S^{\left(k\right)}\right){\mathrm{dr}}_S^{\left(k\right)}\right],$ Wherein, ${\log }_2(1+r_R^{(j-N)})f\left(r_R^{(j-N)}\right){\mathrm{dr}}_R^{(j-N)}$

$H_{\mathrm{SR}}^{\left(k\right)}={(1+r_R^{\left(j\right)})}^{-\frac{1+{\mathrm{\λ}}_j-N}{{\mathrm{\λ}}_k}}-1,$ j∈{N+1,...,2N}；

S23, according to restrictive condition obtain optimum gate limit value λ _k, condition time described restrictive condition is equal in statistical significance into and out of the data of relaying;

S3, start at i-th time slot, information source and via node obtain CSI information, are specially: according to the instant SNR of 2N bar channel with calculate corresponding channel capacity with $C_{\mathrm{RD}}^{\left(k\right)}\left(i\right)={\log }_2(1+r_R^{\left(k\right)}\left(i\right));$

S4, according to S2 optimum gate limit value λ _kobtain the decision function of 2N bar channel described in S3 $V_j\left(i\right)=\left\{\begin{array}{cc}-{\mathrm{\λ}}_j{C}_{\mathrm{SR}}^{\left(j\right)}\left(i\right)& 1\≤j\≤N\\ (1+{\mathrm{\λ}}_{j-N})C_{\mathrm{RD}}^{(j-N)}\left(i\right)& N+1\≤j\≤2N\end{array}\right.;$

S5, according to S4 decision function V _ji (), selects to make V _ji information source that () is maximum or the stay of two nights are carried out sending or receiving, namely

S6, judge whether transfer of data completes, if complete, DTD, if do not complete, proceeds to step S3.

The invention has the beneficial effects as follows:

The present invention, mainly for the buffering alternative routing system of multiple source-multidestination, is deduced the link decisions function in adaptive link preference pattern by the KKT condition in convex optimum theory, obtain optimum link selection algorithm.The present invention can obtain larger throughput compared to traditional relay selection mode.

The present invention is directed to multi-user scene research, have more generality, and for node and the numerous multi-user scene of number of links, the method that the present invention proposes can obtain superior performance.

Accompanying drawing explanation

Fig. 1 is system diagram of the present invention.

Fig. 2 is flow chart of the present invention.

Fig. 3 be the present invention when two couples of users, entire system throughput is along with Ω _sR/ Ω _rDchange curve.

Fig. 4 be the present invention when two couples of users, entire system throughput is along with Ω _sR/ Ω _rDchange curve.

The throughput ratio of the throughput that Fig. 5 obtains for method according to the present invention and max-max algorithm is along with Ω _sR/ Ω _rDchange curve.

Embodiment

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

As shown in Figure 1:

System has N number of information source, and the relaying of N number of stay of two nights and a N number of buffering of outfit, wherein, N is natural number.

In system, a kth information source can only communicate with a kth stay of two nights, and information source needs first information to be sent to relaying carries out decoding and be then stored into during kth cushions, and wherein, k is natural number and 0≤k≤N.

Within the system, the time is divided into infinite multiple isometric time slot, in one time slot, a link can only be had to carry out transfer of data. represent at the i time slot, the Instant SNR (Signal To Noise Ratio, SNR) from a kth information source to relaying, represent from the instant SNR being relayed to a kth stay of two nights, for described expectation, represent that i-th time slot is from a kth information source to the average signal-to-noise ratio of relaying, for described expectation, represent that i-th time slot is from the average SNR being relayed to a kth stay of two nights, for described probability density function, for described probability density function.

If binary variable D _j(i) ∈ 0,1} represents that information source or whether selected the carrying out of the stay of two nights send or receive, wherein, j ∈ 1 ..., N, N+1 ..., 2N}.Work as D _ji ()=1 item represents that a jth information source or the stay of two nights are undertaken sending or receiving by selection, if j ∈ 1 ..., N}, then represent that information source j is sent, if j ∈ is { N+1 by selection, ..., 2N}, then represent the data of the selected reception of (j-N) individual stay of two nights from relaying.

At i-th time slot, suppose that a kth information source is carried out transmission data by selection, the bit number Q in relaying in a kth buffering _ki () represents, a kth information source sends the data of individual bit to relaying, wherein it is the channel capacity of selected link.After a kth information source sends data, the data volume in a relaying place kth buffering is $Q_k\left(i\right)=Q_k(i-1)+C_{\mathrm{SR}}^{\left(k\right)}\left(i\right).$

Similar, at i-th time slot, if the selected reception data of a kth stay of two nights, after relaying sends data, the data volume in a kth buffering is wherein $C_{\mathrm{RD}}^{\left(k\right)}\left(i\right)={\log }_2(1+r_R^{\left(k\right)}\left(i\right)).$

Because buffering does not produce any data, and all message sent from information source all will be transferred to the stay of two nights, so should be equal in statistical significance into and out of the data volume of buffering, can obtain $E\left\{D_k\left(i\right)C_{\mathrm{SR}}^{\left(k\right)}\left(i\right)\right\}=E\left\{D_{k+N}\left(i\right)\min (C_{\mathrm{RD}}^{\left(k\right)}\left(i\right),Q_k(i-1))\right\},$ Wherein, k ∈ 1 ..., N}.Because the data into and out of relaying are equal in statistical significance, so almost can ignore, thus equation can abbreviation be $E\left\{D_k\left(i\right)C_{\mathrm{SR}}^{\left(k\right)}\left(i\right)\right\}=E\left\{D_{k+N}\left(i\right)C_{\mathrm{RD}}^{\left(k\right)}\left(i\right)\right\}.$

Suppose that information source has enough data to transmit all the time, number of timeslots meets M → ∞, and therefore the average transmission rate of information source k can be expressed as the average received throughput of stay of two nights k is expressed as ${\stackrel{\‾}{R}}_{\mathrm{RD}}^{\left(k\right)}=\frac{1}{M}\underset{i=1}{\overset{M}{\mathrm{\Σ}}}{D}_{k+N}\left(i\right)C_{\mathrm{RD}}^{\left(k\right)}\left(i\right)\text{.}$ Wherein k ∈ 1 ..., N}.

Total throughout is defined as the data total amount arriving the stay of two nights, can be expressed as therefore this optimization problem can be expressed as $\begin{array}{c}\max \mathrm{mize}{\mathrm{\Σ}}_{k=1}^N{\stackrel{\‾}{R}}_{\mathrm{RD}}^{\left(k\right)}\\ \mathrm{subjecttoC}1:{\stackrel{\‾}{R}}_{\mathrm{SR}}^{\left(k\right)}={\stackrel{\‾}{R}}_{\mathrm{RD}}^{\left(k\right)},\∀k\\ C2:\underset{j=1}{\overset{2N}{\mathrm{\Σ}}}{D}_j\left(i\right)=1,\∀i\\ C3:D_j\left(i\right)[1-D_j\left(i\right)]=0,\∀i,j\end{array},$ Wherein, C1 is equal in statistical significance into and out of the data of relaying in order to meet, C2 and C3 is to ensure that each time slot only has an information source or the stay of two nights to be selected.

Although D _j(i) be get 0 or 1 a binary variable, suppose described D _ji () consecutive variations between [0,1] carries out scaling to restrictive condition, then restrictive condition is by D _j(i) [1-D _j(i)]=0 become 0≤D _j(i)≤1.Can prove that this optimization problem is at scaling restrictive condition 0≤D _ji the optimal solution behind ()≤1 occurs in D _ji () gets the situation (0 or 1) of boundary value.So the scaling of this binary variable does not affect solving of optimization problem.

After scaling restrictive condition, C3 can be expressed as again $\begin{array}{c}{\mathrm{\β}}_j\left(i\right)[1-D_j\left(i\right)]=0,{\mathrm{\β}}_j\left(i\right)\≥0\\ {\mathrm{\γ}}_j\left(i\right)D_j\left(i\right)=0,{\mathrm{\γ}}_j\left(i\right)\≥0\end{array},$ Wherein, β _j(i) and γ _ji () is Lagrange multiplier.

In order to simplify, select 2 information sources, 2 stays of two nights are carried out setting forth and expand to N number of information source and N number of stay of two nights.

First, we construct the KKT expression formula after scaling:

$\begin{array}{c}P(D_j\left(i\right),{\mathrm{\λ}}_j,\mathrm{\α}\left(i\right),{\mathrm{\β}}_j\left(i\right),{\mathrm{\γ}}_j\left(i\right))=\frac{1}{N}\underset{i=1}{\overset{N}{\mathrm{\Σ}}}[D_3\left(i\right)C_{\mathrm{RD}}^{\left(1\right)}\left(i\right)+D_4\left(i\right)C_{\mathrm{RD}}^{\left(2\right)}\left(i\right)]-\\ {\mathrm{\λ}}_1\frac{1}{N}\underset{i=1}{\overset{N}{\mathrm{\Σ}}}[D_1\left(i\right)C_{\mathrm{SR}}^{\left(1\right)}\left(i\right)-D_3\left(i\right)C_{\mathrm{RD}}^{\left(1\right)}\left(i\right)]-{\mathrm{\λ}}_2\frac{1}{N}\underset{i=1}{\overset{N}{\mathrm{\Σ}}}[D_2\left(i\right)C_{\mathrm{SR}}^{\left(2\right)}\left(i\right)-D_4\left(i\right)C_{\mathrm{RD}}^{\left(2\right)}\left(i\right)]-\end{array},$ By this formula to difference $\underset{1}{\overset{N}{\mathrm{\Σ}}}\mathrm{\α}\left(i\right)[1-\underset{j=1}{\overset{4}{\mathrm{\Σ}}}{D}_j\left(i\right)]+\underset{j=1}{\overset{4}{\mathrm{\Σ}}}\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\β}}_j\left(i\right)[1-D_j\left(i\right)]+\underset{j=1}{\overset{4}{\mathrm{\Σ}}}\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\γ}}_j\left(i\right)[1-D_j\left(i\right)]=0$

D _ji () obtains as differential wherein, j=1 ..., 4.

Link selection can be expressed as:

$D_1^{*}\left(i\right):-{\mathrm{\λ}}_1{C}_{\mathrm{SR}}^{\left(1\right)}\left(i\right)+\mathrm{\α}\left(i\right)-{\mathrm{\β}}_1\left(i\right)+{\mathrm{\γ}}_1\left(i\right)=0$

$D_2^{*}\left(i\right):-{\mathrm{\λ}}_2{C}_{\mathrm{SR}}^{\left(2\right)}\left(i\right)+\mathrm{\α}\left(i\right)-{\mathrm{\β}}_2\left(i\right)+{\mathrm{\γ}}_2\left(i\right)=0$

$D_3^{*}\left(i\right):(1+{\mathrm{\λ}}_1)C_{\mathrm{RD}}^{\left(1\right)}\left(i\right)+\mathrm{\α}\left(i\right)-{\mathrm{\β}}_3\left(i\right)+{\mathrm{\γ}}_3\left(i\right)=0$

$D_4^{*}\left(i\right):(1+{\mathrm{\λ}}_2)C_{\mathrm{RD}}^{\left(2\right)}\left(i\right)+\mathrm{\α}\left(i\right)-{\mathrm{\β}}_4\left(i\right)+{\mathrm{\γ}}_4\left(i\right)=0$

Without loss of generality, we obtain D ₁the necessary condition of (i)=1 and being generalized to j=2, the situation of 3,4.Work as D ₁i ()=1, according to theorem of complementary slackness, can obtain γ ₁(i)=0 and β _j(i)=0, j=2,3,4.Substituted into above formula can obtain:

$\begin{array}{c}-\mathrm{\α}\left(i\right)+{\mathrm{\β}}_1\left(i\right)=-{\mathrm{\λ}}_1{C}_{\mathrm{SR}}^{\left(1\right)}\left(i\right)\stackrel{\mathrm{\Δ}}{=}{V}_1\left(i\right)\\ -\mathrm{\α}\left(i\right)-{\mathrm{\γ}}_2\left(i\right)=-{\mathrm{\λ}}_2{C}_{\mathrm{SR}}^{\left(2\right)}\left(i\right)\stackrel{\mathrm{\Δ}}{=}{V}_2\left(i\right)\\ -\mathrm{\α}\left(i\right)-{\mathrm{\γ}}_3\left(i\right)=(1+{\mathrm{\λ}}_1)C_{\mathrm{RD}}^{\left(1\right)}\left(i\right)\stackrel{\mathrm{\Δ}}{=}{V}_3\left(i\right)\\ -\mathrm{\α}\left(i\right)-{\mathrm{\γ}}_4\left(i\right)=(1+{\mathrm{\λ}}_2)C_{\mathrm{RD}}^{\left(2\right)}\left(i\right)\stackrel{\mathrm{\Δ}}{=}{V}_4\left(i\right)\end{array},$ Wherein, V _ki () is referred to as choice function, deduct all the other equatioies can obtain V with above formula first equation ₁(i)-V _j(i)=β ₁(i)+γ _j(i), j=2,3,4.Due to β _j(i)>=0, γ _j(i)>=0, therefore V ₁(i)>=max{V ₂(i), V ₃(i), V ₄(i) }, in like manner, for we can obtain the necessary condition of corresponding choice function.From above formula, we are easy to obtain-1≤λ _k≤ 0, k=1,2.Situation for N number of information source and N number of stay of two nights also can in like manner obtain.

Embodiment adopts the emulation mode of Monte Carlo, cushions alternative routing sprocket wheel selection algorithm and traditional max-min relay selection algorithm and max-max relay selection algorithm performance compare what mention in the present invention based on multi-user.

Fig. 3 be the present invention when 2 couples of users, entire system throughput is along with Ω _sR/ Ω _rDchange curve, wherein, Ω _sRget 10dB and 20dB respectively.

Theoretical value is by solving

$E\left\{D_j\left(i\right)C_{\mathrm{SR}}^j\left(i\right)\right\}={\∫}_0^{\∞}{\∫}_0^{H_{\mathrm{SR}}^{\left(1\right)}}...{\∫}_0^{H_{\mathrm{SR}}^{(j-1)}}{\∫}_0^{H_{\mathrm{SR}}^{(j+1)}}...{\∫}_0^{H_{\mathrm{SR}}^{\left(N\right)}}{\∫}_0^{H_{\mathrm{RD}}^{\left(1\right)}}...{\∫}_0^{H_{\mathrm{RD}}^{\left(N\right)}}f\left(r_S^{\left(1\right)}\right){\mathrm{dr}}_S^{\left(1\right)}...f\left(r_S^{(j-1)}\right){\mathrm{dr}}_S^{(j-1)}$ $f\left(r_S^{(j+1)}\right){\mathrm{dr}}_S^{(j+1)}...f\left(r_S^{\left(N\right)}\right){\mathrm{dr}}_S^{\left(N\right)}f\left(r_R^{\left(1\right)}\right){\mathrm{dr}}_R^{\left(1\right)}...f\left(r_R^{\left(N\right)}\right){\mathrm{dr}}_R^{\left(N\right)}{\log }_2(1+r_S^{\left(j\right)})f\left(r_S^{\left(j\right)}\right){\mathrm{dr}}_S^{\left(j\right)}$ With

$E\left\{D_j\left(i\right)C_{\mathrm{RD}}^{(j-N)}\left(i\right)\right\}={\∫}_0^{\∞}{\∫}_0^{H_{\mathrm{RD}}^{\left(1\right)}}...{\∫}_0^{H_{\mathrm{RD}}^{(j-1)}}{\∫}_0^{H_{\mathrm{RD}}^{(j+1)}}...{\∫}_0^{H_{\mathrm{RD}}^{\left(N\right)}}{\∫}_0^{H_{\mathrm{SR}}^{\left(1\right)}}...{\∫}_0^{H_{\mathrm{SR}}^{\left(N\right)}}f\left(r_R^{\left(1\right)}\right){\mathrm{dr}}_R^{\left(1\right)}...f\left(r_R^{(j-N-1)}\right){\mathrm{dr}}_R^{(j-N-1)}$ Obtain, $f\left(r_R^{(j-N+1)}\right){\mathrm{dr}}_R^{(j-N+1)}...f\left(r_R^{\left(N\right)}\right){\mathrm{dr}}_R^{\left(N\right)}f\left(r_S^{\left(1\right)}\right){\mathrm{dr}}_S^{\left(1\right)}...f\left(r_S^{\left(N\right)}\right){\mathrm{dr}}_S^{\left(N\right)}{\log }_2(1+r_R^{\left(j\right)})f\left(r_R^{\left(j\right)}\right){\mathrm{dr}}_R^{\left(j\right)}$ Simulation value is by carrying out link selection at each time slot, namely then average throughput is calculated.

Can see that simulation value and theoretical value are coincide very much.The throughput of max-max algorithm is also in analogous diagram simultaneously, can be clear that optimal algorithm of the present invention has the throughput gain of highly significant compared to traditional max-max algorithm.

Fig. 4 be the present invention when 2 couples of users, entire system throughput is along with Ω _sR/ Ω _rDchange curve, wherein we allow Ω _sR+ Ω _rD=20dB and Ω _sRvalue is increased to 19.5dB from 0.5dB always.Can find out at Ω from analogous diagram _sR=Ω _rDtime, throughput of system reaches maximum.Meanwhile, we can find out that the throughput of max-max algorithm is much larger than max-min algorithm, but they are all lower than optimum link selection algorithm proposed by the invention.

Fig. 5 display be that the throughput ratio of the throughput that obtains of algorithm that the present invention proposes and max-max algorithm is along with Ω _sR/ Ω _rDchange curve.In simulations, we allow Ω _sR=1dB, simultaneously Ω _sR/ Ω _rD=[0.1,10], the quantity that user is right is respectively 2 right, and 3 to right with 4.We can see the increase along with the right quantity of user, and the ratio of throughput, in reduction, shows the performance convergence of performance of the present invention to max-max algorithm.Therefore, the performance that the link selection algorithm proposed by the invention system less to user shows can be better.In addition, Ω is worked as _sR=Ω _rDtime, throughput ratio reaches minimum, Ω _sRand Ω _rDdifference larger, the throughput gain that the present invention obtains is also larger.

Claims (2)

1. cushion a link selecting method for alternative routing system based on multi-user, it is characterized in that, comprise the steps:

When S1, t=0, initialization data information, information source and relaying obtain channel condition information (Channel StateInformation, CSI), are specially: according to with obtain with wherein, represent at the i time slot, from a kth information source to the SNR of relaying, represent from the instant SNR being relayed to a kth stay of two nights, for described expectation, represent that i-th time slot is from a kth information source to the average signal-to-noise ratio of relaying, for described expectation, represent that i-th time slot is from the average SNR being relayed to a kth stay of two nights, for described probability density function, for described probability density function, wherein, j represents jth information source, and i represents i-th time slot, k ∈ 1 ..., N}, N represent that system has N number of information source, and N number of stay of two nights and one are equipped with the relaying of N number of buffering, and N is natural number;

S2, according to restrictive condition calculate optimum gate limit value λ _k, wherein, condition time described restrictive condition is equal in statistical significance into and out of the data of relaying;

S3, start at i-th time slot, information source and via node obtain CSI information, are specially: instant according to 2N bar channel with calculate corresponding channel capacity with $C_{\mathrm{RD}}^{\left(k\right)}\left(i\right)={\log }_2(1+r_R^{\left(k\right)}\left(i\right));$

S4, according to S2 optimum gate limit value λ _kobtain the decision function of 2N bar channel described in S3 $V_j\left(i\right)=\left\{\begin{array}{cc}-{\mathrm{\λ}}_j{C}_{\mathrm{SR}}^{\left(j\right)}\left(i\right)& 1\≤j\≤N\\ (1+{\mathrm{\λ}}_{j-N})C_{\mathrm{RD}}^{(j-N)}\left(i\right)& N+1\≤j\≤2N\end{array}\right.;$

S5, according to S4 decision function V _ji (), selects to make V _ji information source that () is maximum or the stay of two nights are carried out sending or receiving, namely

S6, judge whether transfer of data completes, if complete, DTD, if do not complete, proceeds to step S3.

2. a kind of link selecting method cushioning alternative routing system based on multi-user according to claim 1, is characterized in that: calculate optimum gate limit value λ described in S2 _k, specific as follows:

S21、 $E\left\{D_j\left(i\right)C_{\mathrm{SR}}^{\left(j\right)}\left(i\right)\right\}={\∫}_0^{\∞}\left[\underset{k=1,k\≠j}{\overset{N}{\mathrm{\Π}}}{\∫}_0^{H_{\mathrm{SR}}^{\left(k\right)}}f\left(r_s^{\left(k\right)}\right){\mathrm{dr}}_s^{\left(k\right)}\right]\left[\underset{k=1}{\overset{N}{\mathrm{\Π}}}{\∫}_0^{H_{\mathrm{RD}}^{\left(k\right)}}f\left(r_R^{\left(k\right)}\right){\mathrm{dr}}_s^{\left(k\right)}\right],$ Wherein,

${\log }_2(1+r_S^{\left(j\right)})f\left(r_S^{\left(j\right)}\right){\mathrm{dr}}_S^{\left(j\right)}$

$H_{\mathrm{SR}}^{\left(k\right)}={(1+r_S^{\left(j\right)})}^{\frac{{\mathrm{\λ}}_j}{{\mathrm{\λ}}_k}}-1,H_{\mathrm{RD}}^{\left(k\right)}={(1+r_S^{\left(j\right)})}^{-\frac{{\mathrm{\λ}}_j}{1+{\mathrm{\λ}}_k}}-1,j\∈\{1,...,N\};$

S22、 $E\left\{D_j\left(i\right)C_{\mathrm{RD}}^{(j-N)}\left(i\right)\right\}={\∫}_0^{\∞}\left[\underset{k=1,k\≠j-N}{\overset{N}{\mathrm{\Π}}}{\∫}_0^{H_{\mathrm{RD}}^{\left(k\right)}}f\left(r_R^{\left(k\right)}\right){\mathrm{dr}}_s^{\left(k\right)}\right]\left[\underset{k=1}{\overset{N}{\mathrm{\Π}}}{\∫}_0^{H_{\mathrm{SR}}^{\left(k\right)}}f\left(r_S^{\left(k\right)}\right){\mathrm{dr}}_s^{\left(k\right)}\right],$ Wherein,

${\log }_2(1+r_R^{(j-N)})f\left(r_R^{(j-N)}\right){\mathrm{dr}}_R^{(j-N)}$

$H_{\mathrm{SR}}^{\left(k\right)}={(1+r_R^{\left(j\right)})}^{-\frac{1+{\mathrm{\λ}}_j-N}{{\mathrm{\λ}}_k}}-1,j\∈\{N+1,...,2N\};$

S23, according to restrictive condition obtain optimum gate limit value λ _k.