CN102625315B - Horizontal concurrent spectrum allocation method in cognitive wireless system - Google Patents

Abstract

The invention discloses a horizontal concurrent spectrum allocation method in a cognitive wireless system. The invention relates to a spectrum allocation method belonging to the field of communication, and aims to solve the problem of poor fairness of the original concurrent method in spectrum allocation. The spectrum allocation method comprises the following steps of: I, establishing a topological graph according to the wireless network of the cognitive wireless system, and initializing, wherein the available spectrum number is M greater than or equal to 1; II, decomposing the topological graph into M sub-graphs; III, calculating the label number in each sub-graph Gk=(Vk, Ek) according to respective standards, calculating the cognitive user label for M*alpha sub-graphs in M sub-graphs by use of the collaborative maximum system total efficiency standard, and calculating the cognitive user label for the remaining M*(1-alpha) sub-graphs by use of the collaborative maximum proportional fairness standard; IV, allocating the spectrum j to the user realizing maximum label i and label j; V, removing the allocated spectrum and updating the topological graph; VI, judging whether the spectra in the graph are completely allocated, and if so, ending the process; and VII, otherwise, returning to the step II. The method disclosed by the invention is applied to the spectrum allocation in wireless communication.

Description

The frequency spectrum distributing method of horizontal parallel in a kind of cognitive radio system

Technical field

The present invention relates to a kind of frequency spectrum distributing method, belong to the communications field.

Background technology

The algorithm of the spectrum allocation may using is at present parallel algorithm, and it is to be decomposed into multiple subgraphs based on graph theory model topology diagram, then on these subgraphs, carries out independently spectrum allocation may:

In cognitive radio system, owing to being use authority user's the frequency spectrum of selecting a good opportunity, the usable spectrum of cognitive radio users is subject to authorized user operating state, the impact of position and coverage, the characteristic that tool changes at one's leisure.Therefore, in cognitive radio system, adopt graph theory coloring models to carry out spectrum allocation may analysis, need to be to the constraints of spectrum allocation may, distribution target etc. is carried out necessary revision.The prerequisite of cognitive radio users use authority frequency range is: if the authorized user of current channel uses, for fear of the interference to authorized user, this channel can not be used by near cognitive family.Fig. 1 is the network topology structure of a cognitive radio system.

Unified graph theory model is by idle matrix, benefit matrix, and interference matrix and allocation matrix are described.

Idle frequency spectrum refers in certain time and space, the untapped frequency spectrum of primary user.Due to locus and the isoparametric difference of through-put power, for different cognitive user, whether the same frequency band free time may be different.For cognitive user, whether the free time is used idle matrix notation to frequency spectrum.If the distribution time is very short with respect to the environmental change time, each matrix remains unchanged in assignment period.

The benefit benefit matrix notation that user obtains: benefit matrix B={ b _{n, m}} _{p × Q}, b _{n, m}characterize the benefit weight that user n uses frequency spectrum m to bring, as spectrum capabilities etc., P is number of users, and subscript is from 0 to P-1, and Q is total frequency band number, and subscript is from 0 to Q-1.Matrix L is combined with matrix B, can draw the benefit L of effective spectrum _b={ l _{n, m}b _{n, m}} _{p × Q}.

Interference between cognitive radio users represents with interference matrix: interference matrix set C={c _{n, k}| c _{n, k}∈ { 0,1}} _{p × P}, c _{n, k}=1 represents when user n and user k use same frequency spectrum at the same time to produce interference, in the time of n=k, and c _{n, k}=0.

Glitch-free spectrum allocation may matrix A={ a _{n, m}| a _{n, m}∈ { 0,1}} _{p × Q}, a _{n, m}=1 represents that frequency spectrum m is assigned to user n.A must meet noiseless condition:

$a_{n,m}\&CenterDot;a_{k,m}=0\mathrm{if}{c}_{n,k}=1,\&ForAll;n,k<N,m<M---\left(1\right)$

The main standard that parallel method is used is MSB and MPF.

Introduced the benefit of each frequency range, can obtain maximizing the expression formula that frequency spectrum benefit is the optimum allocation standard (Max-Sum-Bandwidth, MSB) of target, the mathematic(al) representation that correspondingly system reaches maximum spectrum benefit becomes:

$\underset{A\&Element;{\mathrm{\&Lambda;}}_{P\&times;Q}}{\max }\underset{n=0}{\overset{P-1}{\mathrm{\&Sigma;}}}\underset{m=0}{\overset{Q-1}{\mathrm{\&Sigma;}}}{a}_{n,m}\&CenterDot;b_{n,m}---\left(2\right)$

Λ in formula _{p × Q}---the set of all noiseless spectrum allocation may matrix A that satisfy condition.

The physical meaning of expression formula is that spectrum allocation may is the maximum spectrum benefit that system is brought, the common representative of consumer of benefit matrix obtainable transmission rate on each frequency band.

In order to compare the fairness of distinct methods, method has also been introduced the module about spectrum allocation may fairness, i.e. maximum ratio fairness tolerance (Max-Proportional-Fair, MPF):

$\underset{A\&Element;{\mathrm{\&Lambda;}}_{N\&times;M}}{\max }{(\underset{n=0}{\overset{P-1}{\mathrm{\&Pi;}}}\underset{m=0}{\overset{Q-1}{\mathrm{\&Sigma;}}}{a}_{n,m}\&CenterDot;b_{n,m})}^{\frac{1}{N}}---\left(3\right)$

Because parallel method is that original topology diagram is decomposed into multiple subgraphs, then on these subgraphs, carry out independently spectrum allocation may, between each subgraph, do not carry out the exchange of information, so each subgraph cannot be considered the distribution condition of other subgraphs in distributing, cannot use the information of having distributed frequency spectrum, this is the poor basic reason of parallel method fairness.

Summary of the invention

While the present invention seeks to for spectrum allocation may, the original poor problem of parallel method fairness, provides the frequency spectrum distributing method of horizontal parallel in a kind of cognitive radio system.

The frequency spectrum distributing method of horizontal parallel in a kind of cognitive radio system of the present invention, realization steps of the method are:

Step 1, set up topological diagram according to the wireless network of cognitive radio system, and this topological diagram is carried out to initialization, the usable spectrum number in this topological diagram is M, M >=1;

Step 2, topological diagram is counted to M by usable spectrum be decomposed into M subgraph;

Step 3, at each subgraph G _k=(V _k, E _k) according to criterion calculation index value separately, adopt collaboration type to maximize system total benefit criterion calculation cognitive user label the M* α in M subgraph subgraph, the individual subgraph of all the other M* (1-α) adopts collaboration type to maximize equitable proportion criterion calculation cognitive user label simultaneously

Wherein, α is defined as horizontal parallel coefficient, 0≤α≤1; V _kfor subgraph G _kthe set on summit, i.e. the set of cognitive user in subgraph; E _kfor subgraph G _kthe set on limit, i.e. subgraph G _kin adjacent a pair of cognitive user between the set of the interference that exists; K=1 ..., M;

Step 4, at each subgraph G _k=(V _k, E _k) in select to make label label _{i, j}the cognitive user of value maximum and frequency spectrum j is distributed to this cognitive user, and 1≤j≤M, 0≤i≤N, N is subgraph G _kmiddle cognitive user number;

Step 5, remove the frequency spectrum being assigned, then upgrade topology diagram;

Step 6, judge whether the frequency spectrum in topological diagram is assigned, if so,, execution step seven, if not,, turns back to step 2;

Step 7, the spectrum allocation may that finishes.

Advantage of the present invention is: the present invention proposes on the basis based on parallel algorithm, and this algorithm uses two kinds of various criterions in allocated channel: collaboration type maximizes system total benefit standard and collaboration type maximizes equitable proportion standard; Make parallel algorithm in the time distributing frequency spectrum, channel of every distribution can carry out an information exchange between subgraph, uses the spectrum information having distributed on other subgraphs, has improved the fairness of spectrum allocation may, wherein i is cognitive user, j is that spectrum number prevents that a large amount of spectrum allocation may is to a cognitive user, as shown in Figure 7, the horizontal parallel algorithm of this paper has been accomplished the increase with spectrum number in addition, time overhead changes little, compares original parallel algorithm, and horizontal parallel algorithm herein shortens the time of spectrum allocation may, improve spectrum allocation may efficiency, as shown in Figure 8.

Brief description of the drawings

Fig. 1 is the graph theory model of cognitive radio system, the i.e. network topology structure of a cognitive radio system, six summit su1～su6 in figure represent six different cognitive user, four summit puA～puD represent four different authorized users, and suppose that the mandate frequency range that each authorized user uses is respectively frequency spectrum A, frequency spectrum B, frequency spectrum C and frequency spectrum D;

Fig. 2 is the flow chart of horizontal parallel algorithm of the present invention;

Fig. 3 is the schematic diagram of system fairness along with the situation of change of α, in figure for number of users N is 10, number of channel M is the value of 10 o'clock α, in figure for number of users N is 10, number of channel M is the value of 20 o'clock α,

In figure counting N for cognitive user is 20, and number of channel M is the value of 10 o'clock α, in figure counting N for cognitive user is 20, and number of channel M is the value of 20 o'clock α;

Fig. 4 is the extreme value of the α obtaining of 50 tests;

Fig. 5 is that cognitive user number is fixed as in 5 situation optimum alpha along with the Changing Pattern emulation schematic diagram of spectrum number;

Fig. 6 is that spectrum number is fixed as in 20 situation optimum alpha along with the emulation schematic diagram of the Changing Pattern of cognitive user number.

Fig. 7 is the fair comparison of horizontal parallel method and original parallel algorithm, in figure for original fairness value, in figure for fairness value of the present invention.

Fig. 8 is the comparison of the time overhead of horizontal parallel method and original parallel algorithm, in figure for the value of time overhead of the present invention, in figure for the value of original time overhead.

Embodiment

Embodiment one: below in conjunction with Fig. 2, present embodiment is described,

A frequency spectrum distributing method for horizontal parallel in cognitive radio system, realization steps of the method are:

Step 1, set up topological diagram according to the wireless network of cognitive radio system, and this topological diagram is carried out to initialization, the usable spectrum number in this topological diagram is M, M >=1;

Step 2, topological diagram is counted to M by usable spectrum be decomposed into M subgraph;

Step 3, at each subgraph G _k=(V _k, E _k) according to criterion calculation index value separately, adopt collaboration type to maximize system total benefit criterion calculation cognitive user label the M* α in M subgraph subgraph, the individual subgraph of all the other M* (1-α) adopts collaboration type to maximize equitable proportion criterion calculation cognitive user label simultaneously

Wherein, α is defined as horizontal parallel coefficient, 0≤α≤1; V _kfor subgraph G _kthe set on summit, i.e. the set of cognitive user in subgraph; E _kfor subgraph G _kthe set on limit, i.e. subgraph G _kin adjacent a pair of cognitive user between the set of the interference that exists; K=1 ..., M;

Step 4, at each subgraph G _k=(V _k, E _k) in select to make label label _{i, j}the cognitive user of value maximum and frequency spectrum j is distributed to this cognitive user, and 1≤j≤M, 0≤i≤N, N is subgraph G _kmiddle cognitive user number;

Step 5, remove the frequency spectrum being assigned, then upgrade topology diagram;

Step 6, judge whether the frequency spectrum in topological diagram is assigned, if so,, execution step seven, if not,, turns back to step 2;

Step 7, the spectrum allocation may that finishes.

Embodiment two: below in conjunction with Fig. 2, present embodiment is described, present embodiment is to the further illustrating of execution mode one,

Horizontal parallel factor alpha in step 3 of the present invention refers to that each takes turns to calculate and uses when label subgraph number of maximization system total benefit standard and this to take turns the ratio of total subgraph number of point timing.

Embodiment three: below in conjunction with Fig. 2, present embodiment is described, present embodiment is to the further illustrating of execution mode one,

The method of calculating cognitive user label when employing of the present invention maximizes system total benefit standard is:

When cognitive user i has been assigned with after frequency spectrum j, the D adjacent with this cognitive user _{i, j}individual node cannot use this frequency spectrum j having distributed, and current spectrum allocation may is b to the contribution margin of whole system total bandwidth _{i, j}/ (D _{i, j}+ 1); ${\mathrm{label}}_{i,j}=\frac{b_{i,j}}{D_{i,j}+1},$

D in formula _{i, j}for the cognitive user number that contains frequency spectrum j in and usable spectrum adjacent with cognitive user i, 0≤i≤N, 0≤j≤M;

B _{i, j}be defined as spectrum capabilities, can be obtained by mountain agriculture formula:

$b_{i,j}={\mathrm{band}}_j\&CenterDot;{\log }_2(1+K\&CenterDot;\frac{p_{i,j}\&CenterDot;{\left|H_{i,j}\right|}^2}{{\mathrm{band}}_j\&CenterDot;{\mathrm{\&sigma;}}_0^2}),$

Band in formula _jfor the bandwidth of frequency spectrum j; p _{i, j}for the transmitting power of cognitive user i on frequency spectrum j;

| H _{i, j}| ²for the spectrum gain of frequency spectrum j to cognitive user i, suppose that every frequency spectrum is all identical to the gain of each cognitive user, has: for the noise power on unit bandwidth;

K is a constant, represents the error rate requirement of cognitive user, and K is expressed as:

$K=-\frac{1.5}{\ln (5\&CenterDot;\mathrm{BER})}.$

Embodiment four: below in conjunction with Fig. 2, present embodiment is described, present embodiment is to the further illustrating of execution mode one,

When employing of the present invention maximizes system total benefit standard distribution frequency spectrum, occur that several nodes have identical label in the time calculating cognitive user label time, the node that bandwidth is little will be given higher index value.

Embodiment five: below in conjunction with Fig. 2, present embodiment is described, present embodiment is to the further illustrating of execution mode one,

The method that employing collaboration type of the present invention maximizes equitable proportion criterion calculation cognitive user label is:

${\mathrm{label}}_{i,j}=\frac{b_{i,j}}{(D_{i,j}+1)\&CenterDot;\underset{j=0}{\overset{M-1}{\mathrm{\&Sigma;}}}{a}_{i,j}\&CenterDot;b_{i,j}}$

B _{i, j}be defined as spectrum capabilities, D _{i, j}for the cognitive user number that contains frequency spectrum j in and usable spectrum adjacent with cognitive user i, 0≤i≤N, 0≤j≤M.

Specific embodiment:

Step 1, set up topological diagram according to the wireless network of cognitive radio system, and this topological diagram is carried out to initialization, the usable spectrum number in this topological diagram is M, M >=1;

Step 2, topological diagram is counted to M by usable spectrum be decomposed into M subgraph;

Step 3, at each subgraph G _k=(V _k, E _k) according to criterion calculation index value separately, adopt collaboration type to maximize system total benefit criterion calculation user label the M* α in M subgraph subgraph, the individual subgraph of all the other M* (1-α) adopts collaboration type to maximize equitable proportion criterion calculation user label simultaneously;

Determining of step 4, horizontal parallel coefficient

In horizontal parallel canonical algorithm, horizontal parallel factor alpha is defined as each and takes turns and calculate the subgraph number that uses CMSB standard when label and take turns the ratio of total subgraph number of point timing with this.

The impact on system fairness for analysis level row coefficient, carries out emulation to system fairness in following four kinds of situations with the situation of change of horizontal parallel coefficient herein.Average system fairness when α gets different value

(1) to count N be 10 to cognitive user, and number of channel M is 10;

(2) to count N be 10 to cognitive user, and number of channel M is 20;

(3) to count N be 20 to cognitive user, and number of channel M is 10;

(4) to count N be 20 to cognitive user, and number of channel M is 20.

1000 topological diagrams of random generation in every kind of situation, when calculating α gets different value, the mean value of the system fairness of 1000 topological diagrams, for a certain fixation case, between 0～1, there is an extreme value that makes system fairness maximum in horizontal parallel factor alpha, for different situations, the extreme value of fairness maximum is also not quite similar.

For example, for situation (2), in the time of α=0.15, can make the fairness maximum of system.In fact, if carry out emulation repeatedly, just can find, even for a certain fixing situation, due to the finiteness of the randomness of its topological structure and the random Topological Numbers producing, make the extreme value of the α of its fairness maximum neither be identical.For example, for situation (2), ask according to the method described above the extreme value of α, duplicate test 50 times, the extreme value of α is not what fix, its value difference in different tests.But the extreme value of α neither continually varying, but between several values value, therefore, in order to determine the extreme value of the α in a certain situation, can in this kind of situation, repeatedly solve the extreme value of α, then get mode in the many numerical value optimum alpha as this kind of situation.In example test as shown in fig. 5, the extreme value of α is got 0.15 test and is occurred 32 times, gets 0.2 test and occurs 6 times, gets 0.3 test and occurs 5 times, and the test of getting other values occurs 7 times, therefore can think that the optimum alpha of situation (2) is 0.15.

Because the optimum alpha of the situation of different cognitive users number and different channels number is also different, cognitive user number is fixed as in 5 situation to optimum alpha along with the Changing Pattern of the number of channel herein, and the number of channel is fixed as in 20 situation optimum alpha along with the Changing Pattern of cognitive user number has carried out emulation.From analogous diagram, can find out, optimum alpha when different cognitive users number is counted with different channels is different, and it is not significantly regular, so in order to determine the optimum alpha of different situations in emulation below, herein by cognitive user number value successively between 5 to 20, the number of channel between 5 to 20 successively the optimum alpha value in value obtain, and its list is preserved, during in order to emulation hereinafter, use.Annex table 1 is shown in by optimum alpha value table.

Optimum level the row coefficient of annex table 1 horizontal parallel method

The present invention is not limited to above-mentioned execution mode, can also be the reasonable combination of technical characterictic described in the respective embodiments described above.

Claims (3)

1. a frequency spectrum distributing method for horizontal parallel in cognitive radio system, is characterized in that: realize and steps of the method are:

Step 1, set up topological diagram according to the wireless network of cognitive radio system, and this topological diagram is carried out to initialization, the usable spectrum number in this topological diagram is M, M >=1;

Step 2, topological diagram is counted to M by usable spectrum be decomposed into M subgraph;

Step 3, at each subgraph G _k=(V _k, E _k) according to criterion calculation index value separately, adopt collaboration type to maximize system total benefit criterion calculation cognitive user label the M* α in M subgraph subgraph, the individual subgraph of all the other M* (1-α) adopts collaboration type to maximize equitable proportion criterion calculation cognitive user label simultaneously

Wherein, α is defined as horizontal parallel coefficient, and horizontal parallel factor alpha refers to that each takes turns to calculate and uses when label subgraph number of maximization system total benefit standard and this to take turns the ratio of total subgraph number of point timing, 0≤α≤1; V _kfor subgraph G _kthe set on summit, i.e. the set of cognitive user in subgraph; E _kfor subgraph G _kthe set on limit, i.e. subgraph G _kin adjacent a pair of cognitive user between the set of the interference that exists; K=1 ..., M;

Step 4, at each subgraph G _k=(V _k, E _k) in select to make label label _i,jthe cognitive user of value maximum and frequency spectrum j is distributed to this cognitive user, and 1≤j≤M, 0≤i≤N, N is subgraph G _kmiddle cognitive user number;

Step 5, remove the frequency spectrum being assigned, then upgrade topology diagram;

Step 6, judge whether the frequency spectrum in topological diagram is assigned, if so,, execution step seven, if not,, turns back to step 2;

Step 7, the spectrum allocation may that finishes.

2. according to the frequency spectrum distributing method of horizontal parallel in a kind of cognitive radio system of claim 1, it is characterized in that: the method for calculating cognitive user label while adopting maximization system total benefit standard is:

${\mathrm{label}}_{i,j}=\frac{b_{i,j}}{D_{i,j}+1},$

D in formula _i,jfor the cognitive user number that contains frequency spectrum j in and usable spectrum adjacent with cognitive user i, 0≤i≤N, 0≤j≤M;

B _i,jfor the spectrum capabilities of user i, can be obtained by mountain agriculture formula:

$b_{i,j}={\mathrm{band}}_j.{\log }_2(1+K.\frac{p_{i,j}.{\left|H_{i,j}\right|}^2}{{\mathrm{band}}_j.{\mathrm{\&sigma;}}_0^2}),$

In formula, band _jfor the bandwidth of frequency spectrum j; p _i,jfor the transmitting power of cognitive user i on frequency spectrum j;

| H _i,j| ²for the spectrum gain of frequency spectrum j to cognitive user i, establish every frequency spectrum all identical to the gain of each cognitive user, have: | H _i,j| ²=| H _j| ², for the noise power on unit bandwidth;

K is a constant, represents the error rate requirement of cognitive user, and K is expressed as:

$K=-\frac{1.5}{1n(5.\mathrm{BER})},$ Wherein, BER is the error rate.

3. according to the frequency spectrum distributing method of horizontal parallel in a kind of cognitive radio system of claim 1, it is characterized in that: while adopting maximization system total benefit standard to distribute frequency spectrum, occur that several nodes have identical label in the time calculating cognitive user label time, the cognitive user to bandwidth minimum by spectrum allocation may.