CN106231610A - Resource allocation methods based on sub-clustering in Femtocell double-layer network - Google Patents

Abstract

The present invention relates to resource allocation methods based on sub-clustering in the Femtocell double-layer network of Femto cell, including utilizing three-wheel subchannel allocation algorithm to be, grand user MUEs distributes subchannel；The object of planning according to power distribution and constraints, using classical water-filling algorithm is that MUEs distributes power；Using the Global Genetic Simulated Annealing Algorithm GASA improved is Femto cell sub-clustering；According to the rate requirement of femto user FUEs, using heuritic approach is that FUEs distributes subchannel；And utilize KKT condition that FUEs is carried out power distribution.The present invention, on the premise of ensureing MUEs proper communication, minimizes the interference between FUEs, improves the availability of frequency spectrum, it is ensured that the service quality of FUEs and MUEs.

Description

Resource allocation methods based on sub-clustering in Femtocell double-layer network

Technical field

The present invention relates to wireless communication technology field, particularly in the Femtocell double-layer network of Femto cell based on The resource allocation methods of sub-clustering.

Background technology

With the third generation (the 3^rdGeneration, 3G) mobile communication system compares, Long Term Evolution (Long Term Evolution, LTE) the spectral frequencies utilization rate of system increases, but its path loss is compared 3G system and has been increased, Therefore LTE system is not provided that good in-door covering.Research shows that the data service of nearly 90% and the speech business of 60% are sent out Raw in indoor and hot spot region, thus how good in-door covering is provided and provides the user satisfied service quality, it is fortune Battalion's current problem demanding prompt solution of business.

By introducing Femto cell in classical macro-cellular coverage, form Femtocell double-layer network, it has also become Solve the effective measures of indoor wireless communication technical problem at present.Femto cell is as short distance, low-power, the family of low cost Community, front yard, is deployed through DSL (Digital Subscriber Loop, DSL) by user or optical fiber is connected to core Heart net, it is possible not only to provide the user the most indoor experience, additionally it is possible to unloading macrocell network flow, and increases network Coverage.Yet with the non-planning of femtocell network, Stochastic accessing and share the characteristics such as frequency spectrum with macrocell, it will Cause the cross-layer interference problem between itself and macrocell and and use same channel other Femto cells between same layer The most how interference problem, reduce above two interference, be the problem needing urgently to study and solve.

Presently relevant document has been proposed for some methods disturbed for reducing cross-layer and disturb with layer, the most centralized It is effective hands of interference in the double-deck Femtocell network of suppression that interference management scheme uses partial frequency multiplexing and power to control Section.It addition, it has been proposed that a kind of packet-based interference management scheme, method particularly includes: orthogonal in group technology is divided into group Orthogonal packet between packet and group, in group, the Femto cell of serious interference is divided at identical group, in identical group by orthogonal group technology Femto cell use different subchannel, different groups can be with the identical subchannel of multiplexing；On the contrary, orthogonal group technology between group Being will not interfere with or disturb the least Femto cell to divide at identical group, the Femto cell in identical group can be identical with multiplexing Subchannel, different component joins different subchannels.

Inventor find, in the prior art, centralized interference management scheme along with the increase of Femto cell quantity, its Computation complexity also can sharply increase so that the method is difficult in the scene of Femto cell dense deployment application；Meanwhile, base In the interference management scheme of packet, in group, orthogonal group technology is to be grouped from the self of each Femto cell, it is difficult to Finding the overall situation to be preferably grouped scheme, meanwhile, the Femto cell number in each group that this packet scheme obtains is the most uneven Weighing apparatus so that a part of femto user (Femtocell User Equipments, FUEs) can not be assigned to abundant son letter Road, thus it is difficult to ensure that the service quality (Quality of Service, QoS) of FUEs.

Summary of the invention

For above problem of the prior art, present invention discusses the resource allocation problem in Femtocell double-layer network, Propose resource allocation methods based on sub-clustering in a kind of Femtocell double-layer network, can effectively suppress cross-layer to disturb and same Layer interference, and grand user (Macrocell User Equipments, MUEs) and the QoS demand of FUEs can be met.

The present invention is a kind of for Femtocell double-layer network resource allocation methods based on sub-clustering, comprises the following steps:

Step 101: utilize three-wheel subchannel allocation algorithm that grand user MUEs performs subchannel distribution；

Step 102: the object of planning distributed according to grand user power and constraints, using classical water-filling algorithm is MUEs Distribution power；

Step 103: using the Global Genetic Simulated Annealing Algorithm GASA improved is Femto cell sub-clustering；

Step 104: according to the rate requirement of femto user FUEs, using heuritic approach is that FUEs distributes subchannel；

Step 105: utilize Caro to need-Ku En-Plutarch (Karush-Kuhn-Tucker, KKT) condition to femto user FUEs carries out power distribution.

Preferably, described step 101 utilizes three-wheel subchannel allocation algorithm that grand user MUEs performs subchannel distribution bag Include: the more new formula quoting the shannon formula grand user's m data rate of modeling isIts In, M is grand total number of users, and K is subchannel sum,For grand user m Signal to Interference plus Noise Ratio in subchannel k, Δ f is channel strip Wide；And then consider the data rate request of grand user, on the premise of meeting grand user rate interval, distribute sub-letter for grand user Road.

Preferably, described on the premise of meeting grand user rate interval, distribute subchannel for grand user, including:

Step 101A: travel through all subchannels, finds out subchannel k making grand user m can obtain maximum channel gain, and will Subchannel k distributes to grand user m, if the speed of the grand user m obtained meets its minimum speed limit demand, the grandest user m no longer joins Add channel distribution, and if then all calculated grand user data rates be satisfied by minimum speed limit demand, then exit circulation；

Step 101B: if subchannel has residue, then repeat step 101A, and grand user rate Rule of judgment becomes sentencing Whether the speed of disconnected calculated corresponding grand user meets its flank speed demand；

Step 101C: if subchannel still has residue, repeat step 101A, no longer carries out grand user data rate and sentences Disconnected.

Preferably, the object of planning that described step 102 is distributed according to grand user power and constraints, use classical water filling Algorithm is that MUEs distribution power includes:

With maximization power system capacity as optimization aim, maximum general power is constraints, builds the power distribution mesh of MUEs Scalar functions:And meet constraints:

$\begin{array}{cc}s.t.& \underset{k=1}{\overset{K}{\Σ}}{p}_k^M\≤P_{total}^M\end{array};$

Using water-filling algorithm is that grand user distributes power, obtains

Wherein,η=Δ f/ ζ ln2 is water line；It is Gain interference ratio in subchannel k,Represent macro base station to grand user m in subchannel k channel gain,Table Show femto base station FBS_jTo grand user m in subchannel k channel gain；Represent femto base station FBS_jIn subchannel Transmitting power on k；σ²For noise power；ζ is Lagrange multiplier, for constant；For macro base station sending out in subchannel k Penetrate power,For total transmitting power, Δ f is channel width, and M is grand total number of users, and K is subchannel sum,.

Preferably, described step 103 uses the Global Genetic Simulated Annealing Algorithm GASA improved to be that Femto cell sub-clustering includes: Using with bunch in femto base station FBSs between interference summation minimum as object function, modeling optimization equation:And meet constraints:C_g∩C_n=Φ (g, n ∈ χ, g ≠ n) And x_in∈{0,1}；Wherein, χ=1 ..., N_ARepresent bunch set, F and N_ARepresent the quantity of femto base station FBSs respectively With bunch quantity；w_ijIt is FBS_iAnd FBS_jBetween interference weights；x_inIt is the sub-clustering oriental matrix of FBSs, works as x_inWhen=1, representing will FBS_iAssign to n-th bunch, work as x_inWhen=0, i.e. represent FBS_iDo not assign to n-th bunch；C_nThe set of FBSs, C in representing n-th bunch_g Represent the set of FBSs in g bunch,For the set of FBSs total in system,And then use something lost Pass simulated annealing and solve this sub-clustering problem.

Preferably, described step 104 is according to the rate requirement of femto user FUEs, and using heuritic approach is that FUEs divides Sub-channel includes: with maximize FUEs data rate as optimization aim:And Meet constraints Wherein, D_jFor FBS_jThe FUEs set of service；Δ f is channel width；For FBS_jThe femto user u of service letter in subchannel k is done Make an uproar ratio；For FBS_jThe rate requirement of the femto user u of service；If a_k,n=1, represent that subchannel k distributes to a bunch C_n, no Then, a_k,n=0；K is subchannel sum.

Preferably, described step 105 utilizes Caro to need-Ku En-Plutarch (Karush-Kuhn-Tucker, KKT) condition pair Femto user FUEs carries out power distribution and includes: set up function model for optimization aim maximizing handling capacity:

And

$\begin{array}{ccc}s.t.& C1& \underset{k=1}{\overset{K}{\Σ}}\underset{u\∈D_j}{\Σ}{p}_{j,k,u}^F\≤P_{total}^F,\∀j\∈C_n\end{array}$

$\begin{array}{cc}C2& \underset{k=1}{\overset{K}{\Σ}}\underset{u\∈D_j}{\Σ}\mathrm{\Δ}f{\log }_2(1+p_{j,k,u}^F{H}_{j,k,u}^F)\≥R_j,\∀j\∈C_n\end{array}$

$\begin{array}{cc}C3& p_{j,k,u}^F{\mathrm{\ρ}}_{j,k,e}^F\≤{\mathrm{\ξ}}_{k,e},\∀j\∈C_n,\∀u,e\∈D_j\end{array}$

Wherein,Represent FBS_jTransmitting power to user u in subchannel k,Represent The set of MUEs,It it is the gain in subchannel k Interference ratio, its at femto user u when subchannel is distributed it has been determined thatFor FBS_jChannel gain in subchannel k,WithIt is respectively FBS_iWith the channel gain of macro base station to femto user u,WithIt is respectively FBS_iAnd macro base station Transmitting power on channel k, σ²For noise power.

In constraints C1,For FBS_jAlways launch power, then C1 represents FBS_jTransmitting merit in all subchannels Rate sum is not more than FBS_jAlways launch power；In constraints C2, R_jFor FBS_jMinimum-rate demand, then C2 represents FBS_j Transfer rate and not less than its minimum-rate demand in all subchannels；In constraints C3, ξ_k,eFemto in expression bunch User u by the interference threshold of other femtos user,Represent femto base station FBS respectively_jTo grand user e in subchannel k On channel gain, then C3 represents FBS_jThe femto user u serviced is by FBS_jThe interference summation of other users serviced It is not more than the interference threshold of femto user u；In constraints C4, ξ_k,mFor femto user u by the interference door of grand user Limit,Represent femto base station FBS respectively_jTo grand user m in subchannel k channel gain, then C4 represents FBS_jInstitute The femto user u of service is disturbed, by grand user, the interference threshold that summation is not more than femto user u.

Preferably, described employing KKT condition FUEs is carried out power distribution farther include: according to FUEs power distribution Optimization object function and constraints, quote KKT condition and obtain:

$p_{j,k,u}^F={\[\frac{\mathrm{\Δ}f\left(1+\mathrm{\β}\right)}{\left(\mathrm{\α}+{\mathrm{\θ\ρ}}_{j,k,u}^F+{\mathrm{\ϵ\ρ}}_{j,k,m}^F\right)\ln 2}-\frac{1}{H_{j,k,u}^F}\]}^{+}$

Wherein, wherein, α, β, θ and ε are Lagrange multipliers, for constant； For water line.

The beneficial effects of the present invention is: the present invention is directed to the resource allocation problem in Femtocell double-layer network, protecting On the premise of demonstrate,proving grand user QoS, resource allocation algorithm based on sub-clustering is used to complete subchannel and the power of MUEs and FUEs Distribution, and restrained effectively cross-layer interference and with layer disturb, the availability of frequency spectrum can not only be improved, more can guarantee that FUEs and The QoS demand of MUEs.

Accompanying drawing explanation

Fig. 1 is resource based on sub-clustering distribution preferred embodiment schematic flow sheet in Femtocell double-layer network of the present invention；

Fig. 2 is MUEs resource allocation algorithm module exemplary plot in Femtocell double-layer network of the present invention；

Fig. 3 is the enforcement of FBSs clustering process in resource based on sub-clustering distribution in Femtocell double-layer network of the present invention Example flow chart；

Fig. 4 is that the present invention emulates comparison diagram with prior art MUEs outage probability；

Fig. 5 is the present invention and prior art MUEs average throughput comparison diagram；

Fig. 6 is that the present invention emulates comparison diagram with prior art Femtocell spectrum efficiency；

Fig. 7 is that the present invention emulates comparison diagram with prior art FUEs fairness；

Fig. 8 is that the present invention emulates comparison diagram with prior art FUEs satisfaction.

Detailed description of the invention

For making the object, technical solutions and advantages of the present invention express clearer, below in conjunction with the accompanying drawings and specifically The present invention is described in further details by case study on implementation.

Resource allocation methods preferred embodiment flow chart based on sub-clustering in Fig. 1 Femtocell of the present invention double-layer network, should Method comprises the following steps:

Step 101: utilize three-wheel subchannel allocation algorithm that grand user MUEs performs subchannel distribution；

Step 102: the object of planning distributed according to grand user power and constraints, using classical water-filling algorithm is MUEs Distribution power；

Step 103: use the Global Genetic Simulated Annealing Algorithm GASA improved that femto base station FBSs is performed sub-clustering；

Step 104: according to the rate requirement of femto user FUEs, uses heuritic approach that FUEs is performed subchannel and divides Join；

Step 105: utilize KKT condition that femto user FUEs is carried out power distribution.

Fig. 2 is MUEs resource allocation function module in Femtocell double-layer network of the present invention, including:

201, MUEs performing subchannel distribution, its concrete function is embodied as:

Assuming that MUEs is carried out subchannel distribution on the premise of power averaging distributes, the grandest user m is in subchannel k Signal to Interference plus Noise Ratio is:

Wherein,WithIt is respectively macro base station and femto base station j (FBS_j) transmitting power in subchannel k； WithRepresent respectively macro base station and FBSj to grand user m in subchannel k channel gain；Table Show the set of FBSs,Represent the set of subchannel, Represent MUEs set,σ²For noise power.

For meeting the data rate request of grand user m, provide a kind of subchannel distribution meeting grand user rate interval and calculate Method, according to formula (1), the more new formula quoting the shannon formula grand user's m data rate of modeling is:

$R_m=\underset{k=1}{\overset{K}{\Σ}}\underset{m=1}{\overset{M}{\Σ}}\mathrm{\Δ}f{\log }_2(1+{\mathrm{\γ}}_{m,k}^M)---\left(2\right)$

WhereinFor grand user m Signal to Interference plus Noise Ratio in subchannel k, Δ f is channel width.

The above-mentioned MUEs subchannel assignment problem be given can be solved by three-wheel subchannel allocation algorithm, and it implements Step includes:

201A: travel through all subchannels, distributes to the grand user that channel gain is maximum by every sub-channels, if certain grand use Family meets the request of its minimum data rate, i.e. R_m′≥R_m,min, then this grand user exits subchannel distribution, wherein R_m′Represent grand The current instant data rate of user m, R_m,minRepresent the minimum data rate request of grand user m；

201B: if subchannel has residue, then continue distribution, if certain grand user meets peak data rate request, i.e. R_m′≥R_m,max, then this grand user exits subchannel distribution, wherein R_m,maxRepresent the peak data rate request of grand user m；

201C: if subchannel still has residue, then proceed subchannel distribution, no longer carry out user data rate judgement.

202, MUEs power distribution, its concrete function is embodied as:

After obtaining MUEs sub-channel assignment result, for improving systematic function further, use classical water-filling algorithm to initially The power of mean allocation re-starts adjustment, and to maximize power system capacity as optimization aim, maximum general power is constraints, structure The power distribution object function building MUEs is:

$\begin{array}{cc}max& \underset{k=1}{\overset{K}{\Σ}}\mathrm{\Δ}f{\log }_2(1+p_k^M{H}_k^M)\end{array}---\left(3\right)$

$\begin{array}{cc}s.t.& \underset{k=1}{\overset{K}{\Σ}}{p}_k^M\≤P_{total}^M\end{array}---\left(4\right)$

Wherein,It is the gain interference ratio in subchannel k,For femto The set of base station,Represent macro base station to grand user m in subchannel k channel gain,Represent femto base station FBS_jTo grand user m in subchannel k channel gain；Represent femto base station FBS_jTransmitting merit in subchannel k Rate；For macro base station transmitting power in subchannel k, constraints (4) is macro base station transmitting merit in all subchannels Rate summation is not more than total transmitting power, i.e.WhereinFor the transmitting power that macro base station is total.

Above-mentioned MUEs power distribution problems is implemented as using classical water-filling algorithm: according to the distribution of MUEs power Optimization object function (3) and constraints (4), and utilize lagrange's method of multipliers build Lagrange's equation be:

$L(p,\mathrm{\ζ})=\underset{k=1}{\overset{K}{\mathrm{\Σ}}}\mathrm{\Δ}f{\log }_2(1+p_k^M{H}_k^M)-\mathrm{\ζ}(\underset{k=1}{\overset{K}{\mathrm{\Σ}}}{p}_k^M-P_{total}^M)---\left(5\right)$

Wherein, ζ is Lagrange multiplier, for constant, by above-mentioned Lagrange's equation (5) to launching powerSolve partially Lead, i.e.Then can get K equation, and it converted, then obtain following relational expression:

$p_k^M={\[\mathrm{\η}-\frac{1}{H_k^M}\]}^{+}---\left(6\right)$

Wherein,η is water line, and its value is η=Δ f/ ζ ln2；Therefore can quickly obtain Through-put power on every sub-channels.,For the set of femto base station FBSs total in system,

FUEs resource allocation algorithm in Femtocell double-layer network of the present invention, including:

301, FBSs being carried out sub-clustering, its concrete function is embodied as:

Definition non-directed graph G=[V, E, W], V is summit, represents F FBSs,E is for linking each summit Limit, W={w_ijThe weights of representative edge, wherein i, j ∈ 1 ..., F}, weights show the most greatly the interference between corresponding femto base station The biggest, and have w_ij=ρ_jip_i/ρ_jjp_j+ρ_ijp_j/ρ_iip_i.The target of sub-clustering optimization is according to w_ijSize will disturb little each other Femto base station FBS divide in same cluster, disturb each other big femto base station FBSs to divide and carry out establishing at different bunches.Profit By graph coloring principle, F FBSs is assigned to N_AIndividual bunch, χ=1 ..., N_A, and minimum with the interference summation between the FBSs in same bunch As object function, it is expressed as:

$\begin{array}{cc}min& \underset{i=1}{\overset{F}{\Σ}}\underset{j=1,j\≠i}{\overset{F}{\Σ}}\underset{n=1}{\overset{N_A}{\Σ}}{w}_{ij}{x}_{in}{x}_{jn}\end{array}---\left(7\right)$

Wherein, F and N_ARespectively represent femto base station FBSs quantity and bunch quantity；w_ijIt is FBS_iAnd FBS_jBetween dry Disturb weights；x_inIt is the sub-clustering oriental matrix of FBSs, works as x_inWhen=1, represent FBS_iAssign to n-th bunch, otherwise, x_in=0, i.e. table Show FBS_iDo not assign to n-th bunch；In constraints C1, C_nThe set of FBSs in representing n-th bunch,For total in system The set of FBSs,Then constraints C1 represents that distribution is to N_AIn individual bunch, the total quantity of FBSs is F；About Bundle condition C 2 represents that all FBSs in Femtocell double-layer network are assigned in different bunches, i.e. requires in each bunch Can not there is the FBSs of overlap；Constraints C3 represents x_inFor binary number, value is 0 or 1.

Based on the above-mentioned object of planning (7) and constraints (8), Global Genetic Simulated Annealing Algorithm is used to proceed from the situation as a whole dynamically Being grouped FBSs, until finding one to be preferably grouped scheme, it implements flow process as it is shown on figure 3, include:

103A: initialize: population at individual size sizepop, maximum evolution number of times MAXGEN, crossover probability P_c, variation is general Rate P_m, initial temperature T of annealing₀, temperature cooling ratio k, final temperature T_end；

103B: stochastic generation initial population Chrom, calculates fitness value f individual in population_i, wherein set f_iFor planning Target；

103C: implement population Chrom to select, intersect and the genetic manipulation such as variation, i.e. produce new FBS sub-clustering result； To new individual calculating its fitness value f produced_i', if f_i′<f_i, then with the old individuality of new individual replacement, otherwise, with probability exp ((f_i-f_i')/T) receive new individuality；

103D: if evolution number of times gen < MAXGEN, then gen=gen+1, go to step 103C, otherwise, go to step 103E；

103E: if T_i<T_end, then algorithm terminates, and returns globally optimal solution, otherwise, performs cooling operation T_i+1=kT_i, go to Step 103B.

302, FUEs performing subchannel distribution, its concrete function is embodied as:

Bringing seriously with layer interference to be prevented effectively from FBSs multipling channel in different bunches, the present invention is different bunches and is just distributing The subchannel handed over, with bunch in FBSs can plan as follows with the identical subchannel of multiplexing:

$\begin{array}{cc}max& \underset{n=1}{\overset{N_A}{\&Sigma;}}\underset{j\&Element;C_n}{\&Sigma;}\underset{u\&Element;D_j}{\&Sigma;}\underset{k=1}{\overset{K}{\&Sigma;}}\mathrm{\&Delta;}f{\log }_2(1+{\mathrm{\&gamma;}}_{j,k,u}^F)\end{array}---\left(9\right)$

$\begin{array}{ccc}s.t.& C1& \underset{n=1}{\overset{N_A}{\&Sigma;}}{a}_{k,n}=1,k\&Element;\{1,...,K\}\end{array}---\left(10\right)$

$\begin{array}{cc}C2& \underset{k=1}{\overset{K}{\&Sigma;}}\mathrm{\&Delta;}f{\log }_2(1+{\mathrm{\&gamma;}}_{j,k,u}^F)a_{k,n}\&GreaterEqual;R_{j,u}^F\end{array}---\left(11\right)$

Wherein, N_AThe quantity represented respectively with K bunch and the sum of subchannel, C_nThe set of FBSs, D in representing n-th bunch_j Represent FBS_jThe FUEs set of service,Represent FBS_jThe femto user FUE u of service letter in subchannel k is dry makes an uproar Ratio；a_k,nRepresent whether subchannel k distributes to a bunch C_n, work as a_k,n=1, subchannel k distributes to a bunch C_n；Otherwise, a_k,n=0, i.e. represent Subchannel k is not assigned to a bunch C_n, then constraints (10) shows that all of subchannel must and be only capable of selecting one group to be allocated；For FBS_jThe rate requirement of the femto user FUE u of service, i.e. constraints (11) shows to distribute to femto user The data rate of FUE u meets himself rate requirement.

According to the object of planning (9) and constraints (10) (11), didactic channel allocation algorithm is used to solve above-mentioned asking Topic, it is as follows that it implements process:

1) input: the data-rate requirements of FUEs isFBS_jThe femto user FUE u of service is in subchannel k Signal to Interference plus Noise Ratio is

2) mean data rate of each bunch is calculated:C_n| it is the individual of FUEs in n-th bunch Number；

3) each bunch of subchannel number needed is determined:

4) calculating the SINR that each available subchannels is at each bunch successively, such as, subchannel k at the SINR of n-th bunch is:

5) judge that subchannel k is maximum, if this bunch is not allocated to enough subchannels, then by son at the SINR of which bunch Channel k distributes to this bunch；

6) updateWith each bunch of allocated number of subchannels.Repeat step 4) 5) until all of son Channel has been allocated.

303, FUEs performing power distribution, its concrete function is embodied as:

After the subchannel of FUEs is assigned, KKT condition is utilized the power of initial mean allocation to be readjusted, with maximum The power system capacity changing all bunches is optimization aim, models as follows:

$\begin{array}{cc}max& \underset{n=1}{\overset{N_A}{\&Sigma;}}\underset{j\&Element;C_n}{\&Sigma;}\underset{u\&Element;D_j}{\&Sigma;}\underset{k=1}{\overset{K}{\&Sigma;}}\mathrm{\&Delta;}f{\log }_2(1+p_{j,k,u}^F{H}_{j,k,u}^F)\end{array}---\left(12\right)$

$\begin{array}{ccc}s.t.& C1& \underset{k=1}{\overset{K}{\&Sigma;}}\underset{u\&Element;D_j}{\&Sigma;}{p}_{j,k,u}^F\&le;P_{total}^F,\&ForAll;j\&Element;C_n\end{array}$

$\begin{array}{cc}C2& \underset{k=1}{\overset{K}{\&Sigma;}}\underset{u\&Element;D_j}{\&Sigma;}\mathrm{\&Delta;}f{\log }_2(1+p_{j,k,u}^F{H}_{j,k,u}^F)\&GreaterEqual;R_j,\&ForAll;j\&Element;C_n\end{array}---\left(13\right)$

$\begin{array}{cc}C3& p_{j,k,u}^F{\mathrm{\&rho;}}_{j,k,e}^F\&le;{\mathrm{\&xi;}}_{k,e},\&ForAll;j\&Element;C_n,\&ForAll;u,e\&Element;D_j\end{array}$

Wherein,It is the gain interference ratio in subchannel k, should Value has determined that when femto user u performs subchannel distribution；For FBS_jGeneral power limits, i.e. constraints C1 shows FBS_jThrough-put power in all subchannels is not more than its general power；R_jFor FBS_jMinimum-rate demand, i.e. constraints C2 table Bright FBS_jThe data rate obtained on all channels to meet its minimum-rate demand；ξ_k,uFor bunch in femto user u by it The interference threshold of he femto user,Represent femto base station FBS respectively_jTo grand user e in subchannel k channel Gain, then constraints C3 shows that femto user u is met its interference threshold by the interference value of other femtos user；ξ_k,mFor Femto user u by the interference threshold of grand user,Represent femto base station FBS respectively_jTo grand user m in subchannel k Channel gain, then constraints C4 shows that the interference value of grand user that femto user u is subject to is not more than its threshold value.

According to the object of planning (12) and constraints (13), KKT condition obtain:

$\begin{array}{c}L\left(p,\mathrm{\&alpha;},\mathrm{\&beta;},\mathrm{\&theta;},\mathrm{\&epsiv;}\right)=\underset{n=1}{\overset{N_A}{\mathrm{\&Sigma;}}}\underset{j\&Element;C_n}{\mathrm{\&Sigma;}}\underset{u\&Element;D_j}{\mathrm{\&Sigma;}}\underset{k=1}{\overset{K}{\mathrm{\&Sigma;}}}\mathrm{\&Delta;}f{\log }_2\left(1+p_{j,k,u}^F{H}_{j,k,u}^F\right)-\mathrm{\&alpha;}\left(\underset{k=1}{\overset{K}{\mathrm{\&Sigma;}}}\underset{u\&Element;D_j}{\mathrm{\&Sigma;}}{p}_{j,k,u}^F-P_{total}^F\right)-\\ \mathrm{\&beta;}\left(R_j-\underset{k=1}{\overset{K}{\mathrm{\&Sigma;}}}\underset{u\&Element;D_j}{\mathrm{\&Sigma;}}\mathrm{\&Delta;}f{\log }_2\left(1+p_{j,k,u}^F{H}_{j,k,u}^F\right)\right)-\mathrm{\&theta;}\left(p_{j,k,u}^F{\mathrm{\&rho;}}_{j,k,e}^F-{\mathrm{\&xi;}}_{k,e}\right)-\\ \mathrm{\&epsiv;}\left(p_{j,k,u}^F{\mathrm{\&rho;}}_{j,k,m}^F-{\mathrm{\&xi;}}_{k,m}\right)\end{array}---\left(14\right)$

Wherein, α, β, θ and ε are Lagrange multipliers, for constant, by above-mentioned KKT equation (14) to launching powerAsk Solve local derviation, i.e.Can obtain:Its In, For water line.

For explanation beneficial effects of the present invention, the present invention use channel model mainly consider path loss, wall penetration loss, Shadow fading and antenna gain, design parameter emulates according to table 1.

Table 1 simulation parameter

In emulation, all of FBSs is all operated in closed mode, the most only allows authorized user to access.The present invention analyzes institute Carry the multinomial performance of algorithm, including between the outage probability of MUEs, MUEs average throughput, the spectrum efficiency of Femtocell, FUEs Fairness, the satisfaction of FUEs.

Fig. 4 shows the outage probability of the MUEs in different chamber under MUEs ratio.In emulation experiment, interference threshold is set Value is-6dB, if actual Signal to Interference plus Noise Ratio is less than-6dB, then it is assumed that MUE interrupts.From fig. 4, it can be seen that ungrouped RRA The MUE outage probability that algorithm obtains constantly can rise along with the increase of indoor MUEs ratio, until close to 100%；But this Invent the MUE outage probability that carried algorithm obtains and be always held at less than 10%.Therefore, the carried algorithm of the present invention effectively reduces The FBSs interference to MUEs so that MUEs disclosure satisfy that its minimum SINR demand, i.e. meets the QoS demand of MUEs.

Fig. 5 describes averagely gulping down of the indoor MUEs that obtained under the deployment density of different Femto cells by various algorithms The amount of telling.Wherein can obtain maximum average throughput in max carrier to interference theory of algorithm, but this algorithm does not accounts between MUEs Fairness, the MUEs that may result in bad channel quality may distribute less than channel.The carried algorithm of the present invention is in FBSs sub-clustering On the basis of for FUEs distribution subchannel and power, taken into account user fairness so that the MUEs of bad channel quality also can obtain simultaneously Obtain preferable communication quality, significantly reduce the FBSs interference to MUEs, and then improve systematic function.

Fig. 6 shows the frequency spectrum service efficiency of the Femtocell under different FBSs density conditions.GASA is dynamically to FBSs Carry out sub-clustering, it is possible to the same layer interference in elimination system effectively so that spectrum efficiency is substantially improved.Orthogonal grouping algorithm in group Relatively herein carried algorithm spectrum efficiency is relatively low, its reason be this algorithm sub-clustering after in each bunch FBSs number differ greatly, But the frequency band size of each bunch of distribution is identical, and this has resulted in availability of frequency spectrum reduction.But, the present invention proposes GASA-HK algorithm is to have carried out subchannel distribution and power distribution on the basis of GASA algorithm, and it is performing son letter to FUEs Road distribution and power can reduce interference on the basis of FBSs sub-clustering further when distributing, improve the Signal to Interference plus Noise Ratio of FBSs, i.e. Improve the systematic function of FBSs.

Fig. 7 describes the fairness between FUEs.Along with Femto cell density improves, RRA algorithm fairness is higher than other Contrast algorithm, but in the case of making community dense distribution due to its randomness, FUEs is disturbed by bigger same layer, so its Between femto user, fairness is relatively low；Between group, in orthogonal grouping algorithm, the FBSs number of each bunch is unbalanced, causes in different bunches FUEs experienced interference difference bigger；The FUEs fairness that GASA algorithm obtains is substantially better than other algorithms.But, institute herein Carry GASA-HK algorithm and compare GASA algorithm, after sub-clustering, based on Max-min fairness, system is carried out channel distribution And power distribution, promote the power of the relatively low subchannel of Signal to Interference plus Noise Ratio, reduce the power of the too high subchannel of Signal to Interference plus Noise Ratio simultaneously, The fairness making FUEs is further promoted.

Fig. 8 describes the satisfaction of FUEs.The carried cluster algorithm of the present invention is an iteration searching process, calculates in conjunction with heredity Method and the advantage of simulated annealing, the FBSs number can being adaptively adjusted in each bunch according to FBSs deployment density, point Bunch performance improves constantly, it is possible to preferably eliminating interference, and compare other algorithms, the carried algorithm of the present invention can make the satisfaction of FUEs Degree is maintained at a higher level.Further, GASA-HK algorithm carries out power adjustment on the basis of GASA algorithm so that more Many FUEs can meet rate requirement.

Claims (8)

1. one kind is used for Femtocell double-layer network resource allocation methods based on sub-clustering, it is characterised in that include following step Rapid:

Step 101: utilize three-wheel subchannel allocation algorithm that grand user MUEs performs subchannel distribution；

Step 102: the object of planning distributed according to grand user power and constraints, using classical water-filling algorithm is MUEs distribution Power；

Step 103: using the Global Genetic Simulated Annealing Algorithm GASA improved is Femto cell sub-clustering；

Step 104: according to the rate requirement of femto user FUEs, using heuritic approach is that FUEs distributes subchannel；

Step 105: utilize Caro to need-Ku En-Plutarch KKT condition that femto user FUEs is carried out power distribution.

Resource allocation methods based on sub-clustering the most according to claim 1, it is characterised in that described step 101 utilizes three Wheel channel allocation algorithm performs subchannel distribution and includes grand user MUEs: quotes shannon formula and models grand user's m data speed The more new formula of rate isWherein, M is grand total number of users, and K is subchannel sum, For grand user m Signal to Interference plus Noise Ratio in subchannel k, Δ f is channel width；And then consider the data rate request of grand user, Subchannel is distributed for grand user on the premise of meeting grand user rate interval.

Resource allocation methods based on sub-clustering the most according to claim 2, it is characterised in that described meeting grand user speed Rate distributes subchannel for grand user on the premise of interval, including:

Step 101A: travel through all subchannels, finds out subchannel k making grand user m can obtain maximum channel gain, and by sub-letter Road k distributes to grand user m, if the speed of the grand user m obtained meets its minimum speed limit demand, the grandest user m no longer participates in letter Road distribute, and if then all calculated grand user data rates be satisfied by minimum speed limit demand, then exit circulation；

Step 101B: if subchannel has residue, then repeat step 101A, and grand user rate Rule of judgment becomes judging Whether the speed of calculated corresponding grand user meets its flank speed demand；

Step 101C: if subchannel still has residue, repeat step 101A, no longer carries out grand user data rate judgement.

Resource allocation methods based on sub-clustering the most according to claim 1, its feature exists, and described step 102 is according to grand use The object of planning of family power distribution and constraints, using classical water-filling algorithm is that MUEs distribution power includes:

With maximization power system capacity as optimization aim, maximum general power is constraints, builds the power distribution target letter of MUEs Number:And meet constraints:

Using water-filling algorithm is that grand user distributes power, obtains

Wherein,η=Δ f/ ζ ln2 is water line；It is Gain interference ratio in subchannel k,Represent macro base station to grand user m in subchannel k channel gain,Table Show femto base station FBS_jTo grand user m in subchannel k channel gain；Represent femto base station FBS_jIn subchannel Transmitting power on k；σ²For noise power；ζ is Lagrange multiplier, for constant；For macro base station sending out in subchannel k Penetrate power,For total transmitting power, Δ f is channel width, and M is grand total number of users, and K is subchannel sum.

Resource allocation methods based on sub-clustering the most according to claim 1, it is characterised in that described step 103 uses and changes The Global Genetic Simulated Annealing Algorithm GASA entered is that Femto cell sub-clustering includes: with the interference between the femto base station FBSs in same bunch Summation is minimum as object function, modeling optimization equation:

And meet constraints:

C_g∩C_n=Φ (g, n ∈ χ, g ≠ n) and x_in∈{0,1}；Wherein, χ=1 ..., N_ARepresent bunch set, F and N_APoint Not Biao Shi femto base station FBSs quantity and bunch quantity；w_ijIt is FBS_iAnd FBS_jBetween interference weights；x_inIt it is dividing of FBSs Bunch oriental matrix, works as x_inWhen=1, represent FBS_iAssign to n-th bunch, work as x_inWhen=0, i.e. represent FBS_iDo not assign to n-th bunch； C_nThe set of FBSs, C in representing n-th bunch_gRepresent the set of FBSs in g bunch,For the set of FBSs total in system,And then use Global Genetic Simulated Annealing Algorithm to solve this sub-clustering problem.

Resource allocation methods based on sub-clustering the most according to claim 5, it is characterised in that described step 104 is according to milli The rate requirement of user FUEs slightly, using heuritic approach is that FUEs distribution subchannel includes: to maximize the data of FUEs Speed is optimization aim:And meet constraintsIts In, D_jFor FBS_jThe FUEs set of service；Δ f is channel width；For FBS_jThe femto user u of service is in subchannel k Signal to Interference plus Noise Ratio；For FBS_jThe rate requirement of the femto user u of service；If a_k,n=1, represent that subchannel k is distributed to bunch C_n, otherwise, a_k,n=0；K is subchannel sum.

Resource allocation methods based on sub-clustering the most according to claim 6, it is characterised in that described step 105 utilizes KKT bar Part carries out power distribution and includes femto user FUEs: set up function model for optimization aim maximizing handling capacity:

And

Wherein,Represent FBS_jTransmitting power to user u in subchannel k,Represent MUEs Set, It is the gain interference ratio in subchannel k, Its at femto user u when subchannel is distributed it has been determined thatFor FBS_jChannel gain in subchannel k,With It is respectively FBS_iWith the channel gain of macro base station to femto user u,WithIt is respectively FBS_iWith macro base station on channel k Transmitting power, σ²For noise power；

In constraints C1,For FBS_jAlways launch power, then C1 represents FBS_jTransmitting power in all subchannels it No more than FBS_jAlways launch power；In constraints C2, R_jFor FBS_jMinimum-rate demand, then C2 represents FBS_jInstitute There is in subchannel transfer rate and not less than its minimum-rate demand；In constraints C3, ξ_k,eFemto user in expression bunch U by the interference threshold of other femtos user,Represent femto base station FBS respectively_jTo grand user e in subchannel k Channel gain, then C3 represents FBS_jThe femto user u serviced is by FBS_jThe interference summation of other users serviced is little Interference threshold in femto user u；In constraints C4, ξ_k,mFor femto user u by the interference threshold of grand user,Represent femto base station FBS respectively_jTo grand user m in subchannel k channel gain, then C4 represents FBS_jServiced Femto user u disturbed, by grand user, the interference threshold that summation is not more than femto user u.

Resource allocation methods based on sub-clustering the most according to claim 7, it is characterised in that described employing KKT condition pair FUEs carries out power distribution and farther includes: according to optimization object function and the constraints of the distribution of FUEs power, quote KKT bar Part obtains:

Wherein,For water line, α, β, θ and ε are Lagrange multipliers, for Constant；