CN106028345A - Small base station capacity and coverage optimization method based on adaptive tabu search - Google Patents

Abstract

The invention relates to a small base station capacity and coverage optimization method based on adaptive tabu search, and belongs to the technical field of wireless communication. The method comprises the steps of: according to initial power of a small base station, calculating an initial value of a capacity and coverage target optimization function; carrying out iteration optimization by adopting a tabu search algorithm, which comprises the steps of calculating a solution vector in a neighbourhood of a current solution, which enables the target optimization function to be optimal, recording as the current solution and adding the current solution into a tabu list if the solution vector enables the value of the target optimization function to be optimal currently or not to be optimal so far, but to be in a non-tabu state, or returning to calculate an optimal solution vector in the neighbourhood after removing the solution vector from the neighbourhood of the current solution, and adaptively updating a tabu length; and after completing iteration optimization, updating transmission power of the small base station, and outputting a current value of the capacity and coverage target optimization function of a small base station system. Capacity and coverage optimization of the small base station system can be implemented, complexity of optimization iteration computation is reduced, optimization and improvement of integral performance of the network are achieved, and global optimization can be achieved.

Description

A kind of little Base Station Compatibility based on self adaptation TABU search and coverage optimization method

Technical field

The invention belongs to wireless communication technology field, particularly to a kind of little Base Station Compatibility based on self adaptation TABU search With coverage optimization method.

Background technology

At present on the one hand, the mobile data traffic business of explosive increase, to mobile communications network bandwidth and capacity and cover Lid is proposed higher requirement, and this makes traditional macro base station coverage mode be difficult to meet user's request now.The opposing party Face, in third generation partner program (3GPP) LTE system, owing to mass data business occurs in indoor, indoor user Data volume is big and skewness, causes that indoor signal is unsatisfactory, local congestion's situation occurs in network.Therefore dispose short distance, The little base station of low-power consumption becomes a kind of effective solution, and little base station is for macro base station, the fundamental form of little base station Formula has multiple, including Home eNodeB, micro-base station, scytoblastema station etc., is deployed under different scene.Little base station system is heterogeneous network, Including macro base station and little base station, its system structure, as it is shown in figure 1, macro base station transmitting power is higher, is deployed in outdoor；Little base station is sent out Penetrate power relatively low, be deployed in indoor or outdoors；One or more little base station bunch, little base station can be had in each macro base station coverage Several little base stations are assembled in bunch；Communicated by back haul link between macro base station with little base station, between the little base station that little base station bunch is interior Communicated by back haul link.Little base station can effectively strengthen outdoor cover, capacity boost, improves in-door covering, and then improves complete Network capacity amount, especially for indoor and outdoor high density business demand region, the deployment of little base station can be effectively improved Consumer's Experience, carry For more efficient service.

Along with the flattening of network structure develops and the introducing of little base station, mobile network more becomes to complicating.Operator in order to Improve network performance, preferably manage network, simplify network design, reduce network operation cost, by the concept of self-organizing network Introduce LTE system so that network can realize self-configuring and self-optimizing.The self-configuring of network and self-optimizing can substitute manually joins Put and optimize operation, reducing the expense of configuration and network management further, and the performance effectively realizing the little base station network of LTE is excellent Change.

Aspect of network self-configuration purpose is to realize automatically configuring of newly deployed little base station so that base station can find automatically With set up neighboring BS relationship, it is possible to automatically configure the physical identification number (ID) of each community, reduce the network planning and networking Participation artificial in management, reduces the maintenance cost of network management, makes the network after configuration disclosure satisfy that primary demand, it is achieved The functional requirement of plug and play.And the purpose of self-optimizing is to promote network performance further, by the monitoring network operation The change of the performance indications in journey and the generation of event of failure, automatically select some optimized algorithm to adjust the ginseng of the network equipment Number, to reach optimum system capability and service area, it is achieved the optimization of overall performance of network promotes.

Capacity and coverage self-optimization belong to of network self-organization, are the optimization problems of the many optimization aim of multiple constraint, by Mutually restricting between capacity and covering, therefore this optimization problem is MPS process and two optimized parameters compromise equilibriums of capacity Process, is combined optimization problem.This problem can be adaptively adjusted respective radio-frequency parameter by certain algorithm, as launched merit Rate, antenna azimuth etc., the optimization reaching capacity and covering problem promotes.This optimization problem should comprise the inspection covering leak Equilibrium etc. between survey, coverage optimization, covering and capacity.The algorithm research that there is now is many based on relatively simple topological network model Assume, preferable channel model assumes and unification adjusts radio frequency parameter so that the application in actual self-optimizing system is relatively For limited.

TABU search (TS) is applied to solve combinatorial optimization problem more, especially when problem dimension higher data amount is bigger, TABU search can be good at reducing the complexity solving optimal solution.It is a kind of heuristic search algorithm, and core concept is mark Remember that more acquired locally optimal solutions avoid (but and not exclusively taboo) these locally optimal solutions as far as possible, and by despising Some are specially pardoned by criterion by avoiding solution, avoid being absorbed in locally optimal solution, thus obtain with diversified search approach Global optimization is searched for.Self adaptation TABU search relates to neighborhood, width neighborhood, taboo list, taboo object, Tabu Length, despises standard The concept such as then；Neighborhood refers to currently solve the value set of certain limit interior (width of this scope is referred to as width neighborhood) around, Neighborhood solves to current and width neighborhood is relevant, and is as currently solving change in search procedure；Taboo list is a kind of square The table of formation formula, according to this matrix ranks label can unique index corresponding may solution, matrix element value represent, if before Certain avoided in dry search procedure solves (i.e. avoiding object) and also needs to be in the number of times of taboo state (element value is 0 generation Table non-taboo state, element span is 0 to Tabu Length, for positive integer), taboo list can be selected for dynamic or fixing , dynamically mean that Tabu Length is variable in search procedure, fix and mean that Tabu Length is always in search procedure The initial value set；Certain solution is in taboo state (taboo list corresponding element value is more than 0) then for taboo object, it is meant that at it The most some (taboo list corresponding element value) secondary search will be avoided (in the case of not meeting aspiration criterion)；Tabu Length When (for positive integer, desirable representative value is 2,5,10 etc.) refers to that avoiding object is deposited in taboo list first, corresponding taboo list cell The value of element, represents and also needs to be in the number of times (the most also needing the iterations leaving in taboo list) of taboo state, afterwards every Secondary iteration, the taboo list element value that each taboo object is corresponding subtracts one, until this taboo object corresponding element value is 0 (i.e. to avoid Object exempts taboo state), it is adaptive change that self adaptation TABU search refers to Tabu Length during chess game optimization Situation；If the target function value that aspiration criterion refers to taboo object corresponding is better than current optimal solution, then ignores it and be in taboo The attribute of state and still adopt it and solve for current.

Existing capacity and coverage optimization algorithm include: particle cluster algorithm, and a kind of particle follows optimal particle in solution space The optimized algorithm of search, solves simple convergence fast, but solves for dispersed problem the best, be easily trapped into local optimum and be difficult to reach Global optimum；Simulated annealing, a kind of mechanism based on probability heuristic search, gradually finely tune parameter to be regulated, can be with Certain probability accepts than the worse solution of current solution to avoid being absorbed in local optimum, is finally reached capacity and the coverage optimization of the overall situation, But its convergence rate is slower.

Summary of the invention

It is an object of the invention to for the combined optimization problem of capacity and covering in little base station system, propose a kind of based on certainly Adapting to little Base Station Compatibility and the coverage optimization method of TABU search, the associating that TABU search is introduced capacity and covering by this method is excellent In change, the self-optimizing of little station system capacity and covering can be realized, reduce the computation complexity of Optimized Iterative, it is achieved network is overall The optimization of performance promotes, and can reach global optimum.

A kind of based on self adaptation TABU search the little Base Station Compatibility of present invention proposition and coverage optimization method, its feature exists In, the method comprises the following steps:

1) initial value of capacity and the objective optimization function of covering it is calculated according to little base station Initial Trans, specifically Calculation procedure is as follows:

If little base station system comprises K macro base station l, M little base station j and the user terminal UE of N number of random distribution_i, wherein, L=1,2 ..., l ..., K, j=1,2 ... j ..., M, i=1,2 ..., i ..., N；The transmitting power of little base station j is p_j, little base Stand j and user terminal UE_iBetween transmission channel gain be g_ij, then it is transferred to user terminal UE from little base station j_iReference signal Receive power P_rx(i, the such as formula of expression formula j) (1):

P_rx(i, j)=p_jg_ij (1)

If system noise is σ², the least base station j is via downlink to user terminal UE_iSignal to Interference plus Noise Ratio SINR_iExpression Formula such as formula (2):

${\mathrm{SINR}}_i=\frac{g_{ij}{p}_j}{{\mathrm{\σ}}^2+\underset{k\≠j}{\Σ}{g}_{ik}{p}_k}---\left(2\right)$

Wherein g_ikFor little base station k and user terminal UE_iBetween transmission channel gain, p_kFor the transmitting power of little base station k, k≠j；

If each pilot power occupies the part that total equalization launching power is fixed, and each channel also takes up always The equal proportion part of power, then user terminal UE_iNormalized throughput t_iSuch as expression formula (3):

$t_i={\log }_2(1+{\mathrm{SINR}}_i)={\log }_2(1+\frac{g_{ij}{p}_j}{{\mathrm{\σ}}^2+\underset{k\≠j}{\Σ}{g}_{ik}{p}_k})---\left(3\right),$

The normalization total throughout T of the least base station j_jSuch as expression formula (4):

$T_j=\underset{{\mathrm{UE}}_i\∈U_j}{\Σ}{\log }_2(1+\frac{g_{ij}{p}_j}{{\mathrm{\σ}}^2+\underset{k\≠j}{\Σ}{g}_{ik}{p}_k})---\left(4\right),$

Wherein, U_jRepresent all user terminals of little base station j service, if the evaluation index of capacity is user in whole network The average throughput of terminal, the expression of its formula is such as formula (5):

$\frac{1}{N}\underset{j=1}{\overset{M}{\Σ}}{T}_j=\frac{1}{N}\underset{j=1}{\overset{M}{\Σ}}\underset{{\mathrm{UE}}_i\∈U_j}{\Σ}{\log }_2\left(1+\frac{g_{ij}{p}_j}{{\mathrm{\σ}}^2+\underset{k\≠j}{\Σ}{g}_{ik}{p}_k}\right)---\left(5\right);$

Represent covering with the minimum p% of the cumulative distribution of user throughput in each macro base station coverage cell, use T_{L, p%}Representing the minimum p% of user throughput cumulative distribution in macro base station l coverage cell, introducing compromise coefficient gamma balances to be covered Lid and two performance indications of capacity, 0 ＜ γ ＜ 1, then M little base station transmitting power is p_j' time, p_j'={ p₁,p₂,..., p_j,...,p_M, network system capacity and the objective optimization function F (p of covering_j') define such as formula (6):

$F\left({p_j}^{\′}\right)=\mathrm{\γ}\frac{1}{\frac{K}{N}\underset{j=1}{\overset{M}{\Σ}}{T}_j}+(1-\mathrm{\γ})\frac{1}{\underset{l=1}{\overset{K}{\Σ}}{T}_{l,p\%}}---\left(6\right);$

Obtain the objective optimization function F (p of capacity and covering_j') initial value；

2) self adaptation tabu search algorithm is used to generate new little base station transmitting power p_j', and calculate objective optimization function New value F (p_j'), specifically comprise the following steps that

2-a) initialize the parameter of self adaptation TABU search: setting taboo list as sky, set Tabu Length as P, P is the most whole Number, width neighborhood, the span of little base station power, maximum iteration time m_iter of search；Make m=1, m=1,2 ..., M ..., m_iter, when the m time iteration starts, make l=1, if the transmitting power of medium and small base station, macro base station l coverage cell is X^{(m ,l)}, it is designated as currently solving；

2-b) calculating is currently solving X^(m,l)Contiguous range in so as to get objective optimization function F (p_j') optimum solution to Amount；And judge whether this solution vector makes objective optimization functional value F (p_j') it is optimum so far, if optimum, perform step 2-e), step 2-c is otherwise performed)；

2-c) judge whether this solution vector is in taboo state, the most then this solution vector is arranged from the current neighborhood solved Remove, and perform step 2-d), otherwise perform step 2-e)；

2-d) judge currently to solve X^(m,l)Neighborhood in the most still have the solution vector being in non-taboo state, if still having, then return Receipt row step 2-b)；Otherwise, the neighborhood that will currently solve is marked as the solution vector of taboo state at first change and be set to non-taboo State, and return execution step 2-b)；

2-e) it is designated as currently solving by this solution vector, and by p_jThe corresponding unit representing medium and small base station, macro base station l coverage cell in ' Element value is updated to X^(m,l)Middle respective value；Update the taboo state of each corresponding element of taboo list, i.e. make objective optimization function F (p_j') optimum solution vector is set to P in the corresponding element value of taboo list, remaining has been in the solution vector of taboo state at taboo list Corresponding element value subtract one；

2-f) judge whether current solution repeats: concrete determination methods is, current goal majorized function value F (p_j') therewith The value of the front objective optimization function of iteration every time compares, if existing equal, is then judged as repeating, performs step 2-g)；If no Exist equal, then current solution did not duplicate, and performed step 2-h)；

2-g) update Tabu Length: the iterations that the iteration residing for record current iteration and a upper repetition values is spaced, It is designated as y_r, r=1,2,3 ..., calculate y_rMeansigma methods be y_rep, update Tabu Length simultaneously, for α ＞ 0, α be P and between y_repScale factor, it is judged that whether the P after renewal is more than α y_repIf, P ＞ α y_rep, then P=max (1, P-1) is made；If P≤α is y_rep, Then make P=P+1；

2-h) judging whether to travel through all K macro base stations, if traveling through, performing step 2-j), it is not fully complete traversal and then performs Step 2-i)；

2-i) make l=l+1, the transmitting power of Nei little base station, macro base station coverage cell is designated as X^(m,l), it is designated as current simultaneously Solve, return and perform step 2-b)；

2-j) judge that m, whether less than m_iter, if being less than, then makes m=m+1, and circulates execution step 2)；Otherwise terminate to follow Ring, performs step 3)；

3), after completing m_iter iteration optimization, little base station transmitting power has been updated to p_j', export the least base station system Capacity and coverage goal majorized function value F (p_j')；Terminate little station system capacity and coverage optimization.

The technical characterstic of the present invention and beneficial effect:

The capacity that self adaptation tabu search algorithm is incorporated into little base station system is creatively asked by the present invention with coverage optimization In topic, and being one group of variable with little base station in each macro base station coverage, the little base station parameter of distributed adjustment, it optimizes Complexity is low, has higher iterative convergence speed compared to existing method；It addition, the self-optimizing algorithm of this capacity and covering can be jumped Going out the predicament of local optimum, preferably approach globally optimal solution, degree of optimization is higher compared to existing method, and by adaptive Should regulate during the method for Tabu Length makes to optimize and can reduce computation complexity further, be good at according to practical situation excellent Change process carries out accommodation.Little Base Station Compatibility based on self adaptation tabu search algorithm and coverage optimization method, can realize Little station system capacity and the self-optimizing of covering, it is achieved the combined optimization target of the network coverage and capacity, it is achieved network globality The optimization of energy promotes.

Accompanying drawing explanation

Fig. 1 is that existing little base station forms structural representation.

Fig. 2 is the method flow block diagram of the present invention.

Detailed description of the invention

A kind of based on self adaptation TABU search the little Base Station Compatibility that the present invention proposes is combined accompanying drawing with coverage optimization method And embodiment is further described below:

A kind of based on self adaptation TABU search the little Base Station Compatibility of present invention proposition and coverage optimization method, be based on little The power parameter of base station adjusts, and in this little base station system, macro base station is disposed with little base station alien frequencies, little base station the user serviced By not by the signal disturbing of macro base station, only by the signal disturbing with the little base station of frequency, the method is as in figure 2 it is shown, include following step Rapid:

1) according to little base station Initial Trans, calculate the initial value of capacity and the objective optimization function of covering, specifically count Calculation step is as follows:

If this little base station system comprises K macro base station l, M little base station j and the user terminal UE of N number of random distribution_i, its In, l=1,2 ..., l ..., K, j=1,2 ... j ..., M, i=1,2 ..., i ..., N；The transmitting power of little base station j is p_j, little Base station j and user terminal UE_iBetween transmission channel gain be g_ij, then it is transferred to user terminal UE from little base station j_iReference letter Number receive power P_rx(i, j) expresses by formula (1):

P_rx(i, j)=p_jg_ij (1)

If system noise is σ², the least base station j is via downlink to user terminal UE_iSignal to Interference plus Noise Ratio SINR_iExpression formula Such as formula (2):

${\mathrm{SINR}}_i=\frac{g_{ij}{p}_j}{{\mathrm{\σ}}^2+\underset{k\≠j}{\Σ}{g}_{ik}{p}_k}---\left(2\right)$

Wherein g_ikFor little base station k and user terminal UE_iBetween transmission channel gain, p_kFor the transmitting power of little base station k, k≠j；

If each pilot power occupies the part that total equalization launching power is fixed, and each channel also takes up always The equal proportion part of power, then user terminal UE_iNormalized throughput t_iIt is represented by formula (3):

$t_i={\log }_2(1+{\mathrm{SINR}}_i)={\log }_2(1+\frac{g_{ij}{p}_j}{{\mathrm{\σ}}^2+\underset{k\≠j}{\Σ}{g}_{ik}{p}_k})---\left(3\right),$

So the normalization total throughout T of little base station j_jFor expression formula (4):

$T_j=\underset{{\mathrm{UE}}_i\∈U_j}{\Σ}{\log }_2(1+\frac{g_{ij}{p}_j}{{\mathrm{\σ}}^2+\underset{k\≠j}{\Σ}{g}_{ik}{p}_k})---\left(4\right),$

Wherein, U_jRepresent all user terminals of little base station j service, if the evaluation index of capacity is user in whole network The average throughput of terminal, the expression of its formula is such as formula (5):

$\frac{1}{N}\underset{j=1}{\overset{M}{\Σ}}{T}_j=\frac{1}{N}\underset{j=1}{\overset{M}{\Σ}}\underset{{\mathrm{UE}}_i\∈U_j}{\Σ}{\log }_2\left(1+\frac{g_{ij}{p}_j}{{\mathrm{\σ}}^2+\underset{k\≠j}{\Σ}{g}_{ik}{p}_k}\right)---\left(5\right);$

With in each macro base station coverage cell the cumulative distribution of user throughput minimum p% (for 5,10 etc., value model Enclosing 0～100, representative value desirable for p is 5) represent covering (being i.e. used for representing edge throughput), use T_{L, p%}Represent macro base station l The minimum p% of user throughput cumulative distribution in coverage cell, introduces compromise coefficient gamma and balances covering and capacity two individual character Energy index, 0 ＜ γ ＜ 1, therefore M little base station transmitting power is p_j' time, p_j'={ p₁,p₂,...,p_j,...,p_M, network system System capacity and the objective optimization function F (p of covering_j') define such as formula (6):

$F\left({p_j}^{\′}\right)=\mathrm{\γ}\frac{1}{\frac{K}{N}\underset{j=1}{\overset{M}{\Σ}}{T}_j}+(1-\mathrm{\γ})\frac{1}{\underset{l=1}{\overset{K}{\Σ}}{T}_{l,p\%}}---\left(6\right);$

Obtain the objective optimization function F (p of capacity and covering_j') initial value；

2) self adaptation tabu search algorithm is used to generate new little base station transmitting power p_j', and calculate objective optimization function New value F (p_j'), specifically comprise the following steps that

2-a) initialize the parameter of self adaptation TABU search: sets taboo list for sky, and setting Tabu Length (it is designated as P, typical Value is 2,5,10 etc., for positive integer), width neighborhood (desirable 1,2,4 ..., for positive integer), the span of little base station power The ginsengs such as maximum iteration time m_iter (desirable representative value is 100, for positive integer) of (the most all desirable performance numbers), search Number；Make m=1, m=1,2 ..., m ..., m_iter, when the m time iteration starts, make l=1, if in macro base station l coverage cell The transmitting power of little base station is X^(m,l), for initial solution vector, it is designated as currently solving simultaneously；

2-b) calculating is currently solving X^(m,l)Contiguous range in so that objective optimization function F (p_j') optimum solution vector； And judge whether (size of comparison object majorized function value) this solution vector makes objective optimization functional value F (p_j') for be at present The most optimum (i.e. maximum), if optimum, performs step 2-e), otherwise perform step 2-c)；

2-c) judge whether this solution vector is in taboo state, the most then this solution vector is arranged from the current neighborhood solved Remove, and perform step 2-d), otherwise perform step 2-e)；

2-d) judge that the current neighborhood solved the most still has the solution vector being in non-taboo state, if still having, then return and hold Row step 2-b)；Otherwise, the neighborhood that will currently solve is marked as the solution vector of taboo state at first change and be set to non-taboo state (i.e. in taboo list, corresponding element value is set to 0), and return execution step 2-b)；

2-e) it is designated as currently solving by this solution vector, and by p_jThe corresponding unit representing medium and small base station, macro base station l coverage cell in ' Element value is updated to X^(m,l)Middle respective value；Update the taboo state of each corresponding element of taboo list, i.e. make objective optimization function F (p_j') optimum solution vector is set to P in the corresponding element value of taboo list, remaining has been in the solution vector of taboo state at taboo list Corresponding element value subtract one (minimum 0, can not be negative)；

2-f) judge whether current solution repeats: concrete determination methods is, current goal majorized function value F (p_j') therewith The value of the front objective optimization function of iteration every time compares, if existing equal, is then judged as repeating, performs step 2-g)；If no Exist equal, then current solution did not duplicate, and performed step 2-h)；

2-g) update Tabu Length: the iterations that the iteration residing for record current iteration and a upper repetition values is spaced, It is designated as y_r, r=1,2,3 ..., calculate y_rMeansigma methods be y_rep, update Tabu Length simultaneously, for α ＞ 0 α ＞ 0, α be P with y_repBetween scale factor (desirable representative value is 0.01,0.1,0.2,0.3 etc., span 0～1), it is judged that the P after renewal Whether more than α y_repIf, P ＞ α y_rep, then P=max (1, P-1) is made；If P≤α is y_rep, then P=P+1 is made；

2-h) judging whether to travel through all K macro base stations, if traveling through, performing step 2-j), it is not fully complete traversal and then performs Step 2-i)；

2-i) make l=l+1, the transmitting power of Nei little base station, macro base station coverage cell is designated as X^(m,l), it is designated as current simultaneously Solve, return and perform step 2-b)；

2-j) judge that m, whether less than m_iter, if being less than, then makes m=m+1, and circulates execution step 2)；Otherwise terminate to follow Ring, performs step 3)；

3), after completing m_iter iteration optimization, little base station transmitting power has been updated to p_j', export the least base station system Capacity and coverage goal majorized function value F (p_j')；Terminate little station system capacity and coverage optimization.

Claims (1)

1. a little Base Station Compatibility based on self adaptation TABU search and coverage optimization method, it is characterised in that the method includes Following steps:

1) it is calculated the initial value of capacity and the objective optimization function of covering according to little base station Initial Trans, specifically calculates Step is as follows:

If little base station system comprises K macro base station l, M little base station j and the user terminal UE of N number of random distribution_i, wherein, l=1, 2 ..., l ..., K, j=1,2 ... j ..., M, i=1,2 ..., i ..., N；The transmitting power of little base station j is p_j, little base station j with User terminal UE_iBetween transmission channel gain be g_ij, then it is transferred to user terminal UE from little base station j_iReference signal receive Power P_rx(i, the such as formula of expression formula j) (1):

P_rx(i, j)=p_jg_ij (1)

If system noise is σ², the least base station j is via downlink to user terminal UE_iSignal to Interference plus Noise Ratio SINR_iExpression formula such as Formula (2):

${\mathrm{SINR}}_i=\frac{g_{ij}{p}_j}{{\mathrm{\σ}}^2+\underset{k\≠j}{\Σ}{g}_{ik}{p}_k}---\left(2\right)$

Wherein g_ikFor little base station k and user terminal UE_iBetween transmission channel gain, p_kFor the transmitting power of little base station k, k ≠ j；

If each pilot power occupies the part that total equalization launching power is fixed, and each channel also takes up general power Equal proportion part, then user terminal UE_iNormalized throughput t_iSuch as expression formula (3):

$t_i={\log }_2(1+{\mathrm{SINR}}_i)={\log }_2(1+\frac{g_{ij}{p}_j}{{\mathrm{\σ}}^2+\underset{k\≠j}{\Σ}{g}_{ik}{p}_k})---\left(3\right),$

The normalization total throughout T of the least base station j_jSuch as expression formula (4):

$T_j=\underset{{\mathrm{UE}}_i\∈U_j}{\Σ}{\log }_2(1+\frac{g_{ij}{p}_j}{{\mathrm{\σ}}^2+\underset{k\≠j}{\Σ}{g}_{ik}{p}_k})---\left(4\right),$

Wherein, U_jRepresent all user terminals of little base station j service, if the evaluation index of capacity is user terminal in whole network Average throughput, its formula express such as formula (5):

$\frac{1}{N}\underset{j=1}{\overset{M}{\Σ}}{T}_j=\frac{1}{N}\underset{j=1}{\overset{M}{\Σ}}\underset{{\mathrm{UE}}_i\∈U_j}{\Σ}{\log }_2(1+\frac{g_{ij}{p}_j}{{\mathrm{\σ}}^2+\underset{k\≠j}{\Σ}{g}_{ik}{p}_k})---\left(5\right);$

Represent covering with the minimum p% of the cumulative distribution of user throughput in each macro base station coverage cell, use T_{L, p%}Table Show the minimum p% of user throughput cumulative distribution in macro base station l coverage cell, introduce compromise coefficient gamma and balance covering and hold Measure two performance indications, 0 ＜ γ ＜ 1, then M little base station transmitting power is p_j' time, p_j'={ p₁,p₂,...,p_j,...,p_M, Network system capacity and the objective optimization function F (p of covering_j') define such as formula (6):

$F\left({p_j}^{\′}\right)=\mathrm{\γ}\frac{1}{\frac{K}{N}\underset{j=1}{\overset{M}{\Σ}}{T}_j}+(1-\mathrm{\γ})\frac{1}{\underset{l=1}{\overset{K}{\Σ}}{T}_{l,p\%}}---\left(6\right);$

Obtain the objective optimization function F (p of capacity and covering_j') initial value；

2) self adaptation tabu search algorithm is used to generate new little base station transmitting power p_j', and calculate new the taking of objective optimization function Value F (p_j'), specifically comprise the following steps that

2-a) initialize the parameter of self adaptation TABU search: sets taboo list as sky, set Tabu Length as P, P be positive integer, Width neighborhood, the span of little base station power, maximum iteration time m_iter of search；Make m=1, m=1,2 ..., M ..., m_iter, when the m time iteration starts, make l=1, if the transmitting power of medium and small base station, macro base station l coverage cell is X^{(m ,l)}, it is designated as currently solving；

2-b) calculating is currently solving X^(m,l)Contiguous range in so as to get objective optimization function F (p_j') optimum solution vector；And Judge whether this solution vector makes objective optimization functional value F (p_j') it is optimum so far, if optimum, perform step 2-e), Otherwise perform step 2-c)；

2-c) judge whether this solution vector is in taboo state, the most then this solution vector is got rid of from the current neighborhood solved, and Perform step 2-d), otherwise perform step 2-e)；

2-d) judge currently to solve X^(m,l)Neighborhood in the most still have the solution vector being in non-taboo state, if still having, then return hold Row step 2-b)；Otherwise, the neighborhood that will currently solve is marked as the solution vector of taboo state at first and changes and be set to non-taboo state, And return execution step 2-b)；

2-e) it is designated as currently solving by this solution vector, and by p_jThe corresponding element value representing medium and small base station, macro base station l coverage cell in ' It is updated to X^(m,l)Middle respective value；Update the taboo state of each corresponding element of taboo list, i.e. make objective optimization function F (p_j') Excellent solution vector is set to P in the corresponding element value of taboo list, and remaining has been in the solution vector correspondence at taboo list of taboo state Element value subtracts one；

2-f) judge whether current solution repeats: concrete determination methods is, current goal majorized function value F (p_j') with the most each The value of the objective optimization function of iteration compares, if existing equal, is then judged as repeating, performs step 2-g)；If there is not phase Deng, then current solution does not duplicate, and performs step 2-h)；

2-g) update Tabu Length: the iterations that the iteration residing for record current iteration and a upper repetition values is spaced, be designated as y_r, r=1,2,3 ..., calculate y_rMeansigma methods be y_rep, update Tabu Length simultaneously, for α ＞ 0, α be P and between y_rep Scale factor, it is judged that whether the P after renewal is more than α y_repIf, P ＞ α y_rep, then P=max (1, P-1) is made；If P≤α is y_rep, then make P=P+1；

2-h) judging whether to travel through all K macro base stations, if traveling through, performing step 2-j), it is not fully complete traversal and then performs step 2-i)；

2-i) make l=l+1, the transmitting power of Nei little base station, macro base station coverage cell is designated as X^(m,l), it is designated as currently solving simultaneously, returns Receipt row step 2-b)；

2-j) judge that m, whether less than m_iter, if being less than, then makes m=m+1, and circulates execution step 2)；Otherwise end loop, Perform step 3)；

3), after completing m_iter iteration optimization, little base station transmitting power has been updated to p_j', export the appearance of the least base station system Amount and coverage goal majorized function value F (p_j')；Terminate little station system capacity and coverage optimization.