CN104581829A - Mobility load balancing method based on AHP (Analytical Hierarchy Process) in LTE system - Google Patents

Abstract

The invention discloses a mobility load balancing method based on an AHP (Analytical Hierarchy Process) in an LTE system. The method mainly comprises the steps of selection of an optimal load transfer cell and determination of optimal CIO (Cell Independent Offset) parameters. According to the method, four modules namely a cell state detection module (1), a user-cell matching pair generation module (2), a load distribution module (3) and a balancing module (4) are adopted, wherein the cell state detection module is mainly used for judging a cell load state and initiating load balancing if a cell is overloaded; the user-cell matching pair generation module is mainly used for selecting an optimal transfer cell of a user through an AHP algorithm; the load distribution module is mainly used for pre-distributing the load quantities of the overloaded part of the overloaded cell to each target cell before the CIO parameters are adjusted; the load of each target cell is smaller, the distributed load quantities are larger; the balancing module is mainly used for adjusting the CIO through a variable step; when each target cell completely receives the distributed load quantities, the optimal CIO of the cell is obtained.

Description

Based on the Mobile load balancing method of step analysis in a kind of LTE system

Technical field

The present invention relates to mobile communication technology field, particularly relate to a kind of Mobile load balancing method based on step analysis in LTE (Long Term Evolution, Long Term Evolution) system.

Background technology

3GPP LTE system adopts the key technologies such as OFDM and MIMO to realize higher data rate, throughput and spectrum efficiency.But, increasing along with user and business datum, some Hot Spot heavier loads, cause higher cutting off rate, reduce Consumer's Experience; And other adjacent idle cell loads are comparatively light, resource is not fully utilized.Therefore, the load balancing how realized in network is problem instantly more and more in the urgent need to address.Load balancing in former gsm system mainly relies on observation to find to be in for a long time the base station of overload, regulates relevant parameter, to reach balanced object.But the human and material resources of the method at substantial of this manual adjustment parameter, efficiency is not high.Therefore, 3GPP working group includes SON (Self-OrganizingNetworks, self-organization network) technology in LTE standard category.SON mainly comprises three zones, i.e. self-configuring, self-optimizing, certainly healing, wherein MLB (Mobility Load Balancing, mobility load balance) is one of important use-case of self-optimizing.

Load balancing is mainly realized by adjustment community independent bias CIO.The defined formula of CIO is as follows:

M _n-M _s>Hyst-CIO _s,n

M in formula _n, M _sbe respectively cell-of-origin and adjacent cell measurement result, Hyst is lag parameter, CIO _s,nfor cell-of-origin and the community independent bias for adjacent cell maintenance.Comprise two aspects to the research of MLB algorithm at present, one is the selection that community is shifted in load; Two is determinations of moving parameter CIO.Select suitable Target cell, contribute to reducing pingpang handoff and handover failure rate; Adjustment associated movement parameters, contributes to effectively realizing load balancing fast, reduces blocking rate.

Current most of load-balancing algorithm is when the load of the community that an eNB serves exceedes predefined load-threshold, determines that this community is community to be adjusted.First from the neighbor cell of this community to be adjusted, the Target cell of load balancing is carried out as with this community to be adjusted in the community choosing least-loaded, secondly in this community to be adjusted, choose the user being suitable for carrying out switching, finally these users are adjusted modes such as (switch) gravity treatment and directed gravity treatments to Target cell to realize the load balancing of minizone.They are when select target community, general only consideration Target cell load or this single factors of Reference Signal Received Power RSRP, the community of such selection often easily causes pingpang handoff or handoff failure, because the Target cell that RSRP is comparatively large or load is lighter, channel condition differs and reserves.In addition when adjusting moving parameter to realize equilibrium, only from the angle of overloaded cells, overloaded cells load being dropped to below predefine thresholding, completes and unload, do not consider that each Target cell receives the ability of load.In addition, how CIO is adjusted also do not provide concrete method of adjustment.

Therefore, need a kind of effective method to overcome the shortcoming in said method load balancing process, at raising cell edge user throughput, while reducing call blocking rate, effective minimizing pingpang handoff, handoff failure, and distribute the load of overloaded cells according to the load-receipt ability of each Target cell, improve the equilibrium degree of minizone.

Summary of the invention

For above deficiency of the prior art, the object of the present invention is to provide a kind ofly effectively can reduce pingpang handoff and handoff failure, the equilibrium degree that contributes to improving minizone, effectively reduce balanced time delay LTE system in based on the Mobile load balancing method of step analysis, technical scheme of the present invention is as follows: based on the Mobile load balancing method of step analysis in a kind of LTE system, it comprises the following steps:

101, initialization, setting judges the height two-stage thresholding of each cell load status, is overload threshold ρ respectively _tHand balanced thresholding ρ _tL, and ρ _tH> ρ _tL, each cell load status of base station eNB periodic test, periodic harvest is reported, as cell load ρ _i> ρ _tHtime, be judged to be overloaded cells, initiate load balancing requests, jump to step 102; Otherwise continue periodic harvest report;

102, the user receiving this cell reference signals received power RSRP minimum value is found out in overloaded cells, as to be switched user; And load in adjacent cell is met ρ _j< ρ _tLthe alternatively target handover cell, community of (j ≠ i);

103, the load capacity R that overloaded cells need be shifted is judged _iwhether be less than the admissible load capacity summation in Target cell, if so, then continue to perform step 103, otherwise perform step 110;

104, step analysis AHP algorithm is adopted to generate user-community proportioning pair, then overloaded cells load capacity is distributed to each Target cell according to overburden portion capacity assignment formula, the according to target cell allocation load descending of arriving, generates target cell list, is designated as L;

105, select first community in L, note j=1, as Target cell, arranges community independent bias CIO _{i, j}=0;

106, judge whether j has exceeded target cell list L size, if j<=size (L), then perform step 107; Otherwise perform step 110.

107, CIO is upgraded _{i, j}value, CIO _{i, j}=CIO _{i, j}+ step, step-length step=(1+ (ρ _i-ρ _tL)) Δ, wherein Δ is predefined value;

108, CIO is judged _{i, j}whether be greater than CIO _maxif, CIO _{i, j}>CIO _max, select next community, return step 106; Otherwise continue to judge, overloaded cells i is at the parameter CIO of current setting _{i, j}whether the fractional load amount of lower transfer is less than the load capacity that Target cell j need receive, and if so, performs step 109; Otherwise, select next community, return step 106;

109, after overloaded cells load void being transferred to Target cell, the more load capacity that need receive of fresh target: Z _j=Z _j-R _hO, upgrade the load value of overloaded cells, namely deduct the fractional load that void transfers to Target cell: ρ _i=ρ _i-ρ _hO, wherein ρ _hOrepresent the empty fractional load transferring to Target cell, judge that whether the load value of overloaded cells after upgrading is lower than balanced thresholding, if perform 110; Otherwise return step 107.;

110, current recorded each CIO is adjusted _{i, j}value, terminates AMLB algorithm, completes mobility load balance.

Further, the computing formula of the load capacity of the overloaded cells i needs transfer in step 103 is: R _i=M _pRB× (ρ _i-ρ _tL), wherein R _irepresent that cell i needs the load capacity of transfer, ρ _irepresent i cell load, M _pRBrepresent the Physical Resource Block PRB quantity of community, suppose each community M _pRBequal; The maximum admissible load capacity computing formula in Target cell is: R _c=M _pRB× (ρ _tL-ρ _c), wherein R _crepresent the maximum admissible load capacity in Target cell, ρ _crepresent Target cell load.

Further, the overburden portion capacity assignment formula in step 104 is: wherein, m (m=1,2 ..., n) represent the goal displacement community of overloaded cells, Z _mrepresent the load capacity that Target cell m is assigned to, R _j(j=1,2 ..., n) represent the maximum admissible load capacity of Target cell j.

Advantage of the present invention and beneficial effect as follows:

The present invention, by the Target cell of AHP algorithms selection, can effectively reduce pingpang handoff and handoff failure; Preassignment overloaded cells load capacity before adjustment parameter CIO, contributes to the equilibrium degree improving minizone; Progressively CIO is adjusted by variable step size, the balanced time delay of effective minimizing, when preallocated load has been received in Target cell, obtain optimal parameter CIO, overloaded cells maximum number of user can be enable to be switched to Target cell and not suffer that Target cell is refused, can make full use of Target cell resource again, and Target cell is not transshipped.

Accompanying drawing explanation

Fig. 1 is the AHP algorithm flow chart according to the preferred embodiment of the present invention;

Fig. 2 is the AHP hierarchical model figure that the present invention builds;

Fig. 3 is module structure drafting of the present invention;

Fig. 4 is AMLB algorithm flow chart of the present invention.

Embodiment

The invention will be further elaborated to provide an infinite embodiment below in conjunction with accompanying drawing.But should be appreciated that, these describe just example, and do not really want to limit the scope of the invention.In addition, in the following description, the description to known features and technology is eliminated, to avoid unnecessarily obscuring concept of the present invention.

Fig. 3 is module structure drafting of the present invention, and the present invention includes 4 modules, each functions of modules is as follows: (1) cell status detection module

The present invention is provided with the height two-stage thresholding of the load condition judging each community from the angle reducing system loading, is overload threshold ρ respectively _tHand balanced thresholding ρ _tL, and ρ _tH> ρ _tL, overloaded cells load must be balanced to ρ _tLbelow.The each cell load status of eNB periodic test, as cell load ρ _i> ρ _tHtime, be judged to be overloaded cells, initiate load balancing requests.

(2) user-community proportioning is to generation module

Overloaded cells relates to the selection of goal displacement community in transferring user process, and the Target cell of selection is unreasonable, very easily causes pingpang handoff even handoff failure.The present invention utilizes AHP algorithm, considers Reference Signal Received Power (RSRP), Target cell available resources (R that overloaded cells user receives Target cell _t), user is switched to several factors such as the signal interference ratio (SINR) behind Target cell, selects the community of maximum weight to shift community as optimal objective.According to each to be switched user select optimal objective community, generate a user-community proportioning pair.

(3) load distribution module

The ability that each Target cell receives load is different.In order to improve the equilibrium degree of minizone, first according to the load-receipt ability preassignment overloaded cells load capacity of each Target cell before adjustment parameter CIO.Concrete distribution method is as follows:

Suppose that overloaded cells is i, its goal displacement community is j, k, m and (ρ _j< ρ _k< ρ _m), Target cell mentioned here refers to that load must lower than balanced thresholding (ρ _c< ρ _tL) community, overloaded cells i need transfer load capacity be:

R _i＝M _PRB×(ρ _i-ρ _TL) (1)

Wherein R _irepresent that cell i needs the load capacity of transfer, ρ _irepresent i cell load, M _pRBrepresent that Physical Resource Block (PRB) quantity of community (supposes each community M _pRBequal).In like manner, can calculate community j, the ultimate load (available resources) that k, m can receive respectively is:

R _c＝M _PRB×(ρ _TL-ρ _c) (2)

Wherein, c represents community j, k, m.Overburden portion load capacity, according to object listing small area loading condition, is done following distribution by overloaded cells i:

$Z_c=\frac{R_c}{R_j+R_k+R_m}\&times;R_i---\left(3\right)$

Wherein, Z _c(c=j, k, m) represents the load capacity that cell allocation arrives.Because R _i<R _j+ R _k+ R _m, otherwise can not load balancing be initiated, so Z _c<R _c.

(4) balance module

The transfer of inter-cell user is mainly realized by the change of parameter CIO.Increase CIO, reduce the thresholding switched, reduce CIO and improve handoff threshold.Visible, the number of users in community can be changed by change parameter CIO, reach balanced.The present invention mainly adjusts parameter CIO by variable step size step, and when overload cell load is larger with the difference between balanced thresholding, step changes faster; Otherwise, change slower.Design variable step size, contributes to reducing balanced time delay.When preallocated load has been received in Target cell, obtain optimal parameter CIO.

Utilize the concrete steps of AHP algorithm: because load balancing relates to the handoff procedure of user, switching cell selection is incorrect easily causes pingpang handoff and handoff failure.AHP algorithm synthesis is utilized to consider several factor, can effective head it off, a user-community proportioning pair can be generated by the present embodiment.Fig. 1 is AHP algorithm flow chart of the present invention, and its concrete implementation step is as follows.

(1) AHP Recurison order hierarchy model is as shown in Figure 2 set up.Optimal objective community as destination layer, by RSRP, SINR, R _tthree factors is as rule layer, and each candidate handover cell is as solution layer.

(2) with 1 ~ 9 comparing dimensional configurations judgment matrix in tournament method and table 1.Each element a in matrix _i,jbe made up of 1 ~ 9 scale or its inverse, meet a _i,j>0 and a _j,i=1/a _i,j, a _ii=1.The judgment matrix A that rule layer (A) Three factors forms relative to the importance of destination layer (O) is such as formula (4)

Table 11 ~ 9 compares yardstick

Note: 2, the median of the adjacent judgement of 4,6,8 expression

(3) Mode of Level Simple Sequence is calculated.According to judgment matrix, determine that the process of each element relative weight in level is called Mode of Level Simple Sequence.For judgment matrix A, calculate and meet formula A ω=λ _maxthe Maximum characteristic root λ of ω _maxand character pair vector ω, ω is normalized and can draws single sequencing weight, ω=[0.6267,0.2797,0.0936].Meanwhile, corresponding coincident indicator CI and Consistency Ratio CR be calculated.The calculating of CR=CI/RI, CI is as formula (5), and Aver-age Random Consistency Index RI is as shown in table 2 with corresponding scale n value.As CR<0.1, think that judgment matrix has satisfied consistency; Otherwise just need to readjust judgment matrix, until there is satisfied consistency.

Table 2 Aver-age Random Consistency Index

$\mathrm{CI}=\frac{{\mathrm{\&lambda;}}_{\max }-n}{n-1}---\left(5\right)$

(4) calculation combination weights (total hierarchial sorting).Calculate the sequencing weight of all indexs of same level for top relative importance, usually successively carried out to lowest level by highest level.If last layer time A (rule layer) comprises A1, A2 ... Am is m (m=3 of the present invention) individual index altogether, and its Single Ordering Weight Value of Hierarchy is [a ₁, a ₂.., a _m], next hierarchical B (solution layer) comprises B1, B2 ..., Bn altogether n index, they for Aj (j=1,2 ..., m) weights of Mode of Level Simple Sequence are respectively [b _1j, b _2j..., b _nj], now B layer total ranking value computing formula is as shown in table 3.

The total ranking value of table 3 B layer calculates

Finally calculate B layer and always sort Consistency Ratio as shown in (6).

$\mathrm{CR}=\frac{\underset{j=1}{\overset{m}{\mathrm{\&Sigma;}}}{a}_j{\mathrm{CI}}_j}{\underset{j=1}{\overset{m}{\mathrm{\&Sigma;}}}{a}_j{\mathrm{RI}}_j}---\left(6\right)$

(5), after drawing total sequencing weight of solution layer B, select the community of maximum weight, be optimal objective community.The optimum subdistrict finding out each to be switched user can generate user-community proportioning pair.

AMLB algorithm concrete steps: Fig. 4 is the concrete AMLB algorithm flow chart realizing load balancing.The present invention, from the height two-stage thresholding being provided with the load condition judging each community of the angle of reduction system loading, is overload threshold ρ respectively _tHand balanced thresholding ρ _tL, and ρ _tH> ρ _tL.If only arrange one-level thresholding, i.e. overload threshold, even if so balanced success in overloaded cells, its load is still close to overload threshold, if at this moment there is burst service this community, can transship again, initiates balanced, increases system loading.On the contrary, if arrange two-stage thresholding, overloaded cells load balancing is to thresholding ρ _tLbelow, the community at this moment after equilibrium have enough can resource to receive the business of burst, equilibrium need not be initiated once again, effectively reduce system loading.

The present embodiment be based upon utilize AHP algorithm basis on, by utilizing AHP algorithm, can generate a crucial user-community proportioning pair, the present embodiment concrete steps are as follows.

(1) eNB periodic harvest measurement report, calculation plot load.As cell load ρ _i> ρ _tHtime, be judged to be overloaded cells, otherwise continue periodic harvest report.

(2) user receiving this community RSRP minimum value is found out in overload region, as to be switched user; Adjacent cell load meets ρ _j< ρ _tLthe alternatively target handover cell, community of (j ≠ i).

(3) judge whether the load capacity that need shift overloaded cells is less than the admissible load capacity summation in Target cell.If so, then continue to perform step (4), otherwise perform step (13).

(4) call AHP algorithm, find out the optimal objective community of to be switched user according to the Combining weights of each candidate cell, generate user-community proportioning pair.Then distribute formula according to load and distribute overloaded cells load capacity to each Target cell, the according to target cell allocation load descending of arriving, generation target cell list, is designated as L.

(5) first community (note j=1) in L is selected, as Target cell.

(6) CIO is set _{i, j}=0.

(7) judge whether j has exceeded target cell list L size.If j<=size (L), then perform step (8); Otherwise perform step (13).

(8) CIO is upgraded _{i, j}value.CIO _{i, j}=CIO _{i, j}+ step, step-length step=(1+ (ρ _i-ρ _tL)) Δ, wherein Δ is predefined value.It is variable that step is set, ρ here _i-ρ _tLbetween difference larger, step changes faster, and time for balance is restrained faster.

(9) CIO is judged _{i, j}whether be greater than CIO _max.If CIO _{i, j}>CIO _max, select next community (j=j+1), return step (6); Otherwise continue to judge, whether the load capacity that overloaded cells i shifts is less than the load capacity (R that Target cell j is assigned to _hO<Z _j, R _hOrepresent under the CIO parameter of current setting, void transfers to the load capacity of Target cell), if so, perform step (10); Otherwise, select next community (j=j+1), return step (6).

(10) after the empty transferring load to Target cell in overloaded cells, the more load capacity that need receive of fresh target: Z _j=Z _j-R _hO

(11) upgrade the load value of overloaded cells, namely deduct the fractional load that void transfers to Target cell: ρ _i=ρ _i-ρ _hO, wherein ρ _hOrepresent the empty fractional load transferring to Target cell.Judge that whether the load value of overloaded cells after upgrading is lower than balanced thresholding, if perform (12); Otherwise return step (8).

(12) current recorded each CIO is adjusted _{i, j}value.

(13) AMLB algorithm is terminated.

These embodiments are interpreted as only being not used in for illustration of the present invention limiting the scope of the invention above.After the content of reading record of the present invention, technical staff can make various changes or modifications the present invention, and these equivalence changes and modification fall into the scope of the claims in the present invention equally.

Claims (3)

1. in LTE system based on a Mobile load balancing method for step analysis, it is characterized in that, comprise the following steps:

101, initialization, setting judges the height two-stage thresholding of each cell load status, is overload threshold ρ respectively _tHand balanced thresholding ρ _tL, and ρ _tH> ρ _tL, each cell load status of base station eNB periodic test, periodic harvest is reported, as cell load ρ _i> ρ _tHtime, be judged to be overloaded cells, initiate load balancing requests, jump to step 102; Otherwise continue periodic harvest report;

102, the user receiving this cell reference signals received power RSRP minimum value is found out in overloaded cells, as to be switched user; And load in adjacent cell is met ρ _j< ρ _tLthe alternatively target handover cell, community of (j ≠ i);

103, the load capacity R that overloaded cells need be shifted is judged _iwhether be less than the maximum admissible load capacity summation in Target cell, if so, then perform step 104; Otherwise jump to step 110;

104, AHP step analysis algorithm is adopted to generate user-community proportioning pair, then overloaded cells load capacity is distributed to each Target cell according to capacity assignment formula, the according to target cell allocation load size descending of arriving, generates target cell list, is designated as L;

105, select first community in L, note j=1, as Target cell, arranges community independent bias CIO _{i, j}=0;

106, judge whether j has exceeded target cell list L size, if j<=size (L), then perform step 107; Otherwise perform step 110;

107, CIO is upgraded _{i, j}value, CIO _{i, j}=CIO _{i, j}+ step, step-length step=(1+ (ρ _i-ρ _tL)) Δ, wherein Δ is predefined value;

108, CIO is judged _{i, j}whether be greater than CIO _maxif, CIO _{i, j}>CIO _max, select next community, return step 106; Otherwise continue to judge, overloaded cells i is at the parameter CIO of current setting _{i, j}whether the fractional load amount of lower transfer is less than the load capacity that Target cell j need receive, and if so, performs step 109; Otherwise, select next community, return step 106;

109, after overloaded cells load void being transferred to Target cell, the more load capacity that need receive of new target cell: Z _j=Z _j-R _hO, upgrade the load value of overloaded cells, namely deduct the fractional load that void transfers to Target cell: ρ _i=ρ _i-ρ _hO, wherein ρ _hOrepresent the empty fractional load transferring to Target cell, judge that whether the load value of overloaded cells after upgrading is lower than balanced thresholding, if perform 110; Otherwise return step 107;

110, current recorded each CIO is adjusted _{i, j}value, terminates AMLB algorithm, completes mobility load balance.

2. in LTE system according to claim 1 based on the Mobile load balancing method of step analysis, it is characterized in that, overloaded cells i in step 103 needs the computing formula of the load capacity of transfer to be:

R _i=M _pRB× (ρ _i-ρ _tL), wherein R _irepresent that cell i needs the load capacity of transfer, ρ _irepresent i cell load, M _pRBrepresent the Physical Resource Block PRB quantity of community, suppose each community M _pRBequal; The maximum admissible load capacity computing formula in Target cell is: R _c=M _pRB× (ρ _tL-ρ _c), wherein R _crepresent the maximum admissible load capacity in Target cell, ρ _crepresent Target cell load.

3. in LTE system according to claim 1 based on the Mobile load balancing method of step analysis, it is characterized in that, the overburden portion capacity assignment formula in step 104 is: wherein, m (m=1,2 ..., n) represent the goal displacement community of overloaded cells, Z _mrepresent the load capacity that Target cell m is assigned to, R _j(j=1,2 ..., n) represent the maximum admissible load capacity of Target cell j.