5G fast load balancing method based on dual lists

Abstract

The present invention discloses a 5G fast load balancing method based on a dual list, which overcomes the problem that the existing technology of load distribution between neighboring areas cannot make the entire system perform at its best performance, including: step 1: calculating user space; step 2: establishing a dual list; step 3: dual list transfer; step 4: cell status update; step 5: convergence judgment. The present invention first calculates the corresponding load space and user space according to the load of each cell, and screens out a forward P list with additional user space and a reverse N list without additional user space; based on the N list, a traversal method is adopted to scan the P list in a point-to-multipoint manner for the overloaded users therein, and a target cell that can tolerate the overloaded user is matched, and a one-time user relocation is implemented in a full or partial transfer manner; by dynamically maintaining the information status of the dual list, rapid convergence can be achieved in the shortest time to meet the goal of efficient load balancing.

Description

Technical Field

[0001] The present invention relates to the field of 5G technology, and in particular to a 5G fast load balancing method based on a dual-list. Background Art

[0002] With the development of 5G technology, the scale of the network has expanded dramatically, data transmission technology has developed rapidly, and the contradiction between wireless links and backend system data processing has become increasingly prominent. Load balancing enables the entire system to perform at its best by distributing load among neighboring areas, thus better realizing mobile broadband and broadband mobility.

[0003] As early as July 9, 2014, the applicant publicly authorized a load balancing method NSLB based on neighboring cell sets for LTE, with the announcement number CN104202770B. The method solves the problem of unbalanced traffic by analyzing the load space that the minimum load cell can accommodate, and then inferring the number of users to be transferred in the neighboring cell set. The idea of ​​NSLB is similar to the current mainstream load balancing algorithm. By paying attention to the load space margin of the neighboring cells, load transfer is implemented between overloaded and lightly loaded cells, thereby achieving the purpose of load balancing. Unfortunately, NSLB is not very concerned about the efficiency of balancing, and may not be suitable for application in scenarios with high throughput and large-scale cell clusters. Summary of the invention

[0004] The present invention aims to improve the convergence efficiency of load balancing as much as possible and provides a 5G fast load balancing method based on a dual list. Starting from estimating the user space, a dual list based on the cell is established through positive and negative angle analysis, and a point-to-multipoint mechanism is used to implement a one-time transfer of users in the list. The load balancing goal can be achieved quickly, which greatly improves the efficiency of load balancing in the 5G system.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] A 5G fast load balancing method based on a dual list includes the following steps:

[0007] Step 1: Calculate the user space, which is calculated based on the number of users in each cell and the load occupied by each user;

[0008] Step 2: Create a double list: sort the cells with user space greater than zero in ascending order and put them into the positive P list, and sort the cells with user space less than zero in descending order and put them into the negative N list;

[0009] Step 3: Double-list transfer: extract cell neighbors from the P list to form a short list, which is divided into full transfer and partial transfer;

[0010] Step 4: Cell status update: Update the user space in the full transfer state and partial transfer state respectively, and update the short list, P list and N list;

[0011] Step 5: Convergence judgment: Set the acceptable balance state, convergence target value and upper limit of convergence times, calculate the load balancing factor and convergence times, and verify whether the N list is empty. If not, loop steps 1 to 4 until the N list is empty.

[0012] Preferably, the method comprises n cells NBS which are neighboring cells n ={gNB 1 , gNB 2 , …, gNB n}, load {Cld 1 , Cld 2 , …, Cld n}, and the total number of users of the corresponding bearer {Unm 1 , Unm 2 , …, Unm n}.

[0013] Preferably, the step 1 comprises the following steps:

[0014] (1-1): For NBS n ={gNB 1 , gNB 2 , …, gNB n}, calculate the sum of all cell loads And the load mean MCld n =MCld all / n;

[0015] (1-2): For each cell gNB i , calculate the load space CldV i =MCld n -Cld i , calculate the load per user CdPu i =Unm i / Cld i , and the user space per cell Wherein, floor(·) represents the floor function, i=1, 2, …, n.

[0016] Preferably, the step 2 includes the following contents:

[0017] Screen out NBS n Satisfy the condition UmV i >0 for all cells, and according to the user space UmV i Sort in ascending order and put them into the forward P list. Among them, j X ≤n; the number of cells in the P list is X; filter out NBS n Satisfy the condition UmV i <0 for all cells, and according to the user space UmV i Sort in descending order and put them into the negative N list. Among them, k Y ≤n; the number of cells in the N list is Y, satisfying Y≤nX.

[0018] Preferably, X+Y≤n.

[0019] Preferably, the step three comprises the following steps:

[0020] (3-1): Take out the first cell gNB in ​​the Nlst list k , start from the head of list P to search for position Nm m ≤X, so that it satisfies and Where abs(·) represents the absolute value function; extract these N from the P list m gNB k The short list of neighbors

[0021] (3-2): If N m If it exists, it is called a full transfer of the cell gNBk; k Transfer to each neighboring area in PShort respectively users; if N m If it does not exist, it is called a cell gNB. k Partial transfer, PShort = Plst, N m =X; by cell gNB k Transfer to each neighboring area in PShort(Plst) respectively users.

[0022] Preferably, the step 4 comprises the following steps:

[0023] (4-1): Update cell gNB in ​​full transfer state k Number of users Unm k =Unm k -abs(UmV k ), update the number of users Unm in each cell in PShort z =Unm z +UmV z in, Update cell gNB k User space UmVk =UmV k +abs(UmV k ), update the user space UmV of each cell in PShort z =UmV z -UmV z =0,

[0024] in,

[0025]

[0026] (4-2): Update cell gNB in ​​partial transfer state k Number of users Update the number of users Unm in each cell in PShort (PLst) z =Unm z +UmV z ; Update cell gNB k User space Update the user space UmV of each cell in PShort z =UmV z -UmV z =0;

[0027] (4-3): Update cell gNB k Load Cld k =Unm k / CdPu k ; Update the load of each cell in PShort Update the P list and N list.

[0028] Preferably, the step five includes the following contents:

[0029] Set the acceptable equilibrium state β 0 , convergence target value ε, upper limit of convergence times Nm th ; For NBS n ={gNB 1 , gNB 2 , …, gNB n}, calculate the load balancing factor Convergence times Nm L =Nm L +1; check whether the condition of Case 1 is met and Case 2 condition Nm L ≥Nm th And Case 3 condition Nlst = NULL, that is, the N list is empty; if it is not satisfied, repeat steps 1 to 4 until one of the conditions is satisfied.

[0030] Therefore, the present invention has the following beneficial effects:

[0031] The present invention can directly calculate the user space of the cell, and then establish corresponding forward lists and reverse lists; based on space matching, for any user in the reverse list, it can find the target cell that meets its carrying requirements in the shortest time, implement a one-time transfer, and dynamically update the information status of the dual lists, thereby achieving rapid convergence of 5G cell load balancing. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a flow chart of the present invention.

[0033] Figure 2 It is a specific flow chart of the present invention.

[0034] Figure 3 This is a comparison chart of the convergence efficiency of the present invention and other algorithms under light overload conditions.

[0035] Figure 4 This is a comparison chart of the convergence efficiency of the present invention and other algorithms under heavy overload conditions.

[0036] Figure 5 This is a comparison chart of the transfer user throughput of the present invention and other algorithms. DETAILED DESCRIPTION

[0037] The present invention is further described below in conjunction with the accompanying drawings and specific embodiments.

[0038] Embodiment 1:

[0039] like Figure 1 The embodiment shown is a 5G fast load balancing method based on a double list, including n cells NBS that are neighboring cells n ={gNB 1 , gNB 2 , …, gNB n}, load {Cld 1 , Cld 2 , …, Cld n}, and the total number of users of the corresponding bearer {Unm 1 , Unm 2 , …, Unm n};

[0040] The steps include:

[0041] Step 1: Calculate the user space, which is calculated based on the number of users in each cell and the load occupied by each user;

[0042] Step 2: Create a double list: sort the cells with user space greater than zero in ascending order and put them into the positive P list, and sort the cells with user space less than zero in descending order and put them into the negative N list;

[0043] Step 3: Double-list transfer: extract cell neighbors from the P list to form a short list, which is divided into full transfer and partial transfer;

[0044] Step 4: Cell status update: Update the user space in the full transfer state and partial transfer state respectively, and update the short list, P list and N list;

[0045] Step 5: Convergence judgment: Set the acceptable balance state, convergence target value and upper limit of convergence times, calculate the load balancing factor and convergence times, and verify whether the N list is empty. If not, loop steps 1 to 4 until the N list is empty.

[0046] Embodiment 2:

[0047] like Figure 2 The embodiment shown is a 5G fast load balancing method based on a double list, including n cells NBS that are neighboring cells n ={gNB 1 , gNB 2 , …, gNB n}, load {Cld 1 , Cld 2 , …, Cld n}, and the total number of users of the corresponding bearer {Unm 1 , Unm 2 , …, Unm n};

[0048] Step 1: Calculate user space;

[0049] Step 1-1: For NBS n ={gNB 1 , gNB 2 , …, gNB n}, calculate the sum of all cell loads And the load mean MCld n =MCld all / n;

[0050] Step 1-2: For each cell gNB i , calculate the load space CldV i =MCld n -Cld i , calculate the load per user CdPu i =Unm i / Cld i, and the user space per cell Where floor(·) represents the floor function, i = 1, 2, …, n;

[0051] Step 2: Create a double list;

[0052] Screen out NBS n Satisfy the condition UmV i >0 for all cells, and according to the user space UmV i Sort in ascending order and put them into the forward P list. Among them, j X ≤n; the number of cells in the P list is X; filter out NBS n Satisfy the condition UmV i <0 for all cells, and according to the user space UmV i Sort in descending order and put them into the negative N list. Among them, k Y ≤n; the number of cells in the N list is Y, satisfying Y≤nX;

[0053] Step 3: Dual list transfer;

[0054] Step 3-1: Take out the first cell gNB in ​​the Nlst list k , start from the head of list P and search for position N m ≤X, so that it satisfies and Where abs(·) represents the absolute value function; extract these N from the P list m gNB k The short list of neighbors

[0055] Step 3-2: If N m If it exists, it is called a cell gNB k Full transfer; by cell gNB k Transfer to each neighboring area in PShort respectively users; if N m If it does not exist, it is called a cell gNB. k Partial transfer, PShort = Plst, N m =X; transferred from cell gNBx to each neighboring cell in PShort(Plst) Users;

[0056] Step 4: Cell status update;

[0057] Step 4-1: Update cell gNB in ​​full transfer state k Number of users Unm k =Umm k-abs(UmV k ), update the number of users Unm in each cell in PShort z =Unm z +UmV z in, Update cell gNB k User space UmV k =UmV k +abs(UnV k ), update the user space UmV of each cell in PShort z =UmV z -UmV z =0,

[0058] in,

[0059]

[0060] Step 4-2: Update cell gNB in ​​partial transfer state k Number of users Update the number of users Unm in each cell in PShort (PLst) z =Unm z +UmV z ; Update cell gNB k User space Update the user space UmV of each cell in PShort z =UmV z -UmV z =0;

[0061] Step 4-3: Update cell gNB k Load Cld k =Unm k / CdPu k ; Update the load Cld of each cell in PShort z =Unm z / CdPu z ; Update P list and N list;

[0062] Step 5: Convergence decision

[0063] Set the acceptable equilibrium state β 0 , convergence target value ε, upper limit of convergence times Nm th ; For NBS n ={gNB 1 , gNB 2 , …, gNB n}, calculate the load balancing factor Convergence times Nm L =Nm L +1; check whether the condition of Case 1 is met and Case 2 condition Nm L ≥Nm th And Case3 condition Nlst = NULL, that is, the N list is empty; if it is not satisfied, repeat steps 1 to 4 until one of the conditions is satisfied.

[0064] Embodiment 3:

[0065] The present invention is specifically described below by taking n=5 as an example. The cell load and the number of users carried by the 5G system are shown in Table 1:

[0066] Table 1

[0067] Community Initial Load Number of users Remark <![CDATA[gNB 1 ]]> 0.48 52 Neighboring areas <![CDATA[gNB 2 ]]> 0.25 37 Neighboring areas <![CDATA[gNB 3 ]]> 0.63 48 Neighboring areas <![CDATA[gNB 4 ]]> 0.32 23 Neighboring areas <![CDATA[gNB 5 ]]> 0.87 66 Neighboring areas <![CDATA[gNB 6 ]]> 0.59 51 Neighboring areas <![CDATA[gNB 7 ]]> 1.23 72 Neighboring areas <![CDATA[gNB 8 ]]> 0.46 49 Neighboring areas

[0068] The basic data is shown in Table 2:

[0069] Table 2

[0070] project data Operating frequency (GHz) 2.6 Operating bandwidth (MHz) 100 Subcarrier bandwidth index 1 <![CDATA[Acceptable equilibrium state β 0 ,]]> 0.999 Convergence target value ε 0.01％ <![CDATA[Upper limit Nm of the number of convergence times th > 100

[0071] This example describes a dual-list-based 5G fast load balancing method, which includes the following steps: calculating user space, establishing a dual-list, dual-list transfer, cell status update, and convergence decision;

[0072] Step 1: Calculate user space;

[0073] Step 1-1: For NBS n ={gNB 1 , gNB 2 , …, gNB n}, calculate the sum of all cell loads And the load average

[0074] Step 1-2: For each cell gNB i , calculate the load space CldV i =NCld n -Cld i ={0.12, 0.35, -0.03, 0.28, -0.27, 0.01, -0.63, 0.14}, calculate the load occupied by each user

[0075]

[0076] And the user space per cell

[0077]

[0078] Step 2: Create a double list;

[0079] Screen out NBS n Satisfy the condition UmV i >0 for all cells, and according to the user space UmV i Sort in ascending order and put them into the forward P list. The number of cells in the P list is X=5; NBS is selected n Satisfy the condition UmV i <0 for all cells, and according to the user space UmV i Sort in descending order and put them into the negative N list. The number of cells in the N list is Y=3;

[0080] Step 3: Dual list transfer;

[0081] Step 3-1: Take out the first cell gNB in ​​the Nlst list k =gNB 3 , start from the head of list P and search for position N m =2≤X=5, so that it satisfies and Extract this N from the P list m =2 gNBs k The short list of neighbors

[0082] Step 3-2: N m Existence, cell gNB k Full transfer occurs; gNBk in the cell transfers to each neighboring cell in PShort Users;

[0083] Step 4: Cell status update;

[0084] Step 4-1: Update cell gNB in ​​full transfer state 3 Number of users Unm 3 =Unm a -abs(UmV 3 )=48-2=46, update the number of users Unm in each cell in PShort 6 =Unm 6 +UmV 6 =51+1=52;Unm 1 =Unm 1 +2-1=52+1=53; Update cell gNB 3 User space UmV 3 =UmV 3 +abs(UmV 3)=-2+2=0, update the user space UmV of each cell in PShort 6 =UmV 6 -UmV 6 =0, UmV 1 =UmV 1 -2+1=13-1=12;

[0085] Step 4-3: Update cell gNB k＝3 Load Update the load of each cell in PShort Updated P list Plst = {gNB 1 , gNB 8 , gNB 4 , gNB 2}, N list Nlst = {gNB 5 , gNB 7};

[0086] Step 5: Convergence decision;

[0087] For NBS n ={gNB 1 , gNB 2 , …, gNB n}, calculate the load balancing factor Convergence times Nm L =Nm L +1=1; does not meet the condition abs(β-β 0 )=0.1935≤ε=0.01%and condition Nm L ≥Nm th and N lists are empty; therefore, steps 1 to 4 are repeated until one of the conditions is met.

[0088] Simulation experiment:

[0089] The DLFLB method of the present invention and the previous NSLB method are simulated on the MATLAB platform. The basic data information is shown in Tables 1 and 2 above. Certain users are randomly added and random services are configured. The results are shown in the attached table. Figures 3 to 5 shown.

[0090] like Figure 3 The load balancing convergence process of the two methods under light overload conditions is shown in Figure 1, that is, only a small number of base stations are in an overloaded state. DLFLB undergoes about 54 convergences, and NSLB undergoes about 63 convergences, and the convergence efficiency is improved by about 14.29%. Figure 4The load balancing convergence process of the two methods under heavy overload conditions is shown, that is, a large number of base stations are in an overloaded state. DLFLB undergoes about 82 convergences, and NSLB undergoes about 123 convergences, and the convergence efficiency is improved by about 33.33%. It can be seen that the DLFLB algorithm can better reflect the convergence advantage in a heavy overload environment, which helps to reduce the system load and improve resource utilization;

[0091] like Figure 5 The throughput curves of the two algorithms for each user transfer are shown. Neither algorithm pays much attention to the performance of the transferred users, so their performances in terms of throughput are similar.

[0092] The specific description of the present invention in the above embodiments is only used to further illustrate the present invention and cannot be understood as limiting the scope of protection of the present invention. Technical engineers in this field may make some non-essential improvements and adjustments to the present invention based on the contents of the above invention, which fall within the scope of protection of the present invention.

Claims

1. A 5G fast load balancing method based on a dual list, characterized in that: The steps include: Step 1: Calculate the user space, which is calculated based on the number of users in each cell and the load occupied by each user; Step 2: Create a double list: sort the cells with user space greater than zero in ascending order and put them into the positive P list, and sort the cells with user space less than zero in descending order and put them into the negative N list; Step 3: Double-list transfer: Extract neighboring cells from the P list to form a short list, which is divided into full transfer and partial transfer, and take out the first cell gNB from the N list k , start from the head of list P and search for position N m , extract this N from the P list m gNB k The neighboring areas of N form a short list. m If it exists, it is called a cell gNB k Full transfer, if N m If it does not exist, it is called a cell gNB. k partial transfer of Step 4: Cell status update: Update the user space in the full transfer state and partial transfer state respectively, and update the short list, P list and N list; Step 5: Convergence judgment: Set the acceptable balance state, convergence target value and upper limit of convergence times, calculate the load balancing factor and convergence times, and verify whether the N list is empty. If not, loop steps 1 to 4 until the N list is empty.

2. According to claim 1, a 5G fast load balancing method based on a dual list is characterized in that: The method includes n cells NBS which are neighboring cells n = {gNB1, gNB2, …, gNB n }, load {Cld1, Cld2, …, Cld n }, and the total number of users of the corresponding bearer {Unm1, Unm2, …, Unm n }.

3. According to claim 2, a 5G fast load balancing method based on a dual list is characterized in that: The step 1 comprises the following steps: (1-1): For NBS n = {gNB1, gNB2, …, gNB n }, calculate the sum of all cell loads And the load mean MCld n =MCld all / n; (1-2): For each cell gNB i , calculate the load space CldV i =MCld n -Cld i , calculate the load per user CdPu i =Unm i / Cld i , And the user space per cell Wherein, floor(·) represents the floor function, i=1, 2, …, n.

4. According to claim 3, a 5G fast load balancing method based on a dual list is characterized in that: The step 2 includes the following contents: Screen out NBS n Satisfy the condition UmV i >0 for all cells, and according to the user space UmV i Sort in ascending order and put them into the forward P list. Among them, j X ≤n; the number of cells in the P list is X; filter out NBS n Satisfy the condition UmV i <0 for all cells, and according to the user space UmV i Sort in descending order and put them into the negative N list. Among them, k Y ≤n; the number of cells in the N list is Y; the above X+Y≤n.

5. A 5G fast load balancing method based on a dual list according to claim 4, characterized in that: The step three comprises the following steps: (3-1): Take out the first cell gNB in ​​the Nlst list k , start from the head of list P and search for position N m ≤X, so that it satisfies and Where abs(·) represents the absolute value function; extract these N from the P list m gNB k The short list of neighbors (3-2): If N m If it exists, it is called a cell gNB k Full transfer; by cell gNB k Transfer to each neighboring area in PShort respectively users; if N m If it does not exist, it is called a cell gNB. k Partial transfer, PShort = Plst, N m =X; by cell gNB k Transfer to each neighboring area in PShort respectively users.

6. A 5G fast load balancing method based on a dual list according to claim 5, characterized in that: The step 4 comprises the following steps: (4-1): Update cell gNB in ​​full transfer state k Number of users Unm k =Unm k -abs(UmV k ), update the number of users Unm in each cell in PShort z =Unm z +UmV z in, Update cell gNB k User space UmV k =UmV k +abs(UmV k ), update the user space UmV of each cell in PShort z =UmV z -UmV z =0, in, (4-2): Update cell gNB in ​​partial transfer state k Number of users Update the number of users Unm in each cell in PShort (PLst) z =Unm z +UmV z ; Update cell gNB k User space Update the user space UmV of each cell in PShort z =UmV z -UmV z =0; (4-3): Update cell gNB k Load Cld k =Unm k / CdPu k ; Update the load Cld of each cell in PShort z =Unm z / CdPu z ; Update P list and N list.

7. A 5G fast load balancing method based on a dual list according to claim 6, characterized in that: The step five includes the following contents: Set the acceptable equilibrium state β0, the convergence target value ε, and the upper limit of the convergence number Nm th ; For NBS n = {gNB1, gNB2, …, gNB n }, calculate the load balancing factor Convergence times Nm L =Nm L +1; check whether the condition of Case 1 is met and Case 2 condition Nm L ≥Nm th And Case 3 condition Nlst = NULL, that is, the N list is empty; if it is not satisfied, repeat steps 1 to 4 until one of the conditions is satisfied.

Images