METHOD FOR ENHANCING NETWORK COVERAGE BASED ON ADAPTIVE GENERATION OF RESOURCE CELL

Abstract

A method for enhancing network coverage based on adaptive generation of resource cell. A traditional cell in a network is taken as an initial state of the network. A correlation matrix in a current network state is generated. Whether |ω.sub.i,j.sup.t−ω.sub.i,j.sup.t-1|/ω.sub.i,j.sup.t-1>ε is determined. Each access point is divided into a plurality of resource cells. A CU-DU network mapping table is generated. A middlehaul link of each of the plurality of resource cells is constructed according to the CU-DU network mapping table.

Claims

1. A method for enhancing network coverage based on adaptive generation of resource cell, comprising: (S1) taking a traditional cell in a network as an initial state of the network; (S2) generating a correlation matrix in a current network state, wherein the step (S2) is performed through steps of: (S2a) monitoring a current network state of a plurality of access points in real time; and performing step (S2b) if a network coverage or traffic changes; (S2b) deploying, by a network resource orchestrator, an airborne access point according to a requirement to an area where a signal/capacity coverage hole occurs; (S2c) quantifying a traffic correlation among the plurality of access points; selecting two of the plurality of access points in the current network state, wherein when the two of the plurality of access points serve the same group of traffic in a local area, the traffic correlation is 1, otherwise the traffic correlation is 0; and combining the traffic correlation between any two of the plurality of access points to form a traffic correlation matrix; (S2d) allowing each of the plurality of access points to broadcast a latitude and longitude information thereof, and receive a latitude and longitude information of the rest of the plurality of access points; calculating a distance between any two of the plurality of access points in the current network state; and normalizing the distance to obtain a network topological matrix of the plurality of access points; (S2e) allowing each of the plurality of access points to receive an interference sent from the rest of the plurality of access points; and normalizing the interference to generate an interference matrix of the plurality of access points; and (S2f) subjecting the traffic correlation matrix, the network topological matrix and the interference matrix to linear weighting to generate a relevance matrix consisting of a relevance between the plurality of access points in the current network state; (S3) determining whether |ω.sub.i,j.sup.t−ω.sub.i,j.sup.t-1|/ω.sub.i,j.sup.t-1>ε, wherein if |ω.sub.i,j.sup.t−ω.sub.i,j.sup.t-1|/ω.sub.i,j.sup.t-1>ε, sending the relevance matrix generated in the current network state to the network resource orchestrator followed by performing step (S4); otherwise proceeding to the step (S2); wherein |•| is an operation of taking an absolute value; ω.sub.i,j.sup.t is a relevance between a ith access point and a jth access point in a relevance matrix at a current iteration; ω.sub.i,j.sup.t-1 is a relevance between a ith access point and a jth access point in a correlation matrix at a previous iteration; and ε is a preset trigger value; (S4) dividing each of the plurality of access points into a plurality of resource cells, wherein the step (S4) is performed through steps of: (S4a) selecting and deleting, by the network resource orchestrator, an access point with a minimum sequence number from an access point set formed by the plurality of access points; (S4b) selecting access points with a relevance higher than a threshold from the rest access points in the access point set to form a resource cell; and deleting the selected access points; and (S4c) determining whether the access point set is empty after treated by step (S4b), wherein if the access point set is empty, completing the division of the plurality of resource cells and then proceeding to step (S5); otherwise proceeding to step (S4a); (S5) generating a CU-DU network mapping table, wherein the step (S5) is performed through steps of: (S5a) calculating an average distance of a control link of each access point in each of the plurality of resource cells according to $l_{k,v}=\underset{m=1}{\overset{n}{.Math.}}{d}_{v,m}/n;$ wherein l.sub.k,v is an average distance of a control link of a vth access point in a kth resource cell; Σ is a summing operation; m is a sequence number of an access point; n is the number of access points in the kth resource cell; and d.sub.v,m is a distance between the vth access point and a mth access point; (S5b) calculating a score value of each access point in each of the plurality of resource cells according to p.sub.k,v=S.sub.k,v/l.sub.k,v, wherein p.sub.k,v is a score value of the v th access point in the kth resource cell; and S.sub.k,v is a throughput of the vth access point in the kth resource cell; (S5c) sorting the score value of each access point in each of the plurality of resource cells from large to small; selecting an access point with a largest score value in each of the plurality of resource cells as an access point for executing a CU function; and taking the remaining access points of each of the plurality of resource cells as access points for executing a DU function; and (S5d) generating a CU-DU network mapping table of a two-dimensional matrix; wherein in a row, if only an element on a leading diagonal is 1, the element on the leading diagonal indicates a traditional access point, and the traditional access point does not fall into the plurality of resource cells; if there are other elements being 1 besides an element on the leading diagonal, the element on the leading diagonal represents an access point that executes a CU function in a resource cell, and the other elements being 1 represent an access point that executes a DU function in a resource cell; and if an element on the leading diagonal is 0, the on the leading diagonal represents an access point that executing a DU function in other resource cells; and (S6) constructing a middlehaul link of each of the plurality of resource cells according to the CU-DU network mapping table, wherein the step (S6) is performed through steps of: (S6a) delivering the CU-DU network mapping table to each of the plurality of access points; (S6b) determining which one of the plurality of resource cells the plurality of access points respectively belongs to, and a function of each of the plurality of access points; wherein if an access point corresponds to a position of a diagonal of the CU-DU network mapping table where an element is 1, a CU function and a DU function of the access point and a backhaul function of the access point with respect to a core network are simultaneously activated; otherwise, the access point only activates a DU function, and an access point of an element being 1 on a diagonal of a row is registered to build a middlehaul link between CU and DU, wherein the row corresponds to a non-zero element on a column in the CU-DU network mapping table corresponding to the access point; (S6c) adjusting, by an access point performing CU function in each of the plurality of resource cells, a deployment parameter of an airborne access point in of the same resource cell; wherein the airborne access point only activates the DU function; and (S6d) orchestrating, by the network resource orchestrator, an available resource for the plurality of resource cells to serve a terminal access device with resource cell as a unit.

2. The method of claim 1, wherein in step (S2a), the current network state comprises network coverage, traffic change, network topology and interference state.

3. The method of claim 1, wherein in step (S3), the trigger value is set according to a congestion rate of a network control link.

4. The method of claim 1, wherein in step (S4b), the threshold is a value in a range of [0,1] set according to the network state; if a hole occurs in a network signal coverage; a network capacity coverage is lower than a capacity requirement of a terminal access device; and the interference becomes worse, the threshold is required to be set to [0,0.5] to make a coverage enhancing effect introduced by a system-level resource management tend to be saturated, wherein the system-level resource management is brought by extending a size of a single resource cell; otherwise the threshold is required to be set to [0.5,1] to make a coverage enhancing effect introduced by a system-level resource management tend to be saturated, wherein the system-level resource management is brought by contracting a size of a single resource cell.

5. The method of claim 1, wherein the step (S6c) comprises: building an objective function of a resource cell to be adjusted, expressed as $f=\frac{C_i}{A_i+\mathrm{.Math.}},$ wherein γ is a utility factor with a value from 0 to 1, representing a proportion of optimization of the objective function; C.sub.i is a network capacity of a ith resource cell in a resource cell set to be adjusted, and $C_i=\underset{j=1}{\overset{M_i}{.Math.}}B.Math.{\log }_2\left(1+{\mathrm{SINR}}_{i,j}\right),$ wherein B is a bandwidth of data transmission; M.sub.i is the number of transmission links served by the ith resource cell; log.sub.2 is a logarithm with base 2; SINR.sub.i,j is a signal to interference plus noise ratio of a jth transmission link of the ith resource cell; $A_i=\frac{\underset{j=1}{\overset{M_i}{.Math.}}{\left(B.Math.{\log }_2\left(1+{\mathrm{SINR}}_{i,j}\right)\right)}^2}{M_i}-{\left(\frac{\underset{j=1}{\overset{M}{.Math.}}B.Math.{\log }_2\left(1+{\mathrm{SINR}}_{i,j}\right)}{M_i}\right)}^2,$ and represents a variance of a transmission rate of all transmission links of the ith resource cell; and ε is a small number to prevent a denominator of the objective function from being 0; and obtaining a deployment parameter of an airborne DU for the resource cell to be adjusted under constraints of flight height, transmit power and energy of the airborne DU.

6. The method of claim 1, wherein in step (S6c), the deployment parameter comprises a location of the airborne access point, a motion mode of the airborne access point and a beam parameter of the airborne access point.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0046] FIG. 1 is a flow chart of a method for enhancing network coverage based on adaptive generation of resource cell according to an embodiment of the present disclosure; and

[0047] FIGS. 2a-d are simulation diagrams of 5G access points and the network resource orchestrator.

DETAILED DESCRIPTION OF EMBODIMENTS

[0048] The present disclosure will be described below in detail with reference to the embodiments and accompanying drawings.

[0049] Referring to an embodiment shown in FIG. 1, a method for enhancing network coverage based on adaptive generation of resource cell is provided, which includes the following steps.

[0050] (S1) A traditional cell in a network is taken as an initial state of the network

[0051] (S2) A correlation matrix in a current network state is generated, where the step (S2) is performed through the following steps.

[0052] A current network state of multiple access points is monitored in real time.

[0053] The current network state includes network coverage, traffic change, network topology and interference state.

[0054] An airborne access point is deployed to an area where a signal/capacity coverage hole occurs by a network resource orchestrator according to a requirement.

[0055] A traffic correlation among the access points is quantified. Two of the access points in the current network state are selected, where when the two of the access points serve the same group of traffic in a local area, the traffic correlation is 1, otherwise the traffic correlation is 0. The traffic correlation between any two of the access points is combined to form a traffic correlation matrix.

[0056] Each of the access points is allowed to broadcast a latitude and longitude information thereof, and receive a latitude and longitude information of the rest of the access points. A distance between any two of access points in the current network state is calculated and then normalized to obtain a network topological matrix of the access points.

[0057] Each of the access points is allowed to receive an interference sent from the rest of the access points. The interference is normalized to generate an interference matrix of the access points.

[0058] The traffic correlation matrix, the network topological matrix and the interference matrix are subjected to linear weighting to generate a relevance matrix consisting of a relevance between access points in the current network state.

[0059] (S3) Whether |ω.sub.i,j.sup.t−ω.sub.i,j.sup.t-1|/ω.sub.i,j.sup.t-1>ε is determined, where if |ω.sub.i,j.sup.t−ω.sub.i,j.sup.t-1|/ω.sub.i,j.sup.t-1>ε, sending the relevance matrix generated in the current network state to the network resource orchestrator followed by performing step (S4), otherwise proceeding to the step (S2). |•| is an operation of taking an absolute value. ω.sub.i,j.sup.t is a relevance between a ith access point and a jth access point in a relevance matrix at a current iteration. ω.sub.i,j.sup.t-1 is a relevance between a ith access point and a jth access point in a correlation matrix at a previous iteration. ε is a preset trigger value.

[0060] The trigger value is set according to a congestion rate of a network control link.

[0061] (S4) Each of the access points is divided into multiple resource cells, where the step (S4) is specifically performed as follows.

[0062] An access point with a minimum sequence number is selected from an access point set formed by the access points and deleted by the network resource orchestrator.

[0063] Access points with a relevance higher than a threshold from the rest access points in the access point set are selected to form a resource cell, and the selected access points are deleted.

[0064] Whether the access point set is empty is determined after treated by the above step, where if the access point set is empty, completing the division of the resource cells and then proceeding to step (S5), otherwise performing the step (S4) again.

[0065] The threshold is a value in a range of [0,1] set according to the network state. If a hole occurs in a network signal coverage. A network capacity coverage is lower than a capacity requirement of a terminal access device. The interference becomes worse, the threshold is required to be set to a low value to make a coverage enhancing effect introduced by a system-level resource management tend to be saturated, where the system-level resource management is brought by extending a size of a single resource cell. Otherwise the threshold is required to be set to a high value to make a coverage enhancing effect introduced by a system-level resource management tend to be saturated, where the system-level resource management is brought by contracting a size of a single resource cell. When the network signal coverage hole occurs, the network capacity coverage is lower than a terminal access device, and the interference becomes worse, the threshold is required to be set to a low value to allow more access points to be divided into a same resource cell, that is, the size of the resource cell is larger and more access points are subject to system-level resource management. Otherwise, the threshold is required to be set to a high value.

[0066] (S5) A CU-DU network mapping table is generated, where the step (S5) is specifically performed as follows.

[0067] An average distance of a control link of each access point in each of the resource cells is calculated according to

[00002]$l_{k,v}=\underset{m=1}{\overset{n}{.Math.}}{d}_{v,m}/n,$

where l.sub.k,v is an average distance of a control link of a vth access point in a kth resource cell. Σ is a summing operation. m is a sequence number of an access point. n is the number of the access points in the kth resource cell. d.sub.v,m is a distance between the vth access point and a mth access point.

[0068] A score value of each access point in each of the resource cells is calculated according to p.sub.k,v=S.sub.k,v/l.sub.k,v, where p.sub.k,v is a score value of the vth access point in the kth resource cell. S.sub.k,v is a throughput of the vth access point in the k th resource cell.

[0069] The score value of each access point in each of the resource cells is sorted from large to small. An access point with a largest score value in each of the resource cells is selected as an access point for executing a CU function. The remaining access points of each of the resource cells are taken as access points for executing a DU function.

[0070] A CU-DU network mapping table of a two-dimensional matrix is generated, where in a row, if only an element on a leading diagonal is 1, the element on the leading diagonal indicates a traditional access point, and the traditional access point does not fall into resource cells. If there are other elements being 1 besides an element on the leading diagonal, the element on the leading diagonal represents an access point that executes a CU function in a resource cell, and the other elements being 1 represent an access point that executes a DU function in a resource cell. If an element on the leading diagonal is 0, the on the leading diagonal represents an access point that executing a DU function in other resource cells.

[0071] (S6) A middlehaul link of each of the resource cells is constructed according to the CU-DU network mapping table, where the step (S6) is specifically performed as follows.

[0072] The CU-DU network mapping table is delivered to each of the access points.

[0073] Which one of the resource cells the access points respectively belongs to, and a function of each of the access points are determined, where if an access point corresponds to a position of a diagonal of the CU-DU network mapping table where an element is 1, a CU function and a DU function of the access point and a backhaul function of the access point with respect to a core network are simultaneously activated; otherwise, the access point only activates a DU function, and an access point of an element being 1 on a diagonal of a row corresponding to a non-zero element on a column in the CU-DU network mapping table corresponding to the access point is registered to build a middlehaul link between CU and DU.

[0074] A deployment parameter of an airborne access point in of the same resource cell is adjusted by an access point performing CU function in each of the resource cells, where the airborne access point only activates the DU function.

[0075] An available resource for the resource cells is orchestrated by the network resource orchestrator to serve a terminal access device with resource cell as a unit.

[0076] The deployment parameter includes a location of the airborne access point, a motion mode of the airborne access point and a beam parameter of the airborne access point.

[0077] A deployment parameter of an airborne access point in of the same resource cell is adjusted by an access point performing CU function in each of the resource cells, where the airborne access point only activates the DU function, performing through the following steps.

[0078] An objective function of a resource cell to be adjusted is built, expressed as

[00003]$f=\frac{C_i}{A_i+\mathrm{.Math.}},$

where γ is a utility factor with a value from 0 to 1, representing a proportion of optimization of the objective function. C.sub.i is a network capacity of a ith resource cell in a resource cell set to be adjusted, and

[00004]$C_i=\underset{j=1}{\overset{M_i}{.Math.}}B.Math.{\log }_2\left(1+{\mathrm{SINR}}_{i,j}\right),$

where B is a bandwidth of data transmission. M.sub.i is the number of transmission links served by the ith resource cell. log.sub.2 is a logarithm with base 2. SINR.sub.i,j is a signal to interference plus noise ratio of a jth transmission link of the ith resource cell.

[00005]$A_i=\frac{\underset{j=1}{\overset{M_i}{.Math.}}{\left(B.Math.{\log }_2\left(1+{\mathrm{SINR}}_{i,j}\right)\right)}^2}{M_i}-{\left(\frac{\underset{j=1}{\overset{M_i}{.Math.}}B.Math.{\log }_2\left(1+{\mathrm{SINR}}_{i,j}\right)}{M_i}\right)}^2,$

and represents a variance of a transmission rate of all transmission links of the ith resource cell. ε is a small number to prevent a denominator of the objective function from being 0.

[0079] A deployment parameter of an airborne DU for the resource cell to be adjusted is obtained under constraints of flight height, transmit power and energy of the airborne DU.

[0080] The technical solutions of the present disclosure will be further described in the following two embodiments.

[0081] Embodiment 1 illustrates an adaptive generation of resource cell of a terrestrial network. Embodiment 2 illustrates an adaptive generation of the resource cells of a three-dimensional dense network.

[0082] Embodiment 1 illustrates a process of enabling adaptive generation of resource cell of a terrestrial network on a stadium with 15.18 square kilometers. As shown in FIG. 2a, 22 access points, all of which are 5G access point, are deployed. An area enclosed by an oval line is an area where the access points deployed, where an area therein enclosed by a rectangular line in center is an area where the network resource orchestrator deployed, and the network resource orchestrator is arranged at center.

[0083] Two rows of the access points are evenly arranged above the area deployed the network resource orchestrator, including a first row and a second row. The first row is evenly arranged four access points numbered 1-4. The second row is evenly arranged four access points numbered 5-8. The access points of the first row and that of the second row in a corresponding location are aligned one by one. Two rows of the access points are evenly arranged below the area deployed the network resource orchestrator, including a third row and a fourth row. The third row is evenly arranged four access points numbered 15-18. The fourth row is evenly arranged four access points numbered 19-22. The access points of the third row and that of the fourth row in a corresponding location are aligned one by one. Three access points, numbered 9, 11 and 13, are evenly arranged at a left of the area deployed the network resource orchestrator. Three access points, numbered 10, 12 and 14, are evenly arranged at a right of the area deployed the network resource orchestrator. The access points have both CU function and DU function, which is activated according to requirements. The network resource orchestrator is configured to control the access points and orchestrate resources. The access points work in a frequency band of 2.6 GHz and a transmit power of 23 dBm.

[0084] In an initial network state, each of the access points serves a terminal access device as independent traditional cell, respectively.

[0085] An average link density of the 22 access points and a downlink traffic of the 22 access points in a current network state are detected in real-time. In this embodiment, when the average link density changes, a correlation matrix of the 22 access points in the current network state is generated, where each element therein represents a correlation between an access point and other access points currently, expressed as follows:

[00006]$\begin{array}{cccccccccccc}& & {\mathrm{AP}}_1& {\mathrm{AP}}_2& {\mathrm{AP}}_3& {\mathrm{AP}}_4& {\mathrm{AP}}_5& {\mathrm{AP}}_6& {\mathrm{AP}}_7& \mathrm{.Math.}& {\mathrm{AP}}_{22}& \\ {\mathrm{AP}}_1& [& 1& 0.89& 0.29& 0.19& 0.95& 0.88& 0.28& \mathrm{.Math.}& 0& ]\end{array}$$\begin{array}{c}\begin{array}{cccccccccccc}& & {\mathrm{AP}}_1& {\mathrm{AP}}_2& {\mathrm{AP}}_3& {\mathrm{AP}}_4& {\mathrm{AP}}_5& {\mathrm{AP}}_6& {\mathrm{AP}}_7& \mathrm{.Math.}& {\mathrm{AP}}_{22}& \\ {\mathrm{AP}}_2& [& 0.88& 1& 0.38& 0.26& 0.86& 0.94& 0.36& \mathrm{.Math.}& 0& ]\end{array}\\ \mathrm{.Math.}\end{array}$$\begin{array}{cccccccccccc}& & {\mathrm{AP}}_1& {\mathrm{AP}}_2& {\mathrm{AP}}_3& {\mathrm{AP}}_4& {\mathrm{AP}}_5& {\mathrm{AP}}_6& {\mathrm{AP}}_7& \mathrm{.Math.}& {\mathrm{AP}}_{22}& \\ {\mathrm{AP}}_5& [& 0.94& 0.87& 0.27& 0.17& 1& 0.89& 0.28& \mathrm{.Math.}& 0.02& ]\end{array}$$\begin{array}{c}\begin{array}{cccccccccccc}& & {\mathrm{AP}}_1& {\mathrm{AP}}_2& {\mathrm{AP}}_3& {\mathrm{AP}}_4& {\mathrm{AP}}_5& {\mathrm{AP}}_6& {\mathrm{AP}}_7& \mathrm{.Math.}& {\mathrm{AP}}_{22}& \\ {\mathrm{AP}}_6& [& 0.85& 0.93& 0.35& 0.22& 0.86& 1& 0.36& \mathrm{.Math.}& 0& ]\end{array}\\ \mathrm{.Math.}\end{array}$$\begin{array}{cccccccccccc}& & {\mathrm{AP}}_1& {\mathrm{AP}}_2& {\mathrm{AP}}_3& {\mathrm{AP}}_4& {\mathrm{AP}}_5& {\mathrm{AP}}_6& {\mathrm{AP}}_7& \mathrm{.Math.}& {\mathrm{AP}}_{22}& \\ {\mathrm{AP}}_{22}& [& 0.03& 0.08& 0.13& 0.06& 0.12& 0.16& 0.18& \mathrm{.Math.}& 1& ].\end{array}$

[0086] Access points with a correlation greater than 0.8 in the correlation matrix are divided into a same resource cell. Such that access point 1, access point 2, access point 5 and access point 6 are divided into the same resource cell, and the rest of 18 access points are not divided into any resource cell, and are configured to serve a terminal access device as a CU independently.

[0087] A CU-DU mapping table is generated by the network resource orchestrator utilizing the correlation matrix in the current network state. Meanwhile, the CU-DU mapping table generated by the network resource orchestrator is delivered to each of the access points. The CU-DU mapping table is expressed as follows:

[00007]$\left[\begin{array}{ccccccccccccccccc}1& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0\\ 0& 1& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0\\ 0& 0& 0& 0& 1& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 1& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0\\ 0& 0& 1& 1& 0& 0& 1& 1& 0& 0& 0& 0& 0& 0& 0& 0& 1\\ 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 0& 0& 1& 0& 0& 0& 0& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 0& 0& 0& 1& 0& 0& 0& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 1& 0& 0& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 1& 0& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 1& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 1& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 1& 0& 0\\ 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 1& 0\\ 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0\end{array}\right].$

[0088] As shown in FIG. 2b, which one of the resource cells the access points respectively belongs to, and a function of each of the access points are determined. A CU function and a DU function of the access point and a backhaul function of the access point with respect to a core network are activated. The access point is registered to CU to build a middlehaul link between CU and DU. In FIG. 2b, access point 1, access point 2, access point 5, and access point 6 are divided into the same resource cell, where access point 2 is the CU of the resource cell, access point 1, access point 5, and access point 6 are the DU of the resource cell, and the rest of the access points serve a terminal access device with resource cell as a unit.

[0089] Embodiment 2 illustrates a process of enabling adaptive generation of the resource cells of a three-dimensional dense network on an outdoor square. As shown in FIG. 2c, 17 access points, all of which are 5G access point, are deployed, including 16 terrestrial access points and an airborne access point. A network resource orchestrator is arranged at center of the FIG. 2c. The 16 terrestrial access points are arranged evenly in four rows and surround the network resource orchestrator, including a first row, a second row, a third row and a fourth row. The first row is evenly arranged four terrestrial access points numbered 1-4. The second row is evenly arranged four terrestrial access points numbered 5-8. The third row is evenly arranged four terrestrial access points numbered 9-12. The fourth row is evenly arranged four terrestrial access points numbered 13-16. The terrestrial access points of the four rows in a corresponding location are aligned one by one. A dotted line therein is a motion trial of the airborne access point numbered 17. All the access points have both CU function and DU function, which is activated according to requirements. The network resource orchestrator is configured to control the access points and orchestrate resources. The access points work in a frequency band of 2.6 GHz. An operating power of the terrestrial access points is 23 dBm. An operating power of the airborne access point is 30 dBm, a circling radius thereof is 20 m-50 m, and a flight height is 100 m-300 m.

[0090] In an initial network state, each of the access points serves a terminal access device as an independent traditional cell, respectively.

[0091] An average link density of the 17 access points and a downlink traffic of the 17 access points in a current network state are detected in real-time. In this embodiment, when the average link density changes, the airborne access point is deployed to an area where a coverage hole occurs by the network resource orchestrator, and a correlation matrix of the 17 access points in the current network state is generated, where each element therein is a correlation between an access point and other access points currently, expressed as follows:

[00008]$\stackrel{\mathrm{.Math.}}{\begin{array}{ccccccccccccc}& & {\mathrm{AP}}_1& {\mathrm{AP}}_2& {\mathrm{AP}}_3& {\mathrm{AP}}_4& {\mathrm{AP}}_5& {\mathrm{AP}}_6& {\mathrm{AP}}_7& {\mathrm{AP}}_8& \mathrm{.Math.}& {\mathrm{AP}}_{17}& \\ {\mathrm{AP}}_3& [& 0.17& 0.34& 1& 0.84& 0.14& 0.29& 0.84& 0.79& \mathrm{.Math.}& 0.91& ]\end{array}}$$\begin{array}{c}\begin{array}{ccccccccccccc}& & {\mathrm{AP}}_1& {\mathrm{AP}}_2& {\mathrm{AP}}_3& {\mathrm{AP}}_4& {\mathrm{AP}}_5& {\mathrm{AP}}_6& {\mathrm{AP}}_7& {\mathrm{AP}}_8& \mathrm{.Math.}& {\mathrm{AP}}_{17}& \\ {\mathrm{AP}}_4& [& 0.10& 0.24& 0.87& 1& 0.99& 0.21& 0.83& 0.87& \mathrm{.Math.}& 0.93& ]\end{array}\\ \mathrm{.Math.}\end{array}$$\begin{array}{ccccccccccccc}& & {\mathrm{AP}}_1& {\mathrm{AP}}_2& {\mathrm{AP}}_3& {\mathrm{AP}}_4& {\mathrm{AP}}_5& {\mathrm{AP}}_6& {\mathrm{AP}}_7& {\mathrm{AP}}_8& \mathrm{.Math.}& {\mathrm{AP}}_{17}& \\ {\mathrm{AP}}_7& [& 0.07& 0.24& 0.80& 0.74& 0.10& 0.30& 1& 0.80& \mathrm{.Math.}& 0.81& ]\end{array}$$\begin{array}{c}\begin{array}{ccccccccccccc}& & {\mathrm{AP}}_1& {\mathrm{AP}}_2& {\mathrm{AP}}_3& {\mathrm{AP}}_4& {\mathrm{AP}}_5& {\mathrm{AP}}_6& {\mathrm{AP}}_7& {\mathrm{AP}}_8& \mathrm{.Math.}& {\mathrm{AP}}_{17}& \\ {\mathrm{AP}}_8& [& 0.04& 0.18& 0.81& 0.85& 0.06& 0.20& 0.85& 1& \mathrm{.Math.}& 0.86& ]\end{array}\\ \mathrm{.Math.}\end{array}$$\begin{array}{ccccccccccccc}& & {\mathrm{AP}}_1& {\mathrm{AP}}_2& {\mathrm{AP}}_3& {\mathrm{AP}}_4& {\mathrm{AP}}_5& {\mathrm{AP}}_6& {\mathrm{AP}}_7& {\mathrm{AP}}_8& \mathrm{.Math.}& {\mathrm{AP}}_{17}& \\ {\mathrm{AP}}_{17}& [& 0.13& 0.27& 0.91& 0.92& 0.11& 0.25& 0.86& 0.86& \mathrm{.Math.}& 1& ].\end{array}$

[0092] Access points with a correlation greater than 0.8 in the correlation matrix are divided into a same resource cell. Such that access point 3, access point 4, access point 7, access point 8 and access point 17 are divided into the same resource cell, and the rest of 12 access points are not divided into any resource cell, and are configured to serve a terminal access device as a CU independently.

[0093] A CU-DU mapping table is generated by the network resource orchestrator utilizing the correlation matrix in the current network state. Meanwhile, the CU-DU mapping table is delivered to each of the access points. The CU-DU mapping table is expressed as follows:

[00009]$\left[\begin{array}{cccccccccccccccccccccc}0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0\\ 1& 1& 0& 0& 1& 1& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0\\ 0& 0& 1& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0\\ 0& 0& 0& 1& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 1& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 0& 1& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 0& 0& 1& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 0& 0& 0& 1& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 1& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 1& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 1& 0& 0& 0& 0& 0& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 1& 0& 0& 0& 0& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 1& 0& 0& 0& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 1& 0& 0& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 1& 0& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 1& 0& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 1& 0& 0& 0\\ 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 1& 0& 0\\ 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 1& 0\\ 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 0& 1\end{array}\right].$

[0094] As shown in FIG. 2d, which one of the resource cells the access points respectively belongs to, and a function of each of the access points are determined. A CU function and a DU function of the access point and a backhaul function of the access point with respect to a core network are activated. The access point is registered to CU to build a middlehaul link between CU and DU. In FIG. 2d, access point 3, access point 4, access point 7, access point 8, and access point 17 are divided into the same resource cell, where access point 7 is the CU of the resource cell, access point 3, access point 4, access point 8, and access point 17 are the DU of the resource cell, and the rest of the access points serve a terminal access device with resource cell as a unit.