Control method, control device, wireless communication system, and program

Abstract

Provided is a control method executed by a control device which determines a relief antenna for relieving communication in a failure area and a tilt of the relief antenna, the control method comprising a relief antenna selection step of acquiring an antenna row obtained by arranging a plurality of antennas in a descending order of priority on the basis of a policy function and sequentially selecting (an) antenna(s) from the head of the antenna row in the descending order as the relief antenna until a designated condition is satisfied, and a tilt calculation step of calculating a tilt of the relief antenna(s) selected in the relief antenna selection step.

Description

Technical Field

[0001] The present invention relates to a technique for relieving an area where communication becomes unavailable due to a failure of a radio base station in a mobile communication network.

Background Art

[0002] When a radio base station fails due to a natural disaster or the like, a communication failure occurs in the area covered by the base station. Since high availability is required for wireless communication, recovery within a short time of about several minutes is required.

[0003] On the other hand, there is a technique for recovering from a failure by changing the depression angle (tilt) of an antenna of a base station around an area where a failure has occurred (failure area) to cover (relieve) the failure area. This technique is called Cell Outage Compensation (COC).

Prior Art Documents

Non-Patent Documents

[0004]

Non-Patent Document 1

Non-Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0005] As a prior art for the relief of a failure area, there is a prior art based on COC (Non-Patent Documents 1 and 2). However, these prior arts have a problem that when a failure area occurs due to a failure of a radio base station or the like, it is impossible to select an appropriate relief antenna according to the situation.

[0006] The present invention has been made in view of the above points, and an object thereof is to provide a technology that enables selection of an appropriate relief antenna according to the situation when a failure area occurs due to a failure of a radio base station or the like.

Means for Solving the Problems

[0007] According to the disclosed technology, there is provided a control method executed by a control device that determines a relief antenna for relieving communication in a failure area and its tilt, a relief antenna selection step of obtaining an antenna array in which a plurality of antennas are arranged in descending order of priority by a policy function, and selecting an antenna as a relief antenna in order from the head of the antenna array until a predetermined condition is satisfied; a tilt calculation step of calculating the tilt of each relief antenna selected by the relief antenna selection step, and The aforementioned policy function has parameters pre-trained using rewards based on improving coverage of the faulty area and suppressing overlap in coverage areas between rescue antennas. a control method is provided.

Effects of the Invention

[0008] According to the disclosed technology, there is provided a technology that enables selection of an appropriate relief antenna according to the situation when a failure area occurs due to a failure of a radio base station or the like.

Brief Description of the Drawings

[0009] [Figure 1] It is a configuration diagram of a system in an embodiment of the present invention. [Figure 2] It is a configuration diagram of a radio base station control device. [Figure 3] It is a diagram showing dmax and dmin. [Figure 4] This is a diagram showing two sectors. [Figure 5] This is a flowchart illustrating the operation of a wireless base station control device. [Figure 6] This is a flowchart illustrating the operation of a wireless base station control device. [Figure 7] This figure shows an example of the device's hardware configuration. [Modes for carrying out the invention]

[0010] Hereinafter, embodiments of the present invention (this embodiment) will be described with reference to the drawings. The embodiments described below are merely examples, and the embodiments to which the present invention is applied are not limited to the embodiments described below.

[0011] (Definition of terms) First, we will explain the definitions of "antenna" and "regional mesh" as used in this specification.

[0012] (1) Antenna A base station consists of multiple antennas, each responsible for communication coverage in a specific direction. Each antenna has a set tilt, and different tilts result in different coverage areas.

[0013] Since there are cases where only some of the antennas at a base station fail, this embodiment considers antenna-level failures rather than base station-level failures.

[0014] (2) Regional Mesh A regional mesh is defined by the Administrative Management Agency (now the Ministry of Internal Affairs and Communications) as "an area divided into meshes of approximately the same size based on latitude and longitude." In this embodiment, a regional mesh is referred to simply as a mesh.

[0015] In this embodiment, all areas, including fault areas and areas covered by rescue antennas, are considered in terms of mesh units.

[0016] There are multiple patterns for the area per mesh, and a larger area indicates a coarser mesh size. Examples include standard mesh (side length approximately 1 km), half mesh (side length approximately 500 m), and 1 / 8 mesh (side length approximately 125 m).

[0017] The coarser the mesh granularity, the less computation is required to calculate the coverage area, thus reducing the time required for coverage area estimation. The mesh granularity can be predetermined, for example, by the operator of the wireless base station control device 100 in this embodiment.

[0018] (Regarding the issues) Here, we will describe in detail the prior art and its problems with respect to the technology of this embodiment. In the following description, the contents disclosed in Non-Patent Documents 1 and 2 are publicly known, but the description of their problems is not publicly known.

[0019] First, I will explain the COC that forms the basis of the technology disclosed in Non-Patent Documents 1 and 2. The COC consists of the following two steps.

[0020] <Rescue Antenna Selection Steps> First, an antenna to be used for rescue is selected from among the antennas available around the faulty area. This step is called the rescue antenna selection step, and the selected antenna is called the rescue antenna. In the rescue antenna selection step, an algorithm is used to select an antenna that is most likely to cover the faulty area.

[0021] <Tilt Optimization Step> Next, the tilt of the rescue antennas is optimized. This step is called the tilt optimization step. In the tilt optimization step, the tilt of the rescue antennas is controlled so that a high coverage rate is obtained for the faulted area and the overlap of the coverage areas of the rescue antennas is minimized.

[0022] (1) Conventional technology and its problems 1 Non-patent document 1 proposes a COC (Corporate Overlap) method aimed at maximizing the number of user connections in a faulty area. In this method, all antennas adjacent to the faulty area are selected as rescue antennas. As a result, the number of rescue antennas becomes excessive, and the solution space to be searched when optimizing the tilt of the rescue antennas becomes large.

[0023] As a result, the accuracy of the optimized solution may decrease, potentially leading to overlapping coverage areas between rescue antennas. This overlapping coverage area can cause radio interference, resulting in a deterioration of communication quality.

[0024] The size of the fault area and the antenna placement vary depending on the situation, and appropriate antenna selection is required for each situation. For example, it is not necessary to control the tilt of all antennas when relieving a small fault area. However, the technology disclosed in Non-Patent Document 1 does not allow for appropriate antenna selection depending on the situation.

[0025] (2) Conventional technology and its challenges 2 In the method disclosed in Non-Patent Document 2, 10 combinations of rescue antennas are prepared in advance from among the antennas adjacent to the faulty area, the tilt optimization is sought for each pattern, and the best solution is adopted.

[0026] Unlike the method disclosed in Non-Patent Document 1, this method selects an appropriate rescue antenna from the antennas surrounding the faulted area. However, this method assumes that the antennas are aligned in a grid at equal intervals and only targets single antenna failures; it cannot select an appropriate antenna for faulted areas that span multiple antennas.

[0027] (3) Conventional technology and its challenges 3 The method disclosed in Non-Patent Document 2 pre-calculates the optimal solution for multiple combinations of rescue antennas. However, in large fault areas spanning multiple antennas, the number of possible rescue antenna combinations becomes enormous, making it difficult to pre-calculate the optimal solution for every possible combination.

[0028] (Technical features (key points) related to the embodiment) In the technology according to this embodiment, in order to solve the above problems, the wireless base station control device 100, which will be described later, selects an appropriate rescue antenna from the antennas of surrounding base stations for the purpose of covering a fault area due to a base station failure, etc., and then controls the tilt of the selected antenna. In particular, the technology according to this embodiment can solve the above problems by having the following features (points). Note that the points described below are points in the technology of the embodiment, and it is not necessary to include all of the points as an invention.

[0029] <Point 1> From among the antennas surrounding the faulty area, a rescue antenna is selected based on the positional relationship between each antenna and the faulty area, as well as the azimuth of the faulty area in the coordinate system of each antenna.

[0030] <Point 2> To select a rescue antenna, a combinatorial optimization method based on Deep Reinforcement Learning (DRL) is used to find an appropriate rescue antenna with minimal computational effort from a vast number of possible combinations. Note that in this embodiment, reinforcement learning methods other than deep reinforcement learning can be used.

[0031] (Effects of the technology according to the embodiment) The technology according to this embodiment, having the points described above, produces the following effects.

[0032] <Effect 1> Point 1 prevents the overlap in coverage areas between the selected rescue antennas, as explained in Problem 1.

[0033] <Effect 2> Point 1 allows for the selection of a rescue antenna regardless of constraints regarding the placement of antennas or failure patterns. This enables rescue even in cases of failures spanning multiple antennas, thus solving Problem 2.

[0034] <Effect 3> Point 2 allows us to determine the appropriate rescue antenna in a practical amount of time without having to calculate the optimal tilt solution for all possible combinations of rescue antennas in advance. This solves Problem 3.

[0035] (Overall system configuration) Figure 1 shows an example of the overall configuration of the wireless communication system in this embodiment. As shown in Figure 1, the system according to this embodiment has a configuration in which a plurality of wireless base stations 10 and a wireless base station control device 100 are connected to a network 200. The network 200 is, for example, a network including a mobile core network. Communication terminals exist under each wireless base station, and each communication terminal communicates with the wireless base station wirelessly.

[0036] The configuration and operation of the wireless base station control device 100 according to this embodiment will be described in detail below.

[0037] (Example configuration of wireless base station control device 100) Figure 2 shows a configuration diagram of the wireless base station control device 100 in an embodiment of the present invention. As shown in Figure 2, the wireless base station control device 100 includes an input receiving unit 110, a data processing unit 120, and a wireless base station control unit 130.

[0038] Furthermore, as shown in Figure 2, the data processing unit 120 includes a rescue priority determination unit 121, a rescue antenna selection unit 122, a parameter database unit 123, a tilt calculation unit 124, and a fault area aggregation unit 125. The wireless base station control unit 130 includes a receiving unit 131 and a transmitting unit 132.

[0039] The wireless base station control device 100 may be a single physical device or a system consisting of multiple physical devices. For example, the "input receiving unit 110 + data processing unit 120" may consist of one device, and the wireless base station control unit 130 may consist of one device. The "input receiving unit 110 + data processing unit 120" or the "data processing unit 120" may also be referred to as the control device. The functions of each part are as follows.

[0040] <Wireless base station control unit 130> The receiving unit 131 and transmitting unit 132 of the wireless base station control unit 130 are each capable of communicating with the wireless base station. The receiving unit 131 receives information from the wireless base station, and the transmitting unit 132 transmits information to the wireless base station.

[0041] <Input reception unit 110> The input receiving unit 110 receives "information about each radio base station," "information about faulty antennas (faulty antennas) within a faulty station," and "important area information" from the radio base station control unit 130.

[0042] "Information regarding each wireless base station" includes, for example, the following information:

[0043] • The location (latitude, longitude), height, and direction of all antennas installed at the wireless base station. • Beam width in the vertical and horizontal directions of each antenna beam • Tilt settings for each antenna

[0044] "Information regarding a faulty antenna within a faulty station (faulty antenna)" includes, for example, the following information. Note that the "area" below is defined as a set of meshes. Furthermore, the "area covered by the faulty antenna" below is defined as the faulty area.

[0045] • Presence or absence of important facilities in the area covered by the malfunctioning antenna • The area covered by the faulty antenna (i.e., the area affected by the malfunction) • Center of gravity of the obstruction area • Number of active users communicating with the faulty antenna • Total traffic handled by the faulty antenna The "Important Area Information" is as follows:

[0046] • Important areas are defined as regional meshes containing important facilities, or sets of meshes designated in advance by the operator. These meshes are managed as important area information by embedding flag information indicating important areas within them.

[0047] <Fault Area Aggregation Unit 125> The fault area aggregation unit 125 receives information from the input reception unit 110 and generates a fault area by aggregating adjacent fault areas into a single fault area. A single fault area formed by aggregating multiple fault areas may also be called an aggregated fault area.

[0048] <Relief Priority Determination Unit 121> The rescue priority determination unit 121 receives information from the input reception unit 110 and the fault area aggregation unit 125 and determines the rescue priority for the fault areas. The rescue priority is defined based on the characteristics of the fault areas. Rescue is considered one by one, starting with the fault areas with the highest rescue priority.

[0049] <Rescue Antenna Selection Section 122> The rescue antenna selection unit 122 selects a rescue antenna from base stations in the vicinity of the fault area to be rescued. A combinatorial optimization algorithm based on DRL is used for selecting the rescue antenna. Specifically, in this embodiment, the combinatorial optimization algorithm disclosed in Reference 1 "I. Bello et al., "Neural combinatorial optimization with reinforcement learning," arXiv preprint arXiv:1611.09940, 2016." is used.

[0050] In this combinatorial optimization algorithm, a reinforcement learning (RL) agent learns through training data to develop a strategy for outputting the optimal sequence of elements that maximizes the reward when given a given sequence of elements as input.

[0051] The rescue antenna selection unit 122 uses a pointer network (Reference 2: O. Vinyals et al., "Pointer networks," in NeurIPS, vol. 28, 2015.) as a function representing this policy (policy function), similar to the technology disclosed in Reference 1.

[0052] The selection algorithm of the rescue antenna selection unit 122 takes information on candidate rescue antennas as input and outputs one or more rescue antennas from the candidates according to the policy function described above. In addition, the parameters of the policy function are learned in advance from training data so that an appropriate antenna is selected in the selection algorithm. This learning is called pre-training. The parameters of the policy function are learned so as to maximize the reward obtained by the action of the policy function (in this case, selection of permutations of rescue antennas).

[0053] The system's behavior during pre-training and its operation after pre-training will be described later.

[0054] <Parameter Database Section 123> The parameter database unit 123 manages the parameters of the algorithm used by the rescue antenna selection unit 122 (specifically, the parameters of the policy function, etc.). In other words, the parameter database unit 123 stores the parameters calculated by the rescue antenna selection unit 122 and transmits the stored parameters in response to inquiries from the rescue antenna selection unit 122.

[0055] <Tilt calculation unit 124> The tilt calculation unit 124 determines a tilt for the rescue antenna selected by the rescue antenna selection unit 122 that provides high coverage of the affected area and minimizes overlap between the coverage areas of the rescue antennas.

[0056] In calculating the tilt, a radio wave propagation range estimation algorithm and a tilt calculation algorithm are used. In this embodiment, the algorithms disclosed in Reference 3, "Masahito Iwamoto, Akihito Suzuki, Masahiro Kobayashi, "Antenna Tilt Control Method by Particle Swarm Optimization for Early Recovery of Faulted Areas," IEICE Technical Report, vol. 121, no. 324, IN2021-24, pp. 1-6, January 2022," are used. However, other radio wave propagation range estimation algorithms and other tilt calculation algorithms may also be used.

[0057] The tilt calculation unit 124 estimates the radio wave propagation range for each combination of the rescue antenna and its tilt, using the radio wave propagation range estimation algorithm disclosed in Reference 3. Specifically, the radio wave propagation range is calculated using a geometric approximation with sectors, and the set of meshes included in the calculated range is defined as the coverage area corresponding to that antenna and tilt combination.

[0058] The tilt calculation unit 124 calculates a combination of tilts for a given one or more rescue antennas that covers the fault area and minimizes the overlap of area coverage between rescue antennas, using the tilt calculation algorithm disclosed in Reference 3.

[0059] Specifically, an objective function is defined using the tilt of each rescue antenna as input, and the combination of tilts that yields the best objective function value is adopted as the calculation result. The objective function is defined by combining "coverage rate of the affected area," "(if a critical area exists) coverage rate of the critical area," and "coverage area size of the rescue antenna." The coverage rate of the affected area, the coverage rate of the critical area, and the coverage area size of the rescue antenna are each calculated using the coverage area calculated by the radio wave propagation range algorithm.

[0060] The tilt calculation unit 124 transmits the determined tilt to the radio base station control unit 130, causing the radio base station to change the tilt of the rescue antenna.

[0061] <Specific example of the operation of the tilt calculation unit 124> The following describes a specific example of the operation of the tilt calculation unit 124.

[0062] First, the tilt calculation unit 124 estimates the radio wave propagation range for each combination of the rescue antenna and tilt. The rescue antenna has a settable tilt range. Within this settable range, the tilt can be changed at a granularity of predetermined angular units, such as 0.5° or 1.0° increments. The tilts prepared for each predetermined angular unit within the settable range are called "each tilt". Then, the radio wave propagation range is estimated independently for all combinations of each rescue antenna and each tilt.

[0063] For example, if the adjustable tilt range of a rescue antenna is 1° to 15°, and the predetermined angle unit is 1°, the radio wave propagation range for that rescue antenna is estimated for each of the 15 tilts {1°, 2°, ... 14°, 15°} using the method described later.

[0064] Generally, the precise radio wave propagation range is calculated using three dimensions: latitude, longitude, and altitude. However, to ensure rapid rescue, it is necessary to estimate the radio wave propagation range in a shorter time. Therefore, in this embodiment, only the area covered by radio waves at 1.5m above the ground is considered as the radio wave propagation range. Here, we assume a communication terminal held in the user's hand and use 1.5m above the ground as an example, but other heights besides 1.5m may also be used.

[0065] This approach reduces the accuracy of the estimation results because the radio wave propagation range is not estimated in three dimensions, but it is expected to improve performance in terms of computation speed.

[0066] Furthermore, the tilt calculation unit 124 in this embodiment calculates the radio wave propagation range using a geometric approximation based on a sector. The specific calculation method is as follows.

[0067] Here, the position of the rescue antenna is (latitude lat, longitude lon), the height of the antenna is h, and the center azimuth of the antenna is az center Let the tilt of the antenna be x, and the vertical and horizontal beam widths (°) of the antenna beam be α and β, respectively. Note that the height h of the antenna is the height from 1.5 m above the ground.

[0068] Also, regarding the adjustable range of the tilt x, let the minimum value be x min and the maximum value be x max .

[0069] At this time, the range of the angle of the antenna beam in the vertical direction is [v min , v max . v min , v max are obtained by the following formulas respectively.

[0070] v min = max{x - α / 2, x min}}, v max = min{x + α / 2, x max} Let the radii of the two sectors used for approximating the radio wave propagation range be d min , d max , then they are obtained by the following formulas respectively.

[0071] d min = h / tan(v max ), d max = h / tan(v min ) When the d min , d max obtained by the above formula are illustrated, it is as shown in Figure 3.

[0072] Let the start azimuth angle and the end azimuth angle of the above two sectors be az start , az end respectively, then these are obtained by the following formulas.

[0073] az start = az center + β / 2, az end = azcenter -β / 2 As stated above, the starting and ending azimuth angles of the two sectors are the same. However, az start az end The domain is 0 ≤ az start ,az end Assume <360°. According to the above formula, az start az end If the calculated value falls outside the defined domain, the remainder modulo 360 is used.

[0074] The center is point (lat, lon), and the radius is d. min , the central angle is az start -az end So, az start from az end Let sector A be a sector facing the direction of . Similarly, let the center be point (lat, lon) and radius be d. max , the central angle is az start -az end So, az start from az end Let sector B be the sector facing the direction of .

[0075] The tilt calculation unit 124 determines sectors A and B through the above calculations, and uses the area within sector B that is not included in sector A as the estimated radio wave propagation range. Sectors A and B are illustrated in Figure 4. Figure 4 shows the two sectors viewed from above.

[0076] The tilt calculation unit 124 performs the above calculations for each rescue antenna and for each tilt to estimate the radio wave propagation range.

[0077] Next, the tilt calculation algorithm for calculating the tilt of the rescue antennas will be described. The tilt calculation unit 124 calculates a combination of tilts for each rescue antenna in question that covers the faulted area and minimizes the overlap of area coverage between the rescue antennas.

[0078] Here, an objective function is defined using the tilt of each rescue antenna as input, and the combination of tilts that yields the best objective function value is adopted as the calculation result. As mentioned above, the objective function is defined by combining "coverage rate of the affected area," "(if a critical area exists) coverage rate of the critical area," and "coverage area size of the rescue antenna." The coverage rate of the affected area, the coverage rate of the critical area, and the coverage area size of the rescue antenna are calculated using the estimated radio wave propagation range corresponding to each antenna and tilt.

[0079] However, as a constraint, an upper limit is set on the number of meshes in the regional area for the coverage area of ​​each antenna, and the tilt is determined within the range that does not exceed this limit.

[0080] In the following sections, we will explain how to calculate the coverage rate of the affected area, the coverage rate of the critical area, and the coverage area size of the rescue antenna, and then explain how to calculate the tilt combination.

[0081] First, let's explain how to calculate the coverage rate of the affected area.

[0082] Here, the radio wave propagation range described above is defined as the cover area. The area originally covered by the rescue antenna is defined as the base area. The base area may be acquired as base station information, or the tilt calculation unit 124 may estimate the base area using the radio wave propagation range estimation method described above. The area consisting of the fault area and the base area is defined as the fault area to be rescued.

[0083] The tilt calculation unit 124 compares the coverage area with the area of ​​damage requiring relief, calculates the proportion of the area of ​​damage requiring relief that is included in the coverage area, and defines this as the damage area coverage rate. The coverage rate is calculated using the number of meshes in the regional mesh.

[0084] If the area within the coverage area that is also included in the area of ​​faults requiring relief is defined as the area of ​​faults that can be relieved, then when using the mesh as described above, the tilt calculation unit 124 calculates the fault area coverage rate using the following formula.

[0085] (Disaster area coverage rate) = (Number of meshes in the disaster area where remediation is possible) / (Number of meshes in the disaster area requiring remediation) Next, I will explain how to calculate the coverage rate of important areas.

[0086] If a critical area exists within a damaged area, the combined area of ​​the critical area and the area directly beneath the user's feet will be defined as the critical area to be rescued.

[0087] The tilt calculation unit 124 compares the coverage area with the critical areas to be relieved, calculates the proportion of the critical areas to be relieved that are included in the coverage area, and defines this as the critical area coverage rate.

[0088] In other words, if the areas within the coverage area that are also included in the critical areas subject to relief are defined as critical areas that can be relieved, the tilt calculation unit 124 calculates the critical area coverage rate using the following formula.

[0089] (Critical area coverage rate) = (Number of meshes in critical areas that can be rescued) / (Number of meshes in critical areas that are subject to rescue) Next, the method for calculating the coverage area size of the rescue antenna will be explained. The tilt calculation unit 124 calculates the coverage area size of each antenna as follows, using the ratio of the calculated number of meshes in the coverage area to the upper limit of the number of meshes in the coverage area.

[0090] (Antenna coverage area size) = ((Number of meshes in the coverage area)) / ((Upper limit of the number of meshes in the coverage area)) The tilt calculation unit 124 calculates this value for each rescue antenna and uses the average value as the (coverage area size of the rescue antenna) as the penalty term of the objective function. This prevents each rescue antenna from unnecessarily expanding its coverage area in order to improve coverage rate and reduces overlap in coverage areas between rescue antennas.

[0091] Next, we will explain how to calculate the tilt. First, define the objective function as follows.

[0092] Let the objective function be α (coverage rate of critical areas) + β (coverage rate of damaged areas) - γ (coverage area size of the rescue antenna).

[0093] α, β, and γ are parameters that weight each item and are set in advance. For example, if the priority during rescue is "maximizing coverage of critical areas," "maximizing coverage of affected areas (excluding critical areas)," and "minimizing overlap of covered areas," then α > β > γ (e.g., α=10, β=5, γ=1) may be used.

[0094] Next, the tilt calculation unit 124 maximizes the objective function using the particle swarm optimization method. The particle swarm optimization method itself is an existing technique. Particle swarm optimization is a metaheuristic method for finding approximate solutions to combinatorial optimization problems. It prepares multiple candidate solutions to the optimization problem, called particles, and updates each particle while sharing the objective function value with each other. Note that using the particle swarm optimization method to maximize the objective function is just one example.

[0095] Here, "all combinations of the tilts of the relief antennas" corresponds to one particle. The i-th particle is x i Let x i The tilt of the j-th relief antenna is θ ij (1 <j<m)とすると、x i =[θ i1 ,θ i2 ,...,θ im ]

[0096] In the particle swarm optimization method, each particle x i Update it according to the following formula.

[0097] Particle update formula: x i (t+1)=x i (t)+v i (t+1) Formula for updating the particle renewal rate: v i (t+1)=wv i (t+1)+c1r1(x i p (t)-x i (t))+c2r2(x g (t)-x i (t)) However, the meaning of each symbol (variable, etc.) is as follows:

[0098] x i (t): Particle at update count t v i (t): Particle velocity at update number t x i p (t): Best solution for particle i up to update number t. x g (t): Best solution for all particles up to update number t w and c1 / c2 are coefficients, and r is a random number.

[0099] The update process ends when the objective function value stops increasing or when a predetermined number of updates is reached. The particle that yields the highest objective function value among all particles obtained up to this point is adopted as the solution.

[0100] In addition to the above processing, if prioritizing coverage of critical areas is desired, the particle with the highest coverage rate of those critical areas among the obtained particles may be selected.

[0101] (Behavior during pre-training) The following describes in detail an example of operation during pre-training in the wireless base station control device 100 having the above configuration, following the steps in the flowchart of Figure 5. As mentioned above, before performing fault relief in the faulted area, the wireless base station control device 100 trains the parameters of the policy function in the selection algorithm through pre-training.

[0102] <s1> In S1, the rescue antenna selection unit 122 receives input of training data necessary for pre-training from the wireless base station control unit 130. The training data consists of "fault area" and "information on candidate rescue antennas". Here, let N be the set of candidate rescue antennas.

[0103] More specifically, the rescue antenna selection unit 122 receives the following information for each antenna n included in N: "the height of the antenna, the distance between the antenna and the center of gravity g of the obstructed area, and the difference in azimuth angles between the center direction of the antenna and the direction from the antenna to the center of gravity g."

[0104] Regarding training data, the system operator prepares multiple fault area occurrence scenarios in advance and creates training data for each scenario. This training data is then input through the wireless base station control unit 130.

[0105] <s2> In S2, the rescue antenna selection unit 122 selects one rescue antenna from the set N. That is, the rescue antenna selection unit 122 inputs the information of N obtained in S1 into the policy function during parameter learning and outputs a permutation of N arranged in order of priority for selection as a rescue antenna. Then, in the loop from S2 to S4, it selects one rescue antenna at a time from the top of that permutation.

[0106] <s3> In S3, the tilt calculation unit 124 optimizes the tilt of all antennas selected in S2 up to that point using a tilt calculation algorithm. Specifically, the rescue antenna selection unit 122 notifies the tilt calculation unit 124 of the selected rescue antenna, and the tilt calculation unit 124 optimizes the tilt of the rescue antenna.

[0107] <s4> In S4, the rescue antenna selection unit 122 determines whether to terminate the antenna selection. Here, the rescue antenna selection unit 122 first receives the optimization results from the tilt calculation unit 124. These results are used to determine the tilt of each antenna during optimization, the coverage rate of the fault area, and the value of the optimization objective function.

[0108] The rescue antenna selection unit 122 terminates antenna selection and proceeds to S5 if the coverage rate of the faulty area during optimization becomes 100%, or if all rescue antenna candidates are selected. If the antenna selection is not terminated, it returns to S2 and continues antenna selection. Note that "100% coverage rate of the faulty area" is an example. Alternatively, "X% coverage rate of the faulty area" may be used, where X is a predetermined integer less than 100.

[0109] <s5> If antenna selection is complete, in S5, the rescue antenna selection unit 122 updates the parameters of the selection algorithm based on the rewards for the antennas selected so far.

[0110] The above reward is an indicator for determining whether the selected antenna is appropriate as a rescue antenna. In this embodiment, the reward is defined as the value of the optimization objective function of the tilt calculation unit 124, and the value calculated in S3 is used as the reward. However, a value other than the value of the optimization objective function of the tilt calculation unit 124 may also be used as the reward.

[0111] The parameter update process in DRL itself is an existing technique, and for example, one can use the methods disclosed in Reference 1 "I. Bello et al., "Neural combinatorial optimization with reinforcement learning," arXiv preprint arXiv:1611.09940, 2016." or Reference 4 "R. Williams et al., "Simple statistical gradient-following algorithms for connectionist reinforcement learning," Mach. Learn., vol. 8, no. 3-4, p.229-256, 1992."

[0112] In pre-training, the optimal parameters of the policy function are obtained by repeatedly selecting a rescue antenna and providing reward feedback according to the procedure shown in Figure 5.

[0113] <S6、S7> In S6 and S7, the rescue antenna selection unit 122 performs a pre-training completion determination process. Specifically, it is as follows.

[0114] The rescue antenna selection unit 122 considers the process from S2 to S5 as one learning cycle and counts how many times this learning cycle has been performed. In S6, the rescue antenna selection unit 122 determines whether the number of learning cycles has reached a preset value or whether the parameter updates have converged, and terminates the learning cycle if either of these conditions is met.

[0115] When the learning process is complete, the rescue antenna selection unit 122 stores the parameters obtained through the learning process in the parameter database unit 123.

[0116] If the learning process does not complete, the selected antenna will be reset, and the process will return to S2 to continue learning.

[0117] (Operation after pre-training) After completing the pre-training of the selection algorithm, an example of its operation when rescuing a faulted area will be explained in detail following the steps in the flowchart in Figure 6.

[0118] <s11> In S11, the fault area aggregation unit 125 receives information necessary for relief from the wireless base station control unit 130. The information necessary for relief includes fault area information, critical area information, and information on potential relief antennas.

[0119] <s12> In S12, the fault area aggregation unit 125 combines adjacent fault areas into one. That is, for each fault area, the fault area aggregation unit 125 redefines adjacent areas as a single fault area. There may be upper limits on the number of areas to be aggregated and the size of the areas.

[0120] At this time, the fault area aggregation unit 125 aggregates the information into a single fault area, including the number of active users who were communicating and the total amount of traffic being processed. <13, S14> If there are multiple fault areas that have gone through the S12 process, the rescue priority determination unit 121 sets a rescue priority and performs rescue operations one by one, starting with the fault area with the highest rescue priority.

[0121] In S13, the rescue priority determination unit 121 checks whether there are any unresolved fault areas, and terminates the process if there are none. If there are unresolved fault areas, the process proceeds to S14, where the rescue priority determination unit 121 determines the rescue priority for each unresolved fault area and selects the fault area with the highest rescue priority among the unresolved fault areas.

[0122] The method by which the rescue priority determination unit 121 determines the rescue priority for each fault area is not limited to a specific method, but for example, the rescue priority can be determined based on indicators corresponding to the urgency of establishing communication, such as "the presence or absence of critical facilities in the fault area," "the size of the critical area within the fault area," "the size of the fault area," "the number of active users communicating with the faulty antenna associated with the fault area," and "the total amount of traffic handled by the faulty antenna associated with the fault area." The specific indicators can be determined as appropriate by the system operator.

[0123] For example, if the rescue priority determination unit 121 determines the rescue priority based on the "number of active users communicating with the faulty antenna," it will set a higher rescue priority the more active users there are. The rescue priority determination unit 121 may also determine the rescue priority using one of the following: "presence or absence of important facilities in the faulty area," "size of the important area within the faulty area," "size of the faulty area," "number of active users communicating with the faulty antenna associated with the faulty area," or "total traffic volume processed by the faulty antenna associated with the faulty area," or it may use a combination of these.

[0124] <s15> In S15, the rescue antenna selection unit 122 selects a rescue antenna for the faulted area from the candidates using a policy function whose parameters have been optimized through pre-training. The procedure for selecting the rescue antenna is the same as the selection procedure during pre-training, but the reward is not calculated and the parameters are not updated. However, the reward may be calculated and the parameters updated in the actual rescue antenna selection phase after pre-training. The specific procedure for S15 is as follows.

[0125] The rescue antenna selection unit 122 receives information about candidate rescue antennas N into the policy function, and the policy function outputs a permutation of the candidate rescue antennas. This permutation is a sequence in which the candidate rescue antennas are arranged from the top in order of priority.

[0126] However, this permutation may also be a column in which rescue antenna candidates are arranged from the top in order of increasing priority. In this case, the rescue antenna selection described below should be performed from the end of the column. The same applies during pre-training.

[0127] Note that "arranging rescue antenna candidates in order of lowest priority from the beginning and selecting a rescue antenna from the end of the column" and "arranging rescue antenna candidates in order of highest priority from the beginning and selecting a rescue antenna from the beginning of the column" are the same thing.

[0128] As information about N, for each antenna n included in N, the antenna height h n , the distance d between the antenna and the center of gravity g of the obstructed area n δ is the difference in azimuth angles between the center direction of the antenna and the direction from the antenna to the center of gravity g. n This information is used to select a rescue antenna. By selecting a rescue antenna based on this information, it is possible to select an appropriate rescue antenna for various rescue antenna configurations and failure patterns.

[0129] The rescue antenna selection unit 122 selects rescue antennas one by one from the beginning of the above permutation. Then, each time the rescue antenna selection unit 122 selects a rescue antenna, it uses a tilt calculation algorithm to calculate the coverage rate of the fault area when the tilt of all rescue antennas selected so far is optimized.

[0130] The rescue antenna selection unit 122 terminates the rescue antenna selection and outputs all rescue antennas selected so far when it achieves 100% coverage of the faulty area or when it has selected all candidate rescue antennas. Note that "100% coverage of the faulty area" is just an example. Alternatively, X may be a predetermined integer less than 100, and "X% coverage of the faulty area" may be used.

[0131] In the selection algorithm according to this embodiment, instead of designating all antennas around the fault area as rescue antennas, the number of rescue antennas is narrowed, thus preventing a decrease in the accuracy of the tilt optimization solution due to an excess of rescue antennas. As a result, it is possible to prevent overlap in the coverage areas of rescue antennas.

[0132] Furthermore, since a combinatorial optimization algorithm based on DRL (Reference 1) is used as the selection algorithm, it is possible to find an appropriate rescue antenna in a practical amount of time for an unknown "fault area" and "candidate rescue antennas located in its vicinity."

[0133] <s16> In S16, the tilt calculation unit 124 estimates the radio wave propagation range for each candidate rescue antenna tilt based on the radio wave propagation range estimation algorithm.

[0134] <s17> In S17, the tilt calculation unit 124 calculates the tilt of each rescue antenna. Here, the radio wave propagation range obtained in S16 is used to obtain an appropriate tilt based on the tilt calculation algorithm.

[0135] <s18> In S18, the tilt calculation unit 124 completes the relief of the fault area currently under relief and transmits information on "each relief antenna and tilt" to the radio base station control unit 130. Then it returns to S13.

[0136] (Example hardware configuration) The wireless base station control device (or control device) in this embodiment can be realized, for example, by having a computer execute a program that describes the processing content described in this embodiment. This "computer" may be a physical machine or a virtual machine on the cloud. When a virtual machine is used, the "hardware" described here is virtual hardware.

[0137] The above program can be recorded on a computer-readable storage medium (such as portable memory), saved, and distributed. It can also be provided via a network, such as the internet or email.

[0138] Figure 7 shows an example of the hardware configuration of the computer described above. The computer in Figure 7 has a drive device 1000, an auxiliary storage device 1002, a memory device 1003, a CPU 1004, an interface device 1005, a display device 1006, an input device 1007, an output device 1008, etc., all of which are interconnected by a bus BS.

[0139] The program that enables processing on the computer is provided, for example, on a recording medium 1001 such as a CD-ROM or memory card. When the recording medium 1001 containing the program is set in the drive device 1000, the program is installed from the recording medium 1001 to the auxiliary storage device 1002 via the drive device 1000. However, the program does not necessarily have to be installed from the recording medium 1001; it may also be downloaded from another computer via a network. The auxiliary storage device 1002 stores the installed program as well as necessary files and data.

[0140] The memory device 1003 reads and stores a program from the auxiliary storage device 1002 when a program startup command is received. The CPU 1004 implements the functions related to the wireless base station control device 100 (control device) according to the program stored in the memory device 1003. The interface device 1005 is used as an interface for connecting to a network. The display device 1006 displays a GUI (Graphical User Interface) etc., generated by a program. The input device 1007 consists of a keyboard and mouse, buttons, or a touch panel etc., and is used to input various operation commands. The output device 1008 outputs the calculation results.

[0141] (Effects of the embodiment) The technology according to this embodiment described above makes it possible to select an appropriate rescue antenna according to the situation when a faulty area occurs due to a failure of a wireless base station or the like.

[0142] (Note) This specification includes control methods, control devices, wireless communication systems, and programs as described in at least the following sections. (Additional note 1) A control method performed by a control device that determines the tilt of a rescue antenna for rescuing communications in a faulty area, A rescue antenna selection step involves obtaining an antenna sequence in which multiple antennas are arranged in descending order of priority using a policy function, and selecting antennas as rescue antennas in order from the beginning of the antenna sequence until a predetermined condition is met. A tilt calculation step which calculates the tilt of each rescue antenna selected in the rescue antenna selection step, and A control method comprising the following features. (Additional note 2) In the pre-training step in which the policy function is pre-trained using training data, the control device: The policy function obtains an antenna sequence in which multiple antennas are arranged in descending order of priority. Antennas are selected as rescue antennas one by one from the beginning of the antenna sequence until a predetermined condition is met. The parameters of the policy function are updated based on the reward obtained when all selected rescue antennas are used. The control method described in Appendix 1. (Additional note 3) The aforementioned predetermined conditions are that the coverage rate of the faulty area becomes 100%, or that all antennas are selected from the antenna array. The control method described in Appendix 1 or 2. (Additional note 4) The aforementioned policy function takes the following inputs for each antenna in the set of candidate rescue antennas: the height of the antenna, the distance between the antenna and the centroid of the faulted area, and the difference in azimuth angles between the antenna's center direction and the direction from the antenna to the centroid. It then outputs the antenna array. The control method described in any one of the appendices 1 to 3. (Additional note 5) A control device for determining the tilt of a rescue antenna that rescues communications in a faulty area, A rescue antenna selection unit obtains an antenna sequence in which multiple antennas are arranged in descending order of priority using a policy function, and selects antennas as rescue antennas in order from the beginning of the antenna sequence until predetermined conditions are met. A tilt calculation unit calculates the tilt of each rescue antenna selected by the rescue antenna selection unit. A control device equipped with the following features. (Additional note 6) A wireless communication system comprising a rescue antenna for rescuing communications in a faulty area, a control device for determining the tilt of the rescue antenna, and one or more wireless base stations, The control device is A rescue antenna selection unit obtains an antenna sequence in which multiple antennas are arranged in descending order of priority using a policy function, and selects antennas as rescue antennas in order from the beginning of the antenna sequence until predetermined conditions are met. A tilt calculation unit calculates the tilt of each rescue antenna selected by the rescue antenna selection unit. A wireless communication system equipped with [the necessary components]. (Additional note 7) A program for causing the computer to function as a component of the control device described in Appendix 5.

[0143] Although this embodiment has been described above, the present invention is not limited to this specific embodiment, and various modifications and changes are possible within the scope of the gist of the invention as described in the claims. [Explanation of Symbols]

[0144] 10 Wireless base stations 100 Wireless base station control devices 110 Input reception section 120 Data Processing Unit 121 Relief Priority Determination Department 122 Relief Antenna Selection Section 123 Parameter Database Section 124 Tilt Calculation Unit 125. Disaster Area Consolidation Section 130 Wireless base station control unit 131 Receiving Unit 132 Transmitter 1000 drive unit 1001 Recording media 1002 Auxiliary storage device 1003 Memory device 1004 CPU 1005 Interface device 1006 Display device 1007 Input device 1008 Output device

Claims

1. A control method performed by a control device that determines the tilt of a rescue antenna for rescuing communications in a faulty area, A rescue antenna selection step involves obtaining an antenna sequence in which multiple antennas are arranged in descending order of priority using a policy function, and selecting antennas as rescue antennas in order from the beginning of the antenna sequence until a predetermined condition is met. The system includes a tilt calculation step that calculates the tilt of each rescue antenna selected in the rescue antenna selection step, The aforementioned policy function has parameters pre-trained using rewards based on improving coverage of the faulty area and suppressing overlap in coverage areas between rescue antennas. Control method.

2. In the pre-training step in which the policy function is pre-trained using training data, the control device: The policy function obtains an antenna sequence in which multiple antennas are arranged in descending order of priority. Antennas are selected as rescue antennas one by one from the beginning of the antenna sequence until a predetermined condition is met. The parameters of the policy function are updated based on the reward obtained when all selected rescue antennas are used. The control method according to claim 1.

3. The aforementioned predetermined conditions are that the coverage rate of the faulty area becomes 100%, or that all antennas are selected from the antenna array. The control method according to claim 1.

4. The aforementioned policy function takes the following inputs for each antenna in the set of candidate rescue antennas: the height of the antenna, the distance between the antenna and the centroid of the faulted area, and the difference in azimuth angles between the antenna's center direction and the direction from the antenna to the centroid. It then outputs the antenna array. The control method according to any one of claims 1 to 3.

5. A control device for determining the tilt of a rescue antenna that rescues communications in a faulty area, A rescue antenna selection unit obtains an antenna sequence in which multiple antennas are arranged in descending order of priority using a policy function, and selects antennas as rescue antennas in order from the beginning of the antenna sequence until predetermined conditions are met. The system includes a tilt calculation unit that calculates the tilt of each rescue antenna selected by the rescue antenna selection unit, The aforementioned policy function has parameters pre-trained using rewards based on improving coverage of the faulty area and suppressing overlap in coverage areas between rescue antennas. Control device.

6. A wireless communication system comprising a rescue antenna for restoring communications in a faulty area, a control device for determining the tilt of the rescue antenna, and one or more wireless base stations, The control device is A rescue antenna selection unit obtains an antenna sequence in which multiple antennas are arranged in descending order of priority using a policy function, and selects antennas as rescue antennas in order from the beginning of the antenna sequence until predetermined conditions are met. The system includes a tilt calculation unit that calculates the tilt of each rescue antenna selected by the rescue antenna selection unit, The aforementioned policy function has parameters pre-trained using rewards based on improving coverage of the faulty area and suppressing overlap in coverage areas between rescue antennas. Wireless communication system.

7. A program for causing a computer to function as a component of the control device described in claim 5.

Images