KR102468852B1 - Method and device for constructing 3d wireless mobile communication coverage map - Google Patents

Abstract

The disclosed technology relates to a method and a device for constructing a 3D wireless mobile communication coverage map. The method comprises the steps of: figuring out, by a model mounted on the unmanned aerial vehicle, a current location of an unmanned aerial vehicle; calculating, by the model, wireless mobile communication coverage performance corresponding to the current location; and predicting, by the model, the wireless mobile communication coverage performance of the next location by applying, as a weight of the result of calculating the wireless mobile communication coverage performance of the current location, an upper confidence bound calculated according to the wireless mobile communication coverage performance at the previous location of the unmanned aerial vehicle. The present invention provides live streaming of a high-definition video using the constructed wireless mobile communication performance map.

Description

3D wireless mobile communication coverage map construction method and apparatus {METHOD AND DEVICE FOR CONSTRUCTING 3D WIRELESS MOBILE COMMUNICATION COVERAGE MAP}

The disclosed technology relates to a method and apparatus for constructing a 3D wireless mobile communication coverage map of an unmanned aerial vehicle.

Due to the development of unmanned aerial vehicles such as drones and the spread of 5G networks, it is possible to perform various flight tasks that provide high-resolution aerial images in real time. For example, live streaming may be performed using a drone. However, in order to smoothly provide such high-resolution images, reliability in the 3D cellular range must be supported.

Recently, in relation to the Korea-Europe 5G project, a method of constructing a performance map for 3D coverage by manually constructing a 3D 5G coverage map has been proposed. As an embodiment, it is possible to build a performance map by estimating 5G coverage performance corresponding to location information of the unmanned aerial vehicle measured at regular intervals and predicting average performance for each location. However, this may be a good method in terms of accuracy of the performance map, but there is a problem in that the efficiency of map construction decreases because the time and cost of map construction increases significantly.

Korean Patent Registration No. 10-2174205

The disclosed technology is to provide a method and apparatus for constructing a 3D wireless mobile communication coverage map of an unmanned aerial vehicle.

A first aspect of the technology disclosed to achieve the above technical problem is the step of determining the current location of the unmanned air vehicle by a model mounted on the unmanned air vehicle, and the step of calculating the wireless mobile communication coverage performance corresponding to the current location by the model. and the model applies an Upper Confidence Bound (UCB) calculated according to the wireless mobile communication coverage performance at the previous location of the unmanned aerial vehicle as a weight of the result of calculating the wireless mobile communication coverage performance at the current location. It is to provide a method for constructing a three-dimensional wireless mobile communication coverage map including the step of predicting the wireless mobile communication coverage performance of the next location by using the same.

The second aspect of the technology disclosed to achieve the above technical problem is a communication device mounted on an unmanned aerial vehicle and wirelessly receiving streaming content stored in a base station, and a GPS mounted on the unmanned aerial vehicle and grasping the position of the unmanned aerial vehicle at regular intervals. A sensor, a storage device mounted on the unmanned aerial vehicle and storing a model for calculating the wireless mobile communication coverage performance of the unmanned aerial vehicle, and a difference calculated according to the wireless mobile communication coverage performance at the previous location of the unmanned aerial vehicle using the model A 3D radio including a processor that predicts the wireless mobile communication coverage performance of the next location by applying the upper confidence bound (UCB) as a weight of the result of calculating the wireless mobile communication coverage performance of the current location of the unmanned aerial vehicle. An apparatus for constructing a mobile communication coverage map is provided.

Embodiments of the disclosed technology may have effects including the following advantages. However, this does not mean that the embodiments of the disclosed technology must include all of them, so the scope of rights of the disclosed technology should not be understood as being limited thereby.

According to an embodiment of the disclosed technology, the method and apparatus for constructing a 3D wireless mobile communication coverage map has an effect of constructing a performance map in real time for an unmanned aerial vehicle to exhibit optimal wireless mobile communication performance.

In addition, there is an effect of automatically finding the optimal location for communication by using the average UCB of the kernel space.

In addition, there is an effect of providing live streaming of high-definition video using the constructed wireless mobile communication performance map.

1 is a diagram illustrating a process of constructing a 3D wireless mobile communication coverage map according to an embodiment of the disclosed technology.
2 is a flowchart of a method for constructing a 3D wireless mobile communication coverage map according to an embodiment of the disclosed technology.
3 is a block diagram of an apparatus for constructing a 3D wireless mobile communication coverage map according to an embodiment of the disclosed technology.
4 is a diagram illustrating path determination according to the UCB algorithm.
5 is a diagram showing a 3D wireless mobile communication coverage map.
6 is a diagram showing evaluation of performance for each algorithm.

Since the present invention can make various changes and have various embodiments, specific embodiments will be illustrated in the drawings and described in detail in the detailed description. However, this is not intended to limit the present invention to specific embodiments, and should be understood to include all modifications, equivalents, and substitutes included in the spirit and scope of the present invention.

Terms such as first, second, A, B, etc. may be used to describe various components, but the components are not limited by the above terms, and are only used for the purpose of distinguishing one component from another. used only as For example, a first element may be termed a second element, and similarly, a second element may be termed a first element, without departing from the scope of the present invention. The terms and/or include any combination of a plurality of related recited items or any of a plurality of related recited items.

In the terms used herein, singular expressions should be understood to include plural expressions unless the context clearly dictates otherwise. And the term "includes" means that the described feature, number, step, operation, component, part, or combination thereof exists, but one or more other features or number, step, operation component, or part. or the possibility of the presence or addition of combinations thereof.

Prior to a detailed description of the drawings, it is to be clarified that the classification of components in the present specification is merely a classification for each main function in charge of each component. That is, two or more components to be described below may be combined into one component, or one component may be divided into two or more for each more subdivided function.

In addition, each component to be described below may additionally perform some or all of the functions of other components in addition to its main function, and some of the main functions of each component may be performed by other components. Of course, it may be dedicated and performed by . Therefore, the existence or nonexistence of each component described through this specification should be interpreted functionally.

1 is a diagram illustrating a process of constructing a 3D wireless mobile communication coverage map according to an embodiment of the disclosed technology. Referring to FIG. 1, a 3D wireless mobile communication coverage map is calculated to maintain wireless mobile communication performance of an unmanned aerial vehicle moving in a 3D air space, such as a drone. Here, the wireless mobile communication may be a mobile network or a next-generation network that guarantees a fast data transmission rate and low latency so as to stream high-definition video or high-capacity contents. Hereinafter, for convenience of description, a description will be made based on a 5G network.

The UAV can load a model to build a 5G coverage map, and based on the calculation result of the model, it can determine a moving path avoiding locations where 5G performance deteriorates. The reason for maintaining 5G performance here is to stream content such as high-definition video transmitted from the base station to a plurality of user terminals in real time. The plurality of user terminals may be terminals of a user located on the ground or terminals possessed by a user in the air, for example, in flight. In particular, when a user's terminal located in the air uses a streaming service, there is a need for a drone to move to an optimal communication position. To this end, we try to avoid locations where 5G performance deteriorates, such as communication shaded sections, and estimate a location where a moving unmanned aerial vehicle can continue to maintain uniform 5G performance.

First, prior to building the 5G coverage map of the unmanned aerial vehicle, the model mounted on the unmanned aerial vehicle identifies the current location of the unmanned aerial vehicle. In one embodiment, the current location may be determined using GPS information of the unmanned aerial vehicle. Similar to the model, a GPS sensor may be mounted on the body of the unmanned aerial vehicle, and the current position of the moving unmanned aerial vehicle may be grasped in real time by receiving GPS information transmitted from the GPS sensor at regular intervals.

Meanwhile, the model calculates the 5G coverage performance corresponding to the current location of the unmanned aerial vehicle according to the GPS information. 5G coverage performance can be calculated using network performance measurement tools such as Iperf-3. For example, coverage performance can be calculated by measuring network bandwidth using Iperf-3 or collecting 5G modem data such as system debug data, SINR, and downlink throughput. Naturally, as the UAV moves from location to location, 5G coverage performance may remain at a similar level to the previous location or may be inferior to the previous location. In particular, in the case of 5G networks, coverage maps are often concentrated in specific areas, and communication shadow areas may exist for each section, and 5G performance may be impaired when passing through these locations.

Therefore, in order to prevent this problem, the 5G coverage performance of the next location to which the drone will move is predicted using the model, and the drone can be moved according to the coverage performance of the location. For example, the 5G coverage performance of the next location can be predicted by applying the Upper Confidence Bound (UCB) calculated according to the 5G coverage performance at the previous location as a weight of the result of calculating the 5G coverage performance of the current location. .

In the prior art, a movement route was manually set or a movement route was set using a statistical model. However, in the case of manual setting, there was a problem that the cost was too high, and in many cases, unsatisfactory performance was shown in the case of using a statistical model. In the disclosed technology, a 5G performance map of an unmanned aerial vehicle is built by applying an upper reliability limit to solve this problem.

UCB is an algorithm based on the conventional multi armed bandits (MAB) problem. The key to MAB is to improve the inaccuracy of statistical methods by appropriately using exploration and exploitation in order to determine a specific slot machine that can obtain the optimal profit among a plurality of slot machines. Just as UCB makes an optimal choice based on its experience of exploration like MAB, it acquires information on the location where the unmanned aerial vehicle is moving at regular intervals and selects the optimal route by using the 5G coverage performance for each location. . The model mounted on the unmanned aerial vehicle stores the UCB for the previous location where the drone moved, and it can be used as a weight for the 5G coverage calculation result of the current location. The performance map calculated according to the UCB algorithm may be composed of a spatial matrix labeled with an upper confidence limit of specific spatial channel data in a round.

Meanwhile, the model may calculate UCB according to Equation 1 below.

는 샘플 평균을 의미하고,

는 공간

의 샘플들의 개수를 의미하고,

는 적게 탐험된(explored) 공간에 제공되는 보너스를 결정하는 파라미터를 의미한다. 즉, 무인비행체에 탑재된 모델은 아직 탐험하지 않은 공간에 가중치를 두어 3차원 공간에서 무인비행체의 이동 경로를 샘플링할 수 있다.From here

is the sample mean,

space

means the number of samples of

denotes a parameter that determines a bonus provided to a less explored space. That is, the model mounted on the unmanned aerial vehicle may sample the movement path of the unmanned aerial vehicle in the 3D space by placing a weight on an unexplored space.

On the other hand, when sampling a 3D cellular coverage map using UCB, there is a problem of poor accuracy in certain situations. For example, when an unmanned aerial vehicle is trapped in a hole in the 5G communication range, it is possible to calculate too short-sighted results. Therefore, in order to overcome this problem, it is possible to prevent the model from calculating a myopic result value by obtaining the average of the average channel state information (CSI) of the space around the space of interest. The model can determine the next position of the UAV using the average UCB in kernel space.

Meanwhile, the model may calculate the average UCB of the kernel space according to Equation 2 below.

는 공간

에 있는 커널의 공간 집합이고

는 커널 내 공간의 개수를 의미한다. 모델은 무인비행체의 현재 위치에 대한 UCB에 주변의 성능치를 가산하여 다음 위치에 대한 5G 성능을 예측할 수 있다.From here

space

is the spatial set of kernels in

denotes the number of spaces in the kernel. The model can predict 5G performance for the next location by adding the performance value of the surroundings to the UCB for the current location of the UAV.

On the other hand, if the process of predicting 5G coverage performance is repeated a certain number of times, a three-dimensional 5G performance map for the unmanned aerial vehicle can be constructed. That is, in the 3D space, the unmanned aerial vehicle can extract (sampling) paths that can support 5G coverage with high performance. The UAV can receive streaming content transmitted from the base station after moving its position according to the prediction result of the model. Since the location of the unmanned aerial vehicle is a location with high 5G coverage performance, live streaming can be performed using the received streaming content.

2 is a flowchart of a method for constructing a 3D 5G coverage map according to an embodiment of the disclosed technology. Referring to FIG. 2 , a method of constructing a 3D 5G coverage map may be performed through a model mounted on a drone. The model corresponds to the UCB algorithm that has improved the conventional MAB algorithm, and a device equipped with the corresponding algorithm can be mounted on an unmanned aerial vehicle. In addition, the following methods may be sequentially performed in real time while the unmanned aerial vehicle is in flight.

In step 210, the model determines the current location of the unmanned aerial vehicle. The position can be determined using GPS information transmitted from a GPS sensor mounted on the unmanned aerial vehicle at regular intervals.

In step 220, the model may calculate 5G coverage performance corresponding to the current location of the unmanned aerial vehicle. 5G coverage performance can be calculated using a network performance measurement tool such as Iperf-3, as described above with reference to FIG.

In step 230, the model applies the Upper Confidence Bound (UCB) calculated according to the 5G coverage performance at the previous location of the UAV as a weight of the result of calculating the 5G coverage performance at the current location to determine the next location of the UAV. The 5G coverage performance of a location can be predicted. The model can determine the next position of the UAV using the average UCB in kernel space. The model may build a 3D 5G performance map for the unmanned aerial vehicle by repeating steps 210 to 230 a certain number of times. In addition, the unmanned aerial vehicle can perform live streaming by receiving streaming content transmitted from the base station after moving the position according to the prediction result of the model.

3 is a block diagram of an apparatus for constructing a 3D 5G coverage map according to an embodiment of the disclosed technology. Referring to FIG. 3 , the 3D 5G coverage map construction apparatus 300 may be a 5G device mounted on an unmanned aerial vehicle and includes a communication device 310, a GPS sensor 320, a storage device 330 and a processor 340. do.

The communication device 310 wirelessly receives streaming content stored in the base station. The streaming content of the base station may be high-definition video, and the communication device 310 may be a device or module that communicates with the base station so that the unmanned aerial vehicle can stream high-definition video in real time.

The GPS sensor 320 is mounted on the unmanned aerial vehicle and can determine the position of the unmanned aerial vehicle at regular intervals. The GPS sensor may use a GPS receiver that receives radio waves from satellites moving along a certain orbit.

The storage device 330 stores a model for calculating 5G coverage performance of an unmanned aerial vehicle. The model stored in the storage device 330 may be a kind of reinforcement learning model learned in advance. The model can determine the next location of the UAV by learning 5G coverage performance and UCB for each moving path of the UAV.

The processor 340 uses the model to set the Upper Confidence Bound (UCB) calculated according to the 5G coverage performance at the previous location of the UAV to the weight of the result of calculating the 5G coverage performance at the current location of the UAV. Apply to predict the 5G coverage performance of the next location. The processor can determine the next position of the UAV using the average UCB in kernel space, and can sample a 3D 5G performance map for the UAV by repeating the calculation process according to the location determination.

Meanwhile, the 3D 5G coverage map building apparatus 300 as described above may be implemented as a program (or application) including an executable algorithm that can be executed on a computer. The program may be stored and provided in a temporary or non-transitory computer readable medium.

A non-transitory readable medium is not a medium that stores data for a short moment, such as a register, cache, or memory, but a medium that stores data semi-permanently and can be read by a device. Specifically, the various applications or programs described above are CD, DVD, hard disk, Blu-ray disk, USB, memory card, ROM (read-only memory), PROM (programmable read only memory), EPROM (Erasable PROM, EPROM) Alternatively, it may be stored and provided in a non-transitory readable medium such as EEPROM (Electrically EPROM) or flash memory.

Temporary readable media include static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), and enhanced SDRAM (Enhanced SDRAM). SDRAM, ESDRAM), Synchronous DRAM (Synclink DRAM, SLDRAM) and Direct Rambus RAM (DRRAM).

4 is a diagram illustrating path determination according to the UCB algorithm. FIG. 4 is a result of simulation based on actual field data, and it is assumed that the unmanned aerial vehicle moves to the first position 401 among movable positions by updating the coverage map according to the UCB. The 5G NSA field test shown in FIG. 4 was performed in a three-dimensional space of 100m x 120m x 100m using 150 GPS altitude spaces of the same size, and about 300 UL bandwidths were measured in each space. It was performed through a test between the device and the server in the base station.

5 is a diagram illustrating a three-dimensional 5G coverage map. 5 (a) shows a map sampled according to a preset manual, (b) shows a map based on statistical sampling, (c) shows a map sampled by a random walk method, (d ) shows the sampled map by setting the UCB parameter to 10, (e) shows the sampled map by setting the UCB parameter to 100, and (f) shows the sampled map using the average UCB of kernel space. it is shown For reference, the sampling method using UCB is called MADUS (Multi Armed Drone UCB Sampling), and the sampling method using average UCB in kernel space is called K-MADUS (Kernelized-Multi Armed Drone UCB Sampling).

Looking at the average uplink processing performance of each space as shown in FIG. 5, the terrestrial coverage map provided by the supplier indicates that the corresponding area is completely covered, but it may actually vary depending on the altitude. Although this is not a problem for general terrestrial users, communication may be restricted for users located in the air. That is, although it is treated as a normal coverage area in a terrestrial cellular network, it may be a shadow area where frame drops occur for users located at a higher altitude than the ground.

On the other hand, as shown in FIG. 5 , a space having a poor channel is displayed as a light-colored sphere, and a space having a good channel is displayed as a dark-colored sphere. The amount of data collected about a space is expressed as the size of a sphere. The larger the sphere, the greater the amount of data. As shown in FIG. 5, the size of the sphere in the case of using the average UCB of the kernel space was tested to be the largest.

6 is a diagram showing evaluation of performance for each algorithm. In order, the first is a statistical sampling result, the second is a random walk-based sampling result, the third is a UCB-based sampling (MADUS) result, and the fourth is a sampling result using average UCB in kernel space (K-MADUS).

As shown in FIG. 6, comparing the MADUS method and the K-MADUS method based on the statistical method (Survey) and the random method, it can be confirmed that sampling is more effective than the standard. The statistical method is sampled in a zig-zag traverse over the entire region of interest covering all points fairly, as indicated above by a consistent sphere size in FIG. 5 . In random walk-based sampling, sampling is performed while moving to randomly adjacent points with equal probability. On the other hand, since MADUS and K-MADUS show less Regret compared to the two standard methods, effective path sampling is performed.

On the other hand, in order to overcome the short-sightedness of MADUS, the K-MADUS kernel can be set to all neighboring points including the current location. The performance of K-MADUS can be highly dependent on the shape of the kernel. The selected kernel compromises the convergence guarantee to the global maximum, enabling ultra-fast convergence to the local maximum where the neighbor exhibits high performance. That is, the 5G coverage performance can be predicted by adding the surrounding performance values of the unmanned aerial vehicle.

The simulation in FIG. 6 was performed using actual field data, and is a result obtained by executing 2,900 sampling rounds. In FIG. 6, it was confirmed that both MADUS and K-MADUS exhibit higher performance than the above two reference methods in terms of throughput. That is, MADUS and K-MADUS can be interpreted as finding a space with a better channel compared to the two algorithms, which can be interpreted as learning the channel information of the space faster.

The method and apparatus for constructing a 3D 5G coverage map according to an embodiment of the disclosed technology have been described with reference to the embodiments shown in the drawings for better understanding, but this is merely an example, and those skilled in the art From this, it will be understood that various modifications and equivalent other embodiments are possible. Therefore, the true technical scope of protection of the disclosed technology should be defined by the appended claims.

Claims (12)

Figuring out the current location of the unmanned aerial vehicle by a model mounted on the unmanned aerial vehicle;
calculating wireless mobile communication coverage performance corresponding to the current location by the model; and
The model applies the Upper Confidence Bound (UCB) calculated according to the wireless mobile communication coverage performance at the previous location of the unmanned aerial vehicle as a weight of the result of calculating the wireless mobile communication coverage performance at the current location A method for constructing a 3D wireless mobile communication coverage map comprising: predicting the wireless mobile communication coverage performance of the next location. According to claim 1,
The model receives GPS information transmitted from a GPS sensor mounted on the unmanned aerial vehicle at regular intervals, and determines the position of the unmanned aerial vehicle according to the GPS information. According to claim 1,
The model is a 3D wireless mobile communication coverage map construction method for calculating the upper reliability limit according to Equation 1 below.
[Equation 1]

(From here

is the sample mean,

space

means the number of samples of

denotes the parameter that determines the bonus given to less explored spaces.) According to claim 1,
The model is a 3D wireless mobile communication coverage map construction method for determining the next location using an average UCB of kernel space. According to claim 4,
The model calculates the average UCB of the kernel space according to Equation 2 below.
[Equation 2]

(From here

space

is the spatial set of kernels in

means the number of spaces in the kernel.) According to claim 1,
The model builds a 3D wireless mobile communication performance map for the unmanned aerial vehicle by repeating the process of estimating the wireless mobile communication coverage performance a certain number of times. According to claim 1,
Wherein the UAV receives streaming content transmitted from a server located in a base station when moving in position according to a prediction result of the model. A communication device mounted on an unmanned aerial vehicle and wirelessly receiving streaming content stored in a server in a base station;
a GPS sensor mounted on the unmanned aerial vehicle and determining the position of the unmanned aerial vehicle at regular intervals;
a storage device mounted on the unmanned aerial vehicle and storing a model for calculating wireless mobile communication coverage performance of the unmanned aerial vehicle; and
The result of calculating the wireless mobile communication coverage performance at the current location of the unmanned aerial vehicle using the upper confidence limit (UCB) calculated according to the wireless mobile communication coverage performance at the previous location of the unmanned aerial vehicle using the model. A processor for predicting the wireless mobile communication coverage performance of the next location by applying as a weight of ; a three-dimensional wireless mobile communication coverage map construction apparatus including a. According to claim 8,
The processor calculates the upper reliability limit according to Equation 1 below.
[Equation 1]

(From here

is the sample mean,

space

means the number of samples of

denotes the parameter that determines the bonus given to less explored spaces.) According to claim 8,
The processor determines the next position of the unmanned aerial vehicle using the average UCB of the kernel space. According to claim 10,
The processor calculates the average UCB of the kernel space according to Equation 2 below.
[Equation 2]

(From here

space

is the spatial set of kernels in

means the number of spaces in the kernel.) According to claim 8,
The processor constructs a 3D wireless mobile communication coverage map for the unmanned aerial vehicle by repeating the process of estimating the wireless mobile communication coverage performance a predetermined number of times.

Images