KR102260094B1 - Control device and method for processing cloud-based big data using unmanned aerial vehicle network - Google Patents

Abstract

The present invention relates to a cloud-based big data processing system using a UAV network including a plurality of UE, at least one UAV and a cloud server, which obtains an environment variable for a change of an atmospheric channel, calculates a target input data rate or data flow based on the obtained variable to allow a user terminal to receive a communication packet, transmitted through a low-altitude platform and a high-altitude platform, within specified time in a given communication environment, and calculates a target worker node number within specified time even though additional time occurs due to an error in a process of processing a big data task requested by the user terminal based on the calculated input data rate or data flow to guarantee performance of a system even in a streaming environment requiring real-time processing about big data.

Description

Cloud-based big data processing control device and method using unmanned aerial vehicle network {CONTROL DEVICE AND METHOD FOR PROCESSING CLOUD-BASED BIG DATA USING UNMANNED AERIAL VEHICLE NETWORK}

The present invention relates to a big data processing control apparatus and method, and to a cloud-based big data processing control apparatus and method using an unmanned aerial vehicle network.

With the advent of next-generation communications such as 5G and 6G, traffic that is difficult to handle with only the existing wired network-based networks is increasing. Accordingly, a network has been proposed that can ensure the QoS (Quality of Service) of the network using an unmanned aerial vehicle (hereinafter referred to as UAV). The UAV-based network replaces the base station and backhaul of the existing wired network-based network with a UAV so that it can respond more quickly to sudden changes in traffic volume.

1 shows a schematic structure of a cloud system using a UAV network.

Referring to FIG. 1 , in a cloud system 100 using a UAV network, a plurality of user equipment (hereinafter referred to as UE) 101 is located in a ground area.

In addition, the UAV network may be divided into a Low Altitude Platform (hereinafter referred to as LAP) and a High Altitude Platform (hereinafter referred to as HAP). At least one low-altitude UAV 102 is located in the LAP with a first predetermined altitude range (eg, 0.1-5 km here), and a second predetermined altitude range higher than the first elevation range (eg 17-22 km here). At least one high-altitude UAV 103 is located in the HAP.

At least one low-altitude UAV 102 serves as a base station for multiple UEs 101 . The low-altitude UAV 102 receives a task from a plurality of UEs 101 and transmits it to the high-altitude UAV 103 .

And at least one high-altitude UAV 103 is a backhaul connecting the at least one low-altitude UAV 102 with the UAV control server 104 and the fixed cloud (general static cloud) server 105 , a backbone ) plays a role. In this case, the at least one high-altitude UAV 103 may communicate with the UAV control server 104 and the fixed cloud server 105 using free space optical (FSO), millimeter wave (mmWave), or the like. At least one high-altitude UAV 103 receives a control signal from the UAV control server 104 to manage at least one low-altitude UAV 102 of the LAP, and transmits the task delivered from the low-altitude UAV 102 to the cloud server 105 ) to pass

The UAV control server 104 controls at least one high-altitude UAV 103 and at least one low-altitude UAV 102 so that an optimal communication environment can be provided to a plurality of UEs 101 . The UAV control server 104 controls the altitude, location, etc. of the low-altitude UAV 102 and the high-altitude UAV 103 located in the HAP and the LAP for a smooth communication environment, and a plurality of UEs 101 in the low-altitude UAV 102 It is possible to control the number of dataflows managed for the .

The cloud server 105 processes tasks transmitted from multiple UEs 101 through the low-altitude UAV 102 and the high-altitude UAV 103 . In particular, the cloud server 105 serves to process tasks that are difficult for the UE 101 or the low-altitude UAV 102 and the high-altitude UAV 103 to process by themselves, such as big data. That is, the cloud server 105 transmits data and tasks collected by the at least one low-altitude UAV 102 from a plurality of UEs 101 to a predetermined cloud server 105 for processing.

In this case, in order for the UAV control server 104 to provide a smooth network environment to the plurality of UEs 101, the movement of the UAVs 102 and 103 must be constantly adjusted. And through this, it is possible to set the most optimal UAV location for data delivery and signaling.

However, the UAVs 102 and 103 perform communication with an FSO (or millimeter wave) using an idle channel to provide a point-to-point connection, and the idle channel is characterized by a continuously changing channel environment. there is this Accordingly, communication performance may be degraded due to various variables during communication between the UAVs 102 and 103 or between the UAV 103 and the UAV control server 104 . Such variables include, for example, misalignment between a transmitter and a receiver, signal interference, atmospheric turbulence, and signal scattering in the stratosphere. Therefore, in order for the UAVs 102 and 103 to maintain stable network performance, it is necessary to optimize the location of the unmanned aerial vehicle by analyzing the channel environment changing in real time.

In particular, in the cloud-based service, there is a constraint that the UAV control server 104 must process within a short processing time range to continuously acquire and analyze data on channel variables in real time to satisfy the required performance.

Korean Patent Publication No. 10-2019-0115536 (published on October 14, 2019)

An object of the present invention is to obtain and analyze a variable for a change in a standby channel in real time to calculate and process an optimal target input data rate or target data flow that can stably provide a service to a plurality of user terminals based on the cloud. An object of the present invention is to provide a cloud-based big data processing control device and method using an unmanned aerial vehicle network that can minimize the delay time during transmission.

Another object of the present invention is to provide an unmanned aerial vehicle network that can accurately estimate the number of worker nodes that can be processed within a specified time by considering additional calculation time that may occur due to communication errors in the data processing process, data processing errors, etc. An object of the present invention is to provide an apparatus and method for controlling cloud-based big data processing.

A cloud-based big data processing control apparatus using an unmanned aerial vehicle network according to an embodiment of the present invention for achieving the above object is a UAV network including a plurality of user terminals (hereinafter referred to as UE) and at least one UAV and a cloud server. In the cloud-based big data processing system, when the environment variable obtained from the environment information collected from the plurality of UAVs changes by more than a predetermined reference change, one of the input data rate and the number of data flows is selected from the obtained environment variable as a control variable a control variable selection unit; a control variable estimator configured to calculate a predetermined time packet reception probability while varying the selected control variable within a predetermined range, and to estimate the control variable for which the calculated predetermined time packet reception probability satisfies a predetermined threshold probability or more; The size of the total input data at a specific point in time is calculated using the input data rate, the number of data flows, and other environmental variables estimated as the control variables, and the estimated time required to process the input data of the calculated size is communicated or data error. It is calculated by reflecting the additional time due to occurrence, but the additional time is calculated by applying the error occurrence probability to the additional time due to communication error, variable recovery time due to data processing error, additional communication time, and additional data processing time, a worker node determining unit for calculating the number of worker nodes indicating the number of UEs capable of satisfying a predetermined target data processing time by using the calculated expected time; and a UAV location setting unit configured to set a location of at least one UAV according to the calculated number of worker nodes.

The worker node determiner calculates the size of the total input data at a specific point in time by multiplying the estimated input data rate and the number of data flows, the total number of UEs, a predetermined batch time, and the packet size. output unit; The additional time due to a _{communication error is calculated as the product of the communication time (T commn} ) and the probability of a communication error (P _ecommn ), and the additional time due to the data processing error is calculated as the variable recovery time (T _evs ) and It is calculated by multiplying the sum of the additional communication time (T _ecommn ) and the additional data processing time (T _eexec ) caused by the data processing error by the data processing error probability (P _{eexec ).} The additional time is calculated as the sum of the additional time, and the expected time (T _DP ) is calculated _{as the sum of the additional time and the variable sharing time (T vs} ) and the calculation time (T _{comp ) for sharing the environment variable} estimated time calculation unit; and a worker node count calculator configured to calculate a target worker node number N ₂ that satisfies the target data processing time T _object by using the expected time T _{DP .}

When the input data rate is selected as the control variable, the control variable estimator calculates a predetermined time packet reception probability while varying the input data rate within a specified input data rate range, and the calculated predetermined time packet reception probability is a predetermined threshold. an input data rate estimator configured to set an input data rate that satisfies the probability or higher as a target input data rate; and when the number of dataflows is selected as the control variable, calculating a predetermined time packet reception probability while varying the number of dataflows within a specified dataflow number range, and the calculated specified time packet reception probability satisfies a predetermined threshold probability or more and a dataflow number estimator configured to set the number of dataflows to the target number of dataflows.

The input data rate estimator calculates the received signal strength (GR _i ) of the i-th UAV, and determines whether data is received from the i-th UAV and a preset UAV threshold reception sensitivity (β) and the calculated received signal strength Using (GR _{i ), a communication success probability (P s} (v, R)) indicating a probability of communication between the i-th UAV and the UE among the at least one UAV is calculated in a predetermined manner, and the calculated communication success Probability (P _s (v, R)) and packet throughput of UAV (μ _g ), ingress server throughput in the cloud server (μ _e ), processing server throughput (μ _p ), output server throughput (μ _o ), database server Based on the throughput (μ _d ) and the database server access probability (δ), six predetermined auxiliary threshold reception sensitivities (β ₁ to β ₆ ) are calculated while varying the input data rate within the specified input data rate range. the six secondary threshold reception (β ₁ ~ β ₆₎ to, UE is the specified time (T _cc) said point-in-time packet reception probability (F (T _cc)) represents the probability of successful reception of the packet within using the It is possible to extract the input data rate (λ) at which the calculated predetermined time packet reception probability (F(T _cc )) is derived to be greater than or equal to the predetermined threshold probability (p).

_{If the input data rate estimator does not derive a predetermined time packet reception probability (F(T cc} )) equal to or greater than a threshold probability (p), the number of dataflows obtained as the environment variable is reduced by a predetermined reference dataflow number ratio unit, , while varying the input data rate, _{it is possible to extract the input data rate λ at which the packet reception probability F(T cc} ) at a predetermined time is derived to be greater than or equal to the predetermined threshold probability p.

The data flow number estimator calculates the received signal strength (GR _i ) of the i-th UAV, and determines whether data is received from the i-th UAV and a preset UAV threshold reception sensitivity (β) and the calculated received signal strength Using (GR _{i ), a communication success probability (P s} (v, R)) indicating a probability of communication between the i-th UAV and the UE among the at least one UAV is calculated in a predetermined manner, and the calculated communication success Probability (P _s (v, R)) and packet throughput of UAV (μ _g ), ingress server throughput in the cloud server (μ _e ), processing server throughput (μ _p ), output server throughput (μ _o ), database server Based on the throughput (μ _d ) and the database server access probability (δ), the number of dataflows is varied within the range of the number of dataflows, and the six predetermined auxiliary threshold reception sensitivities (β ₁ to β ₆ ) are calculated and then calculated. the six secondary threshold reception (β ₁ ~ β ₆₎ to, UE is the specified time (T _cc) said point-in-time packet reception probability (F (T _cc)) represents the probability of successful reception of the packet within using the , and it is possible to extract the number of _{dataflows (N 1} ) in which the calculated probability of receiving packets at a specific time (F(T _cc )) is greater than or equal to a predetermined threshold probability (p).

The dataflow number estimating unit reduces the input data rate obtained as the environment variable by a unit of a predetermined reference input data rate ratio if the predetermined _{time packet reception probability (F(T cc} )) greater than or equal to the threshold probability (p) is not derived. , while varying the number of data flows, the number of _{data flows (N 1} ) in which the _{packet reception probability (F(T cc} )) at a specified time is derived to be greater than or equal to the predetermined threshold probability (p) may be extracted.

A cloud-based big data processing control method using an unmanned aerial vehicle network according to another embodiment of the present invention for achieving the above other object includes a plurality of user terminals (hereinafter referred to as UE) and at least one UAV, a cloud server, and a UAV control server In the big data processing method performed by the UAV control server in a cloud-based big data processing system using a UAV network that selecting one of an input data rate and a number of data flows from the obtained environment variables as a control variable; calculating a predetermined time packet reception probability while varying the selected control variable within a predetermined range, and estimating the control variable for which the calculated predetermined time packet reception probability satisfies a predetermined threshold probability or more; The size of the total input data at a specific time is calculated using the input data rate, the number of data flows, and other environmental variables estimated as the control variables, and the estimated time required to process the input data of the calculated size is communicated or data error. It is calculated by reflecting the additional time due to the occurrence, but the additional time is calculated by applying the error occurrence probability to the additional time due to communication error, variable recovery time due to data processing error, additional communication time, and additional data processing time, determining the number of worker nodes representing the number of UEs capable of satisfying a predetermined target data processing time by using the calculated expected time; and setting the position of the at least one UAV according to the calculated number of worker nodes.

Therefore, the cloud-based big data processing control apparatus and method using an unmanned aerial vehicle network according to an embodiment of the present invention obtains a variable for a change in an air channel, and calculates a target input data rate or data flow based on the obtained variable By doing so, in a given communication environment, it is possible for the user terminal to receive the communication packet delivered through the low-altitude platform and the high-altitude platform within a specified time. In addition, based on the calculated input data rate or data flow, even if additional time due to an error occurs in the process of processing the big data task requested by the user terminal, it calculates the optimal number of target worker nodes so that it can be processed within the specified time. , it is possible to guarantee the performance of the system even in a streaming environment that requires real-time processing of big data.

1 shows a schematic structure of a cloud system using a UAV network.
2 shows a schematic structure of a cloud-based big data processing control apparatus using a UAV network according to an embodiment of the present invention.
FIG. 3 shows a detailed configuration of the variable estimator of FIG. 2 .
4 shows a cloud-based big data processing control method using a UAV network according to an embodiment of the present invention.

In order to fully understand the present invention, the operational advantages of the present invention, and the objects achieved by the practice of the present invention, reference should be made to the accompanying drawings illustrating preferred embodiments of the present invention and the contents described in the accompanying drawings.

Hereinafter, the present invention will be described in detail by describing preferred embodiments of the present invention with reference to the accompanying drawings. However, the present invention may be embodied in various different forms, and is not limited to the described embodiments. In addition, in order to clearly describe the present invention, parts irrelevant to the description are omitted, and the same reference numerals in the drawings indicate the same members.

Throughout the specification, when a part "includes" a certain component, it does not exclude other components unless otherwise stated, but may further include other components. In addition, terms such as "... unit", "... group", "module", and "block" described in the specification mean a unit that processes at least one function or operation, which is hardware, software, or hardware. and a combination of software.

2 shows a schematic structure of a UAV control server according to an embodiment of the present invention, and FIG. 3 shows a detailed configuration of the variable estimator of FIG. 2 .

In this embodiment, it is assumed that the UAV control server 104 of FIG. 1 operates as a cloud-based big data processing control device using a UAV network. Accordingly, FIG. 2 shows a schematic structure of the UAV control server 200 separately.

1 and 2 , the UAV control server 200 according to the present embodiment includes a receiving unit 210, an environment analyzing unit 220, a variable estimating unit 230, a UAV location setting unit 240, and a transmitting unit ( 250) may be included.

The receiver 210 receives network environment information from the low-altitude UAV 102 and the high-altitude UAV 103 . The environment analysis unit 220 obtains environment variables by analyzing the network environment information received by the receiving unit 210 . In addition, the variable estimator 230 performs an optimization operation based on the environment variables obtained from the network environment information analyzed by the environment analyzer 220 to obtain a target input data rate suitable for the current network environment, and a target The number of dataflows and the number of target worker nodes to transmit data to the cloud server 105 among the plurality of UEs 101 are estimated.

The UAV location setting unit 240 is configured to include at least one low-altitude UAV 102 and at least one high-altitude UAV 103 based on the input data rate estimated by the variable estimator 230, the number of data flows, and the number of target worker nodes. ) determines the altitude and location where it should be placed, and generates location information for each UAV.

The transmitter 250 transmits the location information generated by the UAV location setting unit 240 to the at least one low-altitude UAV 102 and the at least one high-altitude UAV 103 , and the low-altitude UAV 102 and the high-altitude UAV 103 . ) to move to the altitude and location corresponding to the location information. At this time, the high-altitude UAV 103 adjusts its own location based on the location information transmitted from the UAV control server 200, and transmits the corrected altitude and location information to the corresponding low-altitude UAV 102 to the low-altitude UAV 102 ) to edit the altitude and location information.

Here, the receiving unit 210 and the transmitting unit 250 may be integrated as a communication unit. In addition, various techniques in which the UAV position setting unit 240 adjusts the positions of the low-altitude UAV 102 and the high-altitude UAV 103 according to the calculated number of target worker nodes have already been disclosed.

Therefore, here, the variable estimator 230 in the UAV control server 200 estimates the input data rate, the number of data flows, and the number of target worker nodes suitable for the current network environment based on the environment variables obtained from the environment analysis unit 220 . The technique will be described in detail.

In this embodiment, the variable estimator 230 _first calculates the input data rate (λ) and the number of data flows (N 1 ) based on the network environment information, and the calculated input data rate (λ) and the number of data flows (N) ₁ ), calculate the number of worker nodes (N _{2 ).}

Referring to FIG. 3 , the variable estimating unit 230 includes an environment change determining unit 231 , a control variable selecting unit 232 , an input data rate estimating unit 233 , a data flow number estimating unit 234 , and an input data size. It may include a calculator 235 , an expected time calculator 236 , and a worker node count calculator 237 .

The environment change determination unit 231 receives the environment variable obtained from the environment analysis unit 220 and determines whether the change in the environment variable is greater than or equal to a predetermined reference change. Here, the environment variable may not be a single variable, but may be composed of a set of a plurality of variables for indicating the state of the standby channel. Here, the environment variable is, for example, the total number of UEs 101 in the network (n), the speed (v) of the UAV 102 and threshold values set in each component of the cloud system, communication signals and server-related variables and input data The rate (λ) and the number of data flows (N ₁ ) may be included.

And the threshold value set for each component of the cloud system may include the UAV threshold reception sensitivity (β), the system response time (τ), the probability that the packet is processed within the specified time (p), and the target data processing time (T _object ). have. In addition, the communication signal and server variables include the frequency of the signal (f), the packet throughput of the low-altitude UAV (μ _g ), the entering server throughput (μ _e ) in the cloud server 105, and the processing server throughput ( μ _p ), output server throughput (μ _o ), database server throughput (μ _d ), and database server access probability (δ) may be included.

In addition, each of the plurality of environment variables may be preset with a different reference change amount corresponding to each environment variable.

The environment change determination unit 231 determines whether at least one environment variable among a plurality of environment variables is equal to or greater than a preset reference change amount, and selects a control variable when it is determined that a change equal to or greater than the reference change amount occurs in the at least one environment variable Passing environment variables to unit 232 allows optimization to be performed. However, if the change amount of each of the plurality of environment variables is equal to or less than the reference change amount, the previously set input data rate (λ), the number of data flows (N ₁ ), and the number of worker nodes (N ₂ ) are maintained. That is, the previously calculated input data rate (λ), the number of data flows (N ₁ ), and the number of worker nodes (N ₂ ) are used as they are without performing a separate optimization process.

The control variable selection unit 232 first selects one of the input data rate λ and the number of data flows N ₁ as a control variable to be optimized among the plurality of environment variables determined by the environment change determination unit 231 . Here, the control variable selection unit 232 may select, for example, a _{variable with a large change amount among the input data rate λ and the number of data flows N 1} as the control variable, but may also randomly select the control variable. When the input data rate λ is selected as the control variable, the control variable selector 232 may activate the input data rate estimator 233 and deactivate the dataflow number estimator 234 . On the other hand, when the number of dataflows N ₁ is selected as the control variable, the dataflow number estimator 234 may be activated and the input data rate estimator 233 may be deactivated.

In the above description, for convenience of explanation, the environment change determining unit 231 and the control variable selecting unit 232 are separately illustrated, but the environmental change determining unit 231 may be included in the control variable selecting unit 232 .

The input data rate estimator 233 is activated by the control variable selection unit 232 , receives an environmental variable, and estimates a target input data rate λ according to the environmental variable. The input data rate estimator 233 may estimate the target input data rate λ by using environment variables including the _{previously set number of data flows N 1 .}

In order to estimate the target input data rate (λ) based on the environment variable, the input data rate estimating unit 233 first determines the communication success probability P _s (v, R)) is calculated.

Assuming that the wireless channel between the cloud server 105 and the UAV 102 is a Rice fading channel, and considering the path loss of electromagnetic waves, the cloud server 105 and the i-th UAV 102 are In communication through a wireless channel, the received signal strength (GR _i ^t ) of the i-th UAV 102 in the t-th time slot may be calculated by Equation (1).

로 표현될 수 있다. 여기서

이고,

이다.At this time, the gain of the transmit/receive antenna is all set to 1, GT is the transmitting power, r _i represents the distance between the i-th UAV 102 and the cloud server 105, and f is the signal represents the frequency, and h _i ^t is a channel fading coefficient of the t-th time slot, and has a rice distribution characteristic by parameters of variance and standard deviation (v, σ), and the operating time of the system is a predetermined reference If the time is exceeded, the time slot index t may be omitted. _{Also, for simplicity, the channel fading coefficient (h i} ) at (v, 1) with a standard deviation of 1 is a zero-mean Gaussian variable of two unit variances (x, y).

can be expressed as here

ego,

to be.

로 표현될 수 있다. 이에 저고도 UAV(102)와 UE 사이에 통신이 이루어질 확률인 통신 성공 확률(P_s(v, R))은 저고도 UAV(102)의 서비스 반경(R)과 전송 세기(GT), 문턱 수신 감도(β), 속도(v)를 기반으로 수학식 2에 따라 계산될 수 있다.In addition, a UAV threshold reception sensitivity β for checking whether data is received by the i-th UAV 102 may be preset. Accordingly, if the received signal strength (GR _i ) is equal to or greater than the UAV threshold reception sensitivity (β) (GR _i ≥ β), the data can be regarded as normally received, whereas the reception signal strength (GR _i ) is the UAV threshold reception sensitivity (β) If it is less than (GR _i < β), it can be considered that data has not been received. And the channel fading coefficient (h _i ), which is a rice distribution variable, is based on the position (x, y) and velocity (v) of the UAV.

can be expressed as _{Accordingly, the communication success probability (P s} (v, R)), which is the probability of communication between the low-altitude UAV 102 and the UE, is the service radius (R) and transmission strength (GT) of the low-altitude UAV 102, the threshold reception sensitivity ( β), it can be calculated according to Equation 2 based on the velocity v.

Here, f(h _i ) represents a probability density function for a channel fading coefficient.

And the calculated communication success probability (P _s (v, R)) and the packet throughput of the low-altitude UAV (μ _g ), the entering server throughput (μ _e ) in the cloud server 105, the processing server (processing server) Based on throughput (μ _p ), output server throughput (μ _o ), database server throughput (μ _d ), and database server access probability (δ), the maximum input data rate (maxλ) that can be processed is _{6 auxiliary threshold reception sensitivity (β 1} ) as shown in Equation 3 while varying the input data rate (λ) within the range (0 ≤ λ ≤ maxλ) of the input data rate (λ) specified by Equation (3) ~ β ₆ ) is divided and calculated.

Using the six auxiliary threshold reception sensitivities (β ₁ ~ β ₆ ) calculated separately, the specified time packet reception probability (F ( F ( ) ) indicating the probability that the UE 101 will successfully receive the packet within the specified time ( T _{cc )} T _cc )) can be calculated according to Equation (4).

Here, f _n (t) may be calculated according to Equation 5.

The specified time packet reception probability (F(T _cc )) calculated by Equation 4 is the communication packet delivered by the UE 101 through the low-altitude UAV 102 and the high-altitude UAV 103 within the _{specified time (T CC ).} is the probability of being delivered.

When the input data rate estimator 233 obtains the specified time packet reception probability F(T _cc ) obtained while varying the input data rate λ, the obtained specified time packet reception probability F(T _cc )) An input data rate (λ) derived from this predetermined threshold probability (p) or more (F(T _{cc ) ≥ p) is extracted.} The input data rate estimation unit 233 includes an input data rate (λ) the received specified time packets obtained while varying probability (F (T _cc)) a threshold probability (p) point-in-time packet reception probability or more of the (F (T _cc) ) does not exist, it means that even if only the input data rate (λ) is changed, the probability of receiving a packet within a specified time is low. Therefore, the previously set number of data flows (N ₁ ) is reduced by a predetermined reference data flow ratio (ψ _N ) (N ₁ (1-ψ _N )), and again according to Equations 2 to 4, the communication success probability (P _s) (v, R)), the auxiliary threshold reception sensitivity (β ₁ ~ β ₆ ), and the point-in-time packet reception probability (F(T _cc )) to satisfy (F(T _cc ) ≥ p). The input data rate (λ) is searched iteratively.

Meanwhile, the dataflow number estimating unit 234 is activated by the control variable selection unit 232 , receives an environmental variable, and estimates _{the target dataflow number N 1 according to the environmental variable.} The input data rate estimator 233 may estimate the target input data rate λ by using environment variables including the _{previously set number of data flows N 1 .}

Similarly to the input data rate estimating unit 233 , the dataflow number estimator 234 also _{estimates the target dataflow number N 1} based on the environment variable, instead of the input data rate λ, the dataflow number N ₁ ) while varying within the range (0 ≤ N ₁ ≤ maxN ₁ ) of the number of dataflows (N ₁ ) specified by the predetermined maximum number of dataflows (maxN ₁ ), communication success probability according to Equations 2 to 4 Calculate (P _s (v, R)), the auxiliary threshold reception sensitivity (β ₁ to β ₆ ), and calculate the point-in-time packet reception probability (F(T _cc )) to calculate (F(T _cc ) ≥ p) Search for the number of dataflows (N _{1 ) that satisfy .} However, if the number of dataflows N _{1 satisfying (F(T cc ) ≥ p) is not found, the input data rate λ} is reduced by a predetermined reference input data rate ratio ψ λ , λ(1- ψ _λ )), and iteratively searches while changing the number of data flows (N _{1 ) again.}

Here, the input data rate estimator 233 and the data flow number estimator 234 may be integrated into a control variable estimator for estimating the control variable selected by the control variable selection unit 232 .

The input data size calculator 235 estimates the target input data rate λ and the number of data flows N _{1 from one of the input data rate estimator 233 or the data flow number estimator 234.} The input data rate (λ) and the number of dataflows (N ₁ ), the total number of UEs 101 in the network (n), the batch time (t) and the packet size (ω) set in the real-time big data processing system (ω) ), the size (s) of the entire input data at a specific point in time is calculated according to Equation (6).

_{The expected time calculator 236 calculates an expected time T DP} required to process input data of a size calculated according to Equation (6). Here, the estimated time (T _DP ) is the variable sharing time (T _vs ) for sharing the acquired environment variables, etc., the calculation time (T _comp ), and the additional time (T) due to errors occurring in the big data processing process. _add ), where the operation time (T _comp ) can be expressed as the sum of the communication time (T _commn ) and the execution time (T _exec ) again, so it can be calculated as in Equation (7).

On the other hand, in Equation 7, the additional time (T _add ) is the communication error probability (P _ecomm ), the additional communication time (T _ecomm ) caused by the communication error, and the data processing error probability (P _eexec ) and the variable in the variable sharing process. It can be calculated by Equation 8 in consideration of the recovery time (T _evs ), the additional communication time (T _ecommn ) and the additional data processing time (T _{eexec ) generated due to a data processing error.}

In Equation 8, the variable recovery time (T _evs ) and the additional communication time (T _ecommn ) are considered together in the variable sharing process. In the cloud-based big data processing system, when an error occurs in a specific worker node, the This is because data is transferred to another worker node for processing, so all processes are performed again. In other words, unlike the communication error process in which only the communication time is added, the error occurring in the big data processing process can be considered that all processes are performed once again, so the variable recovery time (T _evs ) and the additional communication time (T _ecommn ) are reflected together should be

Therefore, the expected time (T _DP ) of Equation 7 may be recalculated as Equation 9.

이고, B는 배치 파일 처리 시간(M_a)(B = M_a) 이며,

이다. coeff는 미리 지정된 계수이고,

와

는 각각 연산 시간의 평균값이고, 변수 공유 시간의 평균값을 나타낸다.where n is the number of worker nodes, i is the number of iterations by error,

, where B is the batch file processing time (M _a ) (B = M _a ),

to be. coeff is a predetermined coefficient,

Wow

is the average value of the calculation time, respectively, and represents the average value of the variable sharing time.

Variable obtained by measurement as average value of variable sharing step time

Assuming that the number of worker nodes (N) is sufficiently large, Equation (9) can be transformed into Equation (10).

이고,

이며,

이다. here

ego,

is,

to be.

The number of worker nodes calculating unit 237 calculates the number of target worker nodes N ₂ satisfying the target data processing time T _{object by using the expected time T DP calculated by the expected time calculation unit 236 .} do.

In order to process data within the batch time in the cloud-based big data processing system, since the expected time T _DP calculated by Equation 10 must be less than or equal to the target data processing time T _object , it may be expressed by Equation 11 .

And Equation 11 can be expressed again as Equation 12.

From Equation 12, the number of worker nodes may _{be derived as three (n 1} , n ₂ , n ₃ ) as in Equation 13.

이다.here

to be.

Thereafter, the worker node count calculator 237 selects the smallest positive real number as shown in Equation 14 for the _{number of three worker nodes (n 1} , n ₂ , n _{3 ) calculated from Equation 13, and the selected worker} Set the minimum natural number greater than or equal to the number of nodes as the number of target worker nodes (N ₂ ).

In FIG. 3 , the input data size calculator 235 , the expected time calculator 236 , and the number of worker nodes calculator 237 may be integrated into the worker node determiner.

The number of worker nodes calculation unit 237 of the variable estimator 230 _{is set when the target number of worker nodes (N 2} ) is set, the estimated target input data rate (λ), the target dataflow number (N ₁ ), and the target worker node The number N ₂ is transmitted to the UAV position setting unit 240 . Accordingly, the UAV location setting unit 240 determines the number of worker nodes based on the target input data rate (λ), the target dataflow number (N ₁ ), and the target worker node number (N ₂ ) transmitted from the worker node number calculation unit 237 . At least one low-altitude UAV 102 and at least one high-altitude UAV 103 are arranged to select a number of UEs corresponding to the number of target worker nodes (N _{2 ) among the UEs 101 and perform communication with the selected UE.} It creates location information about the altitude and location that should be.

At this time, the UAV location setting unit 240 may generate location information by receiving other environment variables in addition to the target input data rate (λ), the target dataflow number (N ₁ ), and the target worker node number (N _{2 ).} .

As described above, the cloud-based big data processing control device using the unmanned aerial vehicle network according to this embodiment uses big data technology such as spark streaming to analyze the channel environment changing in real time for a number of UAVs. _{Estimate the target input data rate (λ) and the target dataflow number (N 1} ) that can satisfy the mini-batch set at the level of 3 seconds, and the estimated target input data rate (λ) and the target dataflow number (N _{1 )} ) based on the number of target worker nodes (N ₂ ), so that the unmanned aerial vehicle network can be optimized.

4 shows a cloud-based big data processing control method using a UAV network according to an embodiment of the present invention.

1 to 3, the cloud-based big data processing control method using the UAV network according to the present embodiment shown in FIG. 4 will be described. First, the environmental information collected from a plurality of UAVs 102 and 103 is applied. Receive and analyze to obtain a designated environment variable (S10). And it is determined whether the obtained environment variable has changed more than the reference change compared to the previously obtained environment variable (S20). If it is determined that the environmental change is greater than or equal to the reference change, one of the input data rate (λ) and the number of data flows (N ₁ ) is first selected as a control variable to be optimized ( S30 ). Here, as the control variable, a variable with a large change amount among the input data rate (λ) and the number of data flows (N ₁ ) may be selected as the control variable, but the control variable may be randomly selected.

Then, it is determined whether the input data rate λ is selected as a control variable (S40). If it is determined that the input data rate λ is selected as the control variable, a target input data rate λ estimation step is performed.

In the target input data rate (λ) estimation step, first, the communication success probability (P _s (v, R)) is calculated _{according to Equation 2 using the previously acquired number of data flows (N 1 ) and other environmental variables (} S51). Then, a range (0 ≤ λ ≤ maxλ) of a predetermined input data rate is analyzed ( S52 ). Then, while varying the input data rate (λ) within the analyzed input data rate range (0 ≤ λ ≤ maxλ), six auxiliary threshold reception sensitivities (β ₁ to β ₆ ) are calculated according to Equation 3 (S53). ). Using the six auxiliary threshold reception sensitivities (β ₁ to β ₆ ) obtained for each of the variable input data rates (λ), the UE 101 at each input data rate (λ) within a _{specified time (T cc )} A predetermined time packet reception probability (F(T _cc )) for successfully receiving a packet is calculated (S54). And the point-in-time packet reception probability (F) that is equal to or greater than a predetermined threshold probability (p) (F(T _cc ) ≥ p) _{among the point-in-time packet reception probabilities (F(T cc )) calculated for each input data rate (λ)} It is determined whether (T _cc )) exists (S55). If the predetermined time packet reception probability F(T _cc )) equal to or greater than the threshold probability p exists, the input data rate λ corresponding to the specified time packet reception probability F(T _cc )) is set as the target input data rate It is estimated as (S56). _{However, if there is no specific time packet reception probability (F(T cc} )) greater than or equal to the threshold probability (p), the previously set number of dataflows (N ₁ ) is reduced by a predetermined reference dataflow ratio (ψ _{N ) (N 1 )} (1-ψ _N )) (S57). Then, the step of estimating the target input data rate (λ) is performed again.

On the other hand, _{if it is determined that the number of dataflows N 1} is selected as the control variable, the target dataflow number N ₁ estimating step is performed.

In the target dataflow number (N _{1 ) estimation step, first, the communication success probability (P s} (v, R)) is calculated according to Equation 2 using the previously obtained input data rate (λ) and other environmental variables ( S61). Then, a range (0 ≤ N ₁ ≤ maxN ₁ ) of a predetermined number of data flows is analyzed (S62). Since the range of the number of the analyzed data flow _{(0 ≤ N 1 ≤ maxN 1} ) with a variable data flow can (N ₁₎ in the calculation of the six auxiliary threshold reception (β ₁ ~ β ₆₎ in accordance with equation (3) do (S63). The number of data flows that are variable (N ₁₎ is a specified time of 6 secondary threshold reception (β ₁ ~ β ₆₎ to, UE (101) from each data flow number (N ₁₎ used is obtained for each (T _{cc ) calculates the packet reception probability (F(T cc} )) of the specified time to successfully receive the packet within (S64). And the point-in- _{time packet reception probability (F(T cc} ) ≥ p) equal to or greater than a predetermined threshold probability (p) among the point-in-time packet reception probabilities (F(T _{cc )) calculated for each dataflow number (N 1 )} It is determined whether F(T _cc )) exists (S65). Ten thousand and one threshold probability (p) at least point-in-time packet reception probability (F (T _cc)) If exists, the point-in-time packet reception probability (F (T _cc)) the target data flow the number of data flows (N ₁₎ corresponding to It is estimated by the number (S66). _{However, if there is no specific time packet reception probability (F(T cc} )) equal to or greater than the threshold probability (p) _{, the previously set input data rate (λ} ) is reduced by the preset reference input data rate ratio (ψ λ) (λ( 1-ψ _λ )) (S67). Then, the target dataflow number (N ₁ ) estimation step is performed again.

On the other hand, when the target input data rate (λ) and the target data flow number (N ₁ ) are estimated, a specific time point according to Equation (6) based on the estimated target input data rate (λ) and the target data flow number (N _{1 )} Calculate the size (s) of the entire input data of (S70).

_{Then, an expected time (T DP} ) required to process input data of the calculated size is calculated ( S80 ). At this time, the expected time (T _DP ) is calculated by adding the additional time (T _add ) generated by the error, and the additional time (T _add ) is generated due to the communication error probability (P _ecomm ) and the communication error, as shown in Equation 8 additional communication time (T _ecomm ), and probability of data processing error (P _eexec ) and variable recovery time (T _evs ) in the process of variable sharing, additional communication time (T _ecommn ) and additional data processing time caused by data processing error It can be calculated according to Equation 10 in consideration of (T _{eexec ).}

When the expected time (T _{DP ) is calculated according to Equation (10), the number of worker nodes (n 1} , n ₂ ) of a specified number that can satisfy the target data processing time (T _object ) using the calculated expected time (T _{DP )} , n ₃ ) is obtained using Equation 13, and the minimum natural number equal to or greater than the smallest positive real number among the _{obtained specified number of worker nodes (n 1} , n ₂ , n _{3 ) is the target number of worker nodes (N 2} ) set to (S90).

When the number of target worker nodes (N ₂ ) is set, based on the number of target worker nodes (N ₂ ), location information about the altitude and location where the UAVs 102 and 103 should be placed is generated, and the generated location information and estimation By transmitting the target input data rate λ and the target data flow number N ₁ to the UAVs 102 and 103, the UAVs 102 and 103 are controlled (S100).

The method according to the present invention may be implemented as a computer program stored in a medium for execution by a computer. Here, the computer-readable medium may be any available medium that can be accessed by a computer, and may include all computer storage media. Computer storage media includes both volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data, and read dedicated memory), RAM (Random Access Memory), CD (Compact Disk)-ROM, DVD (Digital Video Disk)-ROM, magnetic tape, floppy disk, optical data storage, and the like.

Although the present invention has been described with reference to the embodiment shown in the drawings, which is only exemplary, those skilled in the art will understand that various modifications and equivalent other embodiments are possible therefrom.

Accordingly, the true technical protection scope of the present invention should be defined by the technical spirit of the appended claims.

101: UE 102: low altitude UAV
103: high altitude UAV 104: UAV control server
105: cloud server 200: UAV control server
210: receiver 220: environment analysis unit
230: variable estimation unit 240: UAV position setting unit
250: transmitter 231: environmental change determination unit
232: control variable selection unit 233: input data rate estimation unit
234: data flow number estimation unit 235: input data size calculation unit
236: estimated time calculation unit 237: worker node count calculation unit

Claims (20)

In a cloud-based big data processing system using a UAV network including a plurality of user terminals (hereinafter referred to as UE) and at least one UAV and a cloud server
a control variable selection unit that selects, as a control variable, one of an input data rate and a number of data flows from the obtained environment variable when the environment variable obtained from the environment information collected from the plurality of UAVs changes by more than a predetermined reference change;
a control variable estimator configured to calculate a predetermined time packet reception probability while varying the selected control variable within a predetermined range, and to estimate the control variable for which the calculated predetermined time packet reception probability satisfies a predetermined threshold probability or more;
The size of the total input data at a specific point in time is calculated using the input data rate, the number of data flows, and other environmental variables estimated as the control variables, and the estimated time required to process the input data of the calculated size is communicated or data error. It is calculated by reflecting the additional time due to occurrence, but the additional time is calculated by applying the error occurrence probability to the additional time due to communication error, variable recovery time due to data processing error, additional communication time, and additional data processing time, a worker node determining unit for calculating the number of worker nodes indicating the number of UEs capable of satisfying a predetermined target data processing time by using the calculated expected time; and
A cloud-based big data processing control device comprising a UAV location setting unit that sets a location of at least one UAV according to the calculated number of worker nodes. The method of claim 1, wherein the worker node determining unit
an input data size calculator configured to calculate a size of the total input data at a specific point in time by multiplying the estimated input data rate and the number of data flows, the total number of UEs, a predetermined batch time, and a packet size;
The additional time due to a _{communication error is calculated as the product of the communication time (T commn} ) and the probability of a communication error (P _ecommn ), and the additional time due to the data processing error is calculated as the variable recovery time (T _evs ) and It is calculated by multiplying the sum of the additional communication time (T _ecommn ) and the additional data processing time (T _eexec ) caused by the data processing error by the data processing error probability (P _{eexec ).} The additional time is calculated as the sum of the additional time, and the expected time (T _DP ) is calculated _{as the sum of the additional time and the variable sharing time (T vs} ) and the calculation time (T _{comp ) for sharing the environment variable} estimated time calculation unit; and
Cloud-based big data processing control apparatus comprising a worker node count calculator for calculating the number of target worker nodes (N ₂ ) satisfying the target data processing time (T _object ) by using the expected time (T _{DP ).} The method of claim 2, wherein the expected time calculator
Equation for the expected time (T _{DP )}

(here

ego,

is,

to be. where n is the number of worker nodes, i is the number of iterations by error,

, where B is the batch file processing time (M _a ) (B = M _a ),

to be. Also, coeff is a predetermined coefficient,

Wow

is the average value of the calculation time, respectively, and represents the average value of the variable sharing time.)
Cloud-based big data processing control device that calculates by simplifying The method of claim 3, wherein the number of worker nodes calculation unit
formula

In accordance with the three Walker node number _{(n 1, n 2, n} 3) for calculating, and the three workers may node calculation _{(n 1, n 2, n} 3) the least amount of error is greater than or equal to the minimum natural number of targets of Cloud-based big data processing control unit set by the number of worker nodes (N _{2 ).} The method of claim 1, wherein the control variable estimating unit
When the input data rate is selected as the control variable, a predetermined time packet reception probability is calculated while varying the input data rate within a specified input data rate range, and the calculated specified time packet reception probability satisfies a predetermined threshold probability or more an input data rate estimator configured to set the input data rate as a target input data rate; and
When the number of data flows is selected as the control variable, a predetermined time packet reception probability is calculated while varying the number of data flows within a specified data flow number range, and the calculated predetermined time packet reception probability is data satisfying a predetermined threshold probability or more A cloud-based big data processing control device including a dataflow number estimator for setting the number of flows to the target number of dataflows. The method of claim 5, wherein the input data rate estimating unit
Calculate the received signal strength (GR _i ) of the i-th UAV, and use the preset UAV threshold reception sensitivity (β) and the calculated received signal strength (GR _i ) to check whether data is received from the i-th UAV _{to calculate the communication success probability (P s} (v, R)) indicating the probability of communication between the i-th UAV and the UE among the at least one UAV in a predetermined manner,
Calculated communication success probability (P _s (v, R)) and packet throughput of UAV (μ _g ), ingress server throughput (μ _e ) in the cloud server, processing server throughput (μ _p ), output server throughput (μ _{o )} ), the database server throughput (μ _d ), and the database server access probability (δ) while varying the input data rate within the specified input data rate range, calculating _{six predefined auxiliary threshold reception sensitivities (β 1} to β 6 ) Then, using the calculated six auxiliary threshold reception sensitivities (β ₁ to β ₆ ), the predetermined time packet reception probability (F(T) indicating the probability that the UE will successfully receive the packet within the specified time (T _{cc ) cc} )), and extracts the input data rate λ at which the calculated specific time packet reception probability F(T _cc ) is derived to be greater than or equal to a predetermined threshold probability p. 7. The method of claim 6, wherein the input data rate estimating unit
If the predetermined time packet reception probability F(T _cc )) equal to or greater than the threshold probability p is not derived, the number of dataflows obtained as the environment variable is reduced by a predetermined reference dataflow number ratio unit, and the input data rate is variable A cloud-based big data processing control device that extracts an input data rate (λ) at which a packet reception probability (F(T _{cc )) at a specific time is derived above a predetermined threshold probability (p) while doing the same.} The method of claim 5, wherein the data flow number estimating unit
Calculate the received signal strength (GR _i ) of the i-th UAV, and use the preset UAV threshold reception sensitivity (β) and the calculated received signal strength (GR _i ) to check whether data is received from the i-th UAV _{to calculate the communication success probability (P s} (v, R)) indicating the probability of communication between the i-th UAV and the UE among the at least one UAV in a predetermined manner,
Calculated communication success probability (P _s (v, R)) and packet throughput of UAV (μ _g ), ingress server throughput (μ _e ) in the cloud server, processing server throughput (μ _p ), output server throughput (μ _{o )} ), database server throughput (μ _d ), and database server access probability (δ) while varying the number of dataflows within the specified dataflow number range to calculate _{six predefined auxiliary threshold reception sensitivities (β 1} to β 6 ) Then, using the calculated six auxiliary threshold reception sensitivities (β ₁ to β ₆ ), the predetermined time packet reception probability (F(T) indicating the probability that the UE will successfully receive the packet within the specified time (T _{cc ) cc} )), and a cloud-based big data processing control device that extracts the _{number of dataflows (N 1} ) from which the calculated probability of receiving packets at a specified time (F(T _cc )) is greater than or equal to a predetermined threshold probability (p) . The method of claim 8, wherein the data flow number estimating unit
If the predetermined time packet reception probability F(T _cc )) greater than or equal to the threshold probability p is not derived, the input data rate obtained as the environment variable is reduced by a predetermined reference input data rate ratio unit, and the number of data flows is variable A cloud-based big data processing control device that extracts the _{number of data flows (N 1} ) in which the packet reception probability (F(T _cc )) at a specified time is derived above the predetermined threshold probability (p) while doing the same. 9. The method of any one of claims 6 or 8, wherein the six auxiliary threshold receive sensitivities are:
formula

is calculated as
The predetermined time packet reception probability (F(T _cc )) is
formula

(

)
Cloud-based big data processing control unit calculated by A big data processing method performed by the UAV control server in a cloud-based big data processing system using a UAV network including a plurality of user terminals (hereinafter referred to as UE) and at least one UAV, a cloud server, and a UAV control server,
selecting one of an input data rate and a number of data flows from the acquired environment variables as a control variable when the environment variable obtained from the environment information collected from the plurality of UAVs changes by more than a predetermined reference change;
calculating a predetermined time packet reception probability while varying the selected control variable within a predetermined range, and estimating the control variable for which the calculated predetermined time packet reception probability satisfies a predetermined threshold probability or more;
The size of the total input data at a specific point in time is calculated using the input data rate, the number of data flows, and other environmental variables estimated as the control variables, and the estimated time required to process the input data of the calculated size is communicated or data error. It is calculated by reflecting the additional time due to occurrence, but the additional time is calculated by applying the error occurrence probability to the additional time due to communication error, variable recovery time due to data processing error, additional communication time, and additional data processing time, determining the number of worker nodes representing the number of UEs capable of satisfying a predetermined target data processing time by using the calculated expected time; and
A cloud-based big data processing control method comprising the step of setting the position of at least one UAV according to the calculated number of worker nodes. The method of claim 11 , wherein determining the number of worker nodes comprises:
calculating a size of the total input data at a specific point in time as a product of the estimated input data rate and the number of dataflows, the total number of UEs, and a predetermined batch time and packet size;
The additional time due to a _{communication error is calculated as the product of the communication time (T commn} ) and the probability of a communication error (P _ecommn ), and the additional time due to the data processing error is calculated as the variable recovery time (T _evs ) and It is calculated by multiplying the sum of the additional communication time (T _ecommn ) and the additional data processing time (T _eexec ) caused by the data processing error by the data processing error probability (P _{eexec ).} calculating the additional time as the sum of the additional time according to the additional time, and calculating the expected time as the sum of the additional time and the variable sharing time (T _vs ) and the calculation time (T _comp ) for sharing the environment variable; and
Using the expected time (T _DP ), cloud-based big data processing control method comprising the step of calculating the number of target worker nodes that satisfy the target data processing time (T _object). The method of claim 12, wherein calculating the expected time comprises:
Equation for the expected time (T _{DP )}

(here

ego,

is,

to be. where n is the number of worker nodes, i is the number of iterations by error,

, where B is the batch file processing time (M _a ) (B = M _a ),

to be. Also, coeff is a predetermined coefficient,

Wow

is the average value of the calculation time, respectively, and represents the average value of the variable sharing time.)
A cloud-based big data processing control method that is simplified and calculated with The method of claim 13, wherein calculating the number of target worker nodes comprises:
formula

In accordance with the three Walker node number _{(n 1, n 2, n} 3) for calculating, and the three workers may node calculation _{(n 1, n 2, n} 3) the least amount of error is greater than or equal to the minimum natural number of targets of Cloud-based big data processing control method set by the number of worker nodes. 12. The method of claim 11, wherein estimating the control variable comprises:
When the input data rate is selected as the control variable, a predetermined time packet reception probability is calculated while varying the input data rate within a specified input data rate range, and the calculated specified time packet reception probability satisfies a predetermined threshold probability or more setting the input data rate to a target input data rate; and
When the number of data flows is selected as the control variable, a predetermined time packet reception probability is calculated while varying the number of data flows within a specified data flow number range, and the calculated predetermined time packet reception probability is data satisfying a predetermined threshold probability or more A cloud-based big data processing control method comprising the step of setting the number of flows to the number of target data flows. 16. The method of claim 15, wherein the setting of the target input data rate comprises:
calculating the received signal strength (GR _{i ) of the i-th UAV;}
Communication between the i-th UAV among the at least one UAV and the UE by using the UAV threshold reception sensitivity (β) and the calculated reception signal strength (GR _{i ) that are preset to check whether data is received from the i-th UAV calculating a communication success probability (P s} (v, R)) representing a probability of this being made in a predetermined manner;
Calculated communication success probability (P _s (v, R)) and packet throughput of UAV (μ _g ), ingress server throughput (μ _e ) in the cloud server, processing server throughput (μ _p ), output server throughput (μ _{o )} ), the database server throughput (μ _d ), and the database server access probability (δ) while varying the input data rate within the specified input data rate range, calculating _{six predefined auxiliary threshold reception sensitivities (β 1} to β 6 ) to do;
Receiving the calculated six secondary threshold sensitivity (β ₁ ~ β ₆₎ to, UE is the specified time (T _cc) said point-in-time packet reception probability (F (T _cc)) represents the probability of successful reception of the packet within using the calculating ; and
A cloud-based big data processing control method comprising extracting an input data rate at which the calculated specific time packet reception probability (F(T _{cc )) is derived above a predetermined threshold probability (p).} 17. The method of claim 16, wherein the setting of the target input data rate comprises:
In the step of extracting the input data rate, if the predetermined time packet reception probability (F(T _cc )) greater than or equal to the threshold probability (p) is not derived, the number of dataflows obtained as the environment variable is calculated as a unit of the predetermined reference dataflow number ratio. reducing; and
_{Extracting the input data rate (λ) at which the calculated specific time packet reception probability (F(T cc} )) is derived above the predetermined threshold probability (p) while varying the input data rate based on the reduced number of data flows A cloud-based big data processing control method further comprising a. 16. The method of claim 15, wherein the setting of the target dataflow number comprises:
calculating the received signal strength (GR _{i ) of the i-th UAV;}
Communication between the i-th UAV among the at least one UAV and the UE by using the UAV threshold reception sensitivity (β) and the calculated reception signal strength (GR _{i ) that are preset to check whether data is received from the i-th UAV calculating a communication success probability (P s} (v, R)) representing a probability of this being made in a predetermined manner;
Calculated communication success probability (P _s (v, R)) and packet throughput of UAV (μ _g ), ingress server throughput (μ _e ) in the cloud server, processing server throughput (μ _p ), output server throughput (μ _{o )} ), database server throughput (μ _d ), and database server access probability (δ) while varying the number of dataflows within the specified dataflow number range to calculate _{six predefined auxiliary threshold reception sensitivities (β 1} to β 6 ) to do;
Receiving the calculated six secondary threshold sensitivity (β ₁ ~ β ₆₎ to, UE is the specified time (T _cc) said point-in-time packet reception probability (F (T _cc)) represents the probability of successful reception of the packet within using the calculating ; and
A cloud-based big data processing control method comprising extracting the number of dataflows in which the calculated specific time packet reception probability (F(T _cc )) is greater than or equal to a predetermined threshold probability (p). The method of claim 18, wherein the setting of the target dataflow number comprises:
When the predetermined time packet reception probability F(T _cc )) greater than or equal to the threshold probability p is not derived in the step of extracting the number of data flows, the input data rate obtained as the environment variable is set to a predetermined reference input data rate ratio unit. reducing to; and
Based on the reduced input data rate, the number of _{dataflows (N 1} ) in which _{the calculated specific time packet reception probability (F(T cc} )) is derived above the predetermined threshold probability (p) while varying the number of data flows is extracted. Cloud-based big data processing control method further comprising the step. 19. The method of any one of claims 16 or 18, wherein the six auxiliary threshold receive sensitivities are:
formula

is calculated as
The predetermined time packet reception probability (F(T _cc )) is
formula

(

)
Cloud-based big data processing control method calculated by

Images