KR101888584B1 - CDL system based adaptive power control and method thereof - Google Patents

Abstract

Provided are a common data link system for adaptive power control for a reconnaissance airplane and an adaptive power control method thereof. The common data link (CDL) system includes: a reception error ratio prediction model (RERPM) generating device generating an RERPM module predicting the next frame error rate (FER) by learning received signal strength indicators (RSSI) of a plurality of wireless signals and change patterns of frame error rates, which are collected through pre-offline learning, based on a long-short term memory (LSTM) deep learning model; and an adaptive power control device downloading the RERPM module generated through the RERPM generating device and executing the module operably, and then, predicting the next FER by inputting the current FER and current RSSI, which are received from a reconnaissance airplane, into the RERPM module, and controlling the transmission power of the airplane based on the predicted next FER.

Description

Technical Field [0001] The present invention relates to a public data link system for adaptive power control of a surveillance reconnaissance apparatus and an adaptive power control method thereof,

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a public data link system and an adaptive power control method for adaptive power control of a surveillance and reconnaissance device, and more particularly, to a public data link system using an estimation model based on a long-short term memory (LSTM) To a public data link system and an adaptive power control method thereof for adaptive power control of a surveillance reconnaissance device capable of controlling transmission power of a surveillance reconnaissance device.

The military surveillance / reconnaissance system is to improve the operational performance of the military by monitoring enemy types using various sensor cameras and acquiring enemy information through analysis of the corresponding information. This system consists of an aerial / space platform for monitoring / scouting enemy information, and a ground analysis system for control / management of aerial / space platforms and analysis of surveillance / reconnaissance information.

The aerial / space platform uses military satellites, early warning controllers, and unmanned reconnaissance aircraft, and the information gained from these platforms is communicated wirelessly to remote terrestrial analysis systems.

The Common Data Link (CDL) system is one of the communication technologies for performing wireless communication between public / space platform and ground analysis system. It is used to transmit high-capacity surveillance / reconnaissance information to the ground analysis system at high speed. do.

In general, the CDL uses a frequency division multiple access (FDMA) scheme in which a frequency channel is allocated on a link basis in order to transmit monitoring / reconnaissance information stably and in real time.

The network structure supported by the CDL is based on a 1: 1 communication structure between an air node and a ground node, an N: 1 structure in which a plurality of air nodes communicate with one ground node, an air node and a plurality of ground nodes And a 1: N structure for communicating with each other.

In these various network structures, the communication areas between the links are overlapped, which causes the performance degradation due to the frequency interference. This is due to the adjacent channel interference, in which one large power signal interferes with the other weak signal according to the near-far effect at the overlap of the communication area even though different frequency channels are allocated for each link unit.

Therefore, it is necessary to apply a method that can minimize performance degradation due to such frequency interference. To this end, Adaptive Power Control (APC) is used to minimize the frequency interference by adjusting the transmission power according to the propagation environment in various commercial mobile communication networks.

However, there is a problem due to the difference in the operating environment in applying the conventional adaptive power control technique to the CDL system. This is because the intensity of the appropriate transmission power required rapidly changes due to the change of the propagation environment caused by the high mobility of the communication node. Therefore, adaptive power control schemes for CDL systems should be able to cope with rapid changes in the propagation environment.

For this, researches have been proposed to cope with the sudden change of the propagation environment caused by high mobility in real time by setting the calculation cycle of the appropriate transmission power quickly. However, since most processing power is required during CDL operation, a limited computing environment There is a limit to apply to the CDL platform.

In order to solve these problems, researches for predicting appropriate transmission power through game theory or machine learning are underway. However, these studies require a certain operating time or the accuracy of prediction until the prediction results stabilize. Therefore, in CDL, it is necessary to apply predictive-based adaptive power control scheme that guarantees high mobility of public platform and stable communication continuously under limited memory and processing environment.

Korean Patent Publication No. 2002-0038205

SUMMARY OF THE INVENTION The present invention has been made to solve the above problems and it is an object of the present invention to provide a CDL system capable of reducing the link interference between adjacent channels even under limited mobility and limited memory and processing environments, A public data link system and an adaptive power control method for adaptive power control of a surveillance device capable of predicting a channel state corresponding to a propagation environment.

The solution of the present invention is not limited to those mentioned above, and other solutions not mentioned can be clearly understood by those skilled in the art from the following description.

As a means for solving the above-mentioned technical problems, according to an embodiment of the present invention, a common data link (CDL) system for adaptive power control of a surveillance and reconnaissance device includes a plurality of (Long-Short term Memory Reception), which predicts the next FER by learning the change patterns of received signal strength indicators (RSSI) and the frame error rate (FER) of wireless signals based on a long-short term memory (LSTM) An Error Ratio Prediction Model) module; And an LERM module for downloading the LSTM RERPM module generated by the RERPM generation device to be operable and then inputting a current RSSI and a current FER received from the surveillance device to the LSTM RERPM module to predict the next FER, And an adaptive power control apparatus for controlling the transmission power of the surveillance reconnaissance apparatus based on the predicted next FER.

The RERPM generation apparatus generates RSSIs (R = {R ₁ , R ₂ , R ₃ , ..., R _n }) of _n radio signals transmitted from the public and / or space by the surveillance / A normalizer for performing a Min-Max normalization process on the values of the FERs (E = {E ₁ , E ₂ , E ₃ , ..., E _n }); Converts the RSSIs and FERs during the s (sequence length) time before the next time point to be predicted to use the change information according to time as input data into vector input data, and outputs the actual FER at the t point as actual output data Output data conversion unit; Output data (X _{t ,} Y _t ) transformed by the input / output data conversion unit, state information of a previous LSTM cell, and output information learned in the previous LSTM cell, A LSTM layer for outputting final output information in an LSTM cell; and a RERPM module for generating the RERPM module including a fully connected layer for predicting the next FER using the final output information input from the LSTM layer And the like.

The RERPM generation apparatus loads the RSSI and FER history data generated during the mission when the surveillance and reconnaissance probe completes an actual mission through the public data link based on the transmission power calculated by the adaptive power control apparatus And re-learning the RERPM model and downloading the RERPM model evolved by the re-learning to the adaptive power control device so as to apply and operate the common data link, and the adaptive power control device includes: FER can be predicted based on RERPM.

The adaptive power control apparatus includes: a transmission / reception unit that communicates with the surveillance / reconnaissance apparatus for monitoring / scouting enemy states in the air and / or space through the public data link; A RERPM module for learning the current RSSI and the current FER change pattern of the wireless signal received from the surveillance and reconnaissance device on the basis of RERPM deep learning and predicting the next FER; And a power control module for calculating the transmission power of the surveillance unit based on the predicted next FER, and the transmission / reception unit may transmit the calculated transmission power to the surveillance unit.

Wherein the power control module adjusts the transmit power of the surveillance reconnaissance device if the predicted next FER is higher than the maximum allowable error rate and if the predicted next FER is lower than the minimum allowable error rate, Can be adjusted downward.

The public data link system comprises a 1: 1 architecture, a plurality of aerial / space platforms, and a plurality of aerospace platforms, wherein one aerial / space platform corresponding to the surveillance reconnaissance device and one terrestrial analysis system corresponding to the adaptive power control device communicate An N: 1 architecture in which one ground analysis system communicates, a 1: N architecture in which a plurality of ground analysis systems communicate with one aerial / space platform, and one of the plurality of aerial / And a relay network structure for relaying the ground analysis system.

In the case of the relay network architecture, one air / space platform acting as a relay among the plurality of public / space platforms can predict the next FER based on the RERPM module.

According to another aspect of the present invention, there is provided an adaptive power control method of a control apparatus for controlling transmission power of a surveillance and reconnaissance device in a common data link (CDL) system, the method comprising: (A) The following FER is predicted by learning the variation pattern of Received Signal Strength Indicators (RSSI) and the Frame Error Rate (FER) of a plurality of wireless signals collected through off-line learning based on a long-short term memory (LSTM) Generating a RERPM (Long-Short term Memory Reception Error Ratio Prediction Model) module; And (B) the control device downloads the RERPM module generated in the step (A) so as to be operable, inputs the current RSSI and the current FER received from the surveillance device to the RERPM module, Estimating the next FER, and controlling transmission power of the surveillance reconnaissance device based on the predicted next FER.

The method of claim 1, wherein the step (A) further comprises: (A1) the RERPM generation device transmits RSSIs (R = {R ₁ , R ₂ , _{R 3, ..., R n}} ) and the FER _{(E = {E 1, E} 2, E 3, ..., performing a Min-Max Normalization processing for the value of E _n}); (A2) The RERPM generator converts the RSSIs and the FERs for a s (Sequence Length) time before a time point t, which is a next time to predict to use time-based change information as input data, into vector input data, converting the actual FER at time t into actual output data; And (A3) the RERPM generating apparatus includes a plurality of LSTM cells, the vector input / output data converted in the step (A2), state information of a previous LSTM cell, and output information learned in the previous LSTM cell A LSTM layer for outputting final output information in the last LSTM cell, and a Fully Connected Layer for predicting the next FER using the final output information input from the LSTM layer Step.

The step (A) includes: (A4) when the surveillance and reconnaissance device completes the actual mission through the public data link based on the transmission power calculated in the step (B), the RERPM generation device performs the mission Loading the generated RSSI and FER history data to re-learn the RERPM model; And (A5) the RERPM generation apparatus may further include downloading the RERPM model evolved by the re-learning to the adaptive power control apparatus, and processing the RERPM model so that the RERPM model is applied to and operated on the common data link.

(B) comprises the steps of: (B1) the adaptive power control device receiving a radio signal via a public data link from the surveillance / surveillance device monitoring / reconnaissance enemy status in air and / or space; (B2) the adaptive power control apparatus learns a next FER by learning a current RSSI and a current FER change pattern of the received radio signal based on a deep learning model; And (B3) the adaptive power control apparatus calculates a transmission power of the surveillance / reconnaissance device based on the predicted next FER, and transmits the calculated transmission power to the surveillance / reconnaissance device.

Wherein the step (B3) comprises: if the predicted next FER is higher than a maximum allowable error rate, the adaptive power control unit adjusts the transmission power of the surveillance reconnaissance unit so that the predicted next FER is less than a minimum allowable error rate The transmission power of the surveillance unit can be adjusted downward.

According to the present invention, it is possible to quickly and accurately predict a channel state in response to a rapidly changing radio wave environment with high mobility, while simultaneously reducing link interference between adjacent channels even in a limited mobility and processing environment of a public / space platform in a CDL system And the transmission power of data can be adaptively controlled.

The effects of the present invention are not limited to those mentioned above, and other effects not mentioned can be clearly understood by those skilled in the art from the following description.

1 illustrates various network architectures supported by CDL,
Figures 2 and 3 schematically illustrate a CDL for adaptive power control of a surveillance probe according to an embodiment of the present invention;
FIG. 4 is a block diagram illustrating an apparatus for generating a Reception Error Ratio Prediction Model (RERPM) according to an embodiment of the present invention.
FIG. 5 is a diagram illustrating a RERPM configuration based on an LSTM deep learning according to an embodiment of the present invention; FIG.
FIG. 6 is a block diagram illustrating an adaptive power control apparatus according to an embodiment of the present invention shown in FIGS. 2 and 3. FIG.
7 is a diagram for explaining a method of applying an adaptive power control structure to a CDL system,
FIG. 8 is a flowchart for explaining an adaptive power control method in a CDL system according to an embodiment of the present invention;
9 is a flowchart for more specifically explaining steps S835 to S850 in FIG. 8,
10 illustrates a CDL test bed according to an embodiment of the present invention,
11 is a graph comparing RMSE of LSTM RERPM, MNN (Multi-layer Neural Network) and CNN (Convolutional Neural Network) according to an embodiment of the present invention,
12 is a graph showing a result of analyzing a distribution of prediction result values after learning 10000 times,
FIG. 13 is a graph showing a change in the received frame error rate (FER) during the flight test,
14 is a graph showing the change of the RSSI in the flight test.

BRIEF DESCRIPTION OF THE DRAWINGS The above and other objects, features, and advantages of the present invention will become more readily apparent from the following description of preferred embodiments with reference to the accompanying drawings. However, the present invention is not limited to the embodiments described herein but may be embodied in other forms. Rather, the embodiments disclosed herein are provided so that the disclosure can be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.

In this specification, when an element is referred to as being on another element, it may be directly formed on another element, or a third element may be interposed therebetween.

Also, when it is mentioned that the first element (or component) is operated or executed on the second element (or component) ON, the first element (or component) It should be understood that it is operated or executed in an operating or running environment or is operated or executed through direct or indirect interaction with a second element (or component).

It is to be understood that when an element, component, apparatus, or system is referred to as comprising a program or a component made up of software, it is not explicitly stated that the element, component, (E.g., memory, CPU, etc.) or other programs or software (e.g., drivers needed to run an operating system or hardware, etc.).

It is also to be understood that the elements (or elements) may be implemented in software, hardware, or any form of software and hardware, unless the context requires otherwise.

Also, terms used herein are for the purpose of illustrating embodiments and are not intended to limit the invention. In the present specification, the singular form includes plural forms unless otherwise specified in the specification. The terms "comprises" and / or "comprising" used in the specification do not exclude the presence or addition of one or more other elements.

Hereinafter, the present invention will be described in detail with reference to the drawings. In describing the specific embodiments below, various specific details have been set forth in order to explain the invention in greater detail and to assist in understanding it. However, it will be appreciated by those skilled in the art that the present invention may be understood by those skilled in the art without departing from such specific details.

In some instances, it should be noted that portions of the invention that are not commonly known in the description of the invention and are not significantly related to the invention do not describe confusing reasons for explaining the present invention.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, the present invention will be described in detail with reference to the accompanying drawings.

Each configuration of the adaptive power control device 300 for transmission power control of the surveillance and reconnaissance device in the common data link (CDL) system shown in FIGS. 2, 4, 5, And it can easily be deduced that the average expert in the technical field of the present invention does not necessarily mean that each configuration is divided into separate physical devices or written in separate codes.

In this specification, a module may mean a functional and structural combination of hardware for carrying out the technical idea of the present invention and software for driving the hardware. For example, the module may mean a logical unit of a predetermined code and a hardware resource for executing the predetermined code, and it does not necessarily mean a physically connected code or a kind of hardware. Can be easily deduced to the average expert in the field of < / RTI >

The adaptive power control apparatus 300 may be installed in a predetermined data processing apparatus to implement the technical idea of the present invention.

1 is a diagram illustrating various network structures supported by a public data link system.

Referring to FIG. 1, the network structure supported by the CDL system is based on a 1: 1 (point-to-point) communication structure between an air / space platform and a ground node, An N: 1 structure (multiple points) in which a platform communicates with a ground analysis system, a 1: N structure that emits data from one aerial / space platform to the ground and receives corresponding data from multiple ground analysis systems ), And a relay structure in which the public / space platform relays public data to improve operational operability.

The adaptive power control apparatus 300 according to the embodiment of the present invention can be applied to various network structures supported by the CDL system described with reference to FIG. 1, (For example, image transmission through the downlink) degradation can be solved.

FIG. 2 and FIG. 3 are diagrams schematically illustrating a CDL system for adaptive power control of a surveillance and reconnaissance apparatus according to an embodiment of the present invention.

2 and 3, a CDL system for adaptive power control of a surveillance and reconnaissance device according to an embodiment of the present invention includes a RERPM generator 100, a RERPM downloader 200, adaptive power controllers 300, 400).

The RERPM generating apparatus 100 learns channel variation patterns in which the surveillance and reconnaissance apparatus 10 and the adaptive power control apparatus 300 or 400 are operated through the CDL scheme based on a deep learning model, And a RERPM (Reception Error Ratio Prediction Model) for prediction can be designed and generated.

RERPM is based on a long-short term memory (LSTM) deep learning model. Since the RERPM must learn a channel change pattern between the surveillance reconnaissance device 10 and the adaptive power control device 300, And the mission process (Mission Process).

The offline learning process is required in environments with high computing power because the deep learning learning process must be dealt with. Therefore, after performing the mission of the surveillance and reconnaissance system 10 corresponding to the public / space platform, the RERPM generation apparatus 100, which is a ground analysis system, obtains history data of RSSI (Received Signal Strength Indicators) and FER RERPM can be designed by loading (①).

In general, the RSSI information is characteristic data that ultimately represents the channel state on the receiving side because the transmitted signal is the strength of the signal finally received at the receiving end while suffering interference and path loss. The FER information is data indicating a reception quality characteristic that indicates the degree of reception on the reception side. Therefore, the RERPM generating apparatus 100 can design a model for inputting RSSI and FER change information to the deep learning model and predicting the next FER as an output value. In the embodiment of the present invention, this prediction model is called RERPM.

The RERPM generating apparatus 100 can generate a RERPM module capable of predicting the next FER by learning the change patterns of RSSIs and FERs of a plurality of wireless signals collected through the pre-offline learning based on the LSTM deep learning model ).

FIG. 4 is a block diagram illustrating a RERPM generation apparatus 100 according to an embodiment of the present invention, and FIG. 5 is a diagram illustrating a RERPM configuration based on an LSTM deep learning according to an embodiment of the present invention.

Referring to FIG. 4, a RERPM generator 100 according to an embodiment of the present invention may include a normalizer 110, an input / output data converter 120, and a RERPM generator 130.

RERPM requires a preprocessing process of input data as shown in Equation (1).

The normalization unit 110 corresponding to the offline learning process performs a preprocessing process on the input data to be input to the RERPM. The normalization unit 110 outputs n RSSIs as R and FER as E as RERPM input data, as shown in the following Equation (1).

In order to enhance the learning effect of the deep learning model, the normalization unit 110 calculates RSSIs (R = {R ₁ , R (n)) of n radio signals transmitted from the public and / _2, R _3, ..., it is possible to perform the Min-Max Normalization processing for the value of R _n}) and the FER _{(E = {E 1, E} 2, E 3, ..., E n}).

The input / output data conversion unit 120 converts the RSSIs and FERs during the sequence time (t-1 to ts) before the time point t, which is the next time to be predicted, (X _t ) as shown in Equation (2).

In addition, since the input data is data of up to time t-1 from the time point ts, the output data are given as FER vector data (Y _t) of time t as shown in [Equation 3].

Referring to Equation (3), the input / output data conversion unit 120 may convert the actual FER at time t into actual output data (Y _t ). The actual FER at the time t or Yt is the FER of the transmission power actually outputted at the time t at the monitoring and reconnaissance unit 10 corresponding to the air node and is RSSI and FER collected in the off-line training process of FIG.

When the preprocessing process is performed by the normalizer 110 and the input / output data converter 120, a learning process of RERPM is performed next.

To this end, the RERPM generator 130 may generate an RERPM module based on an LSTM deep learning process, which is performed in an order of an LSTM layer and a fully connected layer, as shown in FIG.

In an embodiment of the present invention, RERPM generator 130 is state information of multiple comprise LSTM cell, the vector input data (X _t, Y _t) converted from input and output data converter 120, and the previous LSTM cells ( _{C tm, 0 <m <s} + 1 (m is an integer)), and, by using the output information (h _tm) learning in the previous LSTM cell for outputting a final output information (h _t-1) at the end LSTM cells the RERPM module for using the final output information (h _t-1) input from the LSTM layer and, LSTM layer comprises a fully connected layers of predicting a next FER (P _t) can be generated on the basis of LSTM deep learning model .

In general, the MNN Deep Learning Model and the CNN Deep Learning Model treat sequential input patterns, if any, as independent information that is independent of each other. On the other hand, the LSTM deep-running model models the relationship between input patterns when several signals come in successively and the signals before and after are correlated with each other. Therefore, the LSTM deep learning model is effective in learning correlations with distant temporal information, and thus is mainly used in the field of prediction based on temporal change such as stock price prediction.

This LSTM deep learning model consists of three kinds of gates: an input gate, an output gate, and an erase gate. The input and output gates control the input and output signals, respectively, and the erase gate determines whether to retain or discard the stored information. This makes it possible to memorize meaningful pattern information even if it is old.

)와 출력 정보를 입력으로 사용한다. In detail, since the LSTM deep learning model uses information according to temporal change, in the LSTM layer, each LSTM cell stores the state information of the previous LSTM cell

) And output information as inputs.

), 소거(

), 출력(

) 게이트를 적용한 수학식은 다음과 같다. LSTM Cell's three gates, the input (

), Erasure

), Print(

) The formula for applying the gate is as follows.

X _t in Equations (4) to (7) is input data of Equation (2), and is RSSI and FER that are previously collected and vector-converted for LSTM-based RERPM construction.

는 Sigmoid 함수를 나타내며, Activation Function이다.Equation (6) is a formula for obtaining state information of each cell, and Equation (4) to Equation (7)

Represents Sigmoid function and is an Activation Function.

Finally, in the perfect link layer, the reception error rate (P _t ) can be predicted using Equation (8).

Referring to Equation (8), the RERPM generator 130 can predict the next FER (P _t ) using only the output (h _t-1 ) of the last LSTM cell among a plurality of LSTM cells.

The RERPM generator 130 may generate a RERPM module according to an embodiment of the present invention with reference to Equation (4) to Equation (8).

When the RERPM module is created, the RERPM generator 130 repeatedly performs the following steps to increase the accuracy of the RERPM.

First, the RERPM generator 130 calculates a difference (Loss) between the reception frame error rate (Y _t ) and the prediction result (P _t ) of RERPM using Equation (9).

As the prediction result (P _t ) of the RERPM approaches the reception frame error rate (Y _t ), that is, the less the loss, the higher the accuracy of the RERPM. This is because the received frame error rate (Y _t ) corresponds to the correct answer of RERPM.

Second, in order to reduce the loss, the RERPM generator 130 performs a backpropagation algorithm using the Adam (Adaptive Moment Estimation) slope descent method, thereby minimizing the loss and evolving the RERPM. For example, the RERPM generator 130 may calculate the Adam Optimization slope W _i , b _i , W _f , b _f , W _c , b _c , W _o , and b _{o in} Equations (4) It is possible to find an optimal value by using the down-propagation back propagation algorithm. This makes it possible to generate the optimal RERPM module with the smallest difference between the predicted FER (P _t ) and the actual FER (Y _t ) have.

2 and 3, the RERPM downloader 200 downloads the RERPM module generated by the RERPM generator 100 to the adaptive power control units 300 and 400 (3). This completes the offline learning process. The RERPM generating apparatus 100 and the RERPM downloader 200 can be implemented as one unit, and are shown in FIG. 2, respectively.

The adaptive power control units 300 and 400, which download the RERPM module, install and execute the RERPM module to perform the actual mission, that is, the surveillance unit 10 and the CDL The RSSI and FER generated in the wireless communication are received and collected (4), from which the next FER is predicted, and the transmission power of the monitoring and reconnaissance device 10 can be controlled based on the predicted FER.

In FIG. 2 and FIG. 3, one of the adaptive power control devices 300 is a control node and a ground analysis system provided at a ground node, and the other 400 is a surveillance / It can play a role of relay in the structure.

Hereinafter, adaptive power control operation will be described by taking an adaptive power control apparatus 300 provided at a ground node as an example.

FIG. 6 is a block diagram illustrating an adaptive power control apparatus 300 according to an embodiment of the present invention shown in FIG. 2 and FIG.

6, an adaptive power control apparatus 300 according to an exemplary embodiment of the present invention includes a control receiving unit 310, a power monitor 320, a link quality monitor 330, a RERPM module 340, (350) and a control transmission unit (360).

The control receiver 310, which is one of the transceivers, communicates via the CDL with a surveillance / reconnaissance device 10 that monitors / scouts enemy status in the air and / or space. The control receiving unit 310 receives a radio signal from the surveillance and control unit 10 at every power control period.

The power monitor 320 monitors the received radio signal to check the transmission power value tp _k of the radio signal and transmits the confirmed transmission power value tp _k to the power control module 350.

The link quality monitor 330 can monitor the received radio signal to check RSSI and FER of the radio signal. Hereinafter, the current RSSI and the current FER confirmed by the link quality monitor 330 are referred to as rssi _k and fer _k .

The RERPM module 340 may download the change pattern of the current RSSI (rssi _k ) and the current FER (fer _k ) from the RERPM generating apparatus 100 and learn the next FER by learning based on the RERPM deep learning running. The RERPM module 340 having the structure as shown in FIG. 5 receives the current RSSI (rssi _k ) and the current FER (fer _k ) from the last LSTM cell and predicts a change pattern to predict the next FER (P _{k + 1} ).

The power control module 350 receives the transmission power tp _k of the application node, that is, the surveillance reconnaissance device 10 from the power monitor 320 and monitors the transmission output status of the surveillance surveillance device 10. The power control module 350 may calculate the transmission power tp _{k + 1} of the surveillance and control unit 10 based on the predicted next FER (P _{k + 1} ) from the RERPM module 340.

, 상한 임계값)보다 높으면, 감시정찰기(10)의 송신 전력을 상향조정하도록 명령할 수 있다.More specifically, the power control module 350 determines whether the next predicted FER (Pk _{+ 1} ) is the maximum allowable error rate

, Upper limit threshold value), it is possible to instruct the surveillance and reconnaissance device 10 to adjust the transmission power to be increased.

, 하한 임계값)보다 낮으면, 무선 링크가 충분히 안정된 것으로 판단하고 감시정찰기(10)의 송신 전력을 하향조정하도록 명령할 수 있다.In addition, the power control module 350 determines whether the next predicted FER (Pk _{+ 1} )

, Lower limit threshold value), it can be determined that the radio link is sufficiently stable and command to adjust the transmission power of the surveillance probe 10 downward.

와

사이의 허용 가능한 에러율 값 내에 있으면 제어 명령을 보내지 않고 감시정찰기(10)의 이전 전송 전력을 유지하도록 한다.In addition, the power control module 350 determines whether the next predicted FER (Pk _{+ 1} )

Wow

To keep the previous transmission power of the surveillance reconnaissance device 10 without sending a control command.

The control transmitting unit 360, which is one of the transmitting and receiving units, transmits a request to the surveillance and reconnaissance device 10 to communicate using the finally determined transmission power, for example, the upward or downwardly adjusted transmission power in the power control module 350 send.

The scouting reception unit 11 of the surveillance and reconnaissance device 10 receives the request from the control transmission unit 360 and transfers it to the power setting module 13. [

The power setting module 13 analyzes the received request and sets the transmission power tp _{k + 1} determined by the power control module 350 as the transmission power of the radio signal.

The scouting transmission unit 15 transmits the transmission power tp _{k + 1} requested by the power control module 350 to the adaptive power control unit 35 via the CDL with the transmission power tp _{k +1} set by the power setting module 13, Lt; RTI ID = 0.0 &gt; 300 &lt; / RTI &gt;

Accordingly, the surveillance reconnaissance apparatus 10 can transmit data such as the target photographed image using the predicted transmission power based on the RERPM module 340 in the adaptive power control apparatus 300, The CDL system can perform the mission using the optimal transmission power.

Meanwhile, since the CDL is operated at various speeds in various terrains according to the surveillance / reconnaissance system, it is necessary to learn for each system, and the adaptive power control device 300 and the surveillance and reconnaissance device 10 can perform tasks such as target shooting transmission After doing so, the RERPM can be evolved through an offline re-learning process. Accordingly, in order to apply the RERPM-based adaptive power control structure to the CDL, the offline learning process and the mission execution process can be performed again as shown in FIG.

7 is a diagram for explaining a method of applying an adaptive power control structure to a CDL system.

Referring to FIG. 7, the CDL system includes a RERPM generator 100, a RERPM downloader 200, and adaptive power control units 300 and 400, which are identical to the structure shown in FIG. 2, It is omitted.

However, according to the CDL system shown in FIG. 7, in the offline learning process, when the surveillance and reconnaissance system 10 completes the actual mission through the CDL based on the transmission power calculated by the adaptive power control apparatus 300, The RERPM generating apparatus 100 may reload the RERPM model by loading the RSSI and FER history data generated during the mission execution from the adaptive power control apparatus 300. [

Then, the task execution process is performed. The RERPM downloader 200 downloads the RERPM model evolved by the re-learning to the adaptive power control device 300 so that it is applied to and operated on the CDL.

Accordingly, the adaptive power control apparatus 300 predicts the FER based on the evolved RERPM model, and the surveillance reconnaissance apparatus 10 can monitor and recapture the target with the transmission power calculated based on the predicted FER.

The location of such a control node, i.e., the adaptive power control device 300, may vary depending on the networking architecture. Since the downlink image transmission is mainly performed in the CDL, the ground analysis system acts as a control node using the RERPM module in the 1: 1, N: 1, and 1: N networking structures. However, in the case of a relay networking structure, a relaying aerial / space platform (for example, a surveillance / reconnaissance device) as well as a terrestrial analysis system can perform a role of a control node using a RERPM module.

8 is a flowchart illustrating an adaptive power control method in a CDL system according to an embodiment of the present invention.

The adaptive power control method of FIG. 8 can be implemented by the RERPM generation apparatus 100 and the adaptive power control apparatus 300 of the CDL system described with reference to FIGS. 1 to 7, and therefore, detailed description thereof will be omitted.

Referring to FIG. 8, the RERPM generator 100 normalizes the values of the RSSIs and FERs of the n radio signals received from the surveillance and reconnaissance device 10 by offline learning (S805).

The RERPM generating apparatus 100 transmits RSSIs and FERs during a Sequence Length time (i.e., t-1 to ts) before a time point t, which is a next time to be predicted, to use time-based change information as input data, (X _t ), and the actual FER at time _t can be converted into actual output data (Y _t ) to be vectorized (S 810).

The RERPM generating apparatus 100 includes a plurality of LSTM cells and uses the vector input / output data (X _{t ,} Y _t ) converted in step S 810, the state information of the previous LSTM cell, and the output information learned in the previous LSTM cell An LSTM layer for outputting final output information in the last LSTM cell, and a RERPM module including a complete connection layer for predicting a next FER using the final output information input from the LSTM layer in operation S815.

The RERPM generating apparatus 100 downloads the generated RERPM module to the adaptive power control apparatus 300 (S820).

After the adaptive power control apparatus 300 executes the downloaded RERPM module so as to install and operate the RERPM module, the adaptive power control apparatus 300 enters a communication standby state, that is, an operating state with the surveillance and reconnaissance system 10 (S825).

The surveillance reconnaissance apparatus 10 transmits a radio signal through the CDL for performing mission (for example, image transmission of an enemy target) (S830), and the adaptive power control apparatus 300 transmits a radio signal through the CDL (S835).

The adaptive power control unit 300 predicts the next FER by inputting the current RSSI and the current FER to the RERPM module in step S840. Based on the predicted next FER, the adaptive power control unit 300 transmits the transmission power of the surveillance probe 10 tp _{k + 1} ) (S845).

Then, the adaptive power control apparatus 300 transmits a request to the surveillance and reconnaissance system 10 to perform CDL communication with the transmission power calculated in step S845 (S850).

The surveillance and reconnaissance system 10 analyzes the received request, sets the transmission power received from the step S850 as the transmission power of the radio signal (S855), communicates with the adaptive power control apparatus 300 through the CDL with the set transmission power (S860, S865).

Accordingly, the surveillance reconnaissance apparatus 10 can transmit data such as the target photographed image using the predicted transmission power based on the RERPM module 340 in the adaptive power control apparatus 300, The CDL system can perform the mission using the optimal transmission power.

FIG. 9 is a flowchart for explaining steps S835 to S850 of FIG. 8 in more detail.

9, the adaptive power control apparatus 300 receives a radio signal via a CDL from a surveillance / reconnaissance (i.e., air node) monitoring / reconnaissance enemy condition in the air and / or space, The power tp _k is confirmed (S910).

The adaptive power control apparatus 300 confirms the current RSSI (RSSI _k ) and the current FER (FER _k ) of the received radio signal (S920).

The adaptive power control apparatus 300 may predict the next FER (Pk _{+ 1} ) by learning the current RSSI and the current FER change pattern determined in step S920 based on the RERPM deep learning model (S930).

)보다 높으면(S940-Yes), 감시정찰기(10)의 송신 전력을 상향조정한 후 상향조정된 송신 전력으로 운용하도록 감시정찰기(10)에게 명령할 수 있다(S950, S960).The adaptive power control apparatus 300 determines whether the next FER predicted in step S930 is the maximum allowable error rate

(S940-Yes), the surveillance and reconnaissance system 10 can be instructed to operate with the transmission power of the surveillance reconnaissance system 10 adjusted upward and then adjusted to the upwardly adjusted transmission power (S950, S960).

)보다 낮으면(S970-Yes), 감시정찰기(10)의 송신 전력을 하향조정한 후 하향조정된 송신 전력으로 운용하도록 감시정찰기(10)에게 명령할 수 있다(S980, S960).On the other hand, the adaptive power control apparatus 300 determines whether the next FER predicted in step S930 is the minimum allowable error rate

(S970-Yes), it is possible to instruct the surveillance and reconnaissance apparatus 10 to operate the transmission power of the surveillance / reconnaissance device 10 downwardly and downwardly (S980, S960).

Meanwhile, in order to verify the performance of the public data link system for the adaptive power control of the surveillance apparatus according to the embodiment of the present invention, the CDL test bed can be constructed to analyze the accuracy of the RERPM.

10 is a diagram illustrating a CDL test bed according to an embodiment of the present invention.

Referring to FIG. 10, the CDL test bed includes a ground analysis system (ground node 1000) and a public / space platform (air node 1100).

The ground analysis system 1000 may correspond to the adaptive power control apparatus 300 described above and the aerial / space platform 1100 may correspond to the surveillance and reconnaissance system 10.

 The ground analysis system 1100 and the aerial / space platform 1100 are connected to the RF devices 1010 and 1110, the modem devices 1020 and 1120, the control devices 1030 and 1130, the packet generators 1040 and 1140, (1050, 1150). The packet generators 1040 and 1140 may be separately provided as shown in FIG.

The RF devices 1010 and 1110 use a device capable of controlling power in the range of 60dB and providing a communication distance of 100Km or more at a maximum transmission power of 10W.

The modem devices 1020 and 1120 support multiple access based on Frequency Division Multiple Access (FDMA) and support a modulation scheme of a phase shift keying (PSK) type.

The control devices 1030 and 1130 provide transmission output control functions in real time. In particular, the control device 1030 receives the RSSI and the FER, estimates the next FER using the RERPM module, and can calculate the transmit power to be applied by the public / space platform 1100 based on this.

A variable attenuator 1200 for changing the propagation environment is provided between the RF devices 1010 and 1110 and data is generated at the packet generators 1040 and 1140 so that the ground analysis system 1100 and the air / Space platform 1100 to transmit / receive data.

The performance verification of RERPM and the performance comparison of adaptive power control test are described below.

First, performance verification of RERPM will be described with reference to FIG.

When the test bed as shown in FIG. 10 is constructed, the electronic device can calculate the accuracy of the LSTM RERPM using the Root Mean Square Error (RMSE) as in Equation (10). The electronic device is a computer that receives test results from a test bed and verifies performance.

According to Equation (10), RMSE represents the difference between the FER (P _t ) predicted by the LSTM-based RERPM (i.e., LSTM RERPM) and the actual FER (Y _t ) according to the embodiment of the present invention.

11 is a graph comparing RMSE of MNN and CNN with LSTM RERPM according to an embodiment of the present invention.

The graph of FIG. 11 shows the RMSE change rate during the learning of 1000 by setting the learning rage to 0.01 in the same manner. As a result of the test, it can be seen that the difference between the actual value and the predicted value is small in the order of LSTMCNN? MNN, .

In the MNN model, the RMSE was close to 1 at the beginning, but the accuracy was very low. However, the performance of LSTM RERPM is worse than once. The CNN model initially had an RMSE of about 0.4, and after learning 10 or more times, the accuracy was improved to about 0.2, and there was no significant improvement in accuracy after that. The LSTM RERPM according to the embodiment of the present invention has an accuracy of 0.2 or less from the first learning, and it can be confirmed that the accuracy is gradually increased by learning 800 times. The result is that LSTM RERPM keeps track of pattern changes over time.

The next test is to analyze the RMSE by learning and changing the sequence length in the most accurate LSTM RERPM. As a result, as the sequence length increases from 3 to 7 as shown in [Table 1], the accuracy becomes higher. These results indicate that learning the channel change rate over a long period of time improves the prediction accuracy.

RERPM RMSE MNN RERPM 0.2723 CNN RERPM 0.2282 LSTM RERPM
(sequence length = 3) 0.0878 LSTM RERPM
(sequence length = 5) 0.0762 LSTM RERPM
(sequence length = 7) 0.0713

12 is a graph showing a result of analyzing a distribution of prediction result values after learning 10000 times.

Referring to FIG. 12, the MNN RERPM result, that is, the learning histogram for MNN RERPM generation, is evenly distributed with a reception error rate between 0 and 1 (0 to 100%). On the other hand, the learning histogram for CNN RERPM generation and the learning histogram for LSTM RERPM generation converge to 0 or 1, and the LSTM is better converged. In general, the receiving error rate converges to 0% and 100% based on the receiving sensitivity.

In conclusion, the performance comparison of RERPM shows that the prediction accuracy of LSTM RERPM is the highest, and that the longer the channel state information before the prediction time, the higher the prediction accuracy.

Next, performance of an LERM based RERPM-based adaptive power control (RERPM-APC) and a conventional predictive based adaptive power control (PB-APC) according to an embodiment of the present invention are compared.

Previous PB-APCs have been proposed by KH Lee, DH Lee, DH Lee, SJ Jung and HJ Choi, "A Resource Scheduling Based on Iterative Sorting for Long-Distance Airborne Tactical Communication in Hub Network," Journal of the Korean Institute of Communication Sciences , Vol. 39, No. 12, pp. 1250 ~ 1260, 2014. This is a power control method proposed by the present invention.

는 0.03,

는 0.005로 설정하고, PB-APC에서는 테스트베드에서 수신감도(-75dBm) 정보를 경험으로 획득하고, 상한 링크 품질 임계 값을 -68dBm, 하한 링크 품질 임계 값을 -70dBm으로 설정하여 테스트하였다.LSTM of RERPM-APC

0.03,

(-75dBm) information was experimentally obtained in the test bed in PB-APC, and the upper link quality threshold was set to -68dBm and the lower link quality threshold was set to -70dBm.

The test is a flight test in which the test is performed by changing the attenuation value every 1 second, and the results are shown in Figs. 13 and 14. Fig.

FIG. 13 is a graph showing a change in the reception frame error rate (FER) in the flight test, and FIG. 14 is a graph showing the change in the reception intensity (RSSI) in the flight test.

Referring to FIG. 13, in the conventional PB-APC, a reception error rate of 10 seconds out of 100 seconds is increased to 100%. In the case of FIG. 14, the situation in which the RSSI falls below the reception sensitivity occurs three times in 100 seconds , Communication disconnection occurs for a maximum of 10 seconds or more.

On the other hand, LSTM RERPM-APC does not show any danger of falling below reception sensitivity and guarantees stable communication for 100 seconds compared to PB-APC.

Meanwhile, the adaptive power control method in the public data link system according to the present invention may be provided in a recording medium readable by a computer by tangibly embodying a program of instructions for implementing it, It can be easily understood by engineers.

That is, the adaptive power control method in the public data link system according to the present invention can be implemented in a form of a program that can be performed through various computer means, and can be recorded in a computer-readable recording medium, May include program commands, data files, data structures, etc., alone or in combination. The computer-readable recording medium may be any of various types of media such as magnetic media such as hard disks, optical media such as CD-ROMs and DVDs, and optical disks such as ROMs, RAMs, flash memories, And hardware devices specifically configured to store and execute program instructions.

Accordingly, the present invention provides a program stored on a computer readable recording medium which is executed on a computer that controls at least one component of the public data link system to implement an adaptive power control method in a public data link system .

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is to be understood that the present invention is not limited to the disclosed exemplary embodiments, but, on the contrary, It will be apparent to those skilled in the art that numerous modifications and variations can be made to the present invention without departing from the scope of the present invention. Accordingly, all such modifications and variations are intended to be included within the scope of the present invention. Accordingly, the true scope of the present invention should be determined by the technical idea of the appended claims.

10: Surveillance Reconnaissance Device 100: RERPM Generation Device
200: RERPM downloader 300, 400: Adaptive power control device

Claims (12)

A Common Data Link (CDL) system for adaptive power control of a surveillance reconnaissance system,
The following FER is predicted by learning the variation pattern of Received Signal Strength Indicators (RSSI) and Frame Error Rate (FER) of a plurality of wireless signals collected through pre-offline learning based on a long-short term memory (LSTM) A RERPM generating unit for generating a RERPM (Long-Short term Memory Ratio Prediction Model) module; And
The RERPM module generated by the RERPM generation device is downloaded and operated so as to be operable, and the current RSSI and the current FER are checked from the radio signal received every power control period from the monitoring and reconnaissance device, and the current RSSI and the current FER To the RERPM module to predict the next FER and to control the transmission power of the surveillance unit based on the predicted next FER,
The adaptive power control apparatus includes:
A transceiver for communicating over the public data link with the surveillance probe for monitoring / scouting enemy status in the air and / or space, and receiving the radio signal for each power control period;
A power monitor for monitoring a radio signal received from the surveillance and reconnaissance device to confirm transmission power of the radio signal;
A link quality monitor for monitoring the received radio signal to determine a current RSSI and a current FER of the radio signal;
A RERPM module for receiving a current RSSI of the radio signal and a change pattern of a current FER, which are confirmed in the link quality monitor, and learning the RERPM deep learning basis to predict the next FER; And
Calculating a transmission power of the surveillance reconnaissance device based on the predicted next FER, and if the predicted next FER is higher than a maximum allowable error rate, adjusting a transmission power of the surveillance device identified in the power monitor, And a power control module to down-adjust the transmit power of the identified surveillance probe if the next FER is below a minimum acceptable error rate,
Wherein the transmission / reception unit transmits the transmission power calculated by the power control module to the surveillance / reconnaissance device. The method according to claim 1,
The RERPM generating apparatus includes:
(R = {R ₁ , R ₂ , R ₃ , ..., R _n }) of the _n radio signals transmitted from the public and / or space and the FERs (E = E ₁ , E ₂ , E ₃ ,..., E _n }),
Converts the RSSIs and FERs during the s (sequence length) time before the next time point to be predicted to use the change information according to time as input data into vector input data, and outputs the actual FER at the t point as actual output data Output data conversion unit; And
Output data (X _t, Y _t ) transformed by the input / output data conversion unit, state information of a previous LSTM cell, and output information learned in the previous LSTM cell, A RERPM module for generating the RERPM module including a fully connected layer for predicting the next FER using the final output information input from the LSTM layer; And a dip-running computer that includes a microprocessor and a microprocessor. 3. The method of claim 2,
The RERPM generating apparatus includes:
When the surveillance unit completes the actual mission through the public data link based on the transmission power calculated by the adaptive power control unit, loads the historical data of the RSSI and the FER generated during the mission to load the RERPM module Learning,
The RERPM module evolved by the re-learning is downloaded to the adaptive power control device to be applied to and operated on the common data link,
The adaptive power control apparatus includes:
And the FER is predicted based on the evolved RERPM. The public data link system for adaptive power control of a surveillance and reconnaissance device. delete delete The method according to claim 1,
The public data link system comprises:
A one-to-one structure in which one aerial / space platform corresponding to the surveillance / reconnaissance aircraft communicates with one ground analysis system corresponding to the adaptive power control apparatus, a plurality of aerial / N structure, an 1: N structure in which one aerial / space platform communicates with a number of ground analysis systems, and one of the plurality of aerial / space platforms relaying the terrestrial analysis system to another aerial / And a relay network structure, wherein the shared data link system comprises a relay network structure. The method according to claim 6,
In the case of the relay network structure,
Wherein one public / space platform acting as a relay among the plurality of public / space platforms predicts the next FER based on the RERPM module. &Lt; RTI ID = 0.0 &gt; .
1. An adaptive power control method of a control apparatus for controlling transmission power of a surveillance and reconnaissance apparatus in a common data link (CDL) system,
(A) The RERPM generating apparatus calculates a change pattern of Received Signal Strength Indicators (RSSI) and a Frame Error Rate (FER) of a plurality of radio signals collected through pre-offline learning by using a Long-Short Term Memory (LSTM) And generating a RERPM (Long-Short term Memory Reception Error Ratio Prediction Model) module for predicting a next FER; And
(B) The control device downloads the RERPM module generated in the step (A) so as to be operable, and then checks the current RSSI and the current FER from the radio signal received every power control period from the monitoring and reconnaissance device Inputting the current RSSI and the current FER to the RERPM module to predict the next FER and controlling the transmission power of the surveillance unit based on the predicted next FER,
The step (B)
(B1) the adaptive power control device receiving the radio signal for each power control period via a public data link from the surveillance / surveillance device monitoring / reconnaissance enemy status in air and / or space;
(B2) the adaptive power control unit monitors a radio signal received from the surveillance unit to confirm transmission power of the radio signal;
(B3) the adaptive power control unit monitors the received radio signal to identify a current RSSI and a current FER of the radio signal;
(B4) the adaptive power control apparatus learns the next FER by learning based on the RERPM deep learning based on the current RSSI and the current FER change pattern of the confirmed radio signal; And
(B5) if the predicted next FER is higher than the maximum allowable error rate, the adaptive power control unit adjusts the transmission power of the surveillance reconnaissance unit and transmits the upwardly adjusted transmission power to the surveillance unit; And
(B6) the adaptive power control unit adjusts the transmission power of the surveillance reconnaissance unit to a lower level if the predicted next FER is lower than a minimum acceptable error rate, and transmits the downlink adjusted transmission power to the surveillance unit &Lt; / RTI &gt; wherein the method comprises the steps of: 9. The method of claim 8,
The step (A)
(R ₁ , R ₂ , R ₃ ,..., R _n) of n radio signals transmitted from the surveillance and / or the space through the offline learning by the RERPM generating apparatus, Max normalization processing on the values of the FERs (E = {E ₁ , E ₂ , E ₃ ,..., E _n });
(A2) The RERPM generator converts the RSSIs and the FERs for a s (Sequence Length) time before a time point t, which is a next time to predict to use time-based change information as input data, into vector input data, converting the actual FER at time t into actual output data; And
(A3) The RERPM generating apparatus includes a plurality of LSTM cells, using the vector input / output data converted in the step (A2), the state information of the previous LSTM cell, and the output information learned in the previous LSTM cell Generating a RERPM module including an Fully Connected Layer for predicting the next FER using the final output information input from the LSTM layer, the LSTM layer outputting final output information in a last LSTM cell; &Lt; / RTI &gt; wherein the method comprises the steps of: 10. The method of claim 9,
The step (A)
(A4) If the surveillance unit completes the actual mission through the public data link based on the transmission power calculated in the step (B), the RERPM generating unit generates the history data of the RSSI and the FER generated during the mission Re-learning the RERPM module; And
(A5) The RERPM generation apparatus further includes a step of downloading the RERPM module evolved by the re-learning to the adaptive power control apparatus, and processing the RERPM module to be applied to and operating on the common data link Method of adaptive power control in a public data link system. delete delete

Images