KR102643816B1 - Offloading wireless communication system for improving energy efficiency based on spatial division multiple access - Google Patents

Abstract

The offloading wireless communication system of the present invention receives first processed data and second data, which are results of computational processing on the first data, from a device that performs computational processing on the first data, and operates on the second data. and a server that performs processing to generate second processed data, wherein the server can calculate a plurality of parameters related to offloading based on the first processed data and the second processed data.

Description

Offloading wireless communication system for improving energy efficiency based on spatial division multiple access {OFFLOADING WIRELESS COMMUNICATION SYSTEM FOR IMPROVING ENERGY EFFICIENCY BASED ON SPATIAL DIVISION MULTIPLE ACCESS}

The present invention relates to wireless communication systems, and more particularly, to wireless communication systems and methods for efficient data offloading.

Recently, with the development of the Internet of Things, the popularity of mobile devices such as smartphones, wearable devices, and smart sensors has increased, and users' expectations for running new applications on mobile devices are also increasing. However, due to size limitations, mobile devices have relatively low-performance processors and low-capacity batteries, making it difficult to run new applications that require high-level computation and low-latency services.

To solve this problem, offloading edge computing is being utilized, which allows mobile devices to offload computational tasks to nearby edge servers located at the edge of the network. In offloading edge computing systems, mobile devices generally have limited energy supply, so it is very important to efficiently use the energy consumed for data communication and computing. Therefore, there is a need to optimize several variables used for data offloading.

One object of the present invention relates to an offloading wireless communication system that calculates a plurality of parameters related to offloading.

An offloading wireless communication system according to an embodiment receives first processed data and second data, which are results of operation processing on the first data, from a device that performs operation processing on the first data, and sends the second data to the second data. and a server that generates second processed data by performing computational processing on the data, and the server can calculate a plurality of parameters related to offloading based on the first processed data and the second processed data.

Here, the server transmits information about the plurality of parameters to the device, and the device can set parameters related to data transmission based on the received information about the plurality of parameters.

Here, the server may set parameters related to data reception and parameters related to data processing based on the plurality of parameters.

Here, the plurality of parameters are at least one of a download time of the second data, an offload time of the second data, a transmission power of the device, a frequency of the local CPU, a computing time of the local CPU, and a downlink beamforming vector. may include.

Here, the server determines the number of bits calculated by the device included in the first processed data and the energy consumption of the device, and the number of bits calculated by the server included in the second processed data and the energy consumption of the server. Based on this, the energy efficiency of offloading can be calculated.

는,

(

: 디바이스가 계산한 비트 수,

: 서버가 계산한 비트 수,

: 디바이스의 에너지 소모량,

: 기지국 및 서버의 에너지 소모량,

: 기지국 및 서버 에너지 기여 비율)에 의해 산출되고, 상기 서버 에너지 기여 비율

는 0 내지 1 범위에서 변동 가능할 수 있다.where the energy efficiency

Is,

(

: Number of bits calculated by the device,

: Number of bits calculated by the server,

: Energy consumption of the device,

: Energy consumption of base stations and servers,

: Calculated by the base station and server energy contribution ratio), and the server energy contribution ratio

may be variable in the range of 0 to 1.

An offloading wireless communication method according to an embodiment is an offloading wireless communication method performed by at least one processor, comprising the steps of receiving first processed data and second data that are results of processing the first data from a device. ; generating second processed data by performing computational processing on the second data; and calculating a plurality of parameters related to offloading based on the first processed data and the second processed data.

Here, the method may further include transmitting information about the plurality of parameters to the device so that the device sets parameters related to data transmission based on the plurality of parameters.

Here, the method may further include setting parameters related to data reception and parameters related to data processing based on the plurality of parameters.

Here, the plurality of parameters are at least one of a download time of the second data, an offload time of the second data, a transmission power of the device, a frequency of the local CPU, a computing time of the local CPU, and a downlink beamforming vector. may include.

Here, based on the number of bits calculated by the device and the energy consumption of the device included in the first processed data, and the number of bits calculated by the server and the energy consumption of the server included in the second processed data, A step of calculating the energy efficiency of offloading may be further included.

(

: 디바이스가 계산한 비트 수,

: 서버가 계산한 비트 수,

: 디바이스의 에너지 소모량,

: 기지국 및 서버의 에너지 소모량,

: 기지국 및 서버 에너지 기여 비율)에 의해 에너지 효율성을 산출하는 단계; 및 상기 서버 에너지 기여 비율

를 0 내지 1 범위에서 설정하는 단계를 포함할 수 있다.Here, the step of calculating the energy efficiency of offloading is,

(

: Number of bits calculated by the device,

: Number of bits calculated by the server,

: Energy consumption of the device,

: Energy consumption of base stations and servers,

: Calculating energy efficiency by base station and server energy contribution ratio); and the server energy contribution rate

It may include setting in the range of 0 to 1.

Here, a computer program stored in a computer-readable recording medium may be provided to execute the offloading wireless communication method.

According to an embodiment of the present invention, an offloading wireless communication system can be provided that can analyze data, optimize a plurality of parameters, and increase the efficiency of data offloading by applying the optimized parameters.

1 is an environmental diagram of an offloading wireless communication system according to an embodiment.
Figure 2 is a diagram for explaining the data offloading process of an offloading wireless communication system in time series.
Figure 3 is a flowchart of a method for calculating a plurality of parameters to improve the efficiency of offloading wireless communication according to an embodiment.
Figure 4 is a flowchart of a portion of the process for calculating optimization parameters according to an embodiment.

The embodiments described in this specification are intended to clearly explain the idea of the present invention to those skilled in the art to which the present invention pertains, and the present invention is not limited to the embodiments described in this specification, and the present invention is not limited to the embodiments described in this specification. The scope should be construed to include modifications or variations that do not depart from the spirit of the present invention.

The terms used in this specification are general terms that are currently widely used as much as possible in consideration of their function in the present invention, but this may vary depending on the intention of those skilled in the art, precedents, or the emergence of new technology in the technical field to which the present invention belongs. You can. However, if a specific term is defined and used with an arbitrary meaning, the meaning of the term will be described separately. Therefore, the terms used in this specification should be interpreted based on the actual meaning of the term and the overall content of this specification, not just the name of the term.

The drawings attached to this specification are intended to easily explain the present invention, and the shapes shown in the drawings may be exaggerated as necessary to aid understanding of the present invention, so the present invention is not limited by the drawings.

In this specification, if it is determined that a detailed description of a known configuration or function related to the present invention may obscure the gist of the present invention, the detailed description thereof will be omitted as necessary.

1 is an environmental diagram of an offloading wireless communication system according to an embodiment.

Referring to FIG. 1, the offloading wireless communication system 1000 may include a server 100 and one or more devices 200. Server 100 may communicate with device 200 through base station 110 and one or more antennas of base station 110.

The server 100 is connected to one or more devices 200 and can exchange communication signals with each other. Although not shown in FIG. 1, a device that relays communication between the server 100 and the device 200 may exist. For example, another edge computing device may exist between the server 100 and the device 200 to transmit communication signals between the server 100 and the device 200. That is, the server 100 and the device 200 can communicate with each other without distance restrictions.

According to one embodiment, the server 100 may include a control unit, a communication unit, a storage unit, and an analysis unit. However, these components are not essential, and server 100 may have more or fewer components. Additionally, each component of the server 100 may be physically included in one server, or may be a distributed server distributed for each function.

The control unit may oversee the operation of the server 100. Specifically, the control unit can send control commands to the communication unit, storage unit, and analysis unit to execute the operations of each department.

Unless otherwise specified below, the operation of the server 100 may be interpreted as being performed under the control of the control unit.

The communication unit can connect the server 100 and an external device to communicate. In other words, the communication unit can transmit/receive data with external devices. For example, the communication unit may exchange data with the device 200.

According to one embodiment, the communication unit may receive data to be calculated from the device 200. For example, the data that the communication unit receives from the device 200 may be multimedia data with large capacity and complex data processing, such as data for game execution, data related to video processing, and data related to image processing.

Since data related to recent complex applications cannot be processed in real time inside the device 200, the device 200 transmits part of the data to the server 100 and leaves the data processing to the server 100. You can. Therefore, by simultaneously processing data in the server 100 and the device 200, real-time performance of the application can be guaranteed.

The communication unit may be a communication module that supports a wireless communication method. For example, the communication unit can obtain data from an external device through communication methods such as WiFi, Bluetooth, Zigbee, BLE (Bluetooth Low Energy), and RFID.

The storage unit can store various data and programs necessary for the server 100 to operate. The storage unit may store data acquired by the server 100 from an external device, data calculated by the local computer of the server 100, and power information during the calculation process.

The storage unit can store data temporarily or semi-permanently. Examples of storage include hard disk drive (HDD), solid state drive (SSD), flash memory, read-only memory (ROM), random access memory (RAM), or cloud storage ( Cloud Storage), etc. However, the storage unit is not limited to this and may be implemented with various modules for storing data. The storage unit may be provided in a form built into the server 100 or in a detachable form.

The analysis unit may analyze or calculate data acquired by the communication unit, classify the data, and generate new data. At this time, the analysis unit can perform data operations, such as calculating variables using various mathematical equations.

According to one embodiment, the analysis unit may calculate the number of bits to be operated and the energy used for the operation. Specifically, the analysis unit can calculate the number of bits and energy using arithmetic equations using the local CPU's scanning output and the effective capacitance coefficient and leakage current of the processor chip.

Additionally, the analysis unit can calculate the energy used for offloading and the number of bits that can be transmitted through offloading. Specifically, the analysis unit calculates the energy used for offloading using the circuit power consumed, and calculates the number of bits that can be transmitted through offloading using the system bandwidth, effective uplink channel gain, etc. .

Additionally, the analysis unit can calculate the energy and time consumed by the server when performing calculation tasks. Specifically, the analysis unit can calculate energy and time using a downlink signal, a data signal, and a beamforming vector of the data signal.

Additionally, the analysis unit may calculate the energy efficiency of the offloading wireless communication system 1000. Specifically, energy efficiency can be calculated based on the number of bits calculated in the server 100 and the device 200 and the energy consumed during calculation.

Additionally, the analysis unit can calculate multiple parameters to increase offloading efficiency. Specifically, the analysis unit can calculate optimized values for multiple parameters under various constraint conditions.

The device 200 is an electronic device and may be a mobile device with a relatively small capacity. The device 200 may be provided in the form of an electronic circuit such as a CPU chip that processes electrical signals in hardware and performs various functions (communication, storage, analysis, etc.). Additionally, the device 200 may be provided in the form of a program configured to perform various functions in software.

In the offloading wireless communication system 1000, the device 200 may process some of the data to be processed within the device 200 and transmit the remaining part to the server 100. For example, the device 200 may perform calculation processing on the first data within the device 200 and transmit the second data to the server 100. At this time, the device 200 may transmit the first data transmission signal 10 including the second data to the server 100.

The server 100 may receive the first data transmission signal 10 and perform calculation processing on the second data. The server 100 may generate second processed data as a result of performing computational processing on the second data. The server 100 may transmit second processed data to the device 200 so that the device 200 can execute the application. At this time, the server 100 may transmit the second data transmission signal 20 including second processed data to the device 200.

The server 100 may perform data processing on one or more devices. The server 100 can simultaneously transmit/receive data to and from a plurality of devices, such as the first device 200, second devices through N-th devices, and perform data processing. At this time, the server 100 may allocate memory so that the memory corresponds to each device so that it can process data received from each device.

Figure 2 is a diagram for explaining the data offloading process of an offloading wireless communication system in time series.

Referring to FIG. 2, offloading wireless communication may be performed by an offloading period (D1), a local CPU computing period (D2), and a calculation result downloading period (D3). These periods may progress in time series in the order of D1, D2, and then D3.

The offloading period D1 may be a period in which the device 200 generates data to be transmitted to the server 100 and the server 100 receives the data. At this time, the server 100 can simultaneously receive data from one or more devices using a spatial division multiple access method through a plurality of antennas connected to the base station 110.

The process of generating data to be transmitted by the device 200 to the server 100 may be a process of dividing some of the data to be processed for application execution. Also, at this time, the device 200 may be in the process of generating separate data that includes both data to be processed and information necessary for data processing.

The local CPU computing period (D2) may be a period in which the server 100 performs a data processing process on the received data. The server 100 may select a local CPU during the D2 period and allocate data to the selected local CPU.

During the D2 period, the local CPU may calculate data received from the device 200 and generate data as a result of the calculation. The server 100 or the local CPU may continuously store information about the energy used in the calculation process and the number of calculated bits in the storage.

The calculation result downloading period D3 may be a period in which data processing result data calculated by the local CPU is transmitted to the device 200 through the communication unit of the server 100. Additionally, the D3 period may be a period in which the server 100 receives information about the number of bits calculated by the device 200 and energy consumed for calculation from the device 200 through the communication unit.

The server 100 performs a calculation process to optimize a plurality of parameters related to the offloading system based on the data processing results of the local CPU and the data processing results of the device 200 during or after the D3 period. You can.

Figure 3 is a flowchart of a method for calculating a plurality of parameters to improve the efficiency of offloading wireless communication according to an embodiment.

Referring to FIG. 3, a method of calculating a plurality of parameters according to an embodiment includes receiving first processed data and second data (S110), generating second processed data (S120), and calculating energy efficiency. It may include calculating a step (S130), calculating a plurality of parameters (S140), and setting parameters (S150).

The step of receiving the first processed data and the second data (S110) may be a step in which the server 100 receives the first processed data and the second data from the device 200. At this time, the server 100 may simultaneously receive the first processed data and the second data from the device 200, or may receive some of the two first. Preferably, in order to provide a low-latency service for application execution, the device 200 may transmit the second data to the server 100 before processing the first data.

Specifically, the device 200 may perform arithmetic processing on first data and generate first processed data. At this time, the first processed data may include data calculated based on the first data and information related to energy used to process the first data. For example, the first processed data may include the number of bits calculated based on the first data, the amount of energy used to process the first data, etc.

The second data may be some of the data that needs to be processed to execute a specific application. That is, the device 200 divides the data to be processed to run a specific application into first data and second data, and processes some of the first data within the device 200 and processes the second data into the server. It can be transmitted to the server 100 for processing. Additionally, the device 200 may divide data to be processed to execute a specific application into three or more pieces of data instead of two and process them together with the server 100 or another device.

The step of generating second processed data (S120) may be a step in which the server 100 performs data processing on the second data received from the device 200. Specifically, step S120 may be a step in which the local CPU within the server 100 performs data processing on the second data.

The server 100 or the local CPU may perform computational processing on the second data and generate second processed data. At this time, the second processed data may include data calculated based on the second data and information related to energy used to process the second data. For example, the second processed data may include the number of bits calculated based on the second data, the amount of energy used to process the second data, etc.

In the step of calculating energy efficiency (S130), the server 100 processes data based on the first processed data received from the device 200 and the second processed data that is the result of data processing by the local CPU of the server 100. This may be a step in calculating energy efficiency.

를 산출하는 과정을 이하에서 점진적으로 설명한다. 여기서, 디바이스(200)의 계산 회로는 통신 회로와 분리된 것으로 가정한다. 즉, 디바이스(200)는 컴퓨팅과 오프로딩을 동시에 수행할 수 있는 것으로 가정한다. 또한,

는 는 도 1에 표시된 바와 같이, i번째 디바이스를 의미하는 것일 수 있다.energy efficiency

The process of calculating is explained gradually below. Here, it is assumed that the calculation circuit of the device 200 is separated from the communication circuit. That is, it is assumed that the device 200 can perform computing and offloading simultaneously. also,

may mean the i-th device, as shown in FIG. 1.

이

의 로컬 계산 시간일 때, 로컬 계산 시간은

를 만족할 수 있다.

가

의 계산 작업 1비트를 계산하는데 소모되는 CPU 사이클의 수이고,

가

의 로컬 CPU 주파수일 때,

가 로컬 컴퓨팅을 통해 계산하는 비트 수는 아래 [수학식 1]과 같이 정의되고, 소모 에너지는 아래 [수학식 2]와 같이 정의될 수 있다.

this

When the local computation time is, the local computation time is

can be satisfied.

go

is the number of CPU cycles consumed to calculate 1 bit of computational work,

go

When the local CPU frequency of

The number of bits calculated through local computing can be defined as [Equation 1] below, and the consumed energy can be defined as [Equation 2] below.

[Equation 1]

[Equation 2]

는

에서 프로세서 칩의 유효 정전용량 계수이고,

는 누수 전류에 의해 소비되는 전력이다.here,

Is

is the effective capacitance coefficient of the processor chip,

is the power consumed by leakage current.

와

의 시간이 할당될 수 있다.For offloading and downloading of device calculation tasks, the device 200 may use spatial division multiple access (SDMA). The device 200 can perform simultaneous access as shown in FIG. 2 based on the space division multiple access method. At this time, each offload and download of device data

and

of time can be allocated.

의 전송 전력을

(

,

는

의 최대 전송 전력)라 할 때, 오프로딩을 위해

에 의해 소비되는 에너지는 [수학식 3]과 같이 정의될 수 있다.for off-roading

transmit power of

(

,

Is

(maximum transmission power of), for offloading

The energy consumed by can be defined as [Equation 3].

[Equation 3]

는

가 소비하는 회로 전력이다. 여기서, 기지국(110)은 제로 포싱(zero-forcing) 수신기

를 사용한다고 가정하며,

일 수 있다. 계산 오프로딩을 통해

가 전송할 수 있는 비트 수는 [수학식 4]와 같이 계산될 수 있다.At this time,

Is

is the circuit power consumed. Here, the base station 110 is a zero-forcing receiver.

Assuming you use ,

It can be. Through computation offloading

The number of bits that can be transmitted can be calculated as in [Equation 4].

[Equation 4]

는 시스템 대역폭이고,

는 유효 업링크 채널 이득이며,

과

는 각각 기지국 잡음 세기와

의

번째 행을 의미한다.At this time,

is the system bandwidth,

is the effective uplink channel gain,

class

are the base station noise intensity and

of

It means the second row.

의 에너지를 소비할 수 있다. 여기서,

는 기지국이 신호를 수신하기 위한 기지국 안테나당 회로 전력 소비량이다.During the offload time, the base station 110 is configured to receive offloaded bits.

can consume energy. here,

is the circuit power consumption per base station antenna for the base station to receive signals.

의 작업을 수행하고, 비트의 출력 결과를 생성할 수 있다.

가

의 계산 작업의 단위 입력 비트당 출력 비트 수라고 할 때,

의 계산 작업을 실행하기 위해 서버(100)가 소비하는 에너지는 아래 [수학식 5]로 정의되고, 소비하는 시간은 아래 [수학식 6]으로 정의될 수 있다.Next, the server 100

It can perform operations and produce output results in bits.

go

Given the number of output bits per unit input bit of the calculation operation,

The energy consumed by the server 100 to execute the calculation task can be defined as [Equation 5] below, and the time consumed can be defined as [Equation 6] below.

[Equation 5]

[Equation 6]

일 수 있다. 이때,

는

에게 전송하는 데이터 신호이고,

는 해당 데이터 신호의 빔포밍 벡터이다. 계산 결과를 전송하기 위해 기지국(110)이 소비하는 에너지는 아래 [수학식 7]과 같이 정의될 수 있다.In order to transmit the calculation result, the downlink signal sent from the base station 110 to the device 200 is

It can be. At this time,

Is

It is a data signal transmitted to

is the beamforming vector of the corresponding data signal. The energy consumed by the base station 110 to transmit the calculation result can be defined as [Equation 7] below.

[Equation 7]

는 기지국 안테나당 회로 전력 소비량일 수 있다.here,

may be the circuit power consumption per base station antenna.

동안 기지국(110)이

로 전송할 수 있는 비트의 수는 아래 [수학식 8]과 같이 정의될 수 있다.Also, the download time

While the base station 110

The number of bits that can be transmitted can be defined as [Equation 8] below.

[Equation 8]

가 소비하는 에너지는

일 수 있고, 이때

는

가 수신 신호 처리를 위해 소모하는 전력이다.To receive calculated results

The energy consumed is

It may be, and at this time

Is

This is the power consumed for processing the received signal.

Finally, the energy efficiency of the offloading wireless communication system 1000, which is defined as the ratio of calculation bits to the total energy consumption of the offloading wireless communication system 1000, can be defined as [Equation 9] below.

[Equation 9]

는 에너지 효율성에서 기지국 및 엣지 서버의 에너지 소비량이 전체 엣지 컴퓨팅 시스템의 에너지 소비량에서 차지하는 비율이다. 즉,

는 기지국 및 서버 에너지 기여 비율일 수 있고, 0 내지 1 사이의 값을 가질 수 있다. 예를 들어,

가 0일 경우는 기지국(110) 및 서버(100)의 에너지 소비를 에너지 효율성에 고려하지 않는 경우를 의미하는 것일 수 있다.here,

In energy efficiency, is the ratio of the energy consumption of the base station and edge server to the energy consumption of the entire edge computing system. in other words,

may be the base station and server energy contribution ratio, and may have a value between 0 and 1. for example,

If is 0, this may mean that the energy consumption of the base station 110 and the server 100 is not considered for energy efficiency.

는

로도 쓰일 수 있다. [수학식 9]의 에너지 효율성은 의미를 고려하여, 아래 [수학식 10]으로 간단히 나타낼 수 있다.Energy efficiency of [Equation 9]

Is

It can also be used as . Considering the meaning, the energy efficiency of [Equation 9] can be simply expressed as [Equation 10] below.

[Equation 10]

: 에너지 효율성,

: 디바이스가 계산한 비트 수,

: 서버가 계산한 비트 수,

: 디바이스의 에너지 소모량,

: 기지국 및 서버의 에너지 소모량,

: 기지국 및 서버 에너지 기여 비율)(

: Energy efficiency,

: Number of bits calculated by the device,

: Number of bits calculated by the server,

: Energy consumption of the device,

: Energy consumption of base stations and servers,

: Base station and server energy contribution ratio)

The step of calculating a plurality of parameters (S140) may be a step in which the server 100 calculates a plurality of optimized parameters under constraint conditions based on the energy efficiency calculated in step S130. Specifically, the server 100 may optimize a plurality of parameters based on the energy efficiency of [Equation 9] or [Equation 10] and the [Constraints] below.

At this time, the plurality of parameters are variables related to the energy efficiency of the offloading wireless communication system 1000, such as processing time, power usage, CPU frequency, beamforming vector, etc. of the system 1000 regardless of the type and content of the data being processed. It may be related.

, 오프로드 시간

, 사용자의 전송 전력

, 로컬 CPU 주파수

, 로컬 컴퓨팅 시간

및 다운링크 빔포밍 벡터

를 포함할 수 있다. 즉, 서버(100)가 풀어야 할 최적화 변수의 집합은

로 정의될 수 있다.Specifically, the plurality of parameters is the download time

, off-road time

, the transmission power of the user

, local CPU frequency

, local computing time

and downlink beamforming vectors.

may include. In other words, the set of optimization variables that the server 100 must solve is

It can be defined as:

[Constraints]

는

의 최대 사용 가능한 에너지이고,

는

에 대한 최소 계산 비트 수이며,

는 기지국의 최대 전송 전력이다. 또한 C1, C2, C3, C4, C5, C6는 각 디바이스의 에너지 제약조건, 디바이스의 최소 계산 비트 수 제약조건, 다운로드 비트 수 제약조건, 기지국의 전송 전력 제약조건, 디바이스의 전송 전력 제약조건, 총 시간 제약조건이다.here,

Is

is the maximum available energy,

Is

is the minimum number of computational bits for ,

is the maximum transmission power of the base station. In addition, C1, C2, C3, C4, C5, and C6 are the energy constraints of each device, the device's minimum computational bit number constraint, the download bit number constraint, the base station's transmission power constraint, the device's transmission power constraint, and the total It is a time constraint.

The server 100 may introduce auxiliary variables to solve the non-convex problem before optimizing the energy efficiency of [Equation 9] under the constraints of C1 to C7. Specifically, optimizing the variables for the above energy efficiency is a non-convex problem due to the product term of the fractional objective function and the control variable. Therefore, to solve the non-convex problem, the server 100 may introduce the following [auxiliary variable].

[auxiliary variable]

이다. 위의 보조 변수들을 활용하여 [수학식 9] 및 [제약 조건]의 문제는 아래와 같이 [수학식 11]로 다시 쓸 수 있다. here,

am. Using the above auxiliary variables, the problems of [Equation 9] and [Constraints] can be rewritten as [Equation 11] as follows.

[Equation 11]

이다. 로컬 컴퓨팅 에너지

을 도입하여 [수학식 11]의 문제를 아래 [수학식 12]의 내층 문제와 아래 [수학식 13]의 외층 문제로 나누어 변환할 수 있다.here,

am. local computing energy

By introducing , the problem of [Equation 11] can be converted into the inner layer problem of [Equation 12] below and the outer layer problem of [Equation 13] below.

에 대한 로컬 컴퓨팅 주파수

와 로컬 컴퓨팅 시간

을 최적화하여 로컬 컴퓨팅 비트

을 최대화하기 위한 것이다. 외층 문제는

을 최적화하여 에너지 효율성을 최대화하기 위한 것이다. 구체적으로 내층 문제와 외층 문제는 다음과 같이 쓸 수 있다.The inner layer problem is given local computing energy

local computing frequency for

and local compute time

Optimize your local compute bits

This is to maximize. The outer layer problem is

This is to maximize energy efficiency by optimizing. Specifically, the inner layer problem and the outer layer problem can be written as follows.

[Equation 12]

[Equation 13]

이고, 목표함수의 분자와 분모는 각각 다음과 같다.Here, the optimization variable is

, and the numerator and denominator of the objective function are respectively as follows.

와

, 그리고 최적의 목표함수 값

은 아래의 [수학식 14] 내지 [수학식 16]과 같이 얻을 수 있다.Solution to inner layer problem

and

, and the optimal objective function value

can be obtained as in [Equation 14] to [Equation 16] below.

[Equation 14]

[Equation 15]

[Equation 16]

은 오목(concave) 함수이고 목적함수의 분모는 아핀 함수이기 때문에, 목적함수는 준오목(quasi-concave) 함수가 된다. 따라서, 딘켈바흐(Dinkelbach) 알고리즘을 사용하여 외층 문제를 해결할 수 있으며, 그 과정은 아래와 같다. 외층 문제는

에 대해 아래 [수학식 17]의 문제와 등가일 수 있다.The outer layer problem is expressed as fractional programming.

Since is a concave function and the denominator of the objective function is an affine function, the objective function is a quasi-concave function. Therefore, the outer layer problem can be solved using the Dinkelbach algorithm, and the process is as follows. The outer layer problem is

It may be equivalent to the problem in [Equation 17] below.

[Equation 17]

이고,

이다.here,

ego,

am.

는 [수학식 13]의 문제의 최적 목표함수 값이며, [수학식 17]의 문제의 최적의 목표함수 값을 0으로 만들 수 있다.

와 이에 해당하는 최적의 솔루션

를 얻기 위해서는, [수학식 17]의 문제의 최적의 목표함수 값이 0가 될 때까지 다음의 과정을 반복할 수 있다.also,

is the optimal objective function value of the problem in [Equation 13], and the optimal objective function value of the problem in [Equation 17] can be set to 0.

and the corresponding optimal solution

To obtain, the following process can be repeated until the optimal objective function value of the problem in [Equation 17] becomes 0.

에 대해, [수학식 17]의 문제의 솔루션

를 얻는다.Step 1. Given

For the solution of the problem of [Equation 17]

get

를 이용하여,

를 아래 [수학식 18]로 업데이트한다.Step 2. Obtained in Step 1

Using,

Update to [Equation 18] below.

[Equation 18]

에 대한 [수학식 17]의 문제를 해결하기 위한 방법은 다음과 같다.Step 1, i.e. given

The method to solve the problem of [Equation 17] is as follows.

The objective function of the problem in [Equation 17] is a concave function, but the problem in [Equation 17] is non-convex due to the expression of constraint C3. However, because the non-convex function has a difference-of-concave structure, the problem of [Equation 17] is a difference-of-concave problem. Therefore, the concave difference algorithm is used to solve the problem of [Equation 17].

The concave difference algorithm can be guaranteed to converge to a stopping point that satisfies the Karush-Kuhn-Tucker (KKT) condition of the original problem. The basic idea of the concave difference algorithm is to repeatedly solve the approximate convex problem, which can be obtained by approximating the second concave function of the concave difference structural function with an affine function.

Specifically, to solve the problem of [Equation 17], the problems of [Equation 19] and [Equation 20] below are repeatedly solved. The iterative solving process can be seen in the flowchart of FIG. 4.

에 대하여 반복수 l을 계속적으로 업데이트함으로써 [수학식 17]의 문제를 해결하는 방법을 확인할 수 있다.Figure 4 is a flowchart of a portion of the process for calculating optimization parameters according to an embodiment. Referring to Figure 4, given

It is possible to check how to solve the problem of [Equation 17] by continuously updating the repetition number l.

[Equation 19]

은 반복수를 의미한다.here,

means the number of repetitions.

는 근사 아핀 함수로서, 아래 [수학식 20]과 같이 쓸 수 있다.

is an approximate affine function and can be written as [Equation 20] below.

[Equation 20]

이고,

는

의 에 대한

포인트에서의 기울기(gradient)이다.At this time,

ego,

Is

of for

It is the gradient at the point.

If the server 100 repeatedly solves the problem of [Equation 19] until the solution converges, the server 100 can eventually obtain the optimal solution to the problem of [Equation 17]. At this time, since the problem of [Equation 19] is a convex problem, it can be solved using standard convex optimization techniques such as the interior point method.

The server 100 can ultimately optimize a plurality of parameters using [Equation 11] to [Equation 20] under constraint conditions for [Equation 9] related to energy efficiency. That is, the server 100 can calculate a plurality of optimized parameters.

The parameter setting step (S150) may be a step of applying a plurality of parameters calculated in step S140 to the offloading wireless communication system 1000.

The server 100 may transmit a plurality of calculated parameters to the device 200. The device 200 may update parameters based on a plurality of received parameter information. Specifically, the device 200 may set parameters related to data transmission for transmitting data to the server 100 based on a plurality of received parameter information.

Additionally, the server 100 may update parameters related to the base station 110 and the local CPU based on the calculated plurality of parameters. Specifically, the server 100 may set parameters related to data reception for receiving data from the device 200 and parameters related to data processing for processing data using a local CPU, based on a plurality of parameters.

After step S150, the server 100 and the device 200 may perform the next offloading process based on the set parameters. That is, the server 100 can calculate energy efficiency based on the results of the previous offloading process and optimize a plurality of parameters through energy efficiency. The server 100 and the device 200 reflect the plurality of optimized parameters and perform the next offloading, so that the energy efficiency of the offloading wireless communication system 1000 can be improved compared to before.

A plurality of parameters may be continuously updated by analysis and calculation of the server 100 each time offloading is performed. That is, the offloading wireless communication system 1000 can maximize energy efficiency through optimal parameters by performing energy efficiency calculation and multiple parameter optimization processes each time.

The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded on a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, etc., singly or in combination. Program instructions recorded on the medium may be specially designed and configured for the embodiment or may be known and available to those skilled in the art of computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical media such as CD-ROMs and DVDs, and magnetic media such as floptical disks. -Includes optical media (magneto-optical media) and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, flash memory, etc. Examples of program instructions include machine language code, such as that produced by a compiler, as well as high-level language code that can be executed by a computer using an interpreter, etc. The hardware devices described above may be configured to operate as one or more software modules to perform the operations of the embodiments, and vice versa.

As described above, although the embodiments have been described with limited examples and drawings, various modifications and variations can be made by those skilled in the art from the above description. For example, the described techniques are performed in a different order than the described method, and/or components of the described system, structure, device, circuit, etc. are combined or combined in a different form than the described method, or other components are used. Alternatively, appropriate results may be achieved even if substituted or substituted by an equivalent.

Therefore, other implementations, other embodiments, and equivalents of the claims also fall within the scope of the claims described below.

Claims (13)

Receive first processed data and second data, which are results of computational processing for the first data, from a device that performs computational processing on the first data, and perform computational processing on the second data to produce second processed data. Contains a server to create,
The server calculates a plurality of parameters related to offloading based on the first processed data and the second processed data,
The server calculates offloading energy efficiency using the equation below,

( : Energy efficiency, :Number of bits calculated by the device, :Number of bits calculated by the server, :Energy consumption of the device, :Energy consumption of base stations and servers, : Base station and server energy contribution ratio)
remind is variable in the range of 0 to 1
Off-roading wireless communication system.
According to paragraph 1,
The server transmits information about the plurality of parameters to the device,
The device sets parameters related to data transmission based on the information about the plurality of parameters received.
Off-roading wireless communication system.
According to paragraph 1,
The server sets parameters related to data reception and parameters related to data processing based on the plurality of parameters.
Off-roading wireless communication system.
According to paragraph 1,
The plurality of parameters include at least one of a download time of the second data, an offload time of the second data, a transmission power of the device, a frequency of the local CPU, a computing time of the local CPU, and a downlink beamforming vector. doing
Off-roading wireless communication system.
According to paragraph 1,
The server is based on the number of bits calculated by the device and the energy consumption of the device included in the first processed data, and the number of bits calculated by the server and the energy consumption of the server included in the second processed data. , which calculates the energy efficiency of offloading.
Off-roading wireless communication system.
delete In the server offloading wireless communication method,
Receiving first processed data and second data, which are results of operation processing on the first data, from a device;
generating second processed data by performing computational processing on the second data;
calculating a plurality of parameters related to offloading based on the first processed data and the second processed data; and
Comprising the step of calculating the energy efficiency of the offloading,
The energy efficiency is calculated by the equation below,

( : Energy efficiency, :Number of bits calculated by the device, :Number of bits calculated by the server, :Energy consumption of the device, :Energy consumption of base stations and servers, : Base station and server energy contribution ratio)
remind is variable in the range of 0 to 1
Offloading wireless communication method.
In clause 7,
Further comprising transmitting information about the plurality of parameters to the device so that the device sets parameters related to data transmission based on the plurality of parameters.
Offloading wireless communication method.
In clause 7,
Further comprising setting parameters related to data reception and parameters related to data processing based on the plurality of parameters.
Offloading wireless communication method.
In clause 7,
The plurality of parameters include at least one of a download time of the second data, an offload time of the second data, a transmission power of the device, a frequency of the local CPU, a computing time of the local CPU, and a downlink beamforming vector. doing
Offloading wireless communication method.
In clause 7,
Based on the number of bits calculated by the device and the energy consumption of the device included in the first processed data, and the number of bits calculated by the server and the energy consumption of the server included in the second processed data, offloading Further comprising the step of calculating the energy efficiency of
Offloading wireless communication method.
delete A computer program stored in a computer-readable recording medium to execute the offloading wireless communication method according to any one of claims 7 to 11.