TW202025799A

TW202025799A - Dispatching method and edge computing system

Info

Publication number: TW202025799A
Application number: TW107145893A
Authority: TW
Inventors: 周永明; 朱峰森
Original assignee: 未來市股份有限公司
Priority date: 2018-12-19
Filing date: 2018-12-19
Publication date: 2020-07-01
Also published as: TWI693845B

Abstract

A task dispatching method, applied in an edge computing system comprising an edge computing device and a mobile device, is disclosed. The method comprises sending, by the mobile device, a resource inquiry message to the edge computing device through a connection, wherein the connection comprises a wireless connection, a one-way transmission latency of the connection is less than 10 milliseconds, and the resource inquiry message comprises an inquiry of which resource type the edge computing device is equipped with; sending, by the edge computing device, a resource response message corresponding to the resource inquiry message; and determining, by the mobile device, to dispatch the second type of computing task to the second computing device when the resource response message indicates that the edge computing device comprises the second computing device equipped with the second type of processor.

Description

分派方法及邊緣計算系統Dispatch method and edge computing system

本發明係指一種分派方法及邊緣計算系統，尤指一種可取得邊緣計算的資源型態資訊的分派方法及邊緣計算系統。The present invention refers to an allocation method and edge computing system, and more particularly to an allocation method and edge computing system that can obtain resource type information of edge computing.

邊緣計算（Edge Computing）近年來吸引許多目光，相較於雲端計算，邊緣計算通常設置於基地台附近，而邊緣計算裝置與使用者之間的傳輸延遲（latency）通常較雲端計算裝置與使用者之間的傳輸延遲小很多。因此，邊緣計算裝置適用於進行及時應用程式的運算工作。Edge computing has attracted a lot of attention in recent years. Compared with cloud computing, edge computing is usually located near the base station, and the transmission delay between edge computing devices and users is usually higher than that between cloud computing devices and users. The transmission delay between is much smaller. Therefore, the edge computing device is suitable for performing computing tasks of real-time applications.

於某些情況中，某些邊緣計算裝置可具有額外的處理器，如圖形處理器（Graphic Processing Unit，GPU）或是張量處理器（Tensor Processing Unit，TPU）。然而，行動裝置不確定邊緣計算裝置具有哪種類型的處理器。舉例來說，行動裝置不知道邊緣計算裝置具有GPU而可以有效率地執行三維繪圖運算。為了避免延遲風險，行動裝置不會將三維繪圖運算交給邊緣計算裝置來處理，而使整體邊緣計算系統的效益降低。In some cases, some edge computing devices may have additional processors, such as a graphics processing unit (GPU) or a tensor processing unit (TPU). However, mobile devices are not sure which type of processor the edge computing device has. For example, mobile devices do not know that edge computing devices have GPUs and can efficiently perform 3D graphics operations. In order to avoid the risk of delay, mobile devices will not hand over 3D graphics operations to edge computing devices for processing, which reduces the effectiveness of the overall edge computing system.

因此，本發明之主要目的即在於提供一種可取得邊緣計算的資源型態資訊的分派方法及邊緣計算系統，以改善習知技術的缺點。Therefore, the main purpose of the present invention is to provide a method and an edge computing system that can obtain resource type information of edge computing to improve the shortcomings of the conventional technology.

本發明實施例揭露一種分派方法，應用於一邊緣計算系統，包括一邊緣計算裝置以及一行動裝置，該邊緣計算裝置包括一第一計算裝置及一第二計算裝置中至少一者，該行動裝置具有欲分派的一第一類型計算工作以及一第二類型計算工作，其中該第一計算裝置具有一第一類型處理器但不具有一第二類型處理器，該第二計算裝置具有該第一類型處理器以及該第二類型處理器，該方法包括該行動裝置透過一連線傳送一資源詢問訊息至該邊緣計算裝置，其中該連線包括一無線連線，該連線的一單向傳輸延遲小於10毫秒，該資源詢問訊息用來詢問該邊緣計算裝置具有何種類型的資源；該邊緣計算裝置傳送對應於該資源詢問訊息的一資源回覆訊息，其中該資源回覆訊息指出該邊緣計算裝置是否包括具有該第二類型處理器的該第二計算裝置；以及當該資源回覆訊息指出該邊緣計算裝置包括具有該第二類型處理器的該第二計算裝置時，該行動裝置決定分派該第二類型計算工作至該邊緣計算裝置；其中，相較於該第二類型處理器，該第一類型處理器更適合執行該第一類型計算工作；相較於該第一類型處理器，該第二類型處理器更適合執行該第二類型計算工作。The embodiment of the present invention discloses a dispatch method applied to an edge computing system, including an edge computing device and a mobile device. The edge computing device includes at least one of a first computing device and a second computing device. The mobile device There is a first type of calculation task and a second type of calculation task to be dispatched, wherein the first computing device has a first type of processor but not a second type of processor, and the second computing device has the first type of processor. Type processor and the second type processor, the method includes the mobile device sending a resource query message to the edge computing device through a connection, wherein the connection includes a wireless connection, and a one-way transmission of the connection With a delay of less than 10 milliseconds, the resource inquiry message is used to inquire what type of resource the edge computing device has; the edge computing device sends a resource response message corresponding to the resource inquiry message, wherein the resource response message indicates the edge computing device Whether the second computing device with the second type processor is included; and when the resource response message indicates that the edge computing device includes the second computing device with the second type processor, the mobile device decides to allocate the second computing device The second type of computing work is applied to the edge computing device; wherein, compared to the second type of processor, the first type of processor is more suitable for performing the first type of computing work; compared to the first type of processor, the second type of processor The second type of processor is more suitable for performing the second type of calculation work.

本發明實施例揭露一種邊緣計算系統，包括一行動裝置，具有欲分派的一第一類型計算工作及一第二類型計算工作，用來透過一連線傳送一資源詢問訊息至該邊緣計算裝置，其中該連線包括一無線連線，該連線的一單向傳輸延遲小於10毫秒，該資源詢問訊息用來詢問該邊緣計算裝置具有何種類型的資源；以及一邊緣計算裝置，包括一第一計算裝置及一第二計算裝置中至少一者，用來傳送對應於該資源詢問訊息的一資源回覆訊息，其中該資源回覆訊息指出該邊緣計算裝置是否包括具有該第二類型處理器的該第二計算裝置，該第一計算裝置具有一第一類型處理器但不具有一第二類型處理器，該第二計算裝置具有該第一類型處理器以及該第二類型處理器；其中，當該資源回覆訊息指出該邊緣計算裝置包括具有該第一類型處理器及該第二類型處理器的該第二計算裝置時，該行動裝置決定分派該第二類型計算工作至該邊緣計算裝置；其中，相較於該第二類型處理器，該第一類型處理器更適合執行該第一類型計算工作；相較於該第一類型處理器，該第二類型處理器更適合執行該第二類型計算工作。An embodiment of the present invention discloses an edge computing system, which includes a mobile device with a first type of computing task and a second type of computing task to be assigned, for sending a resource query message to the edge computing device through a connection. The connection includes a wireless connection, a one-way transmission delay of the connection is less than 10 milliseconds, the resource query message is used to inquire what type of resource the edge computing device has; and an edge computing device, including a first At least one of a computing device and a second computing device is used to send a resource response message corresponding to the resource query message, wherein the resource response message indicates whether the edge computing device includes the second type processor A second computing device, the first computing device has a first type of processor but not a second type of processor, the second computing device has the first type of processor and the second type of processor; wherein, when When the resource response message indicates that the edge computing device includes the second computing device with the first type processor and the second type processor, the mobile device decides to assign the second type of computing task to the edge computing device; wherein Compared with the second type of processor, the first type of processor is more suitable for performing the first type of computing work; compared to the first type of processor, the second type of processor is more suitable for performing the second type Calculation work.

本發明實施例揭露一種行動裝置，應用於一邊緣計算系統，其中，該邊緣計算系統包括一邊緣計算裝置，該邊緣計算裝置包括一第一計算裝置及一第二計算裝置中至少一者，該第一計算裝置具有一第一類型處理器但不具有一第二類型處理器，該第二計算裝置具有該第一類型處理器以及該第二類型處理器；其中，該行動裝置透過一連線傳送一資源詢問訊息至該邊緣計算裝置，該連線包括一無線連線，該連線的一單向傳輸延遲小於10毫秒，該資源詢問訊息用來詢問該邊緣計算裝置具有何種類型的資源；其中，該邊緣計算裝置傳送對應於該資源詢問訊息的一資源回覆訊息，其中該資源回覆訊息指出該邊緣計算裝置是否包括具有該第二類型處理器的該第二計算裝置；其中，當該資源回覆訊息指出該邊緣計算裝置包括具有該第二類型處理器的該第二計算裝置時，該行動裝置決定分派該第二類型計算工作該邊緣計算裝置；其中，相較於該第二類型處理器，該第一類型處理器更適合執行該第一類型計算工作；相較於該第一類型處理器，該第二類型處理器更適合執行該第二類型計算工作。An embodiment of the present invention discloses a mobile device applied to an edge computing system. The edge computing system includes an edge computing device. The edge computing device includes at least one of a first computing device and a second computing device. The first computing device has a first type of processor but does not have a second type of processor, the second computing device has the first type of processor and the second type of processor; wherein, the mobile device is connected through a connection Send a resource inquiry message to the edge computing device, the connection includes a wireless connection, and a one-way transmission delay of the connection is less than 10 milliseconds, and the resource inquiry message is used to inquire what type of resource the edge computing device has Wherein, the edge computing device sends a resource response message corresponding to the resource query message, wherein the resource response message indicates whether the edge computing device includes the second computing device with the second type processor; wherein, when the When the resource response message indicates that the edge computing device includes the second computing device with the second type of processor, the mobile device decides to assign the second type of computing task to the edge computing device; wherein, compared to the second type of processing The first type of processor is more suitable for performing the first type of calculation work; compared to the first type of processor, the second type of processor is more suitable for performing the second type of calculation work.

第1圖為本發明實施例一邊緣計算系統10之示意圖。邊緣計算系統10包括一邊緣計算裝置12以及一行動裝置14。行動裝置14可為一智慧型手機、一平板電腦、一筆記型電腦、一虛擬實境（Virtual Reality，VR）裝置、一擴增實境（Augmented Reality，AR）裝置、一混合實境（Mixed Reality，MR）裝置、一延展實境（Extended Reality，XR）裝置或一車用裝置（設置於車上的電子裝置），但不限於此。邊緣計算裝置12可為一電腦叢集（Computer Cluster）或是一叢集計算系統（Clustered Computing System）。邊緣計算裝置12可包括一第一計算裝置120及一第二計算裝置122中至少一者。第一計算裝置120及（或）第二計算裝置122可為一電腦或一伺服器，其可設置於一基站16附近。邊緣計算裝置12與基站16之間可透過有線連線相互連接，也就是說，邊緣計算裝置12與基站16之間可具有一有線連線11。基站16可為一大型基地台（或演進節點B(eNodeB)）、一微型基地台（Femto Base Station）或是一無線存取點（Access Point，AP）。行動裝置14位於基站16的服務區域/範圍中而被基站16服務，也就是說，行動裝置14與基站16透過一空氣介面或一無線連線13相互進行通訊，其中無線連線13可為5G無線通訊系統而具有低傳輸延遲（Latency）。具體來說，邊緣計算裝置12與行動裝置14之間的一連線15可包括有線連線11以及無線連線13，而邊緣計算裝置12與行動裝置14之間的一單向傳輸延遲（其可為由行動裝置14至邊緣計算裝置12的傳輸延遲或是由邊緣計算裝置12至行動裝置14的傳輸延遲）小於10毫秒。因連線15的傳輸延遲夠小，行動裝置14可以將至少一部分的計算工作交給邊緣計算裝置12執行，特別是即時應用程式的計算工作。Figure 1 is a schematic diagram of an edge computing system 10 according to an embodiment of the present invention. The edge computing system 10 includes an edge computing device 12 and a mobile device 14. The mobile device 14 can be a smart phone, a tablet computer, a notebook computer, a virtual reality (VR) device, an augmented reality (AR) device, and a mixed reality (Mixed Reality) device. Reality, MR) device, an Extended Reality (XR) device, or a car device (electronic device installed on the car), but not limited to this. The edge computing device 12 may be a computer cluster (Computer Cluster) or a clustered computing system (Clustered Computing System). The edge computing device 12 may include at least one of a first computing device 120 and a second computing device 122. The first computing device 120 and/or the second computing device 122 can be a computer or a server, which can be installed near a base station 16. The edge computing device 12 and the base station 16 can be connected to each other through a wired connection, that is, there can be a wired connection 11 between the edge computing device 12 and the base station 16. The base station 16 can be a large base station (or evolved node B (eNodeB)), a femto base station (Femto Base Station), or a wireless access point (Access Point, AP). The mobile device 14 is located in the service area/range of the base station 16 and is served by the base station 16, that is, the mobile device 14 and the base station 16 communicate with each other through an air interface or a wireless connection 13, where the wireless connection 13 may be 5G The wireless communication system has low transmission delay (Latency). Specifically, a connection 15 between the edge computing device 12 and the mobile device 14 may include a wired connection 11 and a wireless connection 13, and a one-way transmission delay between the edge computing device 12 and the mobile device 14 (which It can be that the transmission delay from the mobile device 14 to the edge computing device 12 or the transmission delay from the edge computing device 12 to the mobile device 14) is less than 10 milliseconds. Because the transmission delay of the connection 15 is small enough, the mobile device 14 can delegate at least a part of the computing work to the edge computing device 12 to perform, especially the computing work of the real-time application.

於一實施例中，第一計算裝置120可具有一第一類型處理器但不具有一第二類型處理器，第二計算裝置122可同時具有第一類型處理器以及第二類型處理器。第一類型處理器可為一中央處理器(CPU)，而第二類型處理器可為一圖形處理器(GPU)或一張量處理器(TPU)。In an embodiment, the first computing device 120 may have a first type of processor but not a second type of processor, and the second computing device 122 may have both a first type of processor and a second type of processor. The first type of processor may be a central processing unit (CPU), and the second type of processor may be a graphics processing unit (GPU) or a piece of data processing unit (TPU).

在此情形下，行動裝置14可分派一第一類型計算工作（其為一即時應用程式）至邊緣計算裝置12，並讓邊緣計算裝置12利用其CPU執行第一類型計算工作，其中第一類型計算工作可適合用一般用途的CPU來執行。舉例來說，於自動駕駛汽車的應用/場景中，行動裝置14可設置於自駕車上，而第一類型計算工作可為根據自駕車上攝影機捕捉的影像來進行的一碰撞偵測運算，而碰撞偵測運算較適合用一般用途CPU來執行。In this case, the mobile device 14 can dispatch a first type of computing task (which is a real-time application) to the edge computing device 12, and let the edge computing device 12 use its CPU to perform the first type of computing task, where the first type The calculation work can be suitably executed with a general-purpose CPU. For example, in an application/scenario of a self-driving car, the mobile device 14 can be set on a self-driving car, and the first type of calculation task can be a collision detection operation based on an image captured by a camera on the self-driving car, and The collision detection calculation is more suitable for general-purpose CPUs.

另一方面，行動裝置14可取得關於邊緣計算裝置12是否包括第二計算裝置122的資訊，並且，當邊緣計算裝置12包括第二計算裝置122時，行動裝置14可將一第二類型計算工作交給邊緣計算裝置12並讓邊緣計算裝置12的第二計算裝置122利用其第二類型處理器執行第二類型計算工作，其中第二類型計算工作較適合用第二類型處理器來執行。舉例來說，對一VR應用/場景來說，行動裝置14可為一實境呈現裝置（如一VR裝置），而第二類型計算工作為一VR彩現（VR Rendering）運算，其較適合用GPU來執行。如此一來，當行動裝置14知道邊緣計算裝置12具有GPU時，行動裝置14所需耗費的計算量（特別是對應於第二類型計算工作的計算量）可由邊緣計算裝置12來分攤，而可得到更佳的效能(例如降低行動裝置14所消耗的功率或得到較佳的執行效率)。反之，當行動裝置14知道邊緣計算裝置12不具有GPU時，行動裝置14可以不分派第二類型計算工作給邊緣計算裝置12，以例如避免邊緣計算裝置12無法在可容忍的延遲內完成被分派的運算工作。On the other hand, the mobile device 14 can obtain information about whether the edge computing device 12 includes the second computing device 122, and when the edge computing device 12 includes the second computing device 122, the mobile device 14 can perform a second type of computing task It is handed over to the edge computing device 12 and the second computing device 122 of the edge computing device 12 uses its second type of processor to perform the second type of computing work, where the second type of computing work is more suitable to be performed by the second type of processor. For example, for a VR application/scenario, the mobile device 14 may be a reality presentation device (such as a VR device), and the second type of calculation task is a VR rendering (VR Rendering) operation, which is more suitable for use GPU to execute. In this way, when the mobile device 14 knows that the edge computing device 12 has a GPU, the amount of calculation required by the mobile device 14 (especially the amount of calculation corresponding to the second type of computing work) can be shared by the edge computing device 12, but Get better performance (for example, reduce the power consumed by the mobile device 14 or get better execution efficiency). Conversely, when the mobile device 14 knows that the edge computing device 12 does not have a GPU, the mobile device 14 may not assign the second type of computing work to the edge computing device 12, for example, to prevent the edge computing device 12 from being unable to be assigned within a tolerable delay. Computing work.

第2圖為本發明實施例一分派流程20之示意圖，第3圖為分派流程20之示意圖。分派流程20為讓行動裝置14知道邊緣計算裝置12是否具有該第二計算裝置的一規約（protocol）。分派流程20包括以下步驟：FIG. 2 is a schematic diagram of the allocation process 20 according to an embodiment of the present invention, and FIG. 3 is a schematic diagram of the allocation process 20. The dispatching process 20 is to let the mobile device 14 know whether the edge computing device 12 has a protocol of the second computing device. The dispatch process 20 includes the following steps:

步驟202：行動裝置透過一連線，傳送一資源詢問訊息至邊緣計算裝置。Step 202: The mobile device sends a resource inquiry message to the edge computing device through a connection.

步驟204：該邊緣計算裝置傳送對應於該資源詢問訊息的一資源回覆訊息，其中該資源回覆訊息指出該邊緣計算裝置是否具有該第二計算裝置。Step 204: The edge computing device sends a resource response message corresponding to the resource query message, wherein the resource response message indicates whether the edge computing device has the second computing device.

步驟206：當該資源回覆訊息指出該邊緣計算裝置具有該第二計算裝置時，該行動裝置決定分派該第二類型計算工作至該邊緣計算裝置。Step 206: When the resource response message indicates that the edge computing device has the second computing device, the mobile device decides to assign the second type of computing task to the edge computing device.

步驟208： (此步驟為選擇性的)在該行動裝置分派該第二類型計算工作至該邊緣計算裝置之前，該行動裝置傳送一資源請求訊息至該邊緣計算裝置的該第二計算裝置，其中該資源請求訊息包括欲使用該第二類型處理器的一確認請求。Step 208: (This step is optional) Before the mobile device assigns the second type of computing task to the edge computing device, the mobile device sends a resource request message to the second computing device of the edge computing device, where The resource request message includes a confirmation request to use the second type processor.

步驟210： (此步驟為選擇性的)在該行動裝置分派該第二類型計算工作至該邊緣計算裝置之前，該行動裝置傳送對應於該第二類型計算工作的一參數至該邊緣計算裝置的該第二計算裝置。Step 210: (This step is optional) Before the mobile device assigns the second type of computing task to the edge computing device, the mobile device transmits a parameter corresponding to the second type of computing task to the edge computing device The second computing device.

步驟212：該行動裝置分派該第二類型計算工作至該邊緣計算裝置。Step 212: The mobile device assigns the second type of computing task to the edge computing device.

於步驟202中，行動裝置14可透過連線15傳送一資源詢問訊息RI至邊緣計算裝置12。資源詢問訊息RI用來詢問邊緣計算裝置12具有何種類型的資源，也就是說，資源詢問訊息RI 包括一詢問訊息，該詢問訊息用來詢問邊緣計算裝置12是否包括具有第二計算裝置122。資源詢問訊息RI及（或）該詢問訊息的格式並未有所限。In step 202, the mobile device 14 can send a resource query message RI to the edge computing device 12 through the connection 15. The resource inquiry message RI is used to inquire what type of resource the edge computing device 12 has. That is, the resource inquiry message RI includes an inquiry message for inquiring whether the edge computing device 12 includes the second computing device 122. The format of the resource inquiry message RI and/or the inquiry message is not limited.

於步驟204中，在邊緣計算裝置12收到資源詢問訊息RI之後，邊緣計算裝置12做出回應並傳送對應於資源詢問訊息RI的一資源回覆訊息RR。若邊緣計算裝置12具有第一計算裝置120但不具有第二計算裝置122，則資源回覆訊息RR為“否”。若邊緣計算裝置12具有第二計算裝置122但不具有第一計算裝置120，則資源回覆訊息RR為“是”。若邊緣計算裝置12同時具有第一計算裝置及第二計算裝置122，則資源回覆訊息RR為“是”。In step 204, after the edge computing device 12 receives the resource inquiry message RI, the edge computing device 12 responds and sends a resource response message RR corresponding to the resource inquiry message RI. If the edge computing device 12 has the first computing device 120 but not the second computing device 122, the resource response message RR is “No”. If the edge computing device 12 has the second computing device 122 but not the first computing device 120, the resource response message RR is "Yes". If the edge computing device 12 has both the first computing device and the second computing device 122, the resource response message RR is "Yes".

於一實施例中，資源回覆訊息RR指出邊緣計算裝置12包括第一計算裝置120的數量及第二計算裝置122的數量。舉例來說，資源回覆訊息RR指出在邊緣計算裝置12中，第一計算裝置120的數量為五個且第二計算裝置122的數量為八個。藉由準確得知第一計算裝置120的數量及第二計算裝置122的數量，行動裝置14可更精準地估算邊緣計算裝置12的能力。In one embodiment, the resource response message RR indicates that the edge computing device 12 includes the number of first computing devices 120 and the number of second computing devices 122. For example, the resource response message RR indicates that in the edge computing device 12, the number of the first computing device 120 is five and the number of the second computing device 122 is eight. By accurately knowing the number of the first computing device 120 and the number of the second computing device 122, the mobile device 14 can more accurately estimate the capabilities of the edge computing device 12.

於步驟206中，當行動裝置14收到資源回覆訊息RR之後，行動裝置14根據資源回覆訊息，決定是否分派第二類型計算工作至第二計算裝置122。In step 206, after the mobile device 14 receives the resource response message RR, the mobile device 14 determines whether to assign the second type of computing task to the second computing device 122 according to the resource response message.

詳言之，第一類型計算工作較適合由第一類型處理器來執行(例如在前述例子中，自動駕駛汽車的碰撞偵測運算較適合由一般用途CPU處理)，第二類型計算工作較適合由第二類型處理器來執行(例如在前述例子中，VR彩現運算較適合由GPU處理)，且只有第二計算裝置122具有第二類型處理器。因此，當該資源回覆訊息RR指出邊緣計算裝置12具有第二計算裝置122時，就代表邊緣計算裝置12中至少一部分的計算裝置有第二類型處理器，使得邊緣計算裝置12適合被分派第二類型計算工作。反之，當該資源回覆訊息RR指出邊緣計算裝置12不具有第二計算裝置122時，就代表邊緣計算裝置12中所有的計算裝置都沒有第二類型處理器，使得邊緣計算裝置12可能不適合被分派第二類型計算工作。In detail, the first type of calculation work is more suitable to be executed by the first type of processor (for example, in the aforementioned example, the collision detection calculation of an autonomous vehicle is more suitable to be processed by a general-purpose CPU), and the second type of calculation work is more suitable It is executed by the second type processor (for example, in the foregoing example, the VR rendering operation is more suitable for processing by the GPU), and only the second computing device 122 has the second type processor. Therefore, when the resource response message RR indicates that the edge computing device 12 has a second computing device 122, it means that at least a portion of the computing devices in the edge computing device 12 have a second type of processor, so that the edge computing device 12 is suitable for being assigned a second type of processor. Type calculation work. Conversely, when the resource response message RR indicates that the edge computing device 12 does not have the second computing device 122, it means that all computing devices in the edge computing device 12 do not have the second type of processor, so that the edge computing device 12 may not be suitable for allocation. The second type of calculation work.

於步驟208中，在行動裝置14決定要分派第二類型計算工作（如VR彩現運算）至邊緣計算裝置12之後，行動裝置14傳送一資源請求訊息RSR至邊緣計算裝置12，其中資源請求訊息RSR包括欲使用第二類型處理器（如GPU）的一確認請求。當邊緣計算裝置12收到資源請求訊息RSR時，邊緣計算裝置12可檢視其第二類型處理器是否有空閒的運算資源（以執行第二類型計算工作），並傳送對應於資源請求訊息RSR的一確認訊息ACK_R至行動裝置14。藉由該確認請求，行動裝置14可得知第二類型處理器的使用率是否過高，使得邊緣計算裝置12可能不適合被分派第二類型計算工作或只能處理某部分的第二類型計算工作。於另一實施例中，行動裝置14同時傳送資源詢問訊息RI及資源請求訊息RSR。In step 208, after the mobile device 14 decides to assign the second type of computing task (such as VR rendering operation) to the edge computing device 12, the mobile device 14 sends a resource request message RSR to the edge computing device 12, wherein the resource request message The RSR includes a confirmation request to use the second type of processor (such as GPU). When the edge computing device 12 receives the resource request message RSR, the edge computing device 12 can check whether its second-type processor has free computing resources (to perform the second-type computing task), and send a corresponding resource request message RSR A confirmation message ACK_R is sent to the mobile device 14. Through the confirmation request, the mobile device 14 can know whether the usage rate of the second-type processor is too high, so that the edge computing device 12 may not be suitable for being assigned the second-type computing work or can only process a certain part of the second-type computing work . In another embodiment, the mobile device 14 simultaneously transmits the resource inquiry message RI and the resource request message RSR.

於步驟210中，行動裝置14傳送對應於第一類型計算工作的一第一參數及對應於第二類型計算工作的一第二參數中之至少一者至邊緣計算裝置12。舉例來說，假設第一類型計算工作為碰撞偵測運算，該第一參數可為一車輛健康度參數、道路狀況參數或衛星定位參數。再舉一例，假設第二類型計算工作為VR彩現運算，該第二參數可為一螢幕大小或是行動裝置14的一螢幕解析度。藉由該第一參數及該第二參數，邊緣計算裝置12可以更清楚得知行動裝置14的裝置參數或狀況參數，從而提高邊緣計算裝置12的效能(例如藉由統計分析或機器學習的演算法，來加快運算速度或減少消耗)。於另一實施例中，行動裝置14同時傳送資源詢問訊息RI及該些參數。In step 210, the mobile device 14 transmits at least one of a first parameter corresponding to the first type of calculation task and a second parameter corresponding to the second type of calculation task to the edge computing device 12. For example, assuming that the first type of calculation is a collision detection operation, the first parameter can be a vehicle health parameter, a road condition parameter, or a satellite positioning parameter. For another example, assuming that the second type of calculation task is VR rendering calculation, the second parameter can be a screen size or a screen resolution of the mobile device 14. With the first parameter and the second parameter, the edge computing device 12 can more clearly know the device parameters or condition parameters of the mobile device 14, thereby improving the performance of the edge computing device 12 (for example, by statistical analysis or machine learning calculations). Method to speed up the calculation speed or reduce consumption). In another embodiment, the mobile device 14 simultaneously transmits the resource query message RI and the parameters.

於步驟212中，該行動裝置分派第二類型計算工作至該邊緣計算裝置。In step 212, the mobile device assigns the second type of computing task to the edge computing device.

於一實施例中，行動裝置可設置成根據資源回覆訊息RR來決定是否分派第一類型計算工作至該邊緣計算裝置。例如，當資源回覆訊息RR指示邊緣計算裝置12同時具有第一計算裝置及第二計算裝置時，行動裝置14分派第一類型計算工作及第二類型計算工作至邊緣計算裝置12。又例如，當資源回覆訊息RR指示邊緣計算裝置12具有第一計算裝置但不具有第二計算裝置時，行動裝置14分派第一類型計算工作但不分派第二類型計算工作至邊緣計算裝置12。In one embodiment, the mobile device may be configured to determine whether to assign the first type of computing task to the edge computing device according to the resource response message RR. For example, when the resource response message RR indicates that the edge computing device 12 has both the first computing device and the second computing device, the mobile device 14 assigns the first type of computing task and the second type of computing task to the edge computing device 12. For another example, when the resource response message RR indicates that the edge computing device 12 has the first computing device but not the second computing device, the mobile device 14 assigns the first type of computing task but not the second type of computing task to the edge computing device 12.

於另一實施例中，該行動裝置可設置成無論該邊緣運算裝置是否具有第一計算裝置，行動裝置皆分派或皆不分派第一類型計算工作至該邊緣計算裝置。亦即，行動裝置也可設置成不根據資源回覆訊息RR來決定是否分派第一類型計算工作至該邊緣計算裝置。In another embodiment, the mobile device can be configured such that whether or not the edge computing device has the first computing device, the mobile device assigns or does not assign the first type of computing task to the edge computing device. That is, the mobile device can also be configured not to determine whether to assign the first type of computing task to the edge computing device based on the resource response message RR.

現有技術中，行動裝置不確定邊緣計算裝置具有何種類型的處理器。相較之下，本發明的行動裝置可取得邊緣計算裝置的資源類型的資訊，因此，本發明的行動裝置可交付一特定類型計算工作至一特定處理器，其中該特定類型計算工作適合由該特定處理器來執行，如此一來，邊緣計算裝置可分攤行動裝置對應於該特定類型計算工作的運算量，降低行動裝置的功耗。In the prior art, mobile devices are not sure what type of processor the edge computing device has. In contrast, the mobile device of the present invention can obtain information about the resource type of the edge computing device. Therefore, the mobile device of the present invention can deliver a specific type of computing task to a specific processor, wherein the specific type of computing task is suitable for the A specific processor is used to perform the execution. In this way, the edge computing device can share the calculation amount of the mobile device corresponding to the specific type of calculation work, and reduce the power consumption of the mobile device.

綜上所述，本發明的行動裝置可取得邊緣計算裝置的資源類型的資訊並交付特定類型計算工作（例如即時運算應用程式）至特定處理器，其中該特定類型計算工作適合由該特定處理器來執行，如此一來，邊緣計算裝置可分攤行動裝置對應於該特定類型計算工作的運算量，降低行動裝置的功耗。以上所述僅為本發明之較佳實施例，凡依本發明申請專利範圍所做之均等變化與修飾，皆應屬本發明之涵蓋範圍。In summary, the mobile device of the present invention can obtain information on the resource type of the edge computing device and deliver a specific type of computing task (such as a real-time computing application) to a specific processor, where the specific type of computing task is suitable for the specific processor In this way, the edge computing device can share the calculation amount of the mobile device corresponding to the specific type of calculation task, and reduce the power consumption of the mobile device. The above descriptions are only preferred embodiments of the present invention, and all equivalent changes and modifications made in accordance with the scope of the patent application of the present invention shall fall within the scope of the present invention.

10:邊緣計算系統 11:有線連線 12:邊緣計算裝置 120:第一計算裝置 122:第二計算裝置 13:無線連線 14:行動裝置 15:連線 16:基站 20:分派流程 202～212:步驟 10: Edge computing system 11: Wired connection 12: Edge computing device 120: The first computing device 122: second computing device 13: wireless connection 14: mobile device 15: Connect 16: base station 20: Dispatch process 202～212: steps

第1圖為本發明實施例一邊緣計算系統之示意圖。第2圖為本發明實施例一分派流程之示意圖。第3圖為第2圖分派流程之示意圖。Figure 1 is a schematic diagram of an edge computing system according to an embodiment of the present invention. Figure 2 is a schematic diagram of a dispatch process according to an embodiment of the present invention. Figure 3 is a schematic diagram of the allocation process in Figure 2.

20:分派流程 20: Dispatch process

202~212:步驟 202~212: steps

Claims

一種分派方法，應用於一邊緣計算系統，包括一邊緣計算裝置以及一行動裝置，該邊緣計算裝置一第一計算裝置及一第二計算裝置中至少一者，該行動裝置具有欲分派的一第一類型計算工作以及一第二類型計算工作，其中該第一計算裝置具有一第一類型處理器但不具有一第二類型處理器，該第二計算裝置具有該第一類型處理器以及該第二類型處理器，該方法包括：該行動裝置透過一連線傳送一資源詢問訊息至該邊緣計算裝置，其中該連線包括一無線連線，該連線的一單向傳輸延遲小於10毫秒，該資源詢問訊息用來詢問該邊緣計算裝置具有何種類型的資源；該邊緣計算裝置傳送對應於該資源詢問訊息的一資源回覆訊息，其中該資源回覆訊息指出該邊緣計算裝置是否具有該第二計算裝置；以及當該資源回覆訊息指出該邊緣計算裝置包括具有該第二類型處理器的該第二計算裝置時，該行動裝置決定分派該第二類型計算工作至該邊緣計算裝置；其中，相較於該第二類型處理器，該第一類型處理器更適合執行該第一類型計算工作；相較於該第一類型處理器，該第二類型處理器更適合執行該第二類型計算工作。An allocation method is applied to an edge computing system, including an edge computing device and a mobile device. The edge computing device is at least one of a first computing device and a second computing device. The mobile device has a first computing device to be allocated. A type of calculation task and a second type of calculation task, wherein the first computing device has a first type processor but does not have a second type processor, and the second computing device has the first type processor and the second type processor A two-type processor, the method includes: the mobile device transmits a resource query message to the edge computing device through a connection, wherein the connection includes a wireless connection, and the one-way transmission delay of the connection is less than 10 milliseconds, The resource inquiry message is used to inquire what type of resource the edge computing device has; the edge computing device sends a resource response message corresponding to the resource inquiry message, wherein the resource response message indicates whether the edge computing device has the second Computing device; and when the resource response message indicates that the edge computing device includes the second computing device with the second type of processor, the mobile device decides to dispatch the second type of computing task to the edge computing device; Compared with the second type of processor, the first type of processor is more suitable for performing the first type of computing work; compared to the first type of processor, the second type of processor is more suitable for performing the second type of computing work .

如請求項1所述的方法，其中該第一類型處理器為一中央處理器(CPU)，該第二類型處理器為一圖形處理器(GPU)或一張量處理器(TPU)。The method according to claim 1, wherein the first type of processor is a central processing unit (CPU), and the second type of processor is a graphics processing unit (GPU) or a piece of data processing unit (TPU).

如請求項1所述的方法，其中該第一類型計算工作及該第二類型計算工作皆為即時應用程式。The method according to claim 1, wherein the first type of calculation task and the second type of calculation task are both real-time applications.

如請求項1所述的方法，另包括：當該資源回覆訊息指出該邊緣計算裝置具有該第一計算裝置時，分派該第一類型計算工作至該邊緣計算裝置。The method according to claim 1, further comprising: when the resource response message indicates that the edge computing device has the first computing device, allocating the first type of computing task to the edge computing device.

如請求項1所述的方法，另包括：在該行動裝置分派該第二類型計算工作至該邊緣計算裝置之前，該行動裝置傳送一資源請求訊息至該邊緣計算裝置的該第二計算裝置，其中該資源請求訊息包括欲使用該第二類型處理器的一確認請求。The method according to claim 1, further comprising: before the mobile device assigns the second type of computing task to the edge computing device, the mobile device transmits a resource request message to the second computing device of the edge computing device, The resource request message includes a confirmation request to use the second type processor.

如請求項1所述的方法，另包括：在該行動裝置分派該第一類型計算工作或該第二類型計算工作至該邊緣計算裝置之前，該行動裝置傳送對應於該第一類型計算工作的一第一參數或對應於該第二類型計算工作的一第二參數至該邊緣計算裝置。The method according to claim 1, further comprising: before the mobile device assigns the first type of computing task or the second type of computing task to the edge computing device, the mobile device transmits the corresponding to the first type of computing task A first parameter or a second parameter corresponding to the second type of computing task is sent to the edge computing device.

一種邊緣計算系統，包括：一行動裝置，具有欲分派的一第一類型計算工作及一第二類型計算工作，用來透過一連線傳送一資源詢問訊息至該邊緣計算裝置，其中該連線包括一無線連線，該連線的一單向傳輸延遲小於10毫秒，該資源詢問訊息用來詢問該邊緣計算裝置具有何種類型的資源；以及一邊緣計算裝置，包括一第一計算裝置及一第二計算裝置中至少一者，用來傳送對應於該資源詢問訊息的一資源回覆訊息，其中該資源回覆訊息指出該邊緣計算裝置是否具有該第二計算裝置，該第一計算裝置具有一第一類型處理器但不具有一第二類型處理器，該第二計算裝置具有該第一類型處理器以及該第二類型處理器；其中，當該資源回覆訊息指出該邊緣計算裝置具有該第二計算裝置時，該行動裝置決定分派該第二類型計算工作至該邊緣計算裝置；其中，相較於該第二類型處理器，該第一類型處理器更適合執行該第一類型計算工作；相較於該第一類型處理器，該第二類型處理器更適合執行該第二類型計算工作。An edge computing system includes: a mobile device having a first type of computing task and a second type of computing task to be dispatched, for sending a resource query message to the edge computing device through a connection, wherein the connection It includes a wireless connection, the one-way transmission delay of the connection is less than 10 milliseconds, the resource query message is used to inquire what type of resources the edge computing device has; and an edge computing device including a first computing device and At least one of a second computing device is used to send a resource response message corresponding to the resource query message, wherein the resource response message indicates whether the edge computing device has the second computing device, and the first computing device has a The first type of processor but does not have a second type of processor, the second computing device has the first type of processor and the second type of processor; wherein, when the resource response message indicates that the edge computing device has the first type of processor In the second computing device, the mobile device decides to dispatch the second type of computing task to the edge computing device; wherein, compared to the second type of processor, the first type of processor is more suitable for performing the first type of computing task; Compared with the first type of processor, the second type of processor is more suitable for performing the second type of computing work.

如請求項7所述的邊緣計算系統，其中該第一類型處理器為一中央處理器(CPU)，該第二類型處理器為一圖形處理器(GPU)或一張量處理器(TPU)。The edge computing system according to claim 7, wherein the first type of processor is a central processing unit (CPU), and the second type of processor is a graphics processing unit (GPU) or a single quantity processor (TPU) .

如請求項7所述的邊緣計算系統，其中該第一類型計算工作及該第二類型計算工作皆為即時應用程式。The edge computing system according to claim 7, wherein the first type of computing task and the second type of computing task are both real-time applications.

如請求項7所述的邊緣計算系統，其中當該資源回覆訊息指出該邊緣計算裝置具有該第一計算裝置時，該行動裝置分派該第一類型計算工作至該邊緣計算裝置。The edge computing system according to claim 7, wherein when the resource response message indicates that the edge computing device has the first computing device, the mobile device assigns the first type of computing task to the edge computing device.

如請求項7所述的邊緣計算系統，其中在該行動裝置決定分派該第二類型計算工作至該邊緣計算裝置之前，該行動裝置傳送一資源請求訊息至該邊緣計算裝置的該第二計算裝置，該資源請求訊息包括欲使用該第二類型處理器的一請求。The edge computing system of claim 7, wherein before the mobile device decides to dispatch the second type of computing task to the edge computing device, the mobile device transmits a resource request message to the second computing device of the edge computing device , The resource request message includes a request to use the second type processor.

如請求項7所述的邊緣計算系統，其中當該行動裝置決定分派該第一類型計算工作或該第二類型計算工作至該邊緣計算裝置時，該行動裝置傳送對應於該第一類型計算工作的一第一參數或對應於該第二類型計算工作的一第二參數至該邊緣計算裝置。The edge computing system according to claim 7, wherein when the mobile device decides to assign the first type of computing task or the second type of computing task to the edge computing device, the mobile device transmits the computing task corresponding to the first type A first parameter of or a second parameter corresponding to the second type of computing task to the edge computing device.

一種行動裝置，應用於一邊緣計算系統，其中，該邊緣計算系統包括一邊緣計算裝置，該邊緣計算裝置包括一第一計算裝置及一第二計算裝置中至少一者，該第一計算裝置具有一第一類型處理器但不具有一第二類型處理器，該第二計算裝置具有該第一類型處理器以及該第二類型處理器；其中，該行動裝置透過一連線傳送一資源詢問訊息至該邊緣計算裝置，該連線包括一無線連線，該連線的一單向傳輸延遲小於10毫秒，該資源詢問訊息用來詢問該邊緣計算裝置具有何種類型的資源；其中，該邊緣計算裝置傳送對應於該資源詢問訊息的一資源回覆訊息，其中該資源回覆訊息指出該邊緣計算裝置是否具有該第二計算裝置；其中，當該資源回覆訊息指出該邊緣計算裝置包括具有該第二類型處理器的該第二計算裝置時，該行動裝置決定分派該第二類型計算工作至該邊緣計算裝置；其中，相較於該第二類型處理器，該第一類型處理器更適合執行該第一類型計算工作；相較於該第一類型處理器，該第二類型處理器更適合執行該第二類型計算工作。A mobile device applied to an edge computing system, wherein the edge computing system includes an edge computing device, the edge computing device includes at least one of a first computing device and a second computing device, the first computing device has A first type of processor but not a second type of processor, and the second computing device has the first type of processor and the second type of processor; wherein the mobile device transmits a resource inquiry message through a connection To the edge computing device, the connection includes a wireless connection, a one-way transmission delay of the connection is less than 10 milliseconds, and the resource query message is used to inquire what type of resource the edge computing device has; wherein, the edge The computing device sends a resource response message corresponding to the resource query message, wherein the resource response message indicates whether the edge computing device has the second computing device; wherein, when the resource response message indicates that the edge computing device includes the second computing device Type processor of the second computing device, the mobile device decides to dispatch the second type of computing work to the edge computing device; wherein, compared to the second type of processor, the first type of processor is more suitable for executing the The first type of calculation work; compared to the first type of processor, the second type of processor is more suitable for performing the second type of calculation work.

如請求項13所述的行動裝置，其中該第一類型處理器為一中央處理器(CPU)，該第二類型處理器為一圖形處理器(GPU)或一張量處理器(TPU)。The mobile device according to claim 13, wherein the first type of processor is a central processing unit (CPU), and the second type of processor is a graphics processing unit (GPU) or a piece of data processing unit (TPU).

如請求項13所述的行動裝置，其中該第一類型計算工作及該第二類型計算工作皆為即時應用程式。The mobile device according to claim 13, wherein the first type of calculation task and the second type of calculation task are both real-time applications.

如請求項13所述的行動裝置，其中當該資源回覆訊息指出該邊緣計算裝置包括具有該第一計算裝置時，該行動裝置分派該第一類型計算工作至該邊緣計算裝置。The mobile device according to claim 13, wherein when the resource response message indicates that the edge computing device includes the first computing device, the mobile device assigns the first type of computing task to the edge computing device.

如請求項13所述的行動裝置，其中在該行動裝置分派該第二類型計算工作至該邊緣計算裝置之前，該行動裝置傳送一資源請求訊息至該邊緣計算裝置的該第二計算裝置，該資源請求訊息包括欲使用該第二類型處理器的一確認請求。The mobile device according to claim 13, wherein before the mobile device assigns the second type of computing task to the edge computing device, the mobile device transmits a resource request message to the second computing device of the edge computing device, the The resource request message includes a confirmation request to use the second type processor.

如請求項13所述的行動裝置，其中在該行動裝置分派該第一類型計算工作或該第二類型計算工作至該邊緣計算裝置之前，該行動裝置傳送對應於該第一類型計算工作的一第一參數或對應於該第二類型計算工作的一第二參數至該邊緣計算裝置。The mobile device according to claim 13, wherein before the mobile device assigns the first type of computing task or the second type of computing task to the edge computing device, the mobile device transmits a corresponding to the first type of computing task The first parameter or a second parameter corresponding to the second type of computing task is sent to the edge computing device.