CN116661992A

CN116661992A - Terminal Bian Yun collaborative computing method, device, system, medium and program product

Info

Publication number: CN116661992A
Application number: CN202310533292.4A
Authority: CN
Inventors: 丁靓子; 翁欣旦
Original assignee: Alipay Hangzhou Information Technology Co Ltd
Current assignee: Alipay Hangzhou Information Technology Co Ltd
Priority date: 2023-05-09
Filing date: 2023-05-09
Publication date: 2023-08-29

Abstract

The embodiment of the specification discloses a terminal edge cloud collaborative computing method, equipment, a system, a medium and a program product. The method is applied to a mobile terminal and comprises the following steps: responding to target operation of a target user, and judging whether the current computing power of the mobile terminal is sufficient; if not, the target data characteristic information corresponding to the target user is sent to the edge node, so that the edge node operates the target algorithm model to perform reasoning calculation based on the target data characteristic information and the model information, and a first target calculation result is obtained; and receiving a first target calculation result sent by the edge node.

Description

Terminal Bian Yun collaborative computing method, device, system, medium and program product

Technical Field

The present disclosure relates to the field of computer technologies, and in particular, to a method, an apparatus, a system, a medium, and a program product for collaborative computing of a terminal edge cloud.

Background

Deployment and execution of the artificial intelligence scheme are carried out on the mobile terminal, so that the method has obvious advantages in effectiveness, cloud resource consumption and privacy, but is limited by mobile terminal side hardware configuration, and the computing resources are limited. Considering user experience, on a scene application, degradation processing is often required to be performed on a low-end machine, namely, an end-side model is not operated on a machine with poor performance, and meanwhile, the size of the end-side model is strictly limited, namely, a large model cannot be used by an application scene. Therefore, the end intelligent technology faces great challenges in low-end machine coverage, large model support, end cloud collaborative computing power optimization and the like.

Disclosure of Invention

The embodiment of the specification provides a terminal-side cloud collaborative computing method, equipment, a system, a medium and a program product, which are used for carrying out model computation which cannot be operated on a terminal on an edge node through real-time scheduling of mobile terminal computing power, so that the defect of insufficient computing power of a low-end computer is overcome, the problem that the terminal side cannot realize computer large model computation reasoning is solved, and the effect and efficiency of model computation are improved. The technical scheme is as follows:

in a first aspect, an embodiment of the present disclosure provides a method for collaborative computing of a terminal edge cloud, where the method is applied to a mobile terminal, including:

responding to target operation of a target user, and judging whether the current computing power of the mobile terminal is sufficient or not;

if not, the target data characteristic information corresponding to the target user is sent to an edge node, so that the edge node runs a target algorithm model to perform reasoning calculation based on the target data characteristic information and model information, and a first target calculation result is obtained;

and receiving the first target calculation result sent by the edge node.

In one possible implementation manner, the determining whether the current computing force of the mobile terminal is sufficient includes:

Judging whether the current calculation force of the mobile terminal is greater than or equal to a target threshold value;

or (b)

And judging whether the current calculation force of the mobile terminal is greater than or equal to the target calculation force required by the operation of the target algorithm model.

In one possible implementation manner, before the determining whether the current computing force of the mobile terminal is sufficient, the method further includes:

and receiving model information corresponding to the target algorithm model issued by the cloud.

In one possible implementation manner, after the determining, in response to the target operation of the target user, whether the current computing force of the mobile terminal is sufficient, the method further includes:

if yes, running the target algorithm model based on the target data characteristic information and the model information to perform reasoning calculation, and obtaining a second target calculation result.

In one possible implementation, the mobile terminal includes a first computing container;

after receiving the model information corresponding to the target algorithm model issued by the cloud, the method further includes running the target algorithm model based on the target data feature information and the model information to perform inference calculation, and before obtaining a second target calculation result:

Storing model information corresponding to the target algorithm model into the first calculation container;

the performing the inference calculation by running the target algorithm model based on the target data feature information and the model information to obtain a second target calculation result, including:

inputting the target data characteristic information into the target algorithm model in the first calculation container, and carrying out reasoning calculation based on the model information to obtain a second target calculation result.

In one possible implementation manner, the method further includes, after performing the inference calculation on the target algorithm model based on the target data feature information and the model information to obtain a second target calculation result:

and sending the second target calculation result to the cloud end so that the cloud end makes a data decision based on the second target calculation result.

In one possible implementation manner, the sending the target data feature information corresponding to the target user to an edge node includes:

and transmitting the target data characteristic information corresponding to the target user and the model information corresponding to the target algorithm model to the edge node.

In one possible implementation manner, the target data feature information is obtained by performing feature extraction based on target data; the target data includes at least one of: the operation behavior data corresponding to the target user, the equipment data corresponding to the mobile terminal and the space-time data.

In a second aspect, embodiments of the present disclosure provide another end-edge cloud collaborative computing method, where the method is applied to an edge node, and the method includes:

obtaining model information corresponding to a target algorithm model;

receiving target data characteristic information corresponding to a target user sent by the mobile terminal under the condition of insufficient current computing power;

operating the target algorithm model to perform reasoning calculation based on the target data characteristic information and the model information to obtain a first target calculation result;

and sending the first target calculation result to the mobile terminal.

In one possible implementation, the edge node includes a second computing container;

the performing the inference calculation by running the target algorithm model based on the target data feature information and the model information to obtain a first target calculation result, including:

converting the characteristic information of the target data;

And inputting the converted target data characteristic information into the second calculation container, and running the target algorithm model in the second calculation container based on the model information to perform reasoning calculation to obtain a first target calculation result.

In one possible implementation manner, the converting the target data characteristic information includes:

normalizing the target data characteristic information to obtain first target data characteristic information;

converting the first target data characteristic information according to the operation rule of the second computing container to obtain second target data characteristic information;

inputting the converted target data characteristic information into the second calculation container, running the target algorithm model in the second calculation container based on the model information to perform reasoning calculation, and obtaining a first target calculation result, wherein the method comprises the following steps:

and inputting the second target data characteristic information into the second calculation container, and running the target algorithm model in the second calculation container based on the model information to perform reasoning calculation to obtain a first target calculation result.

In a possible implementation, the edge node further includes a hook layer;

The converting the first target data characteristic information according to the operation rule of the second computing container to obtain second target data characteristic information includes:

and converting the first target data characteristic information by using the hook layer according to the operation rule of the second calculation container to obtain second target data characteristic information.

In one possible implementation manner, the obtaining the model information corresponding to the target algorithm model includes:

receiving model information corresponding to a target algorithm model issued by a cloud;

or (b)

And receiving model information corresponding to the target algorithm model, which is sent by the mobile terminal under the condition that the current computing power is insufficient.

In one possible implementation manner, the method further includes, after performing the inference calculation on the target algorithm model based on the target data feature information and the model information to obtain a first target calculation result:

And sending the first target calculation result to the cloud end so that the cloud end makes a data decision based on the first target calculation result.

In a third aspect, embodiments of the present disclosure provide a mobile terminal, where the mobile terminal includes:

the judging module is used for responding to the target operation of the target user and judging whether the current computing force of the mobile terminal is sufficient or not;

the first sending module is used for sending the target data characteristic information corresponding to the target user to an edge node if not, so that the edge node runs a target algorithm model to perform reasoning calculation based on the target data characteristic information and model information to obtain a first target calculation result;

and the first receiving module is used for receiving the first target calculation result sent by the edge node.

In one possible implementation manner, the judging module is specifically configured to:

or (b)

In one possible implementation manner, the mobile terminal further includes:

the second receiving module is used for receiving the model information corresponding to the target algorithm model issued by the cloud.

In one possible implementation manner, the mobile terminal further includes:

and the first reasoning calculation module is used for carrying out reasoning calculation on the target algorithm model based on the target data characteristic information and the model information if so, so as to obtain a second target calculation result.

In one possible implementation, the mobile terminal includes a first computing container; the mobile terminal further includes:

the storage module is used for storing the model information corresponding to the target algorithm model into the first calculation container;

the first reasoning calculation module is specifically configured to:

In one possible implementation manner, the mobile terminal further includes:

and the second sending module is used for sending the second target calculation result to the cloud end so that the cloud end can make a data decision based on the second target calculation result.

In one possible implementation manner, the first sending module is specifically configured to:

In a fourth aspect, embodiments of the present disclosure provide an edge node, where the edge node includes:

the acquisition module is used for acquiring model information corresponding to the target algorithm model;

the third receiving module is used for receiving the target data characteristic information corresponding to the target user sent by the mobile terminal under the condition of insufficient current computing power;

the second reasoning calculation module is used for running a target algorithm model to perform reasoning calculation based on the target data characteristic information and the model information to obtain a first target calculation result;

and the third sending module is used for sending the first target calculation result to the mobile terminal.

the second reasoning calculation module includes:

the conversion unit is used for converting the target data characteristic information;

and the reasoning calculation unit is used for inputting the converted target data characteristic information into the second calculation container, and running the target algorithm model in the second calculation container based on the model information to perform reasoning calculation so as to obtain a first target calculation result.

In one possible implementation, the conversion unit includes:

the normalization subunit is used for normalizing the target data characteristic information to obtain first target data characteristic information;

the conversion subunit is used for converting the first target data characteristic information according to the operation rule of the second computing container to obtain second target data characteristic information;

the above-mentioned reasoning calculation unit is specifically used for:

In a possible implementation, the edge node further includes a hook layer;

the above-mentioned transformant unit is specifically used for:

In one possible implementation manner, the acquiring module is specifically configured to:

or (b)

In one possible implementation manner, the edge node further includes:

and the fourth sending module is used for sending the first target calculation result to the cloud end so that the cloud end can make a data decision based on the first target calculation result.

In a fifth aspect, embodiments of the present disclosure provide a peer-to-peer cloud computing system, where the peer Bian Yun computing system includes a cloud end, a mobile end provided by any one of the first aspect or any one of the possible implementations of the first aspect of embodiments of the present disclosure, and an edge node provided by any one of the second aspect or any one of the possible implementations of the second aspect of embodiments of the present disclosure; the cloud end is used for sending model information corresponding to the target algorithm model to the mobile end and the edge node.

In a sixth aspect, embodiments of the present disclosure provide an electronic device, including: a processor and a memory;

The processor is connected with the memory;

the memory is used for storing executable program codes;

the processor executes a program corresponding to the executable program code stored in the memory by reading the executable program code for performing the method provided by the first aspect of the embodiments of the present specification or any one of the possible implementations of the first aspect or any one of the possible implementations of the second aspect of the embodiments of the present specification.

In a seventh aspect, the present description provides a computer storage medium storing a plurality of instructions adapted to be loaded by a processor and to perform a method as provided by the first aspect or any one of the possible implementations of the second aspect of the present description.

In an eighth aspect, the present description provides a computer program product comprising instructions which, when run on a computer or a processor, cause the computer or the processor to perform a method of collaborative computing by a terminal Bian Yun provided by the first aspect of the present description or any one of the possible implementations of the first aspect or any one of the possible implementations of the second aspect of the present description.

In the embodiment of the specification, the mobile terminal judges whether the current calculation force is sufficient or not by responding to the target operation of the target user; if not, the target data characteristic information corresponding to the target user is sent to the edge node, so that the edge node operates the target algorithm model to perform reasoning calculation based on the target data characteristic information and the model information, and a first target calculation result is obtained; the mobile terminal receives the first target calculation result sent by the edge node, so that the mobile terminal carries out model calculation which cannot be operated on the terminal on the edge node through real-time dispatching of calculation force, the defect of insufficient calculation force of a low-end computer is overcome, the problem that the terminal side cannot realize calculation reasoning of a large computer model is solved, and the effect and efficiency of model calculation are improved.

Drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present description, the drawings that are required in the embodiments will be briefly described below, and it is apparent that the drawings in the following description are only some embodiments of the present description, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.

Fig. 1 is a schematic architecture diagram of an end-cloud collaborative computing scheme in the related art;

FIG. 2 is a schematic diagram of an architecture of a cloud-computing system according to an exemplary embodiment of the present disclosure;

fig. 3 is a schematic flow chart of a method for collaborative computing of a terminal edge cloud according to an exemplary embodiment of the present disclosure;

FIG. 4 is a schematic diagram of an implementation flow for converting target data feature information according to an exemplary embodiment of the present disclosure;

FIG. 5 is a schematic diagram of a process for converting target data feature information according to an exemplary embodiment of the present disclosure;

fig. 6 is a flowchart of another end-edge cloud collaborative computing method according to an exemplary embodiment of the present disclosure;

fig. 7 is a schematic diagram of an implementation process of a method for collaborative computing by using a terminal edge cloud according to an exemplary embodiment of the present disclosure;

fig. 8 is a schematic structural diagram of a mobile terminal according to an exemplary embodiment of the present disclosure;

FIG. 9 is a schematic diagram of an edge node according to an exemplary embodiment of the present disclosure;

fig. 10 is a schematic structural diagram of an electronic device according to an exemplary embodiment of the present disclosure.

Detailed Description

The technical solutions in the embodiments of the present specification will be clearly and completely described below with reference to the drawings in the embodiments of the present specification.

The terms first, second, third and the like in the description and in the claims and in the above drawings, are used for distinguishing between different objects and not necessarily for describing a particular sequential or chronological order. Furthermore, the terms "comprise" and "have," as well as any variations thereof, are intended to cover a non-exclusive inclusion. For example, a process, method, system, article, or apparatus that comprises a list of steps or elements is not limited to only those listed steps or elements but may include other steps or elements not listed or inherent to such process, method, article, or apparatus.

It should be noted that, information (including but not limited to user equipment information, user personal information, etc.), data (including but not limited to data for analysis, stored data, presented data, etc.), and signals according to the embodiments of the present disclosure are all authorized by the user or are fully authorized by the parties, and the collection, use, and processing of relevant data is required to comply with relevant laws and regulations and standards of relevant countries and regions. For example, target data, operation behavior data, device data, and the like referred to in this specification are all acquired with sufficient authorization.

At present, the related terminal intelligent technical scheme generally adopts a terminal cloud collaborative computing mode as shown in fig. 1, and after an algorithm developer develops a target algorithm model, the target algorithm model and related configuration are issued to mobile terminal equipment of a user through a cloud corresponding server. And then triggering end-side artificial intelligent calculation based on user behaviors at the mobile end, namely performing inference calculation on an end-side operation target algorithm model to obtain an end-side calculation result. Meanwhile, bidirectional data communication can be performed between the mobile terminal and the cloud terminal, the mobile terminal can report the terminal side calculation result to the cloud terminal, and the cloud terminal can also send data to the mobile terminal. However, on the one hand, mobile-end artificial intelligence computation is limited by equipment hardware configuration, degradation processing is needed on some low-end machines, and a target algorithm model running on the low-end machines is not supported, and the target algorithm model can cause bad use experiences such as operation jamming and power consumption of users on the low-end machines. On the other hand, because the computational effort stored by the mobile terminal is limited, the operation of the algorithm model with larger size on the mobile terminal consumes long time, occupies large storage space, and is difficult to support the operation on the mobile terminal. In summary, the related end intelligent technical scheme has obvious limitations on a low-end machine and a large model, and a need exists for an end intelligent calculation scheme which can not only overcome the defect of insufficient calculation power of the low-end machine, but also solve the problem that the end side cannot realize the reasoning calculation of the large model of the computer.

Next, please refer to fig. 2, which is a schematic diagram illustrating an architecture of a cloud computing system according to an exemplary embodiment of the present disclosure. As shown in fig. 2, the end-edge cloud collaborative computing system includes: the mobile terminal, the edge node and the cloud. Wherein:

the mobile terminal may include one or more terminals corresponding to the user. The mobile terminal can be provided with user version software for realizing functions of artificial intelligence calculation and the like based on operation behavior data corresponding to a user, equipment data corresponding to the mobile terminal, space-time data and the like. The mobile terminal can establish a data relationship with the network, and establish a data connection relationship with the edge node and the cloud through the network, for example, receive model information corresponding to a target algorithm model issued by the cloud or a first target calculation result sent by the edge node. Meanwhile, the mobile terminal can respond to target operation of a target user, judge whether the current computing power is sufficient, and send target data characteristic information corresponding to the target user to the edge node through the network under the condition that the current computing power is insufficient, so that the edge node operates a target algorithm model to perform reasoning calculation based on the target data characteristic information and model information to obtain a first target calculation result, and receive the first target calculation result sent by the edge node. And under the condition that the current computing power is sufficient, directly operating a target algorithm model based on the target data characteristic information and the model information to perform reasoning calculation, and obtaining a second target calculation result. Any mobile terminal can be, but is not limited to, a mobile phone, a tablet computer, a notebook computer and other devices provided with user software.

An edge node is a platform built on the edge side of the network near the user that can provide storage, computing, networking, etc. resources. The edge node can acquire model information corresponding to a target algorithm model issued by the cloud through a network, receive target data characteristic information corresponding to a target user sent by the mobile terminal under the condition of insufficient current computing power, and operate the target algorithm model to perform reasoning calculation based on the target data characteristic information and the model information to obtain a first target calculation result. After that, the edge node may also select whether to send the first target calculation result to the mobile terminal or the cloud terminal according to an actual application scenario or a data decision requirement through the network, which is not limited in the embodiment of the present disclosure.

Optionally, the edge node may also obtain, through the network, model information corresponding to a target algorithm model sent by the mobile terminal under the condition that the current computing power is insufficient, which is not limited in the embodiment of the present specification.

The cloud end can be one or more servers capable of providing cooperative computation of various ends Bian Yun, the cloud end server can be, but is not limited to, a release and deployment platform for realizing that algorithm developers release model information of a trained target algorithm model to a mobile end or an edge node through a network, and the model information can be, but is not limited to, data such as model files, script texts and the like of the target algorithm model. The cloud end can also receive a second target calculation result sent by the mobile end through the network and make a data decision based on the second target calculation result; or receiving a first target calculation result sent by the edge node through the network, and making a data decision based on the first target calculation result. The servers may be, but are not limited to, hardware servers, virtual servers, cloud servers, and the like.

The network may be a medium for providing a communication link between the cloud end and any one of the mobile ends, any one of the mobile ends and any one of the edge nodes, and between the cloud end and any one of the edge nodes, or may be the internet including network devices and transmission media, which is not limited thereto. The transmission medium may be a wired link, such as, but not limited to, coaxial cable, fiber optic and digital subscriber lines (digital subscriber line, DSL), etc., or a wireless link, such as, but not limited to, wireless internet (wireless fidelity, WIFI), bluetooth, a mobile device network, etc.

It is to be understood that the number of mobile terminals, cloud terminals and edge nodes in the terminal Bian Yun collaborative computing system shown in fig. 2 is merely an example, and in a specific implementation, any number of mobile terminals, cloud terminals and edge nodes may be included in the terminal cloud collaborative computing system, which is not specifically limited in this embodiment of the present disclosure. For example, but not limited to, the mobile terminal may be a mobile terminal cluster formed by a plurality of mobile terminals, and the edge node may be an edge node cluster formed by a plurality of edge nodes.

In order to solve the above-mentioned problems in the related art, a method of collaborative calculation of the terminal Bian Yun provided in the embodiment of the present disclosure is described next with reference to fig. 2. Referring to fig. 3, a flow chart of a method for collaborative computing of a terminal edge cloud according to an exemplary embodiment of the present disclosure is shown. As shown in fig. 3, the end-edge cloud collaborative computing method includes the following steps:

S302, the mobile terminal responds to the target operation of the target user to judge whether the current computing power of the mobile terminal is sufficient.

Specifically, the target operation may be, but not limited to, clicking or sliding operation corresponding to when the target user triggers the mobile terminal to operate the target algorithm model to perform model reasoning calculation. After the mobile terminal receives the target operation of the target user, in order to avoid that the mobile terminal is too low in configuration or the target algorithm model is too large to support the normal operation of the target algorithm model, or cause bad use experience such as operation blocking and power consumption of the target user when the target algorithm model operates at the mobile terminal, whether the current computing power of the mobile terminal is sufficient can be judged in response to the target operation of the target user. The current computing power is the computing power currently carried on the mobile terminal when the target user triggers, and can be determined based on, but not limited to, target application information of a target application corresponding to a target algorithm model on the current mobile terminal and/or device information corresponding to the mobile terminal when the target user triggers. The target application information may include, but is not limited to, the number of threads of the target application, the performance of starting the target application, and the like, and the device information may include, but is not limited to, the device model number of the mobile terminal, the memory, the power condition, the network condition, and the like.

Optionally, when judging whether the current computing force of the mobile terminal is sufficient, whether the current computing force of the mobile terminal is greater than or equal to a target threshold value can be directly judged, if the current computing force is greater than or equal to the target threshold value, the mobile terminal can be considered to have the capability of sufficiently carrying the model operation, and then the current computing force of the mobile terminal can be determined to be sufficient; if the current calculation force is smaller than the target threshold, the mobile terminal can be considered to have no capability of carrying the model operation, and the current calculation force of the mobile terminal can be determined to be insufficient. The target threshold may be set according to the maximum algorithm model running condition that may be used in the application, or may be a larger calculated force value directly, which is not limited in the embodiment of the present disclosure.

Optionally, when judging whether the current calculation force of the mobile terminal is sufficient, in order to ensure the accuracy of judgment, the problem that the mobile terminal cannot support the normal operation of the target algorithm model due to too low configuration of the mobile terminal or too large target algorithm model is avoided, or whether the current calculation force of the mobile terminal is greater than or equal to the target calculation force required by the operation of the target algorithm model can be directly judged, if the current calculation force is greater than or equal to the target calculation force required by the operation of the target algorithm model, the mobile terminal can be considered to be enough to support the normal operation of the target algorithm model, and then the current calculation force of the mobile terminal can be determined to be sufficient; if the current calculation force is smaller than the target calculation force required by the operation of the target algorithm model, the mobile terminal can be considered to not support the normal operation of the target algorithm model, and the current calculation force of the mobile terminal can be determined to be insufficient.

And S304, under the condition that the current computing power is insufficient, the mobile terminal sends the target data characteristic information corresponding to the target user to the edge node.

Specifically, under the condition that the current computing power of the mobile terminal is insufficient, the mobile terminal cannot normally operate the target algorithm model to perform inference calculation, so as to ensure the use experience of the target user and the effect and efficiency of model inference calculation, the mobile terminal can send the target data characteristic information corresponding to the target user to the edge node with stronger computing power, and the edge node operates the target algorithm model to perform inference calculation based on the target data characteristic information and the model information of the target algorithm model, so that a first target calculation result is obtained. The model information may include, but is not limited to, a model file and script text corresponding to the target algorithm model. The target data feature information is obtained by feature extraction based on target data, and the target data is data on which a target algorithm model generated by a target user on a mobile terminal depends during inference calculation, and may include, but is not limited to, at least one of the following: operational behavior data corresponding to the target user, equipment data corresponding to the mobile terminal and space-time data. The operation behavior data corresponding to the target user may include, but not limited to, data generated by clicking, staying time of the target user, etc. due to operation behavior of the target user, data input by the target user, etc. The edge node may be a cdn node or may be another network node such as a router, which is not limited in the embodiment of the present disclosure.

Illustratively, the above-described target data feature information may be, but is not limited to, a target user operation sequence feature extracted from target user operation behavior data, a page exposure sequence feature extracted from page exposure data, or the like.

Optionally, when the mobile terminal sends the target data feature information corresponding to the target user to the edge node, the mobile terminal may also directly send the target data feature information corresponding to the target user and the model information corresponding to the target algorithm model to the edge node together. Model information corresponding to the target algorithm model is issued by a cloud server.

S306, the edge node acquires model information corresponding to the target algorithm model.

Optionally, the edge node operates the target algorithm model to perform inference calculation based on the target data feature information and the model information, and before obtaining the first target calculation result, model information corresponding to the target algorithm model issued by the cloud can be received through the network.

Optionally, the edge node may not only obtain the model information corresponding to the target algorithm model through a cloud end issuing manner, but also directly receive, through the network, the model information corresponding to the target algorithm model sent by the mobile terminal under the condition that the current computing power is insufficient, that is, the S306 and the S304 may be executed synchronously, or may be executed sequentially, which is not limited in the embodiment of the present specification.

And S308, the edge node runs a target algorithm model to perform reasoning calculation based on the target data characteristic information and the model information, and a first target calculation result is obtained.

Specifically, after the edge node receives the target data characteristic information corresponding to the target user sent by the mobile terminal under the condition of insufficient current computing power, the edge node directly operates a target algorithm model based on the target data characteristic information and model information to perform reasoning calculation, and a first target calculation result is obtained, so that a model reasoning calculation effect stronger than that of the mobile terminal is achieved through the edge node with stronger computing power than that of the mobile terminal and lower cost than that of the cloud terminal.

Further, the edge node includes a second computation container. After obtaining the model information corresponding to the target algorithm model, the edge node may directly store the model information corresponding to the target algorithm model into the second computing container. As shown in fig. 4, the implementation process of the edge node running the target algorithm model to perform inference calculation based on the target data feature information and the model information to obtain the first target calculation result may include the following steps:

s402, converting the characteristic information of the target data.

Specifically, after the edge node receives the target data feature information corresponding to the target user sent by the mobile terminal under the condition of insufficient current computing power, in order to keep the second computing container function interface of the edge node consistent with the mobile terminal, the target data feature information needs to be converted first. The conversion process is shown in fig. 5, and the target data characteristic information can be standardized (agreed) first, so that the difference between the target data characteristic information transmitted by the network and the data which can be directly used by the second computing container in the protocol is smoothed, and the first target data characteristic information is obtained. Then, converting the first target data characteristic information according to the operation rule of the second computing container to obtain the second target data characteristic information, and realizing the automatic cross-platform operation of the terminal intelligent scheme, namely, the mobile terminal and the edge node can share the same terminal intelligent scheme file (model information corresponding to the target algorithm model), and a developer does not need to pay attention to platform difference when the target algorithm model is researched and developed, so that a second computing container function interface of the edge node and the mobile terminal can be kept consistent directly through the conversion of the target data characteristic information, and a set of code multi-terminal general purpose is realized.

Optionally, the edge node further comprises a hook layer. In order to ensure the expansibility of the collaborative computing method of the terminal Bian Yun provided in the embodiments of the present disclosure, the edge node may directly utilize the hook layer to convert the first target data feature information according to the operation rule of the second computing container, so as to obtain the second target data feature information.

Alternatively, the edge node may directly convert the first target data feature information according to the operation rule of the second computing container by using a hard-coded conversion manner to obtain the second target data feature information, which is not limited in the embodiments of the present disclosure.

It will be appreciated that the edge node may include 1 or more second computing containers and hook layers, which is not limited by the present embodiments.

S404, inputting the converted target data characteristic information into a second calculation container, and running a target algorithm model in the second calculation container based on the model information to perform reasoning calculation to obtain a first target calculation result.

Specifically, when the edge node converts the target data feature information according to the process shown in fig. 5 to obtain second target data feature information, the second target data feature information may be directly input into a second calculation container, and the target algorithm model is run in the second calculation container based on the model information to perform inference calculation, so as to obtain the first target calculation result.

Next, please refer to fig. 3, in S308, the edge node performs inference calculation by running a target algorithm model based on the target data feature information and the model information, and after obtaining the first target calculation result, the end-edge cloud collaborative calculation method further includes:

s310, the edge node sends the first target calculation result to the mobile terminal.

Specifically, after obtaining the first target calculation result, the edge node sends the first target calculation result to the mobile terminal through the network, so that the mobile terminal makes a data decision based on the first target calculation result sent by the edge node.

Optionally, after the edge node performs inference calculation by running the target algorithm model based on the target data feature information and the model information to obtain a first target calculation result, the first target calculation result may also be sent to the cloud, so that the cloud makes a data decision based on the first target calculation result.

In the embodiment of the specification, a mobile terminal judges whether the current computing power is sufficient or not by responding to the target operation of a target user, and sends target data characteristic information corresponding to the target user to an edge node under the condition that the current computing power is insufficient, so that the edge node operates a target algorithm model to perform reasoning calculation based on the target data characteristic information and model information to obtain a first target calculation result; the mobile terminal receives the first target calculation result sent by the edge node, so that the mobile terminal carries out model calculation which cannot be operated on the terminal on the edge node through real-time dispatching of calculation force, the defect of insufficient calculation force of a low-end computer is overcome, the problem that the terminal side cannot realize calculation reasoning of a large computer model is solved, and the effect and efficiency of model calculation are improved.

Next, please refer to fig. 6, which is a flowchart illustrating another end-edge cloud collaborative computing method according to an exemplary embodiment of the present disclosure. As shown in fig. 6, the end-edge cloud collaborative computing method includes the following steps:

s602, the cloud end transmits model information corresponding to the target algorithm model to the mobile end.

Specifically, after model training is performed by algorithm research personnel based on the feature samples to generate a target algorithm model, model information corresponding to the target algorithm model can be issued to a mobile terminal through a network by a server (release deployment platform) in the cloud.

And S604, the mobile terminal stores model information corresponding to the target algorithm model into a first calculation container.

Specifically, the mobile terminal includes a first computing container. After receiving the model information corresponding to the target algorithm model issued by the cloud, the mobile terminal can store the model information corresponding to the target algorithm model to the first computing container so as to enable the target algorithm model to run in the environment of the first computing container and further improve the model reasoning computing effect in order to avoid the problem of abnormal deployment of the application caused by environmental change in the deployment process of the target algorithm model.

S606, the mobile terminal responds to the target operation of the target user to judge whether the current computing power of the mobile terminal is sufficient.

Specifically, S606 corresponds to S302, and will not be described here.

And S608, if not, the mobile terminal sends the target data characteristic information corresponding to the target user to the edge node.

Specifically, S608 corresponds to S304, and will not be described herein.

S610, the edge node acquires model information corresponding to the target algorithm model.

Specifically, S610 corresponds to S306, and will not be described here again.

And S612, the edge node runs the target algorithm model to perform reasoning calculation based on the target data characteristic information and the model information, and a first target calculation result is obtained.

Specifically, S612 corresponds to S308, and will not be described here.

S614, the edge node sends the first target calculation result to the mobile terminal.

Specifically, S614 corresponds to S310, and will not be described here.

Optionally, when the cloud end is required to make a data decision, after obtaining the first target calculation result, the edge node may directly send the first target calculation result to the cloud end through the network, so that the cloud end makes a data decision based on the first target calculation result.

Next, please continue to refer to fig. 6, in S606, after the mobile terminal determines whether the current computing power of the mobile terminal is sufficient in response to the target operation of the target user, the terminal edge cloud collaborative computing method further includes:

And S616, if yes, the mobile terminal runs the target algorithm model to perform reasoning calculation based on the target data characteristic information and the model information, and a second target calculation result is obtained.

Specifically, if the current computing power of the mobile terminal is sufficient, it can be considered that the mobile terminal has enough capability of bearing the operation of the target algorithm model, the mobile terminal can directly operate the target algorithm model to perform reasoning calculation based on the target data characteristic information and the model information, and a second target calculation result is obtained.

Optionally, when the mobile terminal operates the target algorithm model based on the target data feature information and the model information to perform inference calculation to obtain a second target calculation result, in order to ensure the operation effect of the target algorithm model, the mobile terminal may directly input the target data feature information into the target algorithm model in the first calculation container, and perform inference calculation based on the model information to obtain the second target calculation result.

Optionally, the first computing container may include a virtual machine, so that in order to enable the code logic of the target algorithm model to have a dynamic capability, development of algorithm developers is facilitated, and the target data feature information may be input into the target algorithm model based on the virtual machine in the second computing container to perform inference computation, so as to obtain a second target computing result. The virtual machines may be, but are not limited to, python, js, lua, etc.

Next, please refer to fig. 6, if yes in S616, the mobile terminal runs the target algorithm model to perform inference calculation based on the target data feature information and the model information, and after obtaining the second target calculation result, the end-edge cloud collaborative calculation method may further include:

and S618, the mobile terminal sends the second target calculation result to the cloud.

Specifically, when the cloud end is required to make a data decision, after the mobile end obtains the second target calculation result, the second target calculation result can be directly sent to the cloud end through the network, so that the cloud end makes a data decision based on the second target calculation result.

According to the embodiment of the specification, the real-time intelligent scheduling is carried out on the mobile terminal according to whether the current computing power is sufficient or not, the model calculation which cannot be operated on the terminal is carried out on the edge node, the edge node is introduced into the terminal cloud cooperative technology link, the advantages that the computing power is stronger than that of the mobile terminal and the cost is lower than that of the cloud are brought into play, the defect of insufficient computing power of a low-end computer is overcome, the problem that the computer large model calculation reasoning cannot be realized on the terminal side is solved, and the effect and efficiency of the model calculation are improved.

Next, please refer to fig. 7, which is a schematic diagram illustrating an implementation process of a terminal-edge cloud collaborative computing method according to an exemplary embodiment of the present disclosure. As shown in fig. 7, after model training is performed by algorithm developers based on feature samples to generate a target algorithm model, model information (model files and related configurations) corresponding to the target algorithm model can be issued to a mobile terminal and an edge node of a target user through a release deployment platform (i.e., a cloud server). After the mobile terminal receives the target operation of the target user, the mobile terminal responds to the target operation to trigger the intelligent scheduling module to perform intelligent scheduling based on the hardware level of the current model and the real-time computing power, namely judging whether the current computing power is sufficient or not. When the current computing power of the mobile terminal is sufficient, a target algorithm model is operated in a first computing container based on model information and target data characteristic information of a target user to perform reasoning computation, a second target computing result is obtained, and data decision (result processing) is performed based on the second target computing result or the second target computing result is sent to a cloud for data decision. When the current computing power of the mobile terminal is tense (insufficient), the intelligent scheduling module can send relevant data characteristics (namely target data characteristic information) of the mobile terminal to the edge node. After receiving the target data characteristic information sent by the mobile terminal, the edge node firstly processes (standardizes) the data, then forwards the processed target data characteristic information to a hook layer of the edge node for conversion, and the hook layer is used for realizing that the second computing container function interface is consistent with the mobile terminal. And finally, directly inputting the converted target data characteristic information into a second calculation container, and running a target algorithm model in the second calculation container based on model information to perform reasoning calculation to obtain a first target calculation result. After the edge node obtains the first target calculation result, the edge node can select whether to send the first target calculation result to the mobile terminal for data decision making or to the cloud for data decision making through the network according to actual requirements. After the mobile terminal obtains the second target calculation result corresponding to the terminal side or the first target calculation result corresponding to the edge side (edge node), the mobile terminal can perform bidirectional data communication with the cloud.

Next, please refer to fig. 8, which is a schematic diagram of a mobile terminal according to an exemplary embodiment of the present disclosure. As shown in fig. 8, the mobile terminal 800 includes:

a judging module 810, configured to respond to a target operation of a target user, and judge whether a current computing force of the mobile terminal is sufficient;

the first sending module 820 is configured to send, if not, target data feature information corresponding to the target user to an edge node, so that the edge node operates a target algorithm model to perform inference calculation based on the target data feature information and model information, and obtain a first target calculation result;

a first receiving module 830, configured to receive the first target calculation result sent by the edge node.

In one possible implementation manner, the determining module 810 is specifically configured to:

or (b)

In one possible implementation, the mobile terminal 800 further includes:

In one possible implementation, the mobile terminal 800 includes a first computing container; the mobile terminal 800 further includes:

the first reasoning calculation module is specifically configured to:

In one possible implementation, the mobile terminal 800 further includes:

In one possible implementation manner, the first sending module 820 is specifically configured to:

The above division of each module in the mobile terminal is only used for illustration, and in other embodiments, the mobile terminal may be divided into different modules according to the need, so as to complete all or part of the functions of the mobile terminal. The implementation of each module in the mobile terminal provided in the embodiments of the present specification may be in the form of a computer program. The computer program may run on a terminal or a server. Program modules of the computer program may be stored in the memory of the terminal or server. Which when executed by a processor, performs all or part of the steps of the method of collaborative computing by the terminal Bian Yun described in the embodiments of the present specification.

Reference is next made to fig. 9, which is a schematic structural diagram of an edge node according to an exemplary embodiment of the present disclosure. As shown in fig. 9, the edge node 900 includes:

an obtaining module 910, configured to obtain model information corresponding to a target algorithm model;

A third receiving module 920, configured to receive target data feature information corresponding to a target user sent by the mobile terminal under the condition that current computing power is insufficient;

the second inference calculation module 930 is configured to perform inference calculation by running a target algorithm model based on the target data feature information and the model information, so as to obtain a first target calculation result;

and a third sending module 940, configured to send the first target calculation result to the mobile terminal.

In one possible implementation, the edge node 900 includes a second computing container;

the second reasoning calculation module 930 includes:

In one possible implementation, the conversion unit includes:

the above-mentioned reasoning calculation unit is specifically used for:

In one possible implementation, the edge node 900 further includes a hook layer;

the above-mentioned transformant unit is specifically used for:

In one possible implementation manner, the acquiring module 910 is specifically configured to:

or (b)

In one possible implementation, the edge node 900 further includes:

The above-described division of the modules in the edge node is for illustration only, and in other embodiments, the edge node may be divided into different modules as needed to perform all or part of the above-described functions of the edge node. The implementation of the individual modules in the edge nodes provided in the embodiments of the present description may be in the form of a computer program. The computer program may run on a terminal or a server. Program modules of the computer program may be stored in the memory of the terminal or server. Which when executed by a processor, performs all or part of the steps of the method of collaborative computing by the terminal Bian Yun described in the embodiments of the present specification.

Next, please refer to fig. 10, which is a schematic structural diagram of an electronic device according to an exemplary embodiment of the present disclosure. As shown in fig. 10, the electronic device 1000 may include: at least one processor 1010, at least one communication bus 1020, a user interface 1030, at least one network interface 1040, and a memory 1050. Wherein a communication bus 1020 may be used to enable communication of the connections of the various components described above.

The user interface 1030 may include a Display (Display) and a Camera (Camera), and the optional user interface may also include a standard wired interface, a wireless interface, among others.

The network interface 1040 may optionally include, among other things, a bluetooth module, a near field communication (Near Field Communication, NFC) module, a wireless fidelity (Wireless Fidelity, wi-Fi) module, and the like.

Wherein the processor 1010 may include one or more processing cores. The processor 1010 utilizes various interfaces and lines to connect various portions of the overall electronic device 1000, perform various functions for routing the electronic device 1000 and process data by executing or executing instructions, programs, code sets, or instruction sets stored in the memory 1050, and invoking data stored in the memory 1050. Alternatively, the processor 1010 may be implemented in hardware in at least one of digital signal processing (Digital Signal Processing, DSP), field programmable gate array (Field-Programmable Gate Array, FPGA), programmable logic array (Programmable Logic Array, PLA). The processor 1010 may integrate one or a combination of several of a processor (Central Processing Unit, CPU), an image processor (Graphics Processing Unit, GPU), and a modem, etc. The CPU mainly processes an operating system, a user interface, an application program and the like; the GPU is used for rendering and drawing the content required to be displayed by the display screen; the modem is used to handle wireless communications. It will be appreciated that the modem may not be integrated into the processor 1010 and may be implemented by a single chip.

The Memory 1050 may include a random access Memory (Random Access Memory, RAM) or a Read-Only Memory (ROM). Optionally, the memory 1050 includes a non-transitory computer readable medium. Memory 1050 may be used to store instructions, programs, code, sets of codes, or instruction sets. The memory 1050 may include a stored program area and a stored data area, wherein the stored program area may store instructions for implementing an operating system, instructions for at least one function (such as a judgment function, an inference calculation function, a transmission function, etc.), instructions for implementing the various method embodiments described above, and the like; the storage data area may store data or the like referred to in the above respective method embodiments. Memory 1050 may also optionally be at least one storage device located remotely from the processor 1010. As shown in FIG. 10, the memory 1050, which is a computer storage medium, may include an operating system, a network communication module, a user interface module, and program instructions.

Specifically, the electronic device 1000 is a mobile terminal as mentioned in the foregoing embodiment, and the processor 1010 may be configured to call the program instructions stored in the memory 1050, and specifically perform the following operations:

And responding to the target operation of the target user, and judging whether the current computing power of the mobile terminal is sufficient.

If not, the target data characteristic information corresponding to the target user is sent to an edge node, so that the edge node runs a target algorithm model to perform reasoning calculation based on the target data characteristic information and model information, and a first target calculation result is obtained.

And receiving the first target calculation result sent by the edge node.

In some possible embodiments, the processor 1010 is configured to perform, when determining whether the current computing power of the mobile terminal is sufficient, specifically:

or (b)

In some possible embodiments, before the processor 1010 performs the determining whether the current computing power of the mobile terminal is sufficient, the method further includes:

In some possible embodiments, after the processor 1010 performs the target operation in response to the target user, determining whether the current computing power of the mobile terminal is sufficient, the method further comprises:

In some possible embodiments, the mobile terminal includes a first computing container;

after the processor 1010 executes the model information corresponding to the target algorithm model issued by the cloud, the processor is further configured to execute the performing inference calculation by running the target algorithm model based on the target data feature information and the model information, so as to obtain a second target calculation result, before executing:

and storing model information corresponding to the target algorithm model into the first computing container.

The processor 1010 performs the inference calculation by running the target algorithm model based on the target data feature information and the model information, and is specifically configured to perform:

In some possible embodiments, the processor 1010 executes the target algorithm model to perform an inference calculation based on the target data feature information and the model information, and is further configured to, after obtaining a second target calculation result, perform:

In some possible embodiments, when the processor 1010 performs sending the target data feature information corresponding to the target user to an edge node, the method is specifically configured to perform:

In some possible embodiments, the target data feature information is obtained by feature extraction based on target data; the target data includes at least one of: the operation behavior data corresponding to the target user, the equipment data corresponding to the mobile terminal and the space-time data.

In some possible embodiments, the electronic device 1000 is an edge node as mentioned in the previous embodiments, the processor 1010 may be configured to call the program instructions stored in the memory 1050 and specifically perform the following operations:

and obtaining model information corresponding to the target algorithm model.

And receiving the target data characteristic information corresponding to the target user, which is sent by the mobile terminal under the condition of insufficient current computing power.

And running the target algorithm model to perform reasoning calculation based on the target data characteristic information and the model information to obtain a first target calculation result.

And sending the first target calculation result to the mobile terminal.

In some possible embodiments, the edge node includes a second computation container;

and converting the characteristic information of the target data.

In some possible embodiments, the processor 1010 is specifically configured to perform, when performing the conversion on the target data characteristic information:

and normalizing the target data characteristic information to obtain first target data characteristic information.

And converting the first target data characteristic information according to the operation rule of the second calculation container to obtain second target data characteristic information.

The processor 1010 performs the input of the converted target data feature information into the second calculation container, and performs the inference calculation by running the target algorithm model in the second calculation container based on the model information, so as to obtain a first target calculation result, and is specifically configured to perform:

In some possible embodiments, the edge node further comprises a hook layer;

the processor 1010 is configured to perform the converting the first target data characteristic information according to the operation rule of the second computing container to obtain second target data characteristic information, and is specifically configured to perform:

In some possible embodiments, when the processor 1010 executes to obtain the model information corresponding to the target algorithm model, the method is specifically configured to:

or (b)

In some possible embodiments, the processor 1010 is further configured to perform, after performing the inference calculation based on the target data feature information and the model information and running the target algorithm model to obtain a first target calculation result, perform:

The present description also provides a computer-readable storage medium having instructions stored therein, which when executed on a computer or processor, cause the computer or processor to perform one or more steps of the above embodiments. The above-mentioned constituent modules of the mobile terminal or the edge node may be stored in the above-mentioned computer-readable storage medium if implemented in the form of software functional units and sold or used as independent products.

In the above embodiments, it may be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented in software, may be implemented in whole or in part in the form of a computer program product. The computer program product described above includes one or more computer instructions. When the computer program instructions described above are loaded and executed on a computer, the processes or functions described in accordance with the embodiments of the present specification are all or partially produced. The computer may be a general purpose computer, a special purpose computer, a computer network, or other programmable apparatus. The computer instructions may be stored in or transmitted across a computer-readable storage medium. The computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by a wired (e.g., coaxial cable, fiber optic, digital subscriber line (Digital Subscriber Line, DSL)) or wireless (e.g., infrared, wireless, microwave, etc.). The computer readable storage media may be any available media that can be accessed by a computer or a data storage device such as a server, data center, or the like that contains an integration of one or more available media. The usable medium may be a magnetic medium (e.g., a floppy Disk, a hard Disk, a magnetic tape), an optical medium (e.g., a digital versatile Disk (Digital Versatile Disc, DVD)), or a semiconductor medium (e.g., a Solid State Disk (SSD)), or the like.

Those skilled in the art will appreciate that implementing all or part of the above-described embodiment methods may be accomplished by way of a computer program, which may be stored in a computer-readable storage medium, instructing relevant hardware, and which, when executed, may comprise the embodiment methods as described above. And the aforementioned storage medium includes: various media capable of storing program code, such as ROM, RAM, magnetic or optical disks. The technical features in the present examples and embodiments may be arbitrarily combined without conflict.

The above-described embodiments are merely preferred embodiments of the present disclosure, and do not limit the scope of the disclosure, and various modifications and improvements made by those skilled in the art to the technical solution of the disclosure should fall within the scope of protection defined by the claims.

The foregoing describes specific embodiments of the present disclosure. Other embodiments are within the scope of the following claims. In some cases, the actions or steps recited in the claims and description may be performed in an order different from that in the embodiments recited in the description and still achieve desirable results. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. An end-edge cloud collaborative computing method, which is applied to a mobile end, comprises the following steps:

responding to target operation of a target user, and judging whether the current computing power of the mobile terminal is sufficient;

and receiving the first target calculation result sent by the edge node.

2. The method of claim 1, the determining whether the current computing power of the mobile terminal is sufficient, comprising:

or (b)

3. The method of claim 1, the method further comprising, prior to determining whether the current computing power of the mobile terminal is sufficient:

4. The method of claim 3, after the determining whether the current computing power of the mobile terminal is sufficient in response to the target operation of the target user, the method further comprising:

5. The method of claim 4, the mobile end comprising a first computing container;

after receiving the model information corresponding to the target algorithm model issued by the cloud, running the target algorithm model based on the target data characteristic information and the model information to perform reasoning calculation, and before obtaining a second target calculation result, the method further comprises:

storing model information corresponding to the target algorithm model into the first computing container;

the step of operating the target algorithm model based on the target data characteristic information and the model information to perform reasoning calculation to obtain a second target calculation result comprises the following steps:

6. The method of claim 4, wherein the running the target algorithm model based on the target data feature information and the model information performs inference calculation to obtain a second target calculation result, the method further comprises:

And sending the second target calculation result to the cloud end so that the cloud end can make a data decision based on the second target calculation result.

7. The method of claim 3, wherein the sending the target data characteristic information corresponding to the target user to an edge node includes:

and sending the target data characteristic information corresponding to the target user and the model information corresponding to the target algorithm model to the edge node.

8. The method according to any one of claims 1-7, wherein the target data characteristic information is obtained by characteristic extraction based on target data; the target data includes at least one of: operating behavior data corresponding to the target user, equipment data corresponding to the mobile terminal and space-time data.

9. An end-edge cloud collaborative computing method, the method being applied to an edge node, the method comprising:

obtaining model information corresponding to a target algorithm model;

receiving target data characteristic information corresponding to a target user, which is sent by the mobile terminal under the condition of insufficient current computing power;

operating the target algorithm model based on the target data characteristic information and the model information to perform reasoning calculation so as to obtain a first target calculation result;

And sending the first target calculation result to the mobile terminal.

10. The method of claim 9, the edge node comprising a second computing container;

the step of operating the target algorithm model based on the target data characteristic information and the model information to perform reasoning calculation to obtain a first target calculation result comprises the following steps:

converting the target data characteristic information;

11. The method of claim 10, the converting the target data characteristic information comprising:

12. The method of claim 11, the edge node further comprising a layer of hooks;

and converting the first target data characteristic information by using the hook layer according to the operation rule of the second computing container to obtain second target data characteristic information.

13. The method of claim 9, wherein the obtaining model information corresponding to the target algorithm model comprises:

or (b)

14. The method according to any one of claims 9-13, wherein the target data characteristic information is obtained by characteristic extraction based on target data; the target data includes at least one of: operating behavior data corresponding to the target user, equipment data corresponding to the mobile terminal and space-time data.

15. The method according to any one of claims 9-13, wherein the running the target algorithm model based on the target data feature information and the model information performs inference calculation, and after obtaining a first target calculation result, the method further comprises:

and sending the first target calculation result to the cloud end so that the cloud end can make a data decision based on the first target calculation result.

16. A mobile terminal, the mobile terminal comprising:

17. An edge node, the edge node comprising:

the third receiving module is used for receiving target data characteristic information corresponding to a target user sent by the mobile terminal under the condition of insufficient current computing power;

18. A peer-to-peer cloud co-computing system, the peer Bian Yun co-computing system comprising a cloud, the mobile peer of claim 16, and the edge node of claim 17; the cloud end is used for sending model information corresponding to the target algorithm model to the mobile end and the edge node.

19. An electronic device, comprising: a processor and a memory;

the processor is connected with the memory;

the memory is used for storing executable program codes;

the processor runs a program corresponding to executable program code stored in the memory by reading the executable program code for performing the method according to any one of claims 1-15.

20. A computer storage medium storing a plurality of instructions adapted to be loaded by a processor and to perform the method steps of any of claims 1-15.

21. A computer program product comprising instructions which, when run on a computer or processor, cause the computer or processor to perform the terminal Bian Yun co-computing method of any of claims 1-15.