CN113259472A - Edge node resource allocation method for video analysis task - Google Patents
Edge node resource allocation method for video analysis task Download PDFInfo
- Publication number
- CN113259472A CN113259472A CN202110634436.6A CN202110634436A CN113259472A CN 113259472 A CN113259472 A CN 113259472A CN 202110634436 A CN202110634436 A CN 202110634436A CN 113259472 A CN113259472 A CN 113259472A
- Authority
- CN
- China
- Prior art keywords
- video analysis
- time
- edge node
- delay
- optimization problem
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/50—Network services
- H04L67/60—Scheduling or organising the servicing of application requests, e.g. requests for application data transmissions using the analysis and optimisation of the required network resources
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/10—Protocols in which an application is distributed across nodes in the network
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a video analysis task-oriented edge node resource allocation method, which comprises the steps of after video analysis tasks of all applications are issued, dividing the whole service process into a plurality of time intervals, and determining an optimization problem for each time interval based on Lyapunov optimization; in each time interval, selecting video resolution for all applications by solving an optimization problem, and distributing resources on a plurality of edge nodes; each application selects corresponding video resolution, and unloads a video analysis task to a preset edge node; and the edge node deploys the service, processes the tasks unloaded to the service, and sends the processed tasks to the application or other edge nodes. The invention can effectively shorten the sum of the completion time of all tasks and can ensure that the video analysis accuracy rate meets the application requirement.
Description
Technical Field
The invention belongs to the technical field of edge computing, and particularly relates to a video analysis task-oriented edge node resource allocation method.
Background
Thousands of monitoring cameras are installed in cities, and the cameras are used for traffic control, safety monitoring, factory assembly line monitoring and other scenes. Particularly, in the electric power thing networking, surveillance camera head is used for fields such as equipment inspection, wisdom building site. It is not practical to perform analysis of these videos with the naked eye due to high labor costs, etc. With the rapid progress of technologies such as a convolutional neural network and the like, the accuracy of a computer for completing tasks such as object recognition, image classification and the like is greatly improved, and even the accuracy of the computer exceeds the accuracy of a human in some aspects. Therefore, a large number of targeted video analysis algorithms are beginning to be used for processing real-time video streams, and real-time intelligent monitoring is achieved. Although the video analysis algorithm based on the neural network and the application thereof can realize quite high video analysis accuracy, the requirement on computing resources is very strict, and the video analysis algorithm is difficult to directly run on small-sized equipment such as a mobile phone, a camera and the like. Therefore, the video stream is often required to be imported into the remote cloud data center for performing related video analysis and processing, but the video stream data volume is large, so that not only is the system response delay long, but also the pressure and the overhead of network transmission are greatly increased.
Edge computing is a new type of computing model that deploys computing resources to the edge of a network. Compared with cloud computing, edge computing has the advantages of smaller time delay and smaller network load, and has higher resource utilization rate, so that the edge computing is hopeful to be used for completing video analysis tasks of a plurality of cameras. A typical application scenario for performing video analysis tasks using edge computation is: an wisdom building site deploys edge node and a plurality of camera, and the camera constantly produces video data, utilizes edge node to come real-time processing video, discovers and solves the problem that exists in time (like safety helmet detection, dangerous behavior discernment, material safety guarantee etc.). FIG. 1 is a schematic diagram of a plurality of edge nodes processing a plurality of application tasks; the method shows an example that 2 applications share 3 edge nodes, and the process that the applications complete video analysis tasks by using the edge nodes is divided into 5 steps:
(1) the application generates tasks and their data, ready to be delivered to the edge nodes for processing.
(2) Data of the application is transmitted from the terminal device to the access point.
(3) The access point forwards the data to the edge node.
(4) The edge node processes the task, if there are subsequent processing steps, send the data to the next edge node; otherwise, the data is sent to the access point.
(5) The access point returns the task results to the application.
It is worth noting that edge computing does not simply push the resources of a data center to the edge of a network, and designing a resource allocation strategy is one of the key problems of edge computing, and the following factors are considered to solve the problem:
(1) the edge nodes have fewer computing resources than the data center. Consider a scenario in which: each application submits the tasks to the edge nodes closest to the application in a self-private way, when the geographic positions of the applications are unevenly distributed, the loads of some edge nodes are too high, other edge nodes are in an idle state, and the service quality experienced by the applications is not high at the moment. Therefore, the user experience can be improved only by planning the resource allocation as a whole.
(2) The resource allocation is not a constant. A good resource allocation may deteriorate after a period of time due to network fluctuations, so the resource allocation should be continuously adjusted as the network conditions change.
(3) Deploying services is costly. On the one hand, deployment of functions as a Service (FaaS), Software as a Service (SaaS), and the like consumes time. If the resource allocation is adjusted frequently, the high deployment latency may seriously affect the user experience. On the other hand, the deployment of a service involving several devices in a data center increases the energy consumption of these devices. Today, data centers are very energy intensive and people want to save energy as much as possible, and from an economic point of view, we should avoid frequent service deployment.
At present, there are many research achievements aiming at the problem of resource allocation in the edge computing environment, and the general problem of resource allocation in edge computing mainly focuses on reducing the time delay of applications. Due to the particularity of the video analysis task, it is not desirable to equate the resource allocation problem of the video analysis task with the general problem, because:
(1) applications have certain requirements on the accuracy of video analysis. The accuracy rate of video analysis is highly related to the resolution rate of video frames, and the higher resolution rate is hopeful to improve the accuracy rate of object identification of the convolutional neural network. Meanwhile, the transmission delay is also in positive correlation with the resolution of the video frame, so that the improvement of the accuracy and the reduction of the delay are two contradictory targets to a certain extent. Carefully trade-offs between the two goals to provide better quality of service.
(2) A video analysis task may contain several subtasks. There are dependencies between these subtasks, which we denote with a Directed Acyclic Graph (DAG). Fig. 2 is a schematic diagram of a video processing application of Facebook, in which a plurality of dependent subtasks together form a video processing task. Similarly, the video analysis task of an application may be performed by multiple edge nodes together, and the delay experienced by the user includes the computation delay of the edge nodes and the network delay of the data flow between the edge nodes.
Due to the difference between the resource allocation problem of the video analysis task and the general resource allocation problem, no effective solution is found in the current research results.
Disclosure of Invention
In order to solve the problems mentioned in the prior art, the invention aims to provide a video analysis task-oriented edge node resource allocation method, which aims to reasonably allocate resources of a plurality of edge nodes to a plurality of applications and reduce the sum of time delays of all tasks on the premise of meeting the requirement of the applications on video analysis accuracy as much as possible.
In order to achieve the purpose of the invention, the invention adopts the following technical scheme:
a video analysis task-oriented edge node resource allocation method is characterized by comprising the following steps:
1) after video analysis tasks of all applications are issued, the whole service process is divided into a plurality of time intervals, and an optimization problem is determined for each time interval based on Lyapunov optimization;
2) in each time interval, selecting video resolution for all applications by solving an optimization problem, and distributing resources on a plurality of edge nodes; each application selects corresponding video resolution, and unloads a video analysis task to a preset edge node;
3) and the edge node deploys the service, processes the tasks unloaded to the service, and sends the processed tasks to the application or other edge nodes.
Further, in step 1), the determining an optimization problem for each time interval based on lyapunov optimization includes:
(1) determining the execution sequence of each subtask according to the dependency relationship among the subtasks contained in each video analysis task;
(2) constructing a global optimization problem related to all time intervals by taking the minimum total time delay as a target and taking the requirement of the applied video analysis accuracy as a constraint;
(3) converting the global optimization problem in the step (2) into a global optimization problem in a long time;
(4) and (4) constructing a virtual queue for the global optimization problem in the step (3), decoupling the global optimization problem, and generating an independent optimization problem in each time interval.
The step (2) comprises the following steps:
the network delay consists of transmission delay from the application equipment to the access point, propagation delay from the access point to the edge node and propagation delay between the edge nodes;
calculating time delay is the time spent by calculating a task for the edge node and is obtained by carrying out experimental measurement in advance;
the service deployment time delay consists of the time delay of downloading the service by the edge node and the time delay of installing the service, and the time delay of installing the service is measured by experiments;
in each time interval, the time delay of each application comprises three parts, namely network time delay, calculation time delay and service deployment time delay;
the total delay is the sum of the delays applied to each interval.
In step 2), the selecting the video resolution for all applications by solving the optimization problem, and allocating resources on the plurality of edge nodes includes:
(a) measuring the network state of the current time interval;
(b) solving an optimization problem in the current time interval to obtain a resource allocation scheme of the edge node and an applied resolution selection scheme;
(c) adjusting the resource allocation and the application resolution of the edge nodes in real time;
(d) the virtual queue for the next time interval is calculated.
Compared with the prior art, the invention has the following beneficial effects:
aiming at a system with a plurality of edge nodes and a plurality of applications, an edge node resource allocation and application resolution selection algorithm facing a video analysis task is firstly provided, the algorithm considers the dependency relationship among subtasks, and the effect of meeting the requirement of application accuracy within a long time and effectively reducing time delay is realized by balancing between maintaining the video analysis accuracy and reducing the time delay.
The invention can effectively shorten the sum of the completion time of all tasks and can ensure that the video analysis accuracy rate meets the application requirement. The invention is superior to the most advanced edge node resource allocation strategy in the video analysis task.
Drawings
FIG. 1 is a schematic diagram of a plurality of edge nodes processing a plurality of application tasks;
fig. 2 is a schematic diagram of a video analysis task and its subtasks.
Detailed Description
The technical solution of the present invention is further explained with reference to the drawings and the embodiments.
Aiming at the problem of resource allocation for processing video analysis tasks for multiple applications based on edge nodes under the constraint of video analysis accuracy, the sum of completion time delays of all tasks in a long-term time is optimized by coordinating the use of computing resources of the multiple tasks in the multiple edge nodes. The optimization idea is that the video analysis accuracy rate may be sacrificed in a short time to obtain a lower time delay or a higher time delay to obtain a higher accuracy rate, but the video analysis accuracy rate constraint can be satisfied in a long time and the total time delay of the video analysis task is reduced. For the sake of clarity of the description of the specific method steps, in the present embodiment, the description is made in conjunction with the following network system. Fig. 1 is a schematic diagram of a plurality of edge nodes processing a plurality of application tasks, and illustrates an example of 2 applications sharing 3 edge nodes.
A video analysis task-oriented edge node resource allocation method includes the steps that after all applied video analysis tasks are issued, the whole service process is divided into a plurality of time intervals, and an optimization problem is determined for each time interval based on Lyapunov optimization; in each time interval, selecting video resolution for all applications by solving an optimization problem, and distributing resources on a plurality of edge nodes; each application selects corresponding video resolution, and unloads a video analysis task to a preset edge node; and the edge node deploys the service, processes the tasks unloaded to the service, and sends the processed tasks to the application or other edge nodes. The method comprises the following specific steps:
consider that
The network state and the application requirement are kept relatively stable in one time interval and change in different time intervals. In a marginal environment have
An edge node, the set of edge nodes being represented as
Wherein
Are data centers and the rest are edge nodes. The edge nodes are not essentially different from the data center, and the data center can be considered as calculation resourcesThe source is very abundant, and the video analysis task is either done in the edge node or in the data center. These edge nodes are deployed at various sites in a city, interconnected by a metropolitan area network. Is provided with
An application, the set of applications being
The terminal devices are distributed at various positions and communicate with the edge node through the access point.
Is provided with
A set of tasks being
The tasks of each application are offloaded to the edge nodes for processing. The dependency relationship between tasks is represented by a directed path
. It should be noted that, the general dependency needs to be represented by DAG, and fig. 2 is a schematic diagram of a video analysis task and its subtasks. Modeling the DAG, however, can complicate the problem, and to be able to quickly arrive at a reasonable resource allocation scheme, we serialize the DAG into a directed path.
By using
Representing a time interval
Network latency in (1). When the application submits the task, the data is firstly sent from the terminal equipment to the access point, then sent from the access point to the edge node, and then transmitted between the edge nodes, so the network is realizedThe network delay consists of three parts: the time delay from the terminal equipment to the access point, the time delay from the access point to the edge node and the time delay between the edge nodes.
First consider the delay of the terminal device to the access point. Typically, a terminal device connects to an access point using a wireless network, and the time delay from the terminal device to the access point is determined by the amount of data and the transmission rate. Is provided with
In a time interval
Has an image resolution of
If each pixel comprises
One bit, then one video frame size is
. By using
Representing a time interval
In
And the access point, thereby obtaining the transmission rate
In a time interval
Time delay to access point
Next, the access point to edge node delay and the delay between edge nodes are considered. Without loss of generality, assume to be arbitrary
，
And
the method is a task which is finished on the access point and has zero calculation time delay, so that the access point can be regarded as an edge node, and the method is more convenient and uniform in expression. In order to calculate the time delay between edge nodes, it is critical to find out which pair of edge nodes need to send and receive data. By using
Indicating a time interval
In
Whether or not to process
By using
Indicating a time interval
In
And
whether there is data to be transmitted or received. If present, is
And
satisfy the requirement of
Then
That is, it means that data is transmitted and received between two edge nodes, and there is a propagation delay. In a more formal sense, the first and second,
set time interval
In
And
has a propagation delay of
We get the time interval
Time delay between middle edge nodes
Synthesizing the time delay from the terminal equipment to the access point and the time delay between the edge nodes, and the time interval
The network delay in (1) is expressed as
By using
To represent
In a time interval
The calculated time delay. The computation latency depends on the computational complexity of the task and the amount of computational resources. Assuming that each edge node evenly distributes resources to the tasks it is responsible for, the amount of resources a task occupies is determined by the amount of resources at the edge node, the number of tasks running on the edge node. Is provided with
Is calculated by the amount of computing resources of
Thereby to make
To give
Number of resources allocated
Is composed of
Assuming presence function
Can estimate according to the number of resources and the type of tasksCalculate the time delay, then
Is calculated by
. In fact, if the task type is known, the relationship between the time required for completing the task and the resource quantity can be measured through experiments, and then the relationship is obtained through curve fitting
. The calculated delay for an application is the sum of the calculated delays for all tasks of the application, so
By using
Representing a time interval
Service deployment latency. The edge nodes have limited computing resources, cannot run all services all the time, and adopt a mode of deployment according to needs. If in the time interval
In
Is not responsible for processing
But in time intervals
In
Is responsible for processing
It is considered that
In a time interval
Where the service needs to be deployed. More formally stated, for a given
If present, if present
And
is provided with
And is
Then it is considered as
In a time interval
Where the service needs to be deployed. The service deployment process is that the edge node retrieves the service from the data center and then installs the service, so that the service deployment delay and the propagation delay and the installation delay are in positive correlation. By using
To represent
Mean time to install a service, respectively
And
is shown in the time interval
Time delay for retrieving service and time delay for installing service, some
Service deployment latency is the sum of the latency of retrieving a service and the latency of installing a service
By using
Represents the sum of time delays of all time intervals, and the time delay of each time interval consists of network time delay, calculation time delay and service deployment time delay, so
As the application scenario changes, the requirements of the application on the accuracy of the video analysis also change, for example: for safety, very high accuracy is desirable for the application of analytical theft monitoring; applications that calculate the flow of people may be willing to accept a range of errors and not be critical to the accuracy. Because the requirements of the applications on the accuracy are different, a constraint condition about the accuracy is established for each application. Suppose that
Ideal accuracy rate not less than
And a function exists
Can estimate the accuracy according to the resolution and the task category, thereby
Is that
. And
similarly, it can be obtained by means of experiment and curve fitting
. Expressing the application's requirement for accuracy as
The aim is to reduce the processing delay of the task as much as possible on the premise of meeting the requirement of the application on the accuracy, thereby obtaining the optimization problem of resource allocation and resolution selection
Virtual queue is constructed according to Lyapunov optimization theory
And
introducing custom parameters
Using virtual queues to transfer data to
Decoupling as independent sub-problems within each time interval
Solving for
Can obtain the resource allocation of each edge node for each application, and solve
The video resolution of the application can be obtained. Combining the above analysis, each time interval is now given
The steps of calculating resource allocation and video resolution:
(1) measuring a network state;
(2) solving for
Obtaining a resource allocation scheme;
(3) adjusting the resource allocation of the edge nodes in real time;
(4) solving for
Obtaining a video resolution selection scheme;
(5) adjusting the video resolution of the application in real time;
(6) computing
；
(7) For each one
Computing
。
Claims (4)
1. A video analysis task-oriented edge node resource allocation method is characterized by comprising the following steps:
1) after video analysis tasks of all applications are issued, the whole service process is divided into a plurality of time intervals, and an optimization problem is determined for each time interval based on Lyapunov optimization;
2) in each time interval, selecting video resolution for all applications by solving an optimization problem, and distributing resources on a plurality of edge nodes; each application selects corresponding video resolution, and unloads a video analysis task to a preset edge node;
3) and the edge node deploys the service, processes the tasks unloaded to the service, and sends the processed tasks to the application or other edge nodes.
2. The method for allocating resource of edge node facing to video analysis task as claimed in claim 1, wherein in step 1), said determining an optimization problem for each time interval based on lyapunov optimization comprises the following steps:
(1) determining the execution sequence of each subtask according to the dependency relationship among the subtasks contained in each video analysis task;
(2) constructing a global optimization problem related to all time intervals by taking the minimum total time delay as a target and taking the requirement of the applied video analysis accuracy as a constraint;
(3) converting the global optimization problem in the step (2) into a global optimization problem in a long time;
(4) and (4) constructing a virtual queue for the global optimization problem in the step (3), decoupling the global optimization problem, and generating an independent optimization problem in each time interval.
3. The video analysis task-oriented edge node resource allocation method according to claim 2, wherein in the step (2):
the network delay consists of transmission delay from the application equipment to the access point, propagation delay from the access point to the edge node and propagation delay between the edge nodes;
calculating time delay is the time spent by calculating a task for the edge node and is obtained by carrying out experimental measurement in advance;
the service deployment time delay consists of the time delay of downloading the service by the edge node and the time delay of installing the service, and the time delay of installing the service is measured by experiments;
in each time interval, the time delay of each application comprises three parts, namely network time delay, calculation time delay and service deployment time delay;
the total delay is the sum of the delays applied to each interval.
4. The method for distributing the resources of the edge nodes facing the video analysis task, according to claim 1, wherein in the step 2), the selecting the video resolution for all the applications by solving the optimization problem and distributing the resources on the plurality of edge nodes comprises:
(a) measuring the network state of the current time interval;
(b) solving an optimization problem in the current time interval to obtain a resource allocation scheme of the edge node and an applied resolution selection scheme;
(c) adjusting the resource allocation and the application resolution of the edge nodes in real time;
(d) the virtual queue for the next time interval is calculated.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110634436.6A CN113259472A (en) | 2021-06-08 | 2021-06-08 | Edge node resource allocation method for video analysis task |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110634436.6A CN113259472A (en) | 2021-06-08 | 2021-06-08 | Edge node resource allocation method for video analysis task |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113259472A true CN113259472A (en) | 2021-08-13 |
Family
ID=77186902
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110634436.6A Pending CN113259472A (en) | 2021-06-08 | 2021-06-08 | Edge node resource allocation method for video analysis task |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113259472A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114302233A (en) * | 2021-12-10 | 2022-04-08 | 网络通信与安全紫金山实验室 | Video compression and network service quality joint decision method and device |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109194984A (en) * | 2018-11-09 | 2019-01-11 | 中山大学 | A kind of video frame dispatching method based on edge calculations |
| CN109656703A (en) * | 2018-12-19 | 2019-04-19 | 重庆邮电大学 | A kind of mobile edge calculations auxiliary vehicle task discharging method |
| CN110290011A (en) * | 2019-07-03 | 2019-09-27 | 中山大学 | Dynamic Service laying method based on Lyapunov control optimization in edge calculations |
| CN110557287A (en) * | 2019-09-10 | 2019-12-10 | 北京邮电大学 | Resource allocation method and device based on Lyapunov optimization |
| CN110928691A (en) * | 2019-12-26 | 2020-03-27 | 广东工业大学 | Traffic data-oriented edge collaborative computing unloading method |
| CN111278132A (en) * | 2020-01-19 | 2020-06-12 | 重庆邮电大学 | Resource allocation method for low-delay high-reliability service in mobile edge calculation |
| US20200192598A1 (en) * | 2018-12-18 | 2020-06-18 | Hewlett Packard Enterprise Development Lp | Adiabatic Annealing Scheme and System for Edge Computing |
| CN112148492A (en) * | 2020-09-28 | 2020-12-29 | 南京大学 | Service deployment and resource allocation method considering multi-user mobility |
| CN112347941A (en) * | 2020-11-09 | 2021-02-09 | 南京紫金体育产业股份有限公司 | Motion video collection intelligent generation and distribution method based on 5G MEC |
| CN112601197A (en) * | 2020-12-18 | 2021-04-02 | 重庆邮电大学 | Resource optimization method in train-connected network based on non-orthogonal multiple access |
| CN112804219A (en) * | 2020-12-31 | 2021-05-14 | 中山大学 | Low-delay real-time video analysis method based on edge calculation |
-
2021
- 2021-06-08 CN CN202110634436.6A patent/CN113259472A/en active Pending
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109194984A (en) * | 2018-11-09 | 2019-01-11 | 中山大学 | A kind of video frame dispatching method based on edge calculations |
| US20200192598A1 (en) * | 2018-12-18 | 2020-06-18 | Hewlett Packard Enterprise Development Lp | Adiabatic Annealing Scheme and System for Edge Computing |
| CN109656703A (en) * | 2018-12-19 | 2019-04-19 | 重庆邮电大学 | A kind of mobile edge calculations auxiliary vehicle task discharging method |
| CN110290011A (en) * | 2019-07-03 | 2019-09-27 | 中山大学 | Dynamic Service laying method based on Lyapunov control optimization in edge calculations |
| CN110557287A (en) * | 2019-09-10 | 2019-12-10 | 北京邮电大学 | Resource allocation method and device based on Lyapunov optimization |
| CN110928691A (en) * | 2019-12-26 | 2020-03-27 | 广东工业大学 | Traffic data-oriented edge collaborative computing unloading method |
| CN111278132A (en) * | 2020-01-19 | 2020-06-12 | 重庆邮电大学 | Resource allocation method for low-delay high-reliability service in mobile edge calculation |
| CN112148492A (en) * | 2020-09-28 | 2020-12-29 | 南京大学 | Service deployment and resource allocation method considering multi-user mobility |
| CN112347941A (en) * | 2020-11-09 | 2021-02-09 | 南京紫金体育产业股份有限公司 | Motion video collection intelligent generation and distribution method based on 5G MEC |
| CN112601197A (en) * | 2020-12-18 | 2021-04-02 | 重庆邮电大学 | Resource optimization method in train-connected network based on non-orthogonal multiple access |
| CN112804219A (en) * | 2020-12-31 | 2021-05-14 | 中山大学 | Low-delay real-time video analysis method based on edge calculation |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114302233A (en) * | 2021-12-10 | 2022-04-08 | 网络通信与安全紫金山实验室 | Video compression and network service quality joint decision method and device |
| CN114302233B (en) * | 2021-12-10 | 2023-10-27 | 网络通信与安全紫金山实验室 | Video compression and network service quality joint decision method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhu et al. | Task offloading decision in fog computing system | |
| Chen et al. | Efficiency and fairness oriented dynamic task offloading in internet of vehicles | |
| CN112188442A (en) | Vehicle networking data-driven task unloading system and method based on mobile edge calculation | |
| Zhang et al. | Blockchain-based collaborative edge intelligence for trustworthy and real-time video surveillance | |
| CN112148492B (en) | Service deployment and resource allocation method considering multi-user mobility | |
| Desikan et al. | Topology control in fog computing enabled IoT networks for smart cities | |
| CN112188627B (en) | Dynamic resource allocation strategy based on state prediction | |
| CN113553160A (en) | Task scheduling method and system for edge computing node of artificial intelligence Internet of things | |
| KR102350195B1 (en) | Energy Optimization Scheme of Mobile Devices for Mobile Augmented Reality Applications in Mobile Edge Computing | |
| US20230136612A1 (en) | Optimizing concurrent execution using networked processing units | |
| Naik et al. | Minimizing deadline misses and total run-time with load balancing for a connected car systems in fog computing | |
| Fang et al. | Multi-tenant mobile offloading systems for real-time computer vision applications | |
| CN113965569B (en) | High-energy-efficiency low-delay edge node calculation migration configuration system | |
| CN111131447A (en) | Load balancing method based on intermediate node task allocation | |
| Strelkovskaya et al. | Different extrapolation methods in Problems of Forecasting | |
| CN113259472A (en) | Edge node resource allocation method for video analysis task | |
| CN113504949B (en) | Task unloading and parameter optimization method and system for MAR client in edge calculation | |
| Salehnia et al. | SDN-based optimal task scheduling method in Fog-IoT network using combination of AO and WOA | |
| Li et al. | Task offloading and parameters optimization of MAR in multi-access edge computing | |
| Sopin et al. | The analysis of the computation offloading scheme with two-parameter offloading criterion in fog computing | |
| Li et al. | Self-adaptive load-balancing strategy based on a time series pattern for concurrent user access on Web map service | |
| Chen et al. | Knowledge distillation for mobile edge computation offloading | |
| Wang et al. | Task offloading for edge computing in industrial Internet with joint data compression and security protection | |
| CN111539863B (en) | Intelligent city operation method and system based on multi-source task line | |
| Saranya et al. | An efficient computational offloading framework using HAA optimization-based deep reinforcement learning in edge-based cloud computing architecture |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210813 |
|
| RJ01 | Rejection of invention patent application after publication |