CN111431941B - Real-time video code rate self-adaption method based on mobile edge calculation - Google Patents

Real-time video code rate self-adaption method based on mobile edge calculation Download PDF

Info

Publication number
CN111431941B
CN111431941B CN202010401015.4A CN202010401015A CN111431941B CN 111431941 B CN111431941 B CN 111431941B CN 202010401015 A CN202010401015 A CN 202010401015A CN 111431941 B CN111431941 B CN 111431941B
Authority
CN
China
Prior art keywords
mobile
edge
video
task
qoe
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.)
Active
Application number
CN202010401015.4A
Other languages
Chinese (zh)
Other versions
CN111431941A (en
Inventor
白光伟
肖强
沈航
孙鹏
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nanjing Tech University
Original Assignee
Nanjing Tech University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Nanjing Tech University filed Critical Nanjing Tech University
Priority to CN202010401015.4A priority Critical patent/CN111431941B/en
Publication of CN111431941A publication Critical patent/CN111431941A/en
Application granted granted Critical
Publication of CN111431941B publication Critical patent/CN111431941B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L65/00—Network arrangements, protocols or services for supporting real-time applications in data packet communication
    • H04L65/40—Support for services or applications
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L65/00—Network arrangements, protocols or services for supporting real-time applications in data packet communication
    • H04L65/80—Responding to QoS
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/20—Servers specifically adapted for the distribution of content, e.g. VOD servers; Operations thereof
    • H04N21/21—Server components or server architectures
    • H04N21/218—Source of audio or video content, e.g. local disk arrays
    • H04N21/2187—Live feed
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/20—Servers specifically adapted for the distribution of content, e.g. VOD servers; Operations thereof
    • H04N21/23—Processing of content or additional data; Elementary server operations; Server middleware
    • H04N21/239—Interfacing the upstream path of the transmission network, e.g. prioritizing client content requests
    • H04N21/2393—Interfacing the upstream path of the transmission network, e.g. prioritizing client content requests involving handling client requests
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/20—Servers specifically adapted for the distribution of content, e.g. VOD servers; Operations thereof
    • H04N21/25—Management operations performed by the server for facilitating the content distribution or administrating data related to end-users or client devices, e.g. end-user or client device authentication, learning user preferences for recommending movies
    • H04N21/262—Content or additional data distribution scheduling, e.g. sending additional data at off-peak times, updating software modules, calculating the carousel transmission frequency, delaying a video stream transmission, generating play-lists
    • H04N21/26208—Content or additional data distribution scheduling, e.g. sending additional data at off-peak times, updating software modules, calculating the carousel transmission frequency, delaying a video stream transmission, generating play-lists the scheduling operation being performed under constraints
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/60—Network structure or processes for video distribution between server and client or between remote clients; Control signalling between clients, server and network components; Transmission of management data between server and client, e.g. sending from server to client commands for recording incoming content stream; Communication details between server and client 
    • H04N21/63—Control signaling related to video distribution between client, server and network components; Network processes for video distribution between server and clients or between remote clients, e.g. transmitting basic layer and enhancement layers over different transmission paths, setting up a peer-to-peer communication via Internet between remote STB's; Communication protocols; Addressing
    • H04N21/637—Control signals issued by the client directed to the server or network components
    • H04N21/6373—Control signals issued by the client directed to the server or network components for rate control, e.g. request to the server to modify its transmission rate

Abstract

The invention provides a real-time video code rate self-adaption method based on mobile edge calculation, which comprises the following steps: s1, collecting parameters of video playing, network connection, equipment performance and the like of a user in real time; s2, selecting a task scheduling strategy with optimal performance for the user according to the parameters of S1; s3, automatically switching the working mode according to the selected task scheduling strategy; s4, calculating QoE of the current scheduling strategy according to the video playing condition in the next iteration period, the equipment energy consumption and other parameters, and feeding the QoE back to the edge server; s5, the edge server updates the state transition matrix of the scheduling strategy according to the feedback QoE; s6, repeating the steps, continuously iterating and updating, and finally enabling the performance of the edge scheduling strategy to tend to be optimal; the invention can reduce larger energy consumption for the mobile multimedia equipment on the basis of lower cost, and help the complex video code rate adaptive algorithm learning system to continuously grow and optimize.

Description

Real-time video code rate self-adaption method based on mobile edge calculation
Technical Field
The invention belongs to the field of edge calculation, and particularly relates to a real-time video code rate self-adaption method based on mobile edge calculation.
Background
Due to the rapid development of smart mobile devices and the rise of various video APPs, more and more users start watching videos through mobile devices. Most video service providers today want to provide quality video services to as many users as possible, thereby increasing the viewing time of the users. However, mobile intelligent devices always fluctuate with more complex networks, so that some video bitrate adaptive algorithms based on network conditions are produced at present, and the algorithms select appropriate bitrate for users by analyzing network states or local cache margins, so that blocking is avoided, and video service quality is improved.
Most of the existing mainstream video code rate algorithms concentrate computing tasks on a cloud server, which causes extra energy to be consumed by mobile equipment in the process of counting and summarizing client playing information and transmitting the client playing information to a cloud end. Furthermore, the delay caused by transmission over the wide area network may severely impact the interactivity of the mobile multimedia. On the contrary, due to the characteristics of limited computing power and battery endurance of the mobile device, many defects are also caused by deploying some complex code rate adjustment algorithm strategies at the mobile device end.
MEC (Mobile Edge Computing) is an effective method to solve the above problems. In the MEC framework, cloud computing capabilities are provided within the wireless access network proximate to these mobile devices. In other words, with the MEC, the mobile multimedia bitrate adaptive system can offload its computational tasks to the MEC server at the edge of the network, rather than using the server in the core network or the device itself. The method has the advantages of low delay, high bandwidth and high flexibility in the code rate self-adaption process. If the rate adaptive algorithm involves on-line training, the edge calculation mode has the advantage of higher sample acquisition efficiency than the mode in which the algorithm is deployed at the client, allowing the user not to have to be on-line continuously. Some machine learning algorithms do bring better flexibility and accuracy to mobile video bitrate selection, but the intensive computing requirements also increase the load pressure of a mobile cloud server and the operation pressure of a mobile intelligent device, so that a mode of combining the mobile intelligent device and edge computing can be selected to meet the challenges.
Compared with the traditional Mobile Cloud Computing (MCC), the edge server is closer to the mobile user, and the mobile user can obtain the required computing resource by only one hop of wireless transmission, and the delay is lower compared with the MCC. However, MEC servers have limited computing power and are less scalable. With the explosive growth of mobile users and emerging applications, simply relying on MECs cannot fully meet the computing offload needs of mobile users. Therefore, the MEC should not completely replace cloud computing or local computing of the mobile terminal, and the three should be coordinated and supplemented with each other to better meet the requirements of the mobile user. For example, when latency requirements are low, rather than latency sensitive tasks being transmitted to the cloud computing center for execution, they may be executed by the MEC or locally.
Disclosure of Invention
The invention aims to solve the technical problem of providing a real-time video code rate self-adaption method based on edge calculation, which is based on mobile equipment such as a smart phone and the like, and automatically adjusts code rate self-adaption operation tasks to dynamically migrate in the mobile equipment terminal and an edge calculation node through an MEC according to the processor load and the network connection condition of the mobile equipment terminal, thereby realizing the real-time video code rate self-adaption function under various network conditions.
In order to solve the technical problems, the invention specifically adopts the following technical scheme:
the invention provides a real-time video code rate self-adaptive method based on mobile edge calculation, which comprises the following steps:
s1, the mobile client acquires performance parameters of the video playing user in real time, wherein the performance parameters comprise video playing conditions, multimedia network connection conditions and mobile equipment performance, and defines a code rate self-adaption task;
s2, the edge server selects a task scheduling strategy with optimal performance for the user according to the parameter data collected by the mobile client;
s3, the mobile client automatically switches the working mode according to the selected task scheduling strategy;
s4, the mobile client calculates QoE of the current scheduling strategy according to the video playing condition in the next iteration period and the equipment energy consumption parameter, and feeds the QoE back to the edge server;
s5, the edge server forms a mapping table according to the task mode and the environment state fed back by the mobile client and the user QoE after the task mode is selected, and updates the state transition matrix of the scheduling strategy; when a user needs to perform an edge computing scheduling task next time, a scheduling mode with the highest long-term QoE can be selected from the mapping table.
Further, in the method for adaptive rate of real-time video for mobile edge calculation according to the present invention, in step S1, the video playing condition includes average video quality, average quality variation amplitude, and pause duration; the multimedia network connection condition comprises an average value of user perception throughput and a standard deviation of the user perception throughput in the current time period; the mobile device capabilities include the local computing power of device i and the power consumption of a single cpu cycle.
Further, in step S2, the edge server selects a task scheduling policy with optimal performance for the user according to the parameter data collected by the client, and the method specifically includes the following operations:
s21, expressing the code rate self-adaption task of the mobile equipment as T according to the collected environment informationiThe expression is as follows:
Figure BDA0002489478450000021
wherein d isiIs the size of input data for calculation, including program code, input files; c. CiRepresents the amount of computation required to complete this task, quantified by the number of cpu cycles;
Figure BDA0002489478450000022
is to calculate the task maximum latency, i.e. the delay constraint duration,
Figure BDA0002489478450000023
and the category of the video to be viewed, and the fluctuation degree of the network.
S22, judging whether the mobile video user needs to call the edge service according to the environment state and the reinforcement learning state transition matrix, wherein the judgment is based on whether the QoE is higher than that of local calculation after edge calculation is used;
s23, determining the priority: determining the priority for each video code rate self-adaptive task, wherein the edge server provides service for mobile video users with higher priority preferentially; the priority is used for wireless resource allocation and is determined by wireless communication state, task delay constraint and task property factors;
s24, channel allocation: and allocating the channels to the equipment according to the priority determined in advance, judging whether the energy consumption is lower than that of the original channel when the channels are allocated, replacing the channels if the energy consumption can be effectively reduced, and keeping the original state if the energy consumption is not reduced.
Further, the method for adaptive real-time video bitrate for mobile edge computing provided by the present invention is characterized in that, in step S22, it is determined whether a mobile video user needs to invoke an edge service according to an environment state and a reinforcement learning state transition matrix, and the specific operations are as follows:
(1) the set of a class of devices on the MEC server that perform their computational tasks is denoted GRThen delay constraint
Figure BDA0002489478450000031
The calculation method is as follows:
Figure BDA0002489478450000032
in the above formula, BkRepresenting the remaining length of the video buffer, minus the total time consumed by the communication and calculation tasks; diIs the size of the input data for the calculation, ciIndicating the amount of computation required to accomplish this task,
Figure BDA0002489478450000033
is the computing power of the MEC server, riIs the uplink rate of data transmitted from device i to the edge server;
(2) the set of devices on their local devices that perform their computational tasks is denoted GLThe conditions for determining a device belonging to this type are as follows: if it is not
Figure BDA0002489478450000034
And is
Figure BDA0002489478450000035
This means that tasks are more efficient at local computation when the local computation satisfies the delay constraint and the device energy consumption is lower than invoking edge services over the wireless network, where:
Figure BDA0002489478450000036
wherein d isiIs the size of the input data for the calculation,
Figure BDA0002489478450000037
is a coefficient representing a backhaul transmission time delay of unit data, w represents a channel bandwidth, piIs the power at which the mobile i sends data to the edge server in the unit channel, giIs the channel gain, σ, between the mobile user i and the edge server2Is background noise power, wlog2(..) is actually the uplink data transmission rate, r, obtained by the mobile device i when accessing the edge serveri。
Further, in the real-time video bitrate adaptive method for mobile edge computation provided by the present invention, the step S23 determines the priority specifically as follows:
s231. for the equipment G with insufficient computing capabilityRThe calculation task can be completed only with the assistance of the MEC server, and the wireless resource allocation of the equipment has the highest priority;
s232, for a device set G which can be selectively executed locally or unloaded to an edge server for executionOThe mobile client in (1) should be assigned with different priorities, and the priority of the device i in the radio resource allocation process can be defined as:
Figure BDA0002489478450000041
wherein the content of the first and second substances,
Figure BDA0002489478450000042
hirepresenting the number of eligible channels, α, accessible to device i1、α2Respectively representing a weighting factor, the values being set according to the preferences of the service provider.
Further, in the method for adaptive real-time video bitrate for mobile edge computing provided by the present invention, the step S5 includes that the edge server updates the scheduling policy state transition matrix according to the QoE fed back by the mobile client, and the specific operations are as follows:
s51, the mobile client calculates the comprehensive QoE according to the video playing condition and the energy consumption condition, and the calculation mode is as follows:
q＝Qs,a-λEs,a
wherein Q iss,aThe video quality when the strategy a is selected when the state s is expressed is related to the stability of video code rate, average code rate and pause rate; es,aRepresents the energy consumption of the mobile equipment when the strategy is selected, wherein the higher the energy consumption is, the lower the QOE value is, otherwise, the higher the QOE value is;
s52, the mobile client sends the comprehensive QoE obtained in the step S51 to an edge server;
s53, the edge server updates the state transition matrix according to the comprehensive QoE fed back by the user, and the updating method comprises the following steps:
qs,a＝q(s,a)+γmaxq'(s,a)
wherein gamma represents a weighting factor, the value of gamma should meet gamma ∈ (0,1), the value of gamma means that the learning algorithm pays more attention to instant reward or future reward, if gamma approaches 0, it represents that the learning algorithm pays more attention to instant reward, otherwise, it represents that the learning algorithm pays more attention to future reward;
s54, continuously repeating the steps, and continuously updating Q (s, a) in an iterative manner to finally form a relatively convergent state transition matrix Q; the expression for matrix Q is as follows:
Figure BDA0002489478450000043
where s represents various environment states, a represents the edge scheduling policy selected by the device in the state, and q represents the edge scheduling policy selected by the device in the statesaRepresenting a long-term reward when policy a is selected in the s state.
Further, the method for real-time video bitrate adaptation for mobile edge computation provided by the present invention further includes step S6: and repeating the steps S1 to S5, and continuously iterating and updating, wherein the performance of the final edge scheduling strategy tends to be optimal.
The invention can obtain the following advantages by adopting the technical means:
the invention provides a method for calculating real-time video code rate self-adaption of a mobile edge, which is mainly used for uploading data to an edge server platform to be executed under the conditions of high energy consumption, high load and slow fitting when a video code rate self-adaption machine learning system executes a mobile video client, and selecting a reasonable edge scheduling strategy and method for different users by adopting a reinforcement learning method. The privacy and the safety of video users are improved, the expandability of the video platform is fully considered, and cost reduction and efficiency improvement can be realized for platform operators.
Drawings
Fig. 1 is a timing diagram illustrating steps of a video bitrate adaptive method based on moving edge calculation according to an embodiment of the present invention.
Fig. 2 is a diagram of the steps of an edge scheduling algorithm in an embodiment of the present invention.
FIG. 3 is a bottom level logical diagram of a mobile client and an edge server in an embodiment of the present invention.
Detailed Description
The technical scheme of the invention is further explained by combining the accompanying drawings as follows:
the invention provides a real-time video code rate self-adaption method based on mobile edge calculation, which specifically comprises the following steps as shown in figure 1:
A. the mobile video client is responsible for acquiring and transmitting the processor performance, load and multimedia network connection and playing condition of the mobile equipment;
B. the mobile video client uploads the acquired information to an available edge server;
C. the edge server scheduling module judges and selects an implementation mode of the mobile video user code rate self-adaptive method according to parameters such as the environment state of the mobile equipment, the network connection state, the job task delay constraint (the urgency degree in the video code rate self-adaptive process) and the like, and determines the priority of the user. Since wireless resources and edge service resources are limited, priority is determined for each video playing mobile device, and an edge server preferentially provides service for mobile video users with higher priority. The priority is used for radio resource allocation and is determined by radio communication status and task requirements.
D. The video quality evaluation module calculates the user QoE after the user selects the designated scheduling mode, mainly according to the video average code rate, the video code rate switching frequency and amplitude, the video pause time, the mobile equipment energy consumption and the performance load degree of the video in an iteration period, and feeds back the evaluation result to the edge server.
E. The edge scheduling learning system forms a mapping table by the task mode, the environment state and the user QoE after the task mode is selected, and when a user needs to perform edge calculation scheduling task next time, a scheduling mode with the highest long-term QoE can be selected in the mapping table.
F. And repeating the steps, and continuously iterating and updating to finally enable the performance of the edge scheduling strategy to be optimal.
Further, in the step A: mobile device capabilities include the local computing power f of device ii LAnd energy consumption delta of a single cpu cyclei L(ii) a Multimedia network connection case N can be defined as<μk,σk>In which μkIs the mean, σ, of the user-perceived throughputkThe method not only considers the average value of network bandwidth, but also fully considers the detail change of the network throughput, such as the fluctuation degree of the network and the like. The playing condition mainly comprises average video quality, average quality variation amplitude, pause time and the like. The environment state acquisition submodule is responsible for continuously acquiring the information, and the performance of the equipment can not change greatly along with the change of time, so that the information is acquired only once.
Further, in the step C: the edge server selects a suitable scheduling policy for the mobile device, as shown in fig. 2, and specifically includes the following steps:
C1. representing a code rate adaptation task of a mobile device as T according to collected environment informationi。
Figure BDA0002489478450000061
Wherein d isiIs the size of input data for the calculation, including program code, input files, etc. c. CiRepresents the amount of computation required to complete this task, quantified by the number of cpu cycles.
Figure BDA0002489478450000062
Is to calculate the task maximum latency, i.e., the delay constraint duration.
Figure BDA0002489478450000063
And the category of the video to be viewed, the fluctuation degree of the networkAnd (4) correlating.
C2. And judging whether the mobile video user needs to call the edge service according to the environment state and the reinforcement learning state transition matrix, wherein the judgment is based on whether the QoE is higher than that of local calculation after edge calculation is used.
C3. A priority is determined. Due to the fact that wireless resources and MEC service resources are limited, the priority is determined for each video code rate self-adaption task, and the edge server provides service for mobile video users with higher priority preferentially. The priority is used for radio resource allocation and is determined by radio communication status, task delay constraints, task nature, etc.
C4. And (4) allocating channels. And judging whether the energy consumption is lower than that of the original channel or not when the channel is allocated, if so, replacing the channel, and if not, keeping the original state.
Further, the specific operation method of step C2 is as follows:
according to the delay constraint of the code rate self-adaptive task and the comparison of QoE between different task implementation modes, two different strategies can be probably adopted when the mobile equipment watches the video. The first class of devices are devices that should perform their computing tasks on the MEC server. We denote the set of such devices as GR. For devices that have limited computational resources and cannot themselves meet the task delay constraints, the device needs to choose to offload tasks to the MEC server. Therefore, we can get if
Figure BDA0002489478450000064
Figure BDA0002489478450000065
Device i belongs to GRI ∈ N. Wherein the delay is constrained
Figure BDA0002489478450000066
The calculation method is shown in formula 2:
Figure BDA0002489478450000071
Figure BDA0002489478450000072
is the computing power of the MEC server, riIs the uplink rate of data transmitted from device i to the edge server;
equation 2 shows that the delay constraint of each calculation task should be equal to the remaining length of the video buffer minus the total time consumed by the communication and calculation tasks, so as to avoid video jamming and improve the viewing experience of the user.
The second strategy is that the device should compute tasks on its local device, and we denote the set of devices using this strategy as GL. The conditions for determining a device belonging to this type are as follows: if it is not
Figure BDA0002489478450000073
And is
Figure BDA0002489478450000074
This means that tasks are computed more optimally locally when the local computation satisfies the delay constraint and the device power consumption is lower than invoking edge services over the wireless network. Wherein
Figure BDA0002489478450000075
Wherein d isiIs the size of the input data for the calculation,
Figure BDA0002489478450000076
is a coefficient representing a backhaul transmission time delay of unit data, w represents a channel bandwidth, piIs the power at which the mobile i sends data to the edge server in the unit channel, giIs the channel gain, σ, between the mobile user i and the edge server2Is background noise power, wlog2(..) is actually the uplink data transmission rate, r, obtained by the mobile device i when accessing the edge serveri。
In addition, the edge scheduling algorithm maintains a state transition matrix Q on the edge server, which can map the task delay constraint, the computation amount, the computation power, the network state and other parameters of the mobile device with the optimal policy that we should select. And updating the Q matrix according to the QoE value after the selected policy. Therefore, the complex judgment on the strategy of the equipment is not needed every time, and the selection with the optimal historical performance effect is selected from the Q table according to the current environment state.
The expression for matrix Q is as follows:
Figure BDA0002489478450000077
where s corresponds to various environmental conditions. a represents the edge scheduling policy selected by the device in this state. q. q.ssaRepresenting a long-term reward when policy a is selected in the s state. The calculation mode of the long-term report value is different from the calculation mode of other algorithms, the long-term comprehensive quality of the video and the long-term energy consumption level of the mobile equipment are comprehensively considered, the aim is to improve the video quality as much as possible and simultaneously minimize the energy consumption of the mobile equipment, wherein q iss,aIs calculated as shown in equation 5:
qs,a＝q(s,a)+γmaxq'(s,a) (5)
q(s,a)＝Qs,a-λEs,a (6)
in the formula Qs,aThe video quality when the strategy is selected when the state s is expressed is related to the stability of the video code rate, the average code rate and the pause rate. In the formula Es,aRepresenting the mobile device energy consumption when selecting the policy, a higher energy consumption means a lower QOE value and vice versa.
Further, the method of step C3 is specifically defined as follows:
for the compounds belonging to GRDue to their insufficient computing power, MEC servers are required to assist mobile devices in performing computing tasks. GRShould have the highest priority. Therefore, to reduce the capacity of the unloading systemConsumption, more efficient use of radio network resources, GOThe devices in (1) should be assigned different priorities. The priority of device i in the radio resource allocation process may be defined as:
Figure BDA0002489478450000081
wherein the content of the first and second substances,
Figure BDA0002489478450000082
hirepresenting the number of eligible channels that we can access by device i. The priority definition comprehensively considers delay constraints, wireless resources and equipment energy consumption. In equation (7), the first term represents the impact of the delay constraint on the priority. Devices with more critical delay constraints should have higher priority. The second term in equation (7) represents the impact of radio resource availability on priority. Devices with fewer eligible channels should preferentially allocate radio resources. Otherwise, the device may not be able to transmit the computing task to the MEC server within the delay constraints due to insufficient radio resources. Alpha is alpha1And alpha2The value of (b) is set according to the preference of the service provider, and it is more desirable for the provider to satisfy the delay constraint as much as possible or to ensure high communication quality and high communication efficiency.
Further, the method for generating and updating the mapping table Q in step E specifically includes:
E1. the mobile video client calculates the comprehensive QoE according to the video playing condition, the energy consumption and other conditions, and the calculation mode is as follows:
q＝Qs,a-λEs,a
wherein Qs,aThe video quality when the strategy a is selected when the state s is expressed is related to the stability of the video code rate, the average code rate and the pause rate. In the formula Es,aRepresenting the mobile device energy consumption when selecting the policy, a higher energy consumption means a lower QOE value and vice versa.
E2. And the video client sends the comprehensive QoE obtained in the step S51 to the edge server.
E3. The edge server updates the state transition matrix according to the comprehensive QoE fed back by the user, so that the strategy with the highest long-term benefit can be quickly found in the later period, and the updating method comprises the following steps:
qs,a＝q(s,a)+γmaxq'(s,a)
wherein gamma represents a weighting factor, the value of gamma should satisfy gamma epsilon (0,1), and the value of gamma means that the learning algorithm pays more attention to instant reward or future reward. If γ goes to 0, it indicates that more is considered an instant reward, and vice versa, it indicates that the algorithm will focus on future rewards as well.
E4. Repeating the above steps continuously, and then updating Q (s, a) continuously and iteratively to finally form a relatively convergent state transition matrix Q, wherein the expression of Q is shown as follows
Figure BDA0002489478450000091
The system architecture of the present invention is shown in fig. 3, and the detailed functions are as follows:
the terminal equipment in the system architecture comprises three components of a solution, a proxy and a profiler. The socket module is responsible for providing an interface of unloading decision service, and the proxy module is responsible for executing data transmission and control in the unloading process. Profiler is used to detect applications and collect data of applications and the results of their feedback, such as energy, transmission requirements, etc.
The server side comprises four components of a solution, a proxy, a profiler and a system controller. The first two of which are similar to their functions on the terminal device. The proxy component on the server side would be responsible for periodically optimizing the decision engine. Unlike the terminal device, the server side is additionally provided with a system controller, and the system controller is responsible for processing identity authentication, resource allocation and the like of an incoming request.
The embodiments of the present invention have been described in detail with reference to the drawings, but the present invention is not limited to the above embodiments, and various changes can be made within the knowledge of those skilled in the art without departing from the gist of the present invention.

Claims (1)

1. A real-time video code rate self-adaptive method based on mobile edge calculation is characterized by comprising the following steps:
s1, the mobile client acquires performance parameters of the video playing user in real time, wherein the performance parameters comprise video playing conditions, multimedia network connection conditions and mobile equipment performance, and defines a code rate self-adaption task;
s2, the edge server selects a task scheduling strategy with optimal performance for the user according to the parameter data collected by the mobile client;
s3, the mobile client automatically switches the working mode according to the selected task scheduling strategy;
s4, the mobile client calculates QoE of the current scheduling strategy according to the video playing condition in the next iteration period and the equipment energy consumption parameter, and feeds the QoE back to the edge server;
s5, the edge server forms a mapping table according to the task mode and the environment state fed back by the mobile client and the user QoE after the task mode is selected, and updates the state transition matrix of the scheduling strategy; when a user needs to perform an edge calculation scheduling task next time, a scheduling mode with the highest long-term QoE can be selected from the mapping table;
s6, repeating the steps S1 to S5, and continuously iterating and updating to finally enable the performance of the edge scheduling strategy to be optimal;
wherein:
in the step S1, the video playing condition includes average video quality, average quality variation amplitude, and pause duration; the multimedia network connection condition comprises an average value of user perception throughput and a standard deviation of the user perception throughput in the current time period; the mobile device capabilities include local computing power of device i and energy consumption of a single cpu cycle;
in step S2, the edge server selects a task scheduling policy with optimal performance for the user according to the parameter data collected by the client, and the specific operations are as follows:
s21, according to the collected environment information, the code rate of the mobile equipment is determinedThe adaptation task is denoted TiThe expression is as follows:
Figure FDA0003167073440000011
wherein d isiIs the size of input data for calculation, including program code, input files; c. CiRepresents the amount of computation required to complete this task, quantified by the number of cpu cycles; t is ti maxIs to calculate the maximum waiting time of the task, i.e. the delay constraint duration, ti maxThe category of the watching video and the fluctuation degree of the network are related;
s22, judging whether the mobile video user needs to call the edge service according to the environment state and the reinforcement learning state transition matrix, wherein the judgment is based on whether the QoE is higher than that of local calculation after edge calculation is used;
s23, determining the priority: determining the priority for each video code rate self-adaptive task, wherein the edge server provides service for mobile video users with higher priority preferentially; the priority is used for wireless resource allocation and is determined by wireless communication state, task delay constraint and task property factors;
s24, channel allocation: allocating the channel to the equipment according to the priority determined in advance, judging whether the energy consumption is lower than that of the original channel when the channel is allocated, if the energy consumption can be effectively reduced, replacing the channel, and if not, keeping the original state;
in the step S22, it is determined whether the mobile video user needs to invoke the edge service according to the environment state and the reinforcement learning state transition matrix, and the specific operations are as follows:
(1) the set of a class of devices on the MEC server that perform their computational tasks is denoted GRThen delay constraint ti maxThe calculation method is as follows:
Figure FDA0003167073440000021
in the above formula, BkRepresenting the remaining time of the video buffer, subtracting the total consumed time of communication and calculation tasks; diIs the size of the input data for the calculation, ciRepresenting the amount of computation required to accomplish this task, f0 RIs the computing power of the MEC server, riIs the uplink rate of data transmitted from device i to the edge server;
(2) the set of devices on their local devices that perform their computational tasks is denoted GLThe conditions for determining a device belonging to this type are as follows: if t isi L≤ti maxAnd is
Figure FDA0003167073440000022
This means that tasks are more efficient at local computation when the local computation satisfies the delay constraint and the device energy consumption is lower than invoking edge services over the wireless network, where:
Figure FDA0003167073440000023
wherein d isiIs the size of the input data for the calculation,
Figure FDA0003167073440000024
is a coefficient representing a backhaul transmission time delay of unit data, w represents a channel bandwidth, piIs the power at which the mobile i sends data to the edge server in the unit channel, giIs the channel gain, σ, between the mobile user i and the edge server2Is the background noise power;
the step S23 specifically determines the priority as follows:
s231. for the equipment G with insufficient computing capabilityRThe calculation task can be completed only with the assistance of the MEC server, and the wireless resource allocation of the equipment has the highest priority;
s232. for the execution which can be selected to be executed locally or unloaded to the edge serverDevice set GOThe mobile client in (1) should be assigned with different priorities, and the priority of the device i in the radio resource allocation process can be defined as:
Figure FDA0003167073440000025
wherein the content of the first and second substances,
Figure FDA0003167073440000026
hirepresenting the number of eligible channels, α, accessible to device i1、α2Respectively representing weight factors, and setting the value according to the preference of a service provider;
in step S5, the edge server updates the scheduling policy state transition matrix according to the QoE fed back by the mobile client, which specifically operates as follows:
s51, the mobile client calculates the comprehensive QoE according to the video playing condition and the energy consumption condition, and the calculation mode is as follows:
q＝Qs,a-λEs,a
wherein Q iss,aThe video quality when the strategy a is selected when the state s is expressed is related to the stability of video code rate, average code rate and pause rate; es,aRepresents the energy consumption of the mobile equipment when the strategy is selected, wherein the higher the energy consumption is, the lower the QOE value is, otherwise, the higher the QOE value is;
s52, the mobile client sends the comprehensive QoE obtained in the step S51 to an edge server;
s53, the edge server updates the state transition matrix according to the comprehensive QoE fed back by the user, and the updating method comprises the following steps:
qs,a＝q(s,a)+γmax q'(s,a)
wherein gamma represents a weighting factor, the value of gamma should meet gamma ∈ (0,1), the value of gamma means that the learning algorithm pays more attention to instant reward or future reward, if gamma approaches 0, it represents that the learning algorithm pays more attention to instant reward, otherwise, it represents that the learning algorithm pays more attention to future reward;
s54, continuously repeating the steps, and continuously updating Q (s, a) in an iterative manner to finally form a relatively convergent state transition matrix Q; the expression for matrix Q is as follows:
Figure FDA0003167073440000031
where s represents various environment states, a represents the edge scheduling policy selected by the device in the state, and q represents the edge scheduling policy selected by the device in the statesaRepresenting a long-term reward when policy a is selected in the s state.
CN202010401015.4A 2020-05-13 2020-05-13 Real-time video code rate self-adaption method based on mobile edge calculation Active CN111431941B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010401015.4A CN111431941B (en) 2020-05-13 2020-05-13 Real-time video code rate self-adaption method based on mobile edge calculation

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010401015.4A CN111431941B (en) 2020-05-13 2020-05-13 Real-time video code rate self-adaption method based on mobile edge calculation

Publications (2)

Publication Number Publication Date
CN111431941A CN111431941A (en) 2020-07-17
CN111431941B true CN111431941B (en) 2021-08-27

Family

ID=71552860

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010401015.4A Active CN111431941B (en) 2020-05-13 2020-05-13 Real-time video code rate self-adaption method based on mobile edge calculation

Country Status (1)

Country Link
CN (1) CN111431941B (en)

Families Citing this family (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112260961A (en) * 2020-09-23 2021-01-22 北京金山云网络技术有限公司 Network traffic scheduling method and device, electronic equipment and storage medium
CN112291495B (en) * 2020-10-16 2021-08-03 厦门大学 Wireless video low-delay anti-interference transmission method based on reinforcement learning
CN112492265B (en) * 2020-10-30 2023-05-02 广东电网有限责任公司电力调度控制中心 Uplink and downlink resource joint allocation method applied to smart grid
CN112383791B (en) * 2020-11-12 2023-07-21 咪咕视讯科技有限公司 Media data processing method and device, electronic equipment and storage medium
CN112953922B (en) * 2021-02-03 2022-09-16 西安电子科技大学 Self-adaptive streaming media control method, system, computer equipment and application
CN113114733B (en) * 2021-03-24 2022-07-08 重庆邮电大学 Distributed task unloading and computing resource management method based on energy collection
CN113114756B (en) * 2021-04-08 2022-05-03 广西师范大学 Video cache updating method for self-adaptive code rate selection in mobile edge calculation
CN113645471B (en) * 2021-06-22 2022-06-03 北京邮电大学 Multi-cloud video distribution strategy optimization method and system
CN113660508A (en) * 2021-07-16 2021-11-16 国家石油天然气管网集团有限公司西气东输分公司 Multi-edge computing device cooperation task allocation algorithm for intelligent video identification
CN113806073B (en) * 2021-08-11 2022-09-20 中标慧安信息技术股份有限公司 Computing power distribution scheduling method and system for edge computing platform
CN114257880B (en) * 2022-01-10 2023-11-17 百果园技术(新加坡)有限公司 Code rate policy selection method and device, electronic equipment and storage medium
CN115695390B (en) * 2022-09-23 2024-03-05 昆明理工大学 Mine safety monitoring system mass video data self-adaptive streaming method based on mobile edge calculation
CN116016987A (en) * 2022-12-08 2023-04-25 上海大学 Video code rate self-adaption method based on reinforcement learning and oriented to edge cellular network

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105933329A (en) * 2016-06-12 2016-09-07 浙江大学 Video streaming business code rate self-adaption method based on online study
CN109417534A (en) * 2016-05-02 2019-03-01 华为技术有限公司 Communications network service quality capability opening method and device
CN110149299A (en) * 2018-02-13 2019-08-20 中兴通讯股份有限公司 A kind of method for optimizing video, device and system, terminal and the network equipment
US10405274B2 (en) * 2015-02-26 2019-09-03 Nokia Solutions And Networks Oy Coordinated techniques to improve application, network and device resource utilization of a data stream
CN110213627A (en) * 2019-04-23 2019-09-06 武汉理工大学 Flow medium buffer distributor and its working method based on multiple cell user mobility
CN110418418A (en) * 2019-07-08 2019-11-05 广州海格通信集团股份有限公司 Scheduling method for wireless resource and device based on mobile edge calculations
CN110445866A (en) * 2019-08-12 2019-11-12 南京工业大学 Task immigration and collaborative load-balancing method in a kind of mobile edge calculations environment
CN110971706A (en) * 2019-12-17 2020-04-07 大连理工大学 Approximate optimization and reinforcement learning-based task unloading method in MEC

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10004019B2 (en) * 2015-09-08 2018-06-19 Parallel Wireless, Inc. RAN for multimedia delivery
US11153359B2 (en) * 2015-09-29 2021-10-19 Sony Group Corporation User equipment and media streaming network assistance node
CN108769760B (en) * 2018-05-23 2021-03-12 中国联合网络通信集团有限公司 Code rate adjustment method, UE, MEC and network system
CN108900628A (en) * 2018-07-20 2018-11-27 南京工业大学 Thin cloud computational resource allocation method in edge calculations environment based on pricing mechanism
CN110418416B (en) * 2019-07-26 2023-04-18 东南大学 Resource allocation method based on multi-agent reinforcement learning in mobile edge computing system
CN110913239B (en) * 2019-11-12 2021-03-02 西安交通大学 Video cache updating method for refined mobile edge calculation

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10405274B2 (en) * 2015-02-26 2019-09-03 Nokia Solutions And Networks Oy Coordinated techniques to improve application, network and device resource utilization of a data stream
CN109417534A (en) * 2016-05-02 2019-03-01 华为技术有限公司 Communications network service quality capability opening method and device
CN105933329A (en) * 2016-06-12 2016-09-07 浙江大学 Video streaming business code rate self-adaption method based on online study
CN110149299A (en) * 2018-02-13 2019-08-20 中兴通讯股份有限公司 A kind of method for optimizing video, device and system, terminal and the network equipment
CN110213627A (en) * 2019-04-23 2019-09-06 武汉理工大学 Flow medium buffer distributor and its working method based on multiple cell user mobility
CN110418418A (en) * 2019-07-08 2019-11-05 广州海格通信集团股份有限公司 Scheduling method for wireless resource and device based on mobile edge calculations
CN110445866A (en) * 2019-08-12 2019-11-12 南京工业大学 Task immigration and collaborative load-balancing method in a kind of mobile edge calculations environment
CN110971706A (en) * 2019-12-17 2020-04-07 大连理工大学 Approximate optimization and reinforcement learning-based task unloading method in MEC

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
Mobile Edge Computing Enhanced Adaptive Bitrate Video Delivery With Joint Cache and Radio Resource Allocation;XIAODONG XU et al;《SPECIAL SECTION ON MOBILE EDGE COMPUTING》;20170722;全文 *
Routing in wireless multimedia sensor network s: A survey and challenges ahead;Hang Shen et al;《Journal of Network and Computer Applications》;20161231;全文 *
强化学习在自适应视频码率控制算法中的应用;肖 强等;《小 型 微 型 计 算 机 系 统》;20200228;全文 *
结合计算和缓存的移动通信组网策略优化研究;刘家祥;《中国优秀硕士论文全文库信息科技辑》;20181115;全文 *

Also Published As

Publication number Publication date
CN111431941A (en) 2020-07-17

Similar Documents

Publication Publication Date Title
CN111431941B (en) Real-time video code rate self-adaption method based on mobile edge calculation
CN110493360B (en) Mobile edge computing unloading method for reducing system energy consumption under multiple servers
EP2764750B1 (en) Improving adaptive streaming video quality by optimizing resource allocation
EP3058772B1 (en) Dynamic frequency and power resource allocation with granular policy management
Sun et al. Autonomous resource slicing for virtualized vehicular networks with D2D communications based on deep reinforcement learning
US8737255B2 (en) Methods and arrangements for redistributing resources for use in a radio communication system
Sousa et al. A survey on QoE-oriented wireless resources scheduling
CN107708152B (en) Task unloading method of heterogeneous cellular network
CN107637046B (en) Method and apparatus for controlling multiple connections to increase data transfer rate
CN111885147A (en) Dynamic resource pricing method in edge calculation
CN114007225A (en) BWP allocation method, apparatus, electronic device and computer readable storage medium
CN107820278B (en) Task unloading method for cellular network delay and cost balance
El Haber et al. Computational cost and energy efficient task offloading in hierarchical edge-clouds
Wang et al. Adaptive wireless video streaming: Joint transcoding and transmission resource allocation
US9258557B2 (en) Rate optimization for scalable video transmission
CN111049829B (en) Video streaming transmission method and device, computer equipment and storage medium
CN109600432B (en) Dynamic access method for content center mobile edge network user
Ahmad et al. Battery-aware rate adaptation for extending video streaming playback time
Mahmoodi et al. Spectrum-Aware Mobile Computing
CN113271221B (en) Network capacity opening method and system and electronic equipment
CN113597013A (en) Cooperative task scheduling method in mobile edge computing under user mobile scene
Wu et al. A learning-based expected best offloading strategy in wireless edge networks
CN111107639A (en) Resource allocation method for video data processing and electronic equipment
Guo et al. Joint wireless resource allocation and service function chaining scheduling for Tactile Internet
Luo et al. Adaptive video streaming in software-defined mobile networks: A deep reinforcement learning approach

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
GR01 Patent grant
GR01 Patent grant