CN110460880A - Wireless industrial streaming media self-adapting transmission method based on population and neural network - Google Patents
Wireless industrial streaming media self-adapting transmission method based on population and neural network Download PDFInfo
- Publication number
- CN110460880A CN110460880A CN201910733205.3A CN201910733205A CN110460880A CN 110460880 A CN110460880 A CN 110460880A CN 201910733205 A CN201910733205 A CN 201910733205A CN 110460880 A CN110460880 A CN 110460880A
- Authority
- CN
- China
- Prior art keywords
- neural network
- network
- video
- parameter
- input
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- 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
- H04N21/24—Monitoring of processes or resources, e.g. monitoring of server load, available bandwidth, upstream requests
- H04N21/2402—Monitoring of the downstream path of the transmission network, e.g. bandwidth available
-
- 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
- H04N21/26216—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 involving the channel capacity, e.g. network bandwidth
-
- 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
- H04N21/647—Control signaling between network components and server or clients; Network processes for video distribution between server and clients, e.g. controlling the quality of the video stream, by dropping packets, protecting content from unauthorised alteration within the network, monitoring of network load, bridging between two different networks, e.g. between IP and wireless
- H04N21/64723—Monitoring of network processes or resources, e.g. monitoring of network load
- H04N21/64738—Monitoring network characteristics, e.g. bandwidth, congestion level
Abstract
The invention discloses a kind of wireless industrial streaming media self-adapting transmission method based on population and neural network.The historical data for obtaining specified range from Cloud Server database first completes the training of neural network model, and the various state parameters of real time monitoring wireless channel；Then the wireless network transmissions parameter for keeping quality of experience of video optimal is obtained by particle swarm algorithm；Secondly optimal system set-up parameters are predicted using the mapping function of the neural network model of trained completion, and completes the setting of system；It finally obtains real data and stores, and apply to the training and correction of neural network again.The present invention, which fully considers, to be proposed on the basis of traditional dynamic self-adapting video DASH transport protocol based on HTTP, parameter optimization can be more quickly completed by particle swarm algorithm, and the mode for replacing traditional trial to explore in such a way that neural network directly maps directly obtains system set-up parameters, facilitates system parameter and is more accurately arranged and the more smooth transmission of video.
Description
Technical field
The present invention relates to video technique, specifically a kind of wireless industrial Streaming Media based on population and neural network
Adaptive transmission method.
Background technique
Video technique significantly improves the problem of video data load duration, can be segmented by video technique user
Biggish video data is obtained, while being not necessarily to excessive load duration.Traditional Video service is divided into two major classes: one kind is to use
Real time streaming transport protocol/real-time transport protocol (Real Time Streaming Protocol Real Time Transfer
Protocol, RTSP/RTP) connection-oriented real-time video technology；It is another kind of, it is using hypertext transfer protocol (Hyper
Text Transfer Protocol HTTP) connectionless order video technology.For wireless network and video transmitting procedure
The problem of, scholars propose the dynamic self-adapting transmission of video agreement DASH (Dynamic based on HTTP
Adaptive Streaming over HTTP), to adapt to unstable network environment, improve the user's body of video data
It tests, video data is made to realize high code rate, low fluctuation, the broadcast request without interruption as far as possible.
Typically the adaptive video technology based on HTTP specifically includes that 1) the adaptive biography based on network throughput at present
Transmission method.Adaptive algorithm based on network throughput is mainly to pass through the network bandwidth decision client of estimation subsequent time
Apply for code rate, prevents the interruption in video display process as far as possible.2) based on the adaptive transmission method of buffer control.Based on slow
The Adaptive Transmission algorithm for depositing control is mainly to pass through the media switching rate for changing server transmission rate and client to protect
The video data volume for demonstrate,proving client buffer area is stable as far as possible, and then guarantees user experience quality.
However, the above method is mostly the system setting method that traditional exploration is attempted, although can reach under certain condition
Required effect, but still remain that parameter selection inaccuracy, evaluated error be big, computing resource consumes larger, video tastes
The problems such as unstable quality, and traditional adaptive video transmission method is generally difficult to directly obtain optimal parameter, needs more
It can be only achieved required effect after secondary trial, regulating cycle is longer, while again since the method generally involves the cross-layer of network
Design, and adapt to the dynamic countermeasure of various special circumstances, thus in order to guarantee transmission quality and to system stability and
Reliability needs easier and comprehensive adaptive video method.
Summary of the invention
Lack for intelligent adaptive video transmission technologies and traditional adaptive video technology that there are parameter selections is inaccurate
Really, the status that evaluated error is big, computing resource consumes the problems such as larger, quality of experience of video is unstable, the invention proposes one
Wireless industrial streaming media self-adapting transmission method of the kind based on population and neural network, using particle swarm algorithm and neural network
In conjunction with method, realize the optimal Adaptive Transmission effect of quality of experience of video, technical solution are as follows:
A kind of wireless industrial streaming media self-adapting transmission method based on population and neural network, comprising the following steps:
Step 1: obtaining the historical data of specified range from Cloud Server database, completes the training of neural network model,
And the various state parameters of real time monitoring wireless channel, the specific steps are as follows:
1) connection Cloud Server and application to access the database；
2) it is required to obtain m historical data according to computational accuracy (m is natural number)；
3) the m historical data is carried out deleting processing, so that the variable number in the historical data after deleting meets
The input variable dimension and output variable dimension of neural network；
4) historical data after deleting is smoothed, formula is as follows:
X=a*mean_x+b* (x-mean_x)
Wherein, x indicates that each variable as neural network input, mean_x indicate each variable fetched data
Average value, a, b are coefficient, and a, b value of homologous ray is not different, but total satisfaction a+b=1 | a > b, a > 0, b > 0 } and a > b,
Suitable a, b are selected according to system situation difference；
5) historical data after smoothing processing is combined into according to sequence needed for neural network input matrix and dimension defeated
Incoming vector, and corresponding output vector is obtained, input vector and corresponding output vector input neural network are obtained into nerve net
Network matrix completes the training of neural network model, and reads cloud service by the library function of general network communication library twisted
The various network state parameters of device data receiver port and appointing system file acquisition wireless channel；
Step 2: keep quality of experience of video (Quality of Experience, QoE) optimal by particle swarm algorithm acquisition
Wireless network transmissions parameter so that the video frame rate of subsequent time is maximum, fluctuation is minimum, video is most smooth, specific steps are such as
Under:
1) population initializes, to particle inertia weight w, Studying factors c1、c2, population quantity population_
Size, dimension dim, the number of iterations max_steps, solution space range x_bound, primary group position x, primary speed
V is initialized；
2) fitness function, the calculation formula of the fitness function are designed are as follows:
Fitness=a*Q_m-b*Q_s
Wherein, a, b indicate influence factor coefficient, Q_m indicate directly with average frame per second mkIn logarithm correlativity, Q_s table
Show that specific formula for calculation is as follows directly with the variance s_k of real-time network handling capacity p_i in logarithm correlativity:
Q_m=ln (mk+ε)
Wherein, Q_m and Q_s negative correlation, Q_s is smaller when Q_m gets over Datong District, then illustrates that the video frame per second that is averaged is bigger,
Video is more smooth, and fluctuation is smaller, and video user Quality of experience is better, mkVideo per second in the N/2 second before expression current time
Frame per second fiThe sum of average；σkThe mean square deviation of the network throughput in the N/2 second before expression current time；ε andIndicate with
The relevant constant parameter of system, different system parameters is different, guarantees that Q_m and Q_s, can in the same order of magnitude when calculating
Using ε andIt is adjusted；N indicates the number of data used, fiIndicate video frame rate per second, sk+1Indicate+1 moment of kth net
The variance of network handling capacity, piIndicate the network throughput at the i-th moment,The average value of handling capacity before indicating in N+1 data；
3) renewal speed and position, until iteration terminates or meet the minimum threshold of optimal location, each fitness function
After the completion of calculating, it will be compared to obtain optimal particle position and corresponding fitness value, and more new individual is optimal suitable
Angle value individual_best_fitness and global optimal adaptation angle value global_best_fitness are answered, obtains the overall situation most
X corresponding to good fitness value is that can make handling capacity maximum, the smallest network state parameters of fluctuation；
Step 3: using in the step 1 the mapping function of the neural network model of trained completion predict it is optimal
System set-up parameters, and complete the setting of system, the specific steps are as follows:
1) by reading the Cloud Server database of stored network state parameters, particle swarm algorithm input terminal institute is obtained
The network state parameters needed, and the network state parameters are passed through into the change of the frame per second and picture quality that are introduced into system parameter
Magnitude is extended for the input matrix of required neural network matrix, the variable of frame per second and picture quality in the system parameter
Value can be supplemented with the historical data of previous moment；
2) by the input neural network of the matrix neural network matrix in the step 1), pass through the mapping of neural network
Relationship obtains the predicted value that can make the highest system set-up parameters of quality of experience of video, and is according to the predicted value
System parameter setting；
Step 4: it obtains real data and stores, and apply to the training and correction of neural network again, specific steps are such as
Under:
1) Cloud Server data receiver port and specified system are read by the library function of general network communication library twisted
System file completion continues to monitor grid state, and obtains newest network state parameters and relevant system parameters, and
It is stored；
2) by the video frame rate and picture quality of newest network state parameters and last moment in the step 1)
Parameter is handled by specific steps described in step 1, and according to the input obtained every time with a newest real data to
The rule input neural network for measuring that earliest input vector of replacement time, is periodically corrected neural network.
The beneficial effects of the present invention are:
A kind of wireless industrial streaming media self-adapting transmission method based on population and neural network proposed by the present invention, is adopted
With the problem of parameter selection for the method processing system that particle swarm algorithm and neural network combine, keep system selection parameter more quasi-
Really, react rapider, the wireless industrial streaming media self-adapting transmission method based on population and neural network can cope with wirelessly
The complex situations of the network fluctuation of a variety of bursts under network environment make system with more intelligence, while increasing video playing
Fluency, realize the optimal Adaptive Transmission effect of quality of experience of video.
Detailed description of the invention
Fig. 1 is the wireless industrial streaming media self-adapting transmission side based on population and neural network in this patent embodiment
The flow chart of method.
Fig. 2 is the wireless industrial streaming media self-adapting transmission side based on population and neural network in this patent embodiment
The schematic diagram of method.
Specific embodiment
Here is that technical solution of the present invention is described in detail in conjunction with attached drawing.
Such as the stream of the video frame rate adaptive transmission method based on population and neural network in Fig. 1 this patent embodiment
Shown in journey figure, a kind of video frame rate adaptive transmission method based on population and neural network the following steps are included:
Step 1: obtaining the historical data of specified range from Cloud Server database, completes the training of neural network model,
And the various state parameters of real time monitoring wireless channel, the specific steps are as follows:
1) connection Cloud Server and application to access the database；
2) it is required to obtain m historical data (m is natural number) according to computational accuracy, the value range of m is in the present embodiment
500<m<1000；
3) the m historical data is carried out deleting processing, so that the variable number in the historical data after deleting meets
The input variable dimension and output variable dimension of neural network；
4) historical data after deleting is smoothed, formula is as follows:
X=a*mean_x+b* (x-mean_x)
Wherein, x indicates that each variable as neural network input, mean_x indicate each variable fetched data
Average value, a, b are coefficient, and a, b value of homologous ray is not different, but total satisfaction a+b=1 | a > b, a > 0, b > 0 } and a > b,
Suitable a, b are selected according to system situation difference；
5) historical data after smoothing processing is combined into according to sequence needed for neural network input matrix and dimension defeated
Incoming vector, and corresponding output vector is obtained, input vector and corresponding output vector input neural network are obtained into nerve net
Network matrix completes the training of neural network model, and reads cloud service by the library function of general network communication library twisted
The various network state parameters of device data receiver port and appointing system file acquisition wireless channel, the wherein process of network training
Complex and be difficult to observe, specific process is as shown in Fig. 1, is entering neural network input by pretreated data
After layer, data carry out weight calculation in the hidden layer of neural network, after calculating by specified all hidden layers, obtain defeated
Output data and expected data are carried out error calculation, subtract error to gradient by the method for back transfer later by data out
Small direction carries out, and in this, as the foundation of adjustment neural network weight, and neural network power is updated while reducing error
Value matrix, in calculating process, when reaching trained termination condition, training terminates, such as: error is less than setting error, reaches most
When big study number, when not up to terminating to require, returned data preprocessing part continues to calculating.
Step 2: keep quality of experience of video (Quality of Experience, QoE) optimal by particle swarm algorithm acquisition
Wireless network transmissions parameter so that the video frame rate of subsequent time is maximum, fluctuation is minimum, video is most smooth, specific steps are such as
Under:
1) population initializes, to particle inertia weight w, Studying factors c1、c2, population quantity population_
Size, dimension dim, the number of iterations max_steps, solution space range x_bound, primary group position x, primary speed
V is initialized, in the present embodiment each parameter initialization value are as follows: particle inertia weight w=0.6, Studying factors c1=2, c2=
2, population quantity population_size=100, dimension dim=2, the number of iterations max_steps=10000, solution space model
Enclose 100 groups of random coordinates of x_bound ∈ [- 10,10000], primary group position x, primary speed v random 100
Class value carries out the operation of " initialization particle rapidity, position etc. " first, needs before carrying out population optimizing as shown in Fig. 1
Random particles and its corresponding position and the above relevant parameter information is obtained ahead of time, at the same need to the correlated variables of system into
Row initialization；
2) fitness function is designed, it is the important indicator for influencing algorithm superiority and inferiority that fitness function, which is the pith of algorithm,
The design of fitness function should realize the evaluation to video quality, while consider the occupancy of computing resource again, therefore need
Simply but also video quality can be evaluated not only.The present invention when designing fitness function directly from quality of experience of video QoE, if
Count following fitness function:
Fitness=a*Q_m-b*Q_s
Wherein, a, b indicate influence factor coefficient, Q_m indicate directly with average frame per second mkIn logarithm correlativity, Q_s table
Show that specific formula for calculation is as follows directly with the variance s_k of real-time network handling capacity p_i in logarithm correlativity:
Q_m=ln (mk+ε)
Wherein, Q_m and Q_s negative correlation, Q_s is smaller when Q_m gets over Datong District, then illustrates that the video frame per second that is averaged is bigger,
Video is more smooth, and fluctuation is smaller, and video user Quality of experience is better, mkVideo per second in the N/2 second before expression current time
Frame per second fiThe sum of average；σkThe mean square deviation of the network throughput in the N/2 second before expression current time；ε andIndicate with
The relevant constant parameter of system, different system parameters is different, guarantees that Q_m and Q_s, can in the same order of magnitude when calculating
Using ε andIt is adjusted；N indicates the number of data used, fiIndicate video frame rate per second, sk+1Indicate+1 moment of kth net
The variance of network handling capacity, piIndicate the network throughput at the i-th moment,The average value of handling capacity before indicating in N+1 data,
As shown in Fig. 1, after carrying out initialization and fitness function design, pass through data and adaptation in " calculating fitness " link
It spends function and quantitatively evaluating is carried out to system；
3) renewal speed and position, until iteration terminates or meet the minimum threshold of optimal location, each fitness function
After the completion of calculating, it will be compared to obtain optimal particle position and corresponding fitness value, and more new individual is optimal suitable
Angle value individual_best_fitness and global optimal adaptation angle value global_best_fitness are answered, obtains the overall situation most
X corresponding to good fitness value is that can make handling capacity maximum, the smallest network state parameters of fluctuation, as shown in Fig. 1, In
" it is optimal the to obtain individual " stage is obtained by the calculating of the fitness to each particle when fitness is optimal in previous group particle
Particle individual, and be compared current optimum individual with history optimum individual in " obtain global optimum " stage, with this
Obtain global optimum's individual；It carries out judging whether to continue optimizing operation by termination condition later, if optimizing is not finished,
If " the speed adjustment, position adjustment " stage in attached drawing 1 carries out the speed and position adjustment of each particle, to obtain more preferably
Fitness.
In above-mentioned steps two, particle swarm algorithm meets the fitness function of QoE index request by design, to video
Transmission quality carries out quantification treatment, in particle swarm algorithm part by joining to network states such as frame per second, network throughput, packet loss
The processing modes such as several weightings obtain optimal network state ginseng in the case where meeting play frame rate maximum and fluctuating the smallest requirement
Number, such as the wireless industrial streaming media self-adapting transmission method based on population and neural network in 2 this patent embodiment of attached drawing
Schematic diagram shown in, it is suitable that the network state parameters such as average frame per second, network throughput variance are weighted design by particle swarm algorithm
Response function obtains optimal target network state parameter by particle swarm algorithm, and defeated in this, as next stage neural network
The part initial data entered.
Step 3: using in the step 1 the mapping function of the neural network model of trained completion predict it is optimal
System set-up parameters, and complete the setting of system, the specific steps are as follows:
1) by reading the Cloud Server database of stored network state parameters, particle swarm algorithm input terminal institute is obtained
The network state parameters needed, as shown in Fig. 1, after " output global optimum's particle and fitness ", the particle of acquisition
Relevant information is required target network status information, then by real-time Network Expert Systems, obtains data to population
The operation of " adjustment matrix dimensionality, obtain input vector " is carried out, obtaining with this can be for input square that neural network directly uses
Battle array；
2) by the input neural network of the matrix neural network matrix in the step 1), pass through the mapping of neural network
Relationship obtains the predicted value that can make the highest system set-up parameters of quality of experience of video, and is according to the predicted value
It unites parameter setting, as shown in Fig. 1, target network state parameter is mapped as that " system is set in incoming " neural network " after the stage
Set parameter ", obtained parameter can exactly make that subsequent time network state parameters are best, user experience quality optimal system ginseng
Number, with this " completing system setting "；
Step 4: it obtains real data and stores, and apply to the training and correction of neural network again, specific steps are such as
Under:
1) Cloud Server data receiver port and specified system are read by the library function of general network communication library twisted
System file completion continues to monitor grid state, and obtains newest network state parameters and relevant system parameters, and
It is stored, as shown in Fig. 1, after " completing system setting ", new network state parameters can be generated after system operation, to net
Network is monitored and obtains " real network state parameter " in real time；
2) by the video frame rate and picture quality of newest network state parameters and last moment in the step 1)
Parameter is handled by specific steps described in step 1, and according to the input obtained every time with a newest real data to
The rule input neural network for measuring that earliest input vector of replacement time, is periodically corrected neural network, such as
Shown in attached drawing 1, obtain " real network state parameter " and then it is secondary by " store, delete, smoothly " operate as new data
It re-enters neural network, carries out network calibration to reduce error, this part is as shown in Fig. 2, by optimal network state
After parameter is mapped as optimizer system parameter by neural network, after being configured by obtained system parameter to system
Obtain new network state, training and correction data of the network state parameters as neural network carry out holding for storage and database
Longization operation.
Claims (1)
1. a kind of wireless industrial streaming media self-adapting transmission method based on population and neural network, which is characterized in that including
Following steps:
Step 1: obtaining the historical data of specified range from Cloud Server database, completes the training of neural network model, and real
When monitor wireless channel various state parameters, the specific steps are as follows:
1) connection Cloud Server and application to access the database；
2) it is required to obtain m historical data according to computational accuracy；
3) the m historical data is carried out deleting processing, so that the variable number in the historical data after deleting meets nerve
The input variable dimension and output variable dimension of network；
4) historical data after deleting is smoothed, formula is as follows:
X=a*mean_x+b* (x-mean_x)
Wherein, x indicates that each variable as neural network input, mean_x indicate being averaged for each variable fetched data
Value, a, b are coefficient, and a, b value of homologous ray is not different, but total satisfaction a+b=1 | a > b, a > 0, b > 0 } and a > b, according to
System situation difference selects suitable a, b；
5) by the historical data after smoothing processing according to sequence needed for neural network input matrix and dimension be combined into input to
Amount, and corresponding output vector is obtained, input vector and corresponding output vector input neural network are obtained into neural network square
Battle array completes the training of neural network model, and reads Cloud Server number by the library function of general network communication library twisted
According to the various network state parameters of receiving port and appointing system file acquisition wireless channel；
Step 2: the wireless network transmissions parameter for keeping quality of experience of video optimal is obtained by particle swarm algorithm, so that subsequent time
Video frame rate it is maximum, fluctuation is minimum, video is most smooth, the specific steps are as follows:
1) population initializes, to particle inertia weight w, Studying factors c1、c2, population quantity population_size, dimension
Dim, the number of iterations max_steps, solution space range x_bound, primary group position x, primary speed v is spent to carry out just
Beginningization；
2) fitness function, the calculation formula of the fitness function are designed are as follows:
Fitness=a*Q_m-b*Q_s
Wherein, a, b indicate influence factor coefficient, Q_m indicate directly with average frame per second mkIn logarithm correlativity, Q_s indicates direct
Variance s_k with real-time network handling capacity p_i is in logarithm correlativity, and specific formula for calculation is as follows:
Q_m=ln (mk+ε)
Wherein, Q_m and Q_s negative correlation, Q_s is smaller when Q_m gets over Datong District, then illustrates that the video frame per second that is averaged is bigger, video
More smooth, fluctuation is smaller, and video user Quality of experience is better, mkVideo frame rate f per second in the N/2 second before expression current timei
The sum of average；σkThe mean square deviation of the network throughput in the N/2 second before expression current time；ε andIt indicates and system phase
The constant parameter of pass, different system parameters is different, guarantees that in the same order of magnitude, ε is can be used in Q_m and Q_s when calculating
WithIt is adjusted；N indicates the number of data used, fiIndicate video frame rate per second, sk+1Indicate+1 moment of kth network throughput
The variance of amount, piIndicate the network throughput at the i-th moment,The average value of handling capacity before indicating in N+1 data；
3) renewal speed and position, until iteration terminates or meet the minimum threshold of optimal location, each fitness function is calculated
After the completion, it will be compared to obtain optimal particle position and corresponding fitness value, and more new individual adaptive optimal control degree
Value individual_best_fitness and global optimal adaptation angle value global_best_fitness is obtained global best suitable
Answering x corresponding to angle value is that can make handling capacity maximum, the smallest network state parameters of fluctuation；
Step 3: using the mapping function of the neural network model of trained completion predicts and optimal is in the step 1
System setting parameter, and complete the setting of system, the specific steps are as follows:
1) it by reading the Cloud Server database of stored network state parameters, obtains required for particle swarm algorithm input terminal
Network state parameters, and the network state parameters are passed through to the variate-value of frame per second and picture quality being introduced into system parameter
It is extended for the input matrix of required neural network matrix, the variate-value of frame per second and picture quality in the system parameter can
It is supplemented with the historical data of previous moment；
2) it by the input neural network of the matrix neural network matrix in the step 1), is closed by the mapping of neural network
System obtains the predicted value that can make the highest system set-up parameters of quality of experience of video, and carries out system according to the predicted value
Parameter setting；
Step 4: it obtains real data and stores, and apply to the training and correction of neural network again, the specific steps are as follows:
1) Cloud Server data receiver port and appointing system text are read by the library function of general network communication library twisted
Part completion continues to monitor grid state, and obtains newest network state parameters and other system parameters, and carry out
Storage；
2) by the video frame rate and image quality parameter of newest network state parameters and last moment in the step 1)
It is handled by specific steps described in step 1, and according to being replaced every time with the input vector that a newest real data obtains
Neural network is inputted for time earliest the regular of that input vector, periodically neural network is corrected.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910733205.3A CN110460880B (en) | 2019-08-09 | 2019-08-09 | Industrial wireless streaming media self-adaptive transmission method based on particle swarm and neural network |
| PCT/CN2019/101360 WO2021026944A1 (en) | 2019-08-09 | 2019-08-19 | Adaptive transmission method for industrial wireless streaming media employing particle swarm and neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910733205.3A CN110460880B (en) | 2019-08-09 | 2019-08-09 | Industrial wireless streaming media self-adaptive transmission method based on particle swarm and neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110460880A true CN110460880A (en) | 2019-11-15 |
| CN110460880B CN110460880B (en) | 2021-08-31 |
Family
ID=68485725
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910733205.3A Active CN110460880B (en) | 2019-08-09 | 2019-08-09 | Industrial wireless streaming media self-adaptive transmission method based on particle swarm and neural network |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN110460880B (en) |
| WO (1) | WO2021026944A1 (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112710401A (en) * | 2021-03-29 | 2021-04-27 | 国网江西省电力有限公司供电服务管理中心 | Method for detecting temperature of electric energy meter terminal |
| CN113422751A (en) * | 2020-08-27 | 2021-09-21 | 阿里巴巴集团控股有限公司 | Streaming media processing method and device based on online reinforcement learning and electronic equipment |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113553270A (en) * | 2021-07-28 | 2021-10-26 | 浙江网商银行股份有限公司 | Target object determination method and device |
| CN113515048B (en) * | 2021-08-13 | 2023-04-07 | 华中科技大学 | Method for establishing fuzzy self-adaptive PSO-ELM sound quality prediction model |
| CN113642220B (en) * | 2021-08-26 | 2023-09-22 | 江苏科技大学 | Ship welding process optimization method based on RBF and MOPSO |
| CN115884195A (en) * | 2021-09-28 | 2023-03-31 | 中兴通讯股份有限公司 | Model training method, wireless resource scheduling method and device and electronic equipment |
| CN114553681B (en) * | 2022-03-08 | 2024-02-02 | 中国人民解放军国防科技大学 | Device state abnormality detection method and device and computer device |
| CN114866481A (en) * | 2022-04-29 | 2022-08-05 | 西安交通大学 | Time-varying rerouting method and system based on chaotic dynamic congestion prediction for air-space-ground integrated network |
| CN115473595B (en) * | 2022-08-19 | 2024-04-02 | 三峡大学 | Prediction method for packet loss rate of wireless sensing equipment in transformer substation |
| CN117198418B (en) * | 2023-11-07 | 2024-02-13 | 威海百合生物技术股份有限公司 | Polysaccharide extraction process parameter optimization method and system |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103164742A (en) * | 2013-04-02 | 2013-06-19 | 南京邮电大学 | Server performance prediction method based on particle swarm optimization nerve network |
| CN104361393A (en) * | 2014-09-06 | 2015-02-18 | 华北电力大学 | Method for using improved neural network model based on particle swarm optimization for data prediction |
| CN107547457A (en) * | 2017-09-15 | 2018-01-05 | 重庆大学 | A kind of approach for blind channel equalization based on Modified particle swarm optimization BP neural network |
| CN107909008A (en) * | 2017-10-29 | 2018-04-13 | 北京工业大学 | Video target tracking method based on multichannel convolutive neutral net and particle filter |
| CN108182447A (en) * | 2017-12-14 | 2018-06-19 | 南京航空航天大学 | A kind of adaptive particle filter method for tracking target based on deep learning |
| CN108901058A (en) * | 2018-07-06 | 2018-11-27 | 北方工业大学 | Internet of things node access channel optimization selection method |
| CN109120630A (en) * | 2018-09-03 | 2019-01-01 | 上海海事大学 | A kind of SDN network ddos attack detection method based on Optimized BP Neural Network |
| CN109118013A (en) * | 2018-08-29 | 2019-01-01 | 黑龙江工业学院 | A kind of management data prediction technique, readable storage medium storing program for executing and forecasting system neural network based |
| US20190130188A1 (en) * | 2017-10-26 | 2019-05-02 | Qualcomm Incorporated | Object classification in a video analytics system |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102118803A (en) * | 2011-04-14 | 2011-07-06 | 北京邮电大学 | Video cross-layer scheduling method of mobile communication system on basis of QoE prediction |
| CN102946613B (en) * | 2012-10-10 | 2015-01-21 | 北京邮电大学 | Method for measuring QoE |
| CN109120961B (en) * | 2018-07-20 | 2020-11-03 | 南京邮电大学 | QoE prediction method of IPTV unbalanced data set based on PNN-PSO algorithm |
-
2019
- 2019-08-09 CN CN201910733205.3A patent/CN110460880B/en active Active
- 2019-08-19 WO PCT/CN2019/101360 patent/WO2021026944A1/en active Application Filing
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103164742A (en) * | 2013-04-02 | 2013-06-19 | 南京邮电大学 | Server performance prediction method based on particle swarm optimization nerve network |
| CN104361393A (en) * | 2014-09-06 | 2015-02-18 | 华北电力大学 | Method for using improved neural network model based on particle swarm optimization for data prediction |
| CN107547457A (en) * | 2017-09-15 | 2018-01-05 | 重庆大学 | A kind of approach for blind channel equalization based on Modified particle swarm optimization BP neural network |
| US20190130188A1 (en) * | 2017-10-26 | 2019-05-02 | Qualcomm Incorporated | Object classification in a video analytics system |
| CN107909008A (en) * | 2017-10-29 | 2018-04-13 | 北京工业大学 | Video target tracking method based on multichannel convolutive neutral net and particle filter |
| CN108182447A (en) * | 2017-12-14 | 2018-06-19 | 南京航空航天大学 | A kind of adaptive particle filter method for tracking target based on deep learning |
| CN108901058A (en) * | 2018-07-06 | 2018-11-27 | 北方工业大学 | Internet of things node access channel optimization selection method |
| CN109118013A (en) * | 2018-08-29 | 2019-01-01 | 黑龙江工业学院 | A kind of management data prediction technique, readable storage medium storing program for executing and forecasting system neural network based |
| CN109120630A (en) * | 2018-09-03 | 2019-01-01 | 上海海事大学 | A kind of SDN network ddos attack detection method based on Optimized BP Neural Network |
Non-Patent Citations (1)
| Title |
|---|
| 向涛，涂风华，廖晓峰: "基于神经网络和粒子群算法的MPEG视频传输控制", 《计算机科学》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113422751A (en) * | 2020-08-27 | 2021-09-21 | 阿里巴巴集团控股有限公司 | Streaming media processing method and device based on online reinforcement learning and electronic equipment |
| CN113422751B (en) * | 2020-08-27 | 2023-12-05 | 阿里巴巴集团控股有限公司 | Streaming media processing method and device based on online reinforcement learning and electronic equipment |
| CN112710401A (en) * | 2021-03-29 | 2021-04-27 | 国网江西省电力有限公司供电服务管理中心 | Method for detecting temperature of electric energy meter terminal |
| CN112710401B (en) * | 2021-03-29 | 2021-08-24 | 国网江西省电力有限公司供电服务管理中心 | Method for detecting temperature of electric energy meter terminal |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2021026944A1 (en) | 2021-02-18 |
| CN110460880B (en) | 2021-08-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110460880A (en) | Wireless industrial streaming media self-adapting transmission method based on population and neural network | |
| Lai et al. | Oort: Efficient federated learning via guided participant selection | |
| WO2021012946A1 (en) | Video bit rate determining method and apparatus, electronic device, and storage medium | |
| CN113434212B (en) | Cache auxiliary task cooperative unloading and resource allocation method based on meta reinforcement learning | |
| CN112422644B (en) | Method and system for unloading computing tasks, electronic device and storage medium | |
| CN113052334A (en) | Method and system for realizing federated learning, terminal equipment and readable storage medium | |
| CN113938488A (en) | Load balancing method based on dynamic and static weighted polling | |
| CN114039870B (en) | Deep learning-based real-time bandwidth prediction method for video stream application in cellular network | |
| CN110033081A (en) | A kind of method and apparatus of determining learning rate | |
| CN109688065B (en) | Parameter processing method and device and storage medium | |
| CN116233026A (en) | Intelligent management method and system for data center | |
| CN114051252A (en) | Multi-user intelligent transmitting power control method in wireless access network | |
| CN109039797A (en) | Big stream detection method based on intensified learning | |
| Bhattacharyya et al. | QFlow: A learning approach to high QoE video streaming at the wireless edge | |
| CN114500561B (en) | Power Internet of things network resource allocation decision-making method, system, equipment and medium | |
| WO2023142351A1 (en) | Weight adjustment method and apparatus, and storage medium and electronic apparatus | |
| CN116302569A (en) | Resource partition intelligent scheduling method based on user request information | |
| Cui et al. | Multi-Agent Reinforcement Learning Based Cooperative Multitype Task Offloading Strategy for Internet of Vehicles in B5G/6G Network | |
| CN114401192B (en) | Multi-SDN controller cooperative training method | |
| CN113271221B (en) | Network capacity opening method and system and electronic equipment | |
| CN115129888A (en) | Active content caching method based on network edge knowledge graph | |
| Zhang et al. | Adaptive Frame Rate Optimization Based on Particle Swarm and Neural Network for Industrial Video Stream | |
| CN113971454A (en) | Deep learning model quantification method and related device | |
| CN117834643B (en) | Deep neural network collaborative reasoning method for industrial Internet of things | |
| CN112600869A (en) | Calculation unloading distribution method and device based on TD3 algorithm |
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 |