CN110460880B - Industrial wireless streaming media self-adaptive transmission method based on particle swarm and neural network - Google Patents
Industrial wireless streaming media self-adaptive transmission method based on particle swarm and neural network Download PDFInfo
- Publication number
- CN110460880B CN110460880B CN201910733205.3A CN201910733205A CN110460880B CN 110460880 B CN110460880 B CN 110460880B CN 201910733205 A CN201910733205 A CN 201910733205A CN 110460880 B CN110460880 B CN 110460880B
- Authority
- CN
- China
- Prior art keywords
- neural network
- network
- parameters
- video
- particle swarm
- 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
Links
Images
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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Signal Processing (AREA)
- Multimedia (AREA)
- Biophysics (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Computer Networks & Wireless Communication (AREA)
- Computer Security & Cryptography (AREA)
- Databases & Information Systems (AREA)
- Two-Way Televisions, Distribution Of Moving Picture Or The Like (AREA)
Abstract
The invention discloses an industrial wireless streaming media self-adaptive transmission method based on particle swarm and a neural network. Firstly, acquiring historical data in a specified range from a cloud server database, finishing training of a neural network model, and monitoring various state parameters of a wireless channel in real time; then, wireless network transmission parameters which enable the video experience quality to be optimal are obtained through a particle swarm algorithm; secondly, predicting optimal system setting parameters by using the mapping function of the trained neural network model, and completing the setting of the system; and finally, acquiring and storing actual data, and reapplying the actual data to training and correcting the neural network. The method is provided on the basis of fully considering the traditional dynamic self-adaptive video DASH transmission protocol based on the HTTP, parameter optimization can be completed faster through a particle swarm algorithm, the traditional trial exploration mode is replaced by a neural network direct mapping mode to directly obtain system setting parameters, and more accurate setting of the system parameters and more smooth transmission of videos are facilitated.
Description
Technical Field
The invention relates to a video technology, in particular to an industrial wireless streaming media self-adaptive transmission method based on particle swarm and neural network.
Background
The video technology greatly improves the problem of video data loading time, and a user can acquire larger video data in a segmented manner through the video technology without excessive loading time. Conventional video services fall into two broad categories: one is a connection-oriented Real-Time video technology that employs Real-Time Streaming Protocol/Real-Time transport Protocol (RTSP/RTP); another type is connectionless sequential video technology using the hypertext Transfer Protocol (HTTP). Aiming at the problems in the wireless network and video transmission process, various scholars propose a dynamic Adaptive video transmission protocol DASH (dynamic Adaptive Streaming over HTTP) based on HTTP, which is used for adapting to an unstable network environment, improving the user experience of video data and enabling the video data to realize the playing requirements of high code rate, low fluctuation and no interruption as much as possible.
The typical HTTP-based adaptive video technology at present mainly includes: 1) an adaptive transmission method based on network throughput. The adaptive algorithm based on the network throughput mainly determines the application code rate of the client by estimating the network bandwidth at the next moment, and prevents the interruption in the video playing process as much as possible. 2) An adaptive transmission method based on buffer control. The adaptive transmission algorithm based on cache control mainly ensures that the video data volume of a client buffer area is as stable as possible by changing the sending rate of a server and the media switching rate of a client, thereby ensuring the user experience quality.
However, the above methods are mostly traditional system setting methods of exploration trial, although the required effect can be achieved under certain conditions, the problems of inaccurate parameter selection, large estimation error, large consumption of computing resources, unstable video experience quality and the like still exist, and the traditional adaptive video transmission method is generally difficult to directly obtain the optimal parameters, the required effect can be achieved after multiple trials, the adjustment period is long, and meanwhile, as the method generally relates to the cross-layer design of a network and a dynamic coping strategy for adapting to various special conditions, a simpler, more convenient and more comprehensive adaptive video method is needed for ensuring the transmission quality and the stability and reliability of the system.
Disclosure of Invention
Aiming at the current situations that an intelligent self-adaptive video transmission technology is lacked and the traditional self-adaptive video technology has the problems of inaccurate parameter selection, large estimation error, large calculation resource consumption, unstable video experience quality and the like, the invention provides an industrial wireless streaming media self-adaptive transmission method based on particle swarm and a neural network, which adopts a method of combining the particle swarm algorithm and the neural network to realize the self-adaptive transmission effect with the optimal video experience quality, and the technical scheme is as follows:
an industrial wireless streaming media self-adaptive transmission method based on particle swarm and neural network comprises the following steps:
the method comprises the following steps: acquiring historical data in a specified range from a cloud server database, finishing training of a neural network model, and monitoring various state parameters of a wireless channel in real time, wherein the method comprises the following specific steps:
1) connecting a cloud server and applying for accessing a database;
2) acquiring m pieces of historical data (m is a natural number) according to the calculation precision requirement;
3) deleting the m pieces of historical data to enable the number of variables in the deleted historical data to accord with the input variable dimension and the output variable dimension of the neural network;
4) and smoothing the deleted historical data, wherein the formula is as follows:
X＝a*mean_x+b*(x-mean_x)
wherein X represents each variable input as the neural network, X represents each variable input as the neural network after smoothing, mean _ X represents the average value of data obtained from each variable, a and b are coefficients, a and b of different systems have different values, but always satisfy { a + b ═ 1| a > b, a >0, b >0} and a > b, and appropriate a and b can be selected according to different system conditions;
5) combining the smoothed historical data into input vectors according to the sequence and the dimensionality required by a neural network input matrix, obtaining corresponding output vectors, inputting the input vectors and the corresponding output vectors into the neural network to obtain the neural network matrix, finishing the training of a neural network model, and reading a cloud server data receiving port and an appointed system file through a library function of a universal network communication library twisted to obtain various network state parameters of a wireless channel;
step two: wireless network transmission parameters for optimizing the Quality of Experience (QoE) of the video are obtained by a particle swarm algorithm, so that the video frame rate at the next moment is maximum, the fluctuation is minimum, and the video is most smooth, and the method specifically comprises the following steps:
1) initializing a particle swarm, and calculating the inertia weight w and the learning factor c of the particle1、c2Initializing the particle swarm number output _ size, dimension dim, iteration times max _ steps, a solution space range x _ bound, an initial particle swarm position x', and an initial particle velocity v;
2) designing a fitness function, wherein a calculation formula of the fitness function is as follows:
fitness＝a*Q_m-b*Q_s
wherein a 'and b' represent influence factor coefficients, and Q _ m represents direct and average frame rate mkAnd Q _ s represents that the variance s _ k directly related to the real-time network throughput p _ i is in a logarithmic correlation relationship, and the specific calculation formula is as follows:
Q_m＝ln(mk+ε)
the Q _ m and the Q _ s are in a negative correlation relationship, the larger the Q _ m is, and the smaller the Q _ s is, the larger the average frame rate of the video is, the smoother the video is, the smaller the fluctuation is, the better the experience quality of the video user is, and the m iskRepresenting the video frame rate f per second in N/2 seconds before the current timeiAn average of the sums; sigmakMean square error representing network throughput in N/2 seconds before the current time; ε and
representing system-dependent constant parameters, different system parameters being different, e and Q may be used when calculating in order to ensure that Q _ m and Q _ s are of the same order
Carrying out adjustment; n denotes the number of data strips used, fiRepresenting the video frame rate, s, per secondk+1Represents the variance of the network throughput at time k +1, p-i represents the network throughput at time i,
represents the average of the throughput in the first N +1 pieces of data;
3) updating the speed and the position until iteration is finished or the minimum threshold value of the optimal position is met, comparing to obtain the optimal particle position and the corresponding fitness value after the fitness function is calculated each time, updating the individual optimal fitness value indivisual _ best _ fit and the global optimal fitness value global _ best _ fit, and obtaining x' corresponding to the global optimal fitness value, namely the network state parameter which can enable the throughput to be maximum and the fluctuation to be minimum;
step three: predicting optimal system setting parameters by using the mapping function of the trained neural network model in the step one, and completing the setting of the system, wherein the specific steps are as follows:
1) the method comprises the steps of obtaining network state parameters required by an input end of a particle swarm algorithm by reading a cloud server database storing the network state parameters, inputting the network state parameters into the particle swarm algorithm to obtain globally optimal particles and fitness, expanding the values of frame rate and image quality variable introduced into system parameters into an input matrix of a required neural network matrix, wherein the values of the frame rate and image quality variable in the system parameters can be supplemented by historical data at the previous moment;
2) inputting the input matrix in the step three 1) into a neural network, obtaining a predicted value of a system setting parameter which can enable the video experience quality to be highest through the mapping relation of the neural network, and setting the system parameter according to the predicted value;
step four: actual data are obtained and stored, and are reapplied to training and correcting of the neural network, and the method comprises the following specific steps:
1) reading a data receiving port of a cloud server and a specified system file through a library function of a universal network communication library twisted to complete continuous monitoring of the system network state, and obtaining and storing the latest network state parameters and related system parameters;
2) and (2) processing the latest network state parameter in the step four 1) and the video frame rate and the image quality parameter at the last moment according to the specific step of the step one, inputting the latest network state parameter, the video frame rate and the image quality parameter into the neural network according to a rule that an input vector obtained by using the latest actual data replaces the input vector with the earliest time, and periodically correcting the neural network.
The invention has the beneficial effects that:
according to the industrial wireless streaming media self-adaptive transmission method based on the particle swarm and the neural network, the parameter selection problem of the system is processed by adopting a method of combining the particle swarm algorithm and the neural network, so that the system selection parameter is more accurate, the response is quicker, the industrial wireless streaming media self-adaptive transmission method based on the particle swarm and the neural network can cope with the complex situation of various sudden network fluctuations under the wireless network environment, the system is more intelligent, meanwhile, the smoothness of video playing is increased, and the self-adaptive transmission effect with the best video experience quality is realized.
Drawings
Fig. 1 is a flowchart of an industrial wireless streaming media adaptive transmission method based on a particle swarm and a neural network in an embodiment of the present patent.
Fig. 2 is a schematic diagram of an industrial wireless streaming media adaptive transmission method based on particle swarm and neural network in the embodiment of the patent.
Detailed Description
The following is a detailed description of the technical solution of the present invention with reference to the accompanying drawings.
As shown in fig. 1, which is a flowchart of a video frame rate adaptive transmission method based on a particle swarm and a neural network in an embodiment of the present patent, a video frame rate adaptive transmission method based on a particle swarm and a neural network includes the following steps:
the method comprises the following steps: acquiring historical data in a specified range from a cloud server database, finishing training of a neural network model, and monitoring various state parameters of a wireless channel in real time, wherein the method comprises the following specific steps:
1) connecting a cloud server and applying for accessing a database;
2) acquiring m pieces of historical data (m is a natural number) according to the calculation precision requirement, wherein the value range of m in the embodiment is 500< m < 1000;
3) deleting the m pieces of historical data to enable the number of variables in the deleted historical data to accord with the input variable dimension and the output variable dimension of the neural network;
4) and smoothing the deleted historical data, wherein the formula is as follows:
X＝a*mean_x+b*(x-mean_x)
wherein X represents each variable input as the neural network, X represents each variable input as the neural network after smoothing, mean _ X represents the average value of data obtained from each variable, a and b are coefficients, a and b of different systems have different values, but always satisfy { a + b ═ 1| a > b, a >0, b >0} and a > b, and appropriate a and b can be selected according to different system conditions;
5) combining the smoothed historical data into input vectors according to the sequence and the dimensionality required by a neural network input matrix, obtaining corresponding output vectors, inputting the input vectors and the corresponding output vectors into the neural network to obtain the neural network matrix, finishing the training of a neural network model, reading a cloud server data receiving port and a specified system file through a library function of a universal network communication library twisted to obtain various network state parameters of a wireless channel, wherein the network training process is complex and difficult to observe, the specific process is shown as figure 1, after the preprocessed data enter a neural network input layer, performing weight calculation on the data in a hidden layer of the neural network, obtaining output data after calculating all the specified hidden layers, performing error calculation on the output data and expected data, and then performing error calculation in a gradient reducing direction through a reverse transfer method, taking the weight as a basis for adjusting the weight of the neural network, updating the weight matrix of the neural network while reducing the error, and ending the training when a training termination condition is reached in the calculation process, for example: when the error is less than the set error and the maximum learning frequency is reached, returning to the data preprocessing part to continue to calculate when the end requirement is not reached.
Step two: wireless network transmission parameters for optimizing the Quality of Experience (QoE) of the video are obtained by a particle swarm algorithm, so that the video frame rate at the next moment is maximum, the fluctuation is minimum, and the video is most smooth, and the method specifically comprises the following steps:
1) initializing a particle swarm, and calculating the inertia weight w and the learning factor c of the particle1、c2Initializing the particle swarm number output _ size, dimension dim, iteration times max _ steps, solution space range x _ bound, initial particle swarm position x, and initial particle velocity v, wherein the initialization values of the parameters in this embodiment are: particle inertia weight w equal to 0.6, learning factor c1＝2、c22, the particle group number output _ size is 100, the dimension dim is 2, the iteration number max _ steps is 10000, and the solution space range x _ bound ∈ [ -10,10000]As shown in fig. 1, firstly, operations of initializing particle speed, position and the like are performed, before performing particle swarm optimization, random particles and corresponding positions and relevant parameter information above are required to be obtained in advance, and relevant variables of a system are required to be initialized;
2) and designing a fitness function, wherein the fitness function is an important part of the algorithm and is an important index influencing the quality of the algorithm, and the fitness function is designed not only to realize the evaluation of the video quality, but also to consider the occupancy rate of computing resources, so that the video quality needs to be evaluated simply. The invention directly starts from the QoE (quality of experience) of the video when designing the fitness function, and designs the following fitness function:
fitness＝a*Q_m-b*Q_s
wherein a 'and b' represent influence factor coefficients, and Q _ m represents direct and average frame rate mkIs logarithmicAnd Q _ s represents that the correlation directly presents a logarithmic correlation with the variance s _ k of the real-time network throughput p _ i, and the specific calculation formula is as follows:
Q_m＝ln(mk+ε)
the Q _ m and the Q _ s are in a negative correlation relationship, the larger the Q _ m is, and the smaller the Q _ s is, the larger the average frame rate of the video is, the smoother the video is, the smaller the fluctuation is, the better the experience quality of the video user is, and the m iskRepresenting the video frame rate f per second in N/2 seconds before the current timeiAn average of the sums; sigmakMean square error representing network throughput in N/2 seconds before the current time; ε and
representing system-dependent constant parameters, different system parameters being different, e and Q may be used when calculating in order to ensure that Q _ m and Q _ s are of the same order
Carrying out adjustment; n denotes the number of data strips used, fiRepresenting the video frame rate, s, per secondk+1Represents the variance of the network throughput at time k +1, p-i represents the network throughput at time i,
the average value of the throughput in the first N +1 pieces of data is represented, as shown in fig. 1, after initialization and fitness function design, the fitness is calculated in a step of 'calculating the fitness' through a data and fitness function alignment systemCarrying out quantitative evaluation on the system;
3) updating the speed and the position until iteration is finished or the minimum threshold value of the optimal position is met, comparing to obtain the optimal particle position and the corresponding fitness value after the fitness function is calculated each time, updating the individual optimal fitness value individual _ best _ fit and the global optimal fitness value global _ best _ fit, and obtaining x' corresponding to the global optimal fitness value, namely the network state parameter which can enable throughput to be maximum and fluctuation to be minimum, as shown in figure 1, in an "obtaining individual optimum" stage, obtaining an individual particle with optimal fitness in the current group of particles through calculation of the fitness of each particle, and in an "obtaining global optimum" stage, comparing the current individual with the historical individual to obtain the global optimum individual; and then judging whether to continue the optimizing operation or not through a termination condition, and if the optimizing operation is not finished, adjusting the speed and the position of each particle in a speed adjustment and position adjustment stage in the attached figure 1 so as to obtain better fitness.
In the second step, the particle swarm algorithm designs a fitness function meeting the QoE index requirement to perform quantization processing on the transmission quality of the video, and obtains an optimal network state parameter by weighting and other processing modes of the network state parameters such as the frame rate, the network throughput, the packet loss rate and the like in the particle swarm algorithm part under the condition of meeting the requirements of the maximum playing frame rate and the minimum fluctuation, as shown in a schematic diagram of the industrial wireless streaming media adaptive transmission method based on the particle swarm and the neural network in the embodiment of fig. 2, the particle swarm algorithm performs weighting design on the network state parameters such as the average frame rate, the network throughput variance and the like to obtain the optimal target network state parameter, and the optimal target network state parameter is obtained by the particle swarm algorithm and is used as part of the original data input by the neural network of the next stage.
Step three: predicting optimal system setting parameters by using the mapping function of the trained neural network model in the step one, and completing the setting of the system, wherein the specific steps are as follows:
1) acquiring network state parameters required by an input end of a particle swarm algorithm by reading a cloud server database storing the network state parameters, as shown in figure 1, after outputting globally optimal particles and fitness, acquiring related information of the particles, namely required target network state information, and performing operations of adjusting matrix dimensionality and acquiring input vectors on data acquired by the particle swarm through real-time network state monitoring so as to acquire an input matrix which can be directly used by a neural network;
2) inputting the input matrix in the step three 1) into a neural network, obtaining a predicted value of a system setting parameter which can enable the video experience quality to be highest through a mapping relation of the neural network, and setting the system parameter according to the predicted value, as shown in the attached drawing 1, mapping a target network state parameter into the system setting parameter after the target network state parameter is transmitted into the neural network, wherein the obtained parameter is the system parameter which can enable the network state parameter to be the best at the next moment and the user experience quality to be the best, and thus, completing the system setting;
step four: actual data are obtained and stored, and are reapplied to training and correcting of the neural network, and the method comprises the following specific steps:
1) the method comprises the steps of reading a cloud server data receiving port and an appointed system file through a library function of a universal network communication library twisted to complete continuous monitoring of a system network state, obtaining latest network state parameters and related system parameters, and storing the latest network state parameters and the related system parameters, wherein as shown in figure 1, after system setting is completed, a new network state parameter is generated after a system runs, a network is monitored, and an actual network state parameter is obtained in real time;
2) processing the latest network state parameter in the step four 1) and the video frame rate and the image quality parameter at the last moment according to the concrete steps of the step one, and periodically inputting the input vector obtained by the latest actual data to the neural network according to the rule that the input vector replaces the input vector with the earliest time, correcting the neural network, as shown in fig. 1, after obtaining the "actual network state parameters", the "storing, deleting and smoothing" operations are again used as new data to be input into the neural network again, and the network correction is performed to reduce the error, as shown in fig. 2, after the optimal network state parameters are mapped into the optimal system parameters through the neural network, a new network state can be obtained after the system is set through the obtained system parameters, and the network state parameters are used as training and correcting data of the neural network to carry out storage and persistence operation of a database.
Claims (1)
1. An industrial wireless streaming media self-adaptive transmission method based on particle swarm and neural network is characterized by comprising the following steps:
the method comprises the following steps: acquiring historical data in a specified range from a cloud server database, finishing training of a neural network model, and monitoring various state parameters of a wireless channel in real time, wherein the method comprises the following specific steps:
1) connecting a cloud server and applying for accessing a database;
2) acquiring m pieces of historical data according to the calculation precision requirement;
3) deleting the m pieces of historical data to enable the number of variables in the deleted historical data to accord with the input variable dimension and the output variable dimension of the neural network;
4) and smoothing the deleted historical data, wherein the formula is as follows:
X＝a*mean_x+b*(x-mean_x)
wherein X represents each variable input as the neural network, X represents each variable input as the neural network after smoothing, mean _ X represents the average value of data obtained from each variable, a and b are coefficients, a and b of different systems have different values, but always satisfy { a + b ═ 1| a > b, a >0, b >0} and a > b, and appropriate a and b can be selected according to different system conditions;
5) combining the smoothed historical data into input vectors according to the sequence and the dimensionality required by a neural network input matrix, obtaining corresponding output vectors, inputting the input vectors and the corresponding output vectors into the neural network to obtain the neural network matrix, finishing the training of a neural network model, and reading a cloud server data receiving port and an appointed system file through a library function of a universal network communication library twisted to obtain various network state parameters of a wireless channel;
step two: the method comprises the following steps of obtaining wireless network transmission parameters enabling the video experience quality to be optimal through a particle swarm algorithm, enabling the video frame rate at the next moment to be maximum, the fluctuation to be minimum and the video to be smooth, and specifically comprising the following steps:
1) initializing a particle swarm, and calculating the inertia weight w and the learning factor c of the particle1、c2Initializing the particle swarm number output _ size, dimension dim, iteration times max _ steps, a solution space range x _ bound, an initial particle swarm position x', and an initial particle velocity v;
2) designing a fitness function, wherein a calculation formula of the fitness function is as follows:
fitness＝a'*Q_m-b'*Q_s
wherein a 'and b' represent influence factor coefficients, and Q _ m represents direct and average frame rate mkAnd Q _ s represents that the variance s _ k directly related to the real-time network throughput p _ i is in a logarithmic correlation relationship, and the specific calculation formula is as follows:
Q_m＝ln(mk+ε)
the Q _ m and the Q _ s are in a negative correlation relationship, the larger the Q _ m is, and the smaller the Q _ s is, the larger the average frame rate of the video is, the smoother the video is, the smaller the fluctuation is, the better the experience quality of the video user is, and the m iskRepresenting the video frame rate f per second in N/2 seconds before the current timeiAn average of the sums; sigmakIndicating within N/2 seconds before the current timeMean square error of network throughput; ε and
representing system-dependent constant parameters, different system parameters being different, e and Q may be used when calculating in order to ensure that Q _ m and Q _ s are of the same order
Carrying out adjustment; n denotes the number of data strips used, fiRepresenting the video frame rate, s, per secondk+1Represents the variance of the network throughput at time k +1, p-i represents the network throughput at time i,
represents the average of the throughput in the first N +1 pieces of data;
3) updating the speed and the position until iteration is finished or the minimum threshold value of the optimal position is met, comparing to obtain the optimal particle position and the corresponding fitness value after the fitness function is calculated each time, updating the individual optimal fitness value indivisual _ best _ fit and the global optimal fitness value global _ best _ fit, and obtaining x' corresponding to the global optimal fitness value, namely the network state parameter which can enable the throughput to be maximum and the fluctuation to be minimum;
step three: predicting optimal system setting parameters by using the mapping function of the trained neural network model in the step one, and completing the setting of the system, wherein the specific steps are as follows:
1) the method comprises the steps of obtaining network state parameters required by an input end of a particle swarm algorithm by reading a cloud server database storing the network state parameters, inputting the network state parameters into the particle swarm algorithm to obtain globally optimal particles and fitness, expanding the values of frame rate and image quality variable introduced into system parameters into an input matrix of a required neural network matrix, wherein the values of the frame rate and image quality variable in the system parameters can be supplemented by historical data at the previous moment;
2) inputting the input matrix in the step three 1) into a neural network, obtaining a predicted value of a system setting parameter which can enable the video experience quality to be highest through the mapping relation of the neural network, and setting the system parameter according to the predicted value;
step four: actual data are obtained and stored, and are reapplied to training and correcting of the neural network, and the method comprises the following specific steps:
1) reading a data receiving port of a cloud server and a specified system file through a library function of a universal network communication library twisted to complete continuous monitoring of the system network state, and obtaining and storing the latest network state parameters and other system parameters;
2) and (2) processing the latest network state parameter in the step four 1) and the video frame rate and the image quality parameter at the last moment according to the specific step of the step one, inputting the latest network state parameter, the video frame rate and the image quality parameter into the neural network according to a rule that an input vector obtained by using the latest actual data replaces the input vector with the earliest time, and periodically correcting the neural network.
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 CN110460880A (en) | 2019-11-15 |
| CN110460880B true 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) |
Families Citing this family (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114079820A (en) * | 2020-08-19 | 2022-02-22 | 安霸国际有限合伙企业 | Interval shooting video generation centered on an event/object of interest input on a camera device by means of a neural network |
| 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 |
| 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 |
| CN113553270B (en) * | 2021-07-28 | 2024-07-26 | 浙江网商银行股份有限公司 | Target object determining 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 |
| CN114764413B (en) * | 2022-03-21 | 2024-10-18 | 中国电子科技集团公司第五十四研究所 | Self-adaptive engine system for interfering electromagnetic signals |
| 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 |
| CN115022595B (en) * | 2022-06-02 | 2024-08-02 | 厦门大学 | Unmanned aerial vehicle group intelligent anti-interference transmission method for high-quality video transmission |
| 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 |
| CN117724446B (en) * | 2023-12-14 | 2024-07-05 | 广州智业节能科技有限公司 | Early warning monitoring system, method and device |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| 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 |
| 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 |
Family Cites Families (8)
| 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 |
| CN103164742B (en) * | 2013-04-02 | 2016-02-17 | 南京邮电大学 | A kind of server performance Forecasting Methodology based on particle group optimizing neural network |
| CN104361393B (en) * | 2014-09-06 | 2018-02-27 | 华北电力大学 | Data predication method is used for based on the improved neural network model of particle swarm optimization algorithm |
| US20190130188A1 (en) * | 2017-10-26 | 2019-05-02 | Qualcomm Incorporated | Object classification in a video analytics system |
| CN108901058A (en) * | 2018-07-06 | 2018-11-27 | 北方工业大学 | Internet of things node access channel optimization selection method |
| 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 |
| 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 |
-
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 (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| 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 |
| 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 |
|---|
| 基于神经网络和粒子群算法的MPEG视频传输控制;向涛，涂风华，廖晓峰;《计算机科学》;20051231;第32卷(第9期);134-135、186 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110460880A (en) | 2019-11-15 |
| WO2021026944A1 (en) | 2021-02-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110460880B (en) | Industrial wireless streaming media self-adaptive transmission method based on particle swarm and neural network | |
| Zhang et al. | A multi-agent reinforcement learning approach for efficient client selection in federated learning | |
| CN111092823B (en) | Method and system for adaptively adjusting congestion control initial window | |
| CN113434212B (en) | Cache auxiliary task cooperative unloading and resource allocation method based on meta reinforcement learning | |
| CN113537514B (en) | Digital twinning-based federal learning framework with high energy efficiency | |
| CN113141317B (en) | Streaming media server load balancing method, system, computer equipment and terminal | |
| CN110955463B (en) | Internet of things multi-user computing unloading method supporting edge computing | |
| CN114125785A (en) | Low-delay high-reliability transmission method, device, equipment and medium for digital twin network | |
| CN114500561B (en) | Power Internet of things network resource allocation decision-making method, system, equipment and medium | |
| CN113778691B (en) | Task migration decision method, device and system | |
| CN114205353B (en) | Calculation unloading method based on hybrid action space reinforcement learning algorithm | |
| CN114528987A (en) | Neural network edge-cloud collaborative computing segmentation deployment method | |
| CN116489712B (en) | Mobile edge computing task unloading method based on deep reinforcement learning | |
| CN113868808B (en) | Road network approach detection time delay optimization method, device and system | |
| CN117707795B (en) | Graph-based model partitioning side collaborative reasoning method and system | |
| CN112600869B (en) | Calculation unloading distribution method and device based on TD3 algorithm | |
| CN117749796A (en) | Cloud edge computing power network system calculation unloading method and system | |
| CN113504949A (en) | Task unloading and parameter optimization method and system for MAR client in edge computing | |
| CN111930435A (en) | Task unloading decision method based on PD-BPSO technology | |
| CN116306915A (en) | Wireless federal learning method for large-scale Internet of things collaborative intelligence | |
| KR20190129422A (en) | Method and device for variational interference using neural network | |
| CN114942799A (en) | Workflow scheduling method based on reinforcement learning under cloud edge environment | |
| CN114449536A (en) | 5G ultra-dense network multi-user access selection method based on deep reinforcement learning | |
| Liu et al. | Throughput Prediction-Enhanced RL for Low-Delay Video Application | |
| CN113743012A (en) | Cloud-edge collaborative mode task unloading optimization method under multi-user scene |
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 |