阅读说明：本技术 一种基于分层缓存的实时业务时延优化方法 (Real-time service delay optimization method based on layered cache ) 是由 邹虹 王青青 张鸿 李职杜 吴大鹏 王汝言 于 2020-03-09 设计创作，主要内容包括：本发明涉及一种基于分层缓存的实时业务时延优化方法,属于通信技术领域,特别是实时数据处理技术领域。该方法针对实时业务中由于无线链路资源有限造成用户时延过大的问题,首先,考虑到用户观看视频等实时业务过程中会有后退,快进等行为,提出采用光域和无线域分层缓存的方式为流行视频内容提供缓存,在光域利用协作缓存完整的视频文件,无线域缓存流行度高的视频片段；进而,根据用户获取视频片段的不同位置构造最小化传输时延问题,结合可伸缩视频流的特性采用粒子群算法为每个视频层分配最佳的传输速率达到最小化传输时延的目的。该方法可有效地减少用户的传输时延,具有广阔的运用前景。(The invention relates to a real-time service delay optimization method based on layered caching, belongs to the technical field of communication, and particularly relates to the technical field of real-time data processing. The method aims at the problem that the time delay of a user is overlarge due to limited wireless link resources in real-time services, firstly, in consideration of the behaviors of retreating, fast forwarding and the like in the process of watching videos and other real-time services of the user, the method provides a mode of adopting optical domain and wireless domain hierarchical caching to cache popular video contents, complete video files are cached in an optical domain in a cooperation mode, and video clips with high popularity are cached in a wireless domain; and then, constructing a problem of minimizing transmission delay according to different positions of the video clip obtained by the user, and distributing an optimal transmission rate for each video layer by adopting a particle swarm algorithm by combining the characteristics of the scalable video stream to achieve the purpose of minimizing the transmission delay. The method can effectively reduce the transmission delay of the user and has wide application prospect.)

1. A real-time service delay optimization method based on a layered cache is characterized in that: the method comprises the steps of firstly, dynamically pre-caching popular video files and video clips on two layers of an optical domain and a wireless domain by analyzing and calculating the caching value and popularity of real-time services; further, a minimum time delay function is constructed according to the specific mode of obtaining the video clips by the user, and a proper transmission rate is distributed to each video layer through a particle swarm algorithm; the method specifically comprises the following steps:

s1: optical wireless domain layered caching: analyzing the popularity of the complete video file and the video clip, and performing layered caching on the video content with higher popularity in an optical domain and a wireless domain;

s2: optical domain ONU cooperation buffering: caching a video file with high popularity at an ONU node of an optical domain, and assisting a heavy-load ONU to perform video pre-caching by using a light-load ONU according to the caching value of the video file;

s3: wireless domain video clip caching: each video clip in a video file has independent popularity, and a plurality of video clips can be repeatedly sent under the condition that a user backs or fast forwards, so that the video clips with high popularity are cached at a router in a wireless network, a Markov model is constructed to analyze the popularity of the video clips, and the network cost is analyzed by combining the distance between the user and the router, so that the video clips are cached in a proper router;

s4: analyzing service delay: after finishing caching the video files and the video clips according to the steps S2 and S3, establishing a minimum time transmission delay model by analyzing a specific path of video content acquired by a user according to cache hit rates of optical domain ONU and wireless domain router nodes;

s5: video layer rate allocation: according to the characteristic of scalable video coding, a user has to correctly decode a low video layer when receiving a high video layer, and an optimal rate allocation scheme of the video layer is obtained based on a particle swarm optimization under the constraint condition that the total time delay of the user is minimized.

2. The method according to claim 1, wherein the method for optimizing the delay of the real-time service based on the hierarchical cache comprises: the step S2 specifically includes:

s21: the number of times that the user clicks the video and the popularity of the video both obey Zipf distribution, and the popularity of the video file is represented by the Zipf distribution;

s22: residual buffer space C combined with optical domain ONU_nSize S of video file v_vSelecting direct cache or replacement cache for the cache value of the video file, and judging whether the video file v meets the cache condition;

s23: for the heavy-load ONU, calculating the cache value of the video file v according to the probability of the user requesting the video file v under the ONU node and the popularity of the video file v, and caching according to the step S22; for the light-load ONU, the light-load ONU is utilized to cooperatively cache the video file which does not meet the caching condition of the step S22 but has high request probability in the heavy-load ONU; and calculating the caching value of the video file v in the light-load ONU according to the request probability and the popularity of the video file v under the light-load ONU and the probability of the video file v needing to be cooperatively cached in the heavy-load ONU, and caching according to the step S22.

3. The method according to claim 1, wherein the method for optimizing the delay of the real-time service based on the hierarchical cache comprises: in step S3, the method specifically includes:

s31: establishing a Markov model to analyze the popularity of the video segments, wherein in order to reduce the complexity of calculation and ensure the accuracy of prediction, the analyzed video file is the video content pre-cached in the optical domain ONU, the video segments continuously watched by a user are used as a user access sequence, and the popularity of the user access sequence is analyzed by the Markov model, so that the request probability of the video segments can be obtained;

s32: in order to reduce the delay, the pre-cached video content should be as close to the user side as possible; calculating the network overhead of the user for acquiring the video clip according to the size of the user access sequence and the number of router hops transmitted to the user from the buffer position;

s33: and according to the request probability of each video clip obtained in the step S31 and the quotient of the network cost calculated in the step S32, the probability that the wireless domain router node caches each user access sequence is represented, and the user access sequence with the highest router cache popularity and the lowest total cost is selected by adopting a method of traversing the router nodes.

4. The method according to claim 1, wherein the method for optimizing the delay of the real-time service based on the hierarchical cache comprises: in step S4, the method specifically includes:

s41: after a user sends a request, in a wireless domain, firstly judging whether an adjacent router of the user is hit, and if the adjacent router of the user is hit in a cache, returning the content to the user; if the adjacent router is not cached and hit, continuing to forward the request, and judging whether the non-adjacent router and the ONU connected with the user hit; if the cache hits, selecting according to the hop count between the node and the user; if only one node cache is hit, returning the content to the user; otherwise, continuing to forward the request;

s42: in the optical domain, judging whether an ONU connected with a user is in cache hit or not, if so, returning video content to the user, and if not, detecting whether a coordinated ONU node is in cache hit or not; if the optical domain and the wireless domain are not cached and hit, the server provides service for the user;

s43: analyzing time delay when the wireless domain, the optical domain and the server respectively provide services for the user according to the cache hit rate of the ONU node, the cache hit rate of the wireless domain router node, the distance between the service node and the user and the size of the video clip;

s44: and constructing a minimized time delay function for obtaining the complete video file by the user according to three different paths for providing services for the user.

5. The method according to claim 1, wherein the method for optimizing the delay of the real-time service based on the hierarchical cache comprises: in step S5, the method specifically includes:

s51: initializing a deployment strategy, providing rates corresponding to four different modulation coding modes, randomly generating I particles, wherein each particle is an E-dimensional vector and the number of iteration is limited to J;

s52: updating the particle sequence, and evaluating the particles in the population by using a fitness function;

s53: optimizing the particle swarm, updating and iterating the position and the speed of the particles to obtain a better video layer rate distribution strategy; if the particle fitness value after iteration is smaller than the fitness value of the individual extreme value, replacing the individual extreme value with the position of the iteration, otherwise, continuously updating the individual extreme value; similarly, the global extreme value of the whole particle swarm represents the optimal position for stopping the current iteration, if the individual extreme value of the current particle is smaller than the global extreme value, the position of the particle is used for replacing the global extreme value, otherwise, the global extreme value is not updated;

s54: selecting an optimal solution, wherein if the global extreme value fitness of the particles after multiple iterations is smaller than a certain set range relative to the global extreme value change amplitude before the iterations, the extreme value is very close to the optimal extreme value, and the iterations are stopped, namely the optimal transmission rate is allocated to the video layer under the constraint condition of the minimum time delay function; otherwise, repeating the above steps until reaching the maximum iteration number.

Technical Field

The invention belongs to the technical field of communication, particularly relates to the technical field of real-time data processing, and relates to a real-time service delay optimization method based on hierarchical cache.

Background

With the rapid growth of the number of mobile users and wireless multimedia applications, limited network resources and increasing traffic demands have become major issues in mobile communication networks. The proliferation of real-time traffic occupies more resources in the mobile network, and especially in densely populated areas and during peak periods of user requests, congestion of the transmission link is easily caused. This will place higher demands on the next generation radio access networks, such as low delay, high peak rate and better network coverage. The FiWi network conforms to the future development requirements of the network, integrates the characteristics of high capacity, high speed, low power consumption of optical fiber access, mobility, flexibility and the like of wireless access, and can provide lower use cost, higher data rate, better experience quality and wider coverage range for users. FiWi networks have become one of the most promising technologies for the next generation broadband access networks.

A great deal of research shows that the frequency of clicking videos by users is closely related to the popularity of the videos, and the frequency and the popularity of the videos are subject to Zipf distribution. There are a large number of repeated requests in video services, for example, in some large video websites, 20% of the videos located in the top of popularity rank account for nearly 80% of the click-through rate. Therefore, the Content Server (CS) repeatedly transmits the same video Content to different users, which results in a rapid decrease in link utilization. And since the video service is a very typical delay sensitive service, if the link condition is poor, the transmission delay of the user is obviously increased, and even the possibility of interruption is caused. Therefore, an efficient caching strategy is carried out in the FiWi network in advance to avoid repeated transmission of the same content, and the network performance can be effectively improved to achieve the purpose of reducing time delay.

However, according to the network environment of dynamic change, the traditional video coding method can not flexibly select proper quality for users, the scalable video coding technology encodes the video content into a Base layer (Base L layer, B L) and one or more Enhancement layers (Enhancement L layers, E L s), the Base layer provides the most basic viewing quality, the data of the upper layer depends on the data of the lower layer, and the users must correctly decode the data of the lower layer to receive the data of the upper layer.

Disclosure of Invention

In view of the above, the present invention provides a real-time service delay optimization method based on hierarchical caching, which dynamically pre-caches popular video files and video clips on both optical domain and wireless domain by analyzing and calculating the caching value and popularity of real-time services, and first, aiming at the problems that a dynamically changing network environment and the delay sensitivity of real-time services are likely to cause interruption events for users, and the problems that the video quality adaptive to the current network state cannot be dynamically selected for users by using the conventional video coding technology; furthermore, in order to effectively reduce the transmission delay, a minimum delay function is constructed according to the specific mode of acquiring the video clips by the user, and an appropriate transmission rate is allocated to each video layer through a particle swarm algorithm. The method can effectively reduce the transmission delay of the user.

In order to achieve the purpose, the invention provides the following technical scheme:

a real-time service time delay optimization method based on hierarchical cache comprises the steps of firstly dynamically pre-caching popular video files and video clips on an optical domain layer and a wireless domain layer by analyzing and calculating the cache value and popularity of a real-time service; further, a minimum time delay function is constructed according to the specific mode of obtaining the video clips by the user, and a proper transmission rate is distributed to each video layer through a particle swarm algorithm; the method specifically comprises the following steps:

s1: optical wireless domain layered caching: analyzing the popularity of the complete video file and the video clip, and performing layered caching on the video content with higher popularity in an optical domain and a wireless domain;

s2: optical domain ONU cooperation buffering: caching a video file with high popularity at an ONU node of an optical domain, and assisting a heavy-load ONU to perform video pre-caching by using a light-load ONU according to the caching value of the video file;

s3: wireless domain video clip caching: each video clip in a video file has independent popularity, and a plurality of video clips can be repeatedly sent under the condition that a user backs or fast forwards, so that the video clips with high popularity are cached at a router in a wireless network, a Markov model is constructed to analyze the popularity of the video clips, and the network cost is analyzed by combining the distance between the user and the router, so that the video clips are cached in a proper router;

s4: analyzing service delay: after finishing caching the video files and the video clips according to the steps S2 and S3, establishing a minimum time transmission delay model by analyzing a specific path of video content acquired by a user according to cache hit rates of optical domain ONU and wireless domain router nodes;

s5: video layer rate allocation: according to the characteristic of scalable video coding, a user has to correctly decode a low video layer when receiving a high video layer, and an optimal rate allocation scheme of the video layer is obtained based on a particle swarm optimization under the constraint condition that the total time delay of the user is minimized.

Further, the step S2 specifically includes:

s21: the number of times that the user clicks the video and the popularity of the video both obey Zipf distribution, and the popularity of the video file is represented by the Zipf distribution;

s22: residual buffer space C combined with optical domain ONU_nSize S of video file v_vAnd anSelecting direct caching or replacement caching for the caching value of the video file, and judging whether the video file v meets the caching condition;

s23: for the heavy-load ONU, calculating the cache value of the video file v according to the probability of the user requesting the video file v under the ONU node and the popularity of the video file v, and caching according to the step S22; for the light-load ONU, the light-load ONU is utilized to cooperatively cache the video file which does not meet the caching condition of the step S22 but has high request probability in the heavy-load ONU; and calculating the caching value of the video file v in the light-load ONU according to the request probability and the popularity of the video file v under the light-load ONU and the probability of the video file v needing to be cooperatively cached in the heavy-load ONU, and caching according to the step S22.

Further, in step S3, the method specifically includes:

s31: establishing a Markov model to analyze the popularity of the video segments, wherein in order to reduce the complexity of calculation and ensure the accuracy of prediction, the analyzed video file is the video content pre-cached in the optical domain ONU, the video segments continuously watched by a user are used as a user access sequence, and the popularity of the user access sequence is analyzed by the Markov model, so that the request probability of the video segments can be obtained;

s32: in order to reduce the delay, the pre-cached video content should be as close to the user side as possible; calculating the network overhead of the user for acquiring the video clip according to the size of the user access sequence and the number of router hops transmitted to the user from the buffer position;

s33: and according to the request probability of each video clip obtained in the step S31 and the quotient of the network cost calculated in the step S32, the probability that the wireless domain router node caches each user access sequence is represented, and the user access sequence with the highest router cache popularity and the lowest total cost is selected by adopting a method of traversing the router nodes.

Further, in step S4, the method specifically includes:

s41: after a user sends a request, in a wireless domain, firstly judging whether an adjacent router of the user is hit, and if the adjacent router of the user is hit in a cache, returning the content to the user; if the adjacent router is not cached and hit, continuing to forward the request, and judging whether the non-adjacent router and the ONU connected with the user hit; if the cache hits, selecting according to the hop count between the node and the user; if only one node cache is hit, returning the content to the user; otherwise, continuing to forward the request;

s42: in the optical domain, judging whether an ONU connected with a user is in cache hit or not, if so, returning video content to the user, and if not, detecting whether a coordinated ONU node is in cache hit or not; if the optical domain and the wireless domain are not cached and hit, the server provides service for the user;

s43: analyzing time delay when the wireless domain, the optical domain and the server respectively provide services for the user according to the cache hit rate of the ONU node, the cache hit rate of the wireless domain router node, the distance between the service node and the user and the size of the video clip;

s44: and constructing a minimized time delay function for obtaining the complete video file by the user according to three different paths for providing services for the user.

Further, in step S5, the method specifically includes:

s51: initializing a deployment strategy, providing rates corresponding to four different modulation coding modes, randomly generating I particles, wherein each particle is an E-dimensional vector and the number of iteration is limited to J;

s52: updating the particle sequence, and evaluating the particles in the population by using a fitness function;

s53: optimizing the particle swarm, updating and iterating the position and the speed of the particles to obtain a better video layer rate distribution strategy; if the particle fitness value after iteration is smaller than the fitness value of the individual extreme value, replacing the individual extreme value with the position of the iteration, otherwise, continuously updating the individual extreme value; similarly, the global extreme value of the whole particle swarm represents the optimal position for stopping the current iteration, if the individual extreme value of the current particle is smaller than the global extreme value, the position of the particle is used for replacing the global extreme value, otherwise, the global extreme value is not updated;

s54: selecting an optimal solution, wherein if the global extreme value fitness of the particles after multiple iterations is smaller than a certain set range relative to the global extreme value change amplitude before the iterations, the extreme value is very close to the optimal extreme value, and the iterations are stopped, namely the optimal transmission rate is allocated to the video layer under the constraint condition of the minimum time delay function; otherwise, repeating the above steps until reaching the maximum iteration number.

The invention has the beneficial effects that: aiming at the problems that a user is easy to have an interrupt event due to time delay sensitivity of a dynamically changing network environment and a real-time service, and the problems that the video quality adaptive to the current network state cannot be dynamically selected for the user by using the traditional video coding technology, the invention provides a real-time service time delay optimization method based on layered cache, and popular video files and video clips are dynamically pre-cached on two layers of an optical domain and a wireless domain by analyzing and calculating the cache value and popularity of the real-time service; furthermore, in order to effectively reduce the transmission delay, a minimum delay function is constructed according to the specific mode of acquiring the video clips by the user, and an appropriate transmission rate is allocated to each video layer through a particle swarm algorithm. The method provided by the invention can effectively reduce the transmission delay of the user and has wide application prospect.

Additional advantages, objects, and features of the invention will be set forth in part in the description which follows and in part will become apparent to those having ordinary skill in the art upon examination of the following or may be learned from practice of the invention. The objectives and other advantages of the invention may be realized and attained by the means of the instrumentalities and combinations particularly pointed out hereinafter.

Drawings

For the purposes of promoting a better understanding of the objects, aspects and advantages of the invention, reference will now be made to the following detailed description taken in conjunction with the accompanying drawings in which:

FIG. 1 is a system architecture diagram of the present invention;

FIG. 2 is a flow chart of the optical domain and wireless domain cooperative caching in the present invention;

FIG. 3 is a flow chart of service delay analysis in the present invention;

fig. 4 is a flow chart of a video layer rate allocation scheme in the present invention.

Detailed Description

The following detailed description of specific embodiments of the invention refers to the accompanying drawings.

As shown in fig. 1, a method for optimizing a real-time service delay based on a hierarchical cache preferably includes the following steps:

step one, optical wireless domain layered caching: and analyzing the popularity of the video files and video clips, and performing layered caching on the video content with higher popularity in an optical domain and a wireless domain.

Step two, optical domain ONU cooperation caching: the video file v with higher popularity is cached at the ONUn of the optical domain, the caching value of the video file v is calculated, the light-load ONU is utilized to assist the heavy-load ONU in caching the video file with high request probability, and the transmission delay of a user under the heavy-load ONU is reduced. The preferable method specifically comprises the following steps:

step two (one), the number of times that the user clicks the video and the popularity of the video obey Zipf distribution, and the popularity of the video file v is represented by the Zipf distribution:

the V is as follows: the content server contains V video files;

the α is that the video popularity gradient parameter, α -0 means that the video popularity obeys even distribution, and the bigger α is, the video requests are concentrated on fewer popular video files;

step two, combining the residual cache space C of the optical domain ONU node_nAnd size S of video file v_vSelecting a direct cache or a replacement cache: 1) if the remaining space C is cached_n> 0, and satisfy C_n＞S_vThe cache can be directly realized; 2) if the direct caching condition is not met, the caching value of the video file v' cached in the ONU is smaller than that of the video v, and S is met_v′＞S_vOr the sum of the caching values of the y video files is smaller than that of the video v and meets the requirementA replaceable cache;

step two (third), aiming at the heavy load ONUn', according to the probability p of the user u requesting the video file v under the ONU_v,uPopularity f with video files_vCalculating the caching value of the video file v under the heavy load ONUAnd carrying out cache judgment according to the second step (II);

the above-mentioned_n"is: a collection of heavily loaded ONUn "serving users;

step two (four), aiming at the light load ONUn', the light load ONU is used for cooperatively caching the cache condition which does not meet the step two in the heavy load ONU, but the user request probability p_v,uHigh video file v. Calculating the caching value of the video file v under the light load ONU as follows:the method comprises the steps that the caching value of a user request video file v under the light load ONU and the probability of the video file v needing to be cooperatively cached in the heavy load ONU are included, and caching judgment is carried out according to the second step;

the above-mentionedComprises the following steps:the probability that the video file v needs to be cooperatively cached in the heavily loaded ONU is expressed;

said N is₁Comprises the following steps: co-exist of N₁A heavily loaded ONU;

the above-mentionedThe video file v is a binary variable and indicates whether the video file v meets the condition of caching in a heavy load ONUn';

step three, caching the video clips in the wireless domain: each video clip k in a video file has its own popularity, which can result in multiple video clips being sent repeatedly in case of a user backing off or fast forwarding. Therefore, the invention caches the video segment k with high popularity at the router node m of the wireless domain, and preferably specifically comprises the following steps:

and step three (one), in order to reduce the complexity of calculation and ensure the accuracy of prediction, the analyzed video is a total video file pre-cached in the optical domain ONU. And taking the video clips continuously watched by the user as a user access sequence B, wherein the B comprises a plurality of video clips. Establishing Markov model to analyze video segment popularity, and calculating state transition probability p of Markov chain by using historical data_ijAnd an initial state probability distribution θ, the state transition matrix H representing the transition probability between any two states in the markov chain:

predicting according to the established Markov model, wherein the state with the maximum probability value is the user access sequence most possibly requested by the user, namely the popularity f of the user access sequence_B,v；

Step three (two), according to the size s of the user access sequence B_BThe number of router hops from the router m to the user u, and the network overhead of the user for acquiring the user access sequence from the router m are calculated

D is_(m,u)Comprises the following steps: the hop count from router m to user u;

d is_(m,n)Comprises the following steps: hop count of routers m to ONUn;

the above-mentioned_m，′_mComprises the following steps: by considering the idea of the semi-aversion P-bit problem, i.e. m weight for the router node of' preference type_mIs positive for "aversion type'_mThe value of (1) is negative, so that the distance between the selected cache node and the user is closer to the greatest extent, and the network overhead is smaller;

step three (three), according to the popularity f of each user access sequence obtained by the step three (one)_B,vAnd the network overhead calculated in the third step (second step)ComputingAnd caching the probability of the user accessing the sequence B for the wireless domain router m. Selecting a user access sequence with the lowest total cost and the highest cache popularity of the router node by adopting a method of traversing the router node;

step four, service delay analysis: after finishing caching the video files and the video clips in the second step and the third step, analyzing cache hit rates of the optical domain ONUn and the wireless router m, and analyzing the total time delay of the user for obtaining the complete video file v according to the specific path of the user for obtaining the video content. The preferable method specifically comprises the following steps:

step four (one), after the user sends the request, firstly, judging whether the adjacent router m of the user hits in the wireless domain, if so, if m is 1, returning the video content to the user; if the adjacent router m is not in the cache hit when the adjacent router m is 0, continuing to forward the request and judging the non-adjacent routerAnd whether the ONUn connected with the user hits, if all, the ONUn is cached to hitAnd n is 1, selecting according to the hop count between the node and the user: 1) when in useWhen it is acquired from the router m, 2) when it isThen, obtaining from ONUn; if only one node cache is hit, returning the content to the user; otherwise, continuing to forward the request;

step four, in the optical domain, judging whether the ONUn connected with the user is in cache hit, and if the cache hit n is 1, returning the video content to the user; if the miss n is 0, detecting cooperationWhether hit occurs; if both the optical domain and the wireless domain miss the cache hit m is 0,n＝0，the server provides a service ser equal to 1.

Step four (step three), according to the cache hit ratio of the optical domain ONUnCache hit rate for wireless domain router mDistance d between service node and user_(m/n/ser,u)And size of video segmentAnalyzing the time delay when the wireless domain, the optical domain and the server respectively provide service for the user:

wherein said Comprises the following steps: binary number, which represents ONU, router cache hit rate;

the L_kThe video clip k contains L_kA video layer;

b is_lComprises the following steps: size of video layer l is b_l；

The above-mentionedComprises the following steps: transmitting the speed of the video layer l in the video clip k from the router m, ONUn and the server respectively;

k is₁，k₂，k₃Comprises the following steps: the wireless domain, the optical domain and the server respectively provide the total number of the video clips of the service for the user;

step four, constructing a minimum time delay function for the user to obtain the complete video file v according to three different paths for providing services for the user

Step five, video layer rate allocation: depending on the nature of scalable video coding, a high quality layer relies on a low quality layer, and a user must correctly decode the low layer data in order to receive the high layer data. And under the constraint of minimum total time delay of the acquired complete video file v, distributing different sending rates to each video layer according to a particle swarm algorithm. The preferable method specifically comprises the following steps:

step five (one), the deployment strategy is initialized, and the rates corresponding to four different modulation coding modes are provided: BPSK, QPSK, 16-QAM, 64-QAM, correspond to the number 1-4 separately, produce I particle at random, each particle is a E dimensional vector, limit the iteration number to J;

step five (step two), particle sequence updating, utilizing fitness function F_nAnd evaluating the particles in the population, wherein the particles with smaller fitness represent that the time delay of the user is smaller under the scheme. The present invention is based on the analysis of video clips,namely, the method can be decomposed into three independent time delay minimization problems of a server, an ONU and a router for solving, and the fitness function is constructed by taking the service provided by the ONU as an example

Z is_nComprises the following steps: and a dynamically changing penalty function, after the particle selects a good rate for each layer of video, calculating the total bandwidth consumed by transmitting all video layers, and if the total bandwidth is greater than the maximum available bandwidth of the ONU, subtracting the penalty function Z after the fitness function_n(ii) a Penalty function Z if the total bandwidth required is less than the maximum available bandwidth of the ONU_nAt 0, the larger the amplitude of the consumed network bandwidth is than the link bandwidth, the penalty function is also increased proportionally.

And fifthly, optimizing the particle swarm, wherein each available scheme in the particle swarm algorithm is represented by one particle, each particle is an E-dimensional vector and has two characteristics of position and speed, and the objective function value corresponding to the current position of the particle is the fitness value of the particle_kThe current position of the ith particle in the video layerRepresenting an allocation scheme in which each elementRepresenting a video layer l_kThe assigned rate. For the optimum position currently searched by the ith particleShowing that the optimum position searched by the whole particle group isAnd (4) showing. The particle update rate at the next moment in the iterative process is related to the current position of the particle, the optimal position searched by the particle and the optimal position searched by the particle swarm. The particle updating speed calculation method comprises the following steps:

wherein μ is: the coefficient of inertia of the particles is such that,

the above-mentioned₁，₂Comprises the following steps: the weight coefficient is used for adjusting the weight relation between the self-searched optimal position and the global optimal position,

the rand is as follows: random constants with the value range of [0,1 ];

the calculation method of the position of the particle at the next moment comprises the following steps:and obtaining a better video layer rate distribution strategy according to the position and speed updating iteration of the particles. If the particle fitness value after iteration is larger than the optimal position of the particle i under the current iteration, namely the individual extreme value O_iIf the fitness value is small, the position of the iteration is used for replacing the individual extreme value, otherwise, the individual extreme value is continuously updated. Similarly, the global extreme value G of the whole particle swarm represents the optimal position of the current iteration, if the individual extreme value of the current particle i is smaller than the global extreme value, the position of the particle is used for replacing the global extreme value, otherwise, the global extreme value is not updated.

And step five (four), selecting an optimal solution, wherein along with the continuous change of the speed and the position of the particles, the speed and the movement track of the particles are influenced by the individual extreme value of the particles and the global extreme value of the whole particle swarm, so that all the particles approach towards the direction of the objective function, and if the global extreme value fitness F of the particles after the jth iteration is carried out, the global extreme value fitness F of the particles is_n(G_j) If F is smaller than a predetermined range Delta from the global extreme before iteration_n(G_j)/F_n(G)≤Δ,G^*＝G_jThen, the extreme value is close to the optimal extreme value, and the iteration is stopped;

the G is^*Comprises the following steps: stopping the optimal position of the particles during iteration, namely the optimal video layer rate distribution scheme selected under the constraint condition of the minimum time delay function, otherwise, repeating the steps until the maximum iteration times is reached.

Finally, the above embodiments are only intended to illustrate the technical solutions of the present invention and not to limit the present invention, and although the present invention has been described in detail with reference to the preferred embodiments, it will be understood by those skilled in the art that modifications or equivalent substitutions may be made on the technical solutions of the present invention without departing from the spirit and scope of the technical solutions, and all of them should be covered by the claims of the present invention.