CN117560327A - Burst traffic oriented service quality adjustment method under limited network - Google Patents
Burst traffic oriented service quality adjustment method under limited network Download PDFInfo
- Publication number
- CN117560327A CN117560327A CN202311815910.0A CN202311815910A CN117560327A CN 117560327 A CN117560327 A CN 117560327A CN 202311815910 A CN202311815910 A CN 202311815910A CN 117560327 A CN117560327 A CN 117560327A
- Authority
- CN
- China
- Prior art keywords
- data
- burst
- traffic
- user
- transmission
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 39
- 230000005540 biological transmission Effects 0.000 claims abstract description 55
- 238000006731 degradation reaction Methods 0.000 claims abstract description 22
- 230000015556 catabolic process Effects 0.000 claims abstract description 19
- 230000007246 mechanism Effects 0.000 claims abstract description 12
- 230000003139 buffering effect Effects 0.000 claims abstract description 5
- 238000007906 compression Methods 0.000 claims description 23
- 230000006835 compression Effects 0.000 claims description 23
- 238000013144 data compression Methods 0.000 claims description 16
- 238000012549 training Methods 0.000 claims description 13
- 230000008569 process Effects 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 7
- 230000000593 degrading effect Effects 0.000 claims description 7
- 238000013461 design Methods 0.000 claims description 5
- 238000005070 sampling Methods 0.000 claims description 5
- 230000003993 interaction Effects 0.000 claims description 4
- 238000012544 monitoring process Methods 0.000 claims description 3
- 238000012545 processing Methods 0.000 claims description 3
- 230000006870 function Effects 0.000 description 16
- 230000006399 behavior Effects 0.000 description 12
- 238000005516 engineering process Methods 0.000 description 3
- 238000004891 communication Methods 0.000 description 2
- 230000001133 acceleration Effects 0.000 description 1
- 230000003542 behavioural effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000003111 delayed effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000013467 fragmentation Methods 0.000 description 1
- 238000006062 fragmentation reaction Methods 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000000638 stimulation Effects 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L47/00—Traffic control in data switching networks
- H04L47/70—Admission control; Resource allocation
- H04L47/74—Admission control; Resource allocation measures in reaction to resource unavailability
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L47/00—Traffic control in data switching networks
- H04L47/70—Admission control; Resource allocation
- H04L47/83—Admission control; Resource allocation based on usage prediction
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention discloses a burst traffic-oriented service quality adjustment method under a limited network, which predicts burst traffic conditions among cloud edges by designing a model for predicting the burst traffic of data among the cloud edges based on users; when the burst traffic is predicted to far exceed the available bandwidth among cloud edges, a mechanism for calculating the optimal degradation proportion is designed, and degradation is carried out on transmission data of different modes, so that a user is ensured to receive data resources after the degradation in an acceptable range; and (3) designing a buffer retransmission mechanism, when the degraded data still is difficult to send, measuring the importance of the data, buffering the data with lower importance, and continuing to transmit when the network is better, so that the cloud-edge data resource integrated cooperative application capability is improved.
Description
Technical Field
The invention belongs to the field of cloud edge coordination of a data center, and particularly relates to a burst traffic-oriented service quality self-adaptive adjustment method under a limited network.
Background
Along with the arrival of the information age, the demands of the end users on the cloud for data and services are increasing, and the traditional cloud-end architecture transmits the demands of the users on the data and services to the cloud and then transmits the access results of the data and services on the cloud to the users, so that the requirements of the end users on the data and services in high real-time are limited by the states of available network communication resources between the cloud, and the requirements of the end users on the data and services are difficult to meet. The information processing systems in the current fields mostly adopt a cloud-edge-end three-level architecture, and edge nodes close to end users are utilized to provide data and service guarantee for users nearby, but due to limited storage and calculation capacity of the edge nodes, quick interaction and cooperative application of cloud edge data resources still need to be considered.
At present, most of methods for guaranteeing the transmission quality of data resources in a limited network are based on a reliable UDP technology, and the transmission efficiency of the data resources is improved by counting the round trip time of data packets and adaptively adjusting the size and the transmission rate of data packets. The method has obvious effect on the data flow with low change rate, but for the burst flow generated by the temporary requirement of the end user between cloud edges, the adjustment of the data packet size and the sending rate generally needs a plurality of data packet round trip periods, and when the adjustment is finished, the burst flow is already transmitted, so that the data transmission adjustment process under the limited network is far delayed from the actual requirement, and the network is congested. Therefore, a mechanism capable of predicting the burst traffic transmission time and transmission node information between cloud edge data centers is required to be designed, and a set of service quality self-adaptive adjustment method is specially formulated for the burst traffic.
Disclosure of Invention
The invention aims to provide a burst traffic-oriented service quality adjustment method under a limited network, which provides reliability guarantee for transmission of burst traffic among cloud edges from three aspects of cloud edge burst traffic prediction, flexible data degradation and cache retransmission control and optimizes the efficiency of cloud edge cooperative application.
The technical solution for achieving the purpose of the invention comprises the following steps.
A burst traffic oriented service quality adjustment method under a limited network includes:
step S1, a mechanism for predicting burst traffic between cloud edges according to the requirements of end users is designed:
predicting burst flow L appearing between cloud edges according to inherent attributes and behavior records of an end user group P and event time T, constructing a burst flow prediction model, training the burst flow prediction model, and predicting cloud edge burst flow L in the next stage of the current time of an event according to the trained prediction model;
step S2, designing a strategy for flexibly degrading the burst traffic according to the network state:
after the cloud edge burst flow L is predicted, comparing the cloud edge burst flow L with a cloud edge network state D, calculating the percentage difference between available bandwidth and actual interaction flow between cloud edges, degrading the burst flow according to the difference and the mode of the flow to be transmitted, and transmitting next-level data under the condition of meeting the basic requirement of a user so as to achieve the balance of the requirement of the user and the bandwidth limitation; when L is less than or equal to D, data degradation is not needed; when L is more than D, monitoring whether the user receives the identification of the degraded data resource, if so, quantifying the minimum requirement of the user on the data resource, executing the degradation process, and if not, skipping the step;
step S3, designing a mechanism for automatically caching burst traffic according to the network state: and counting historical data application results, comparing the importance degree of various data in the burst traffic to the user, selecting data with lower priority to buffer the data, and continuing to send the buffered data after the transmission of the burst traffic is finished, so that the smooth transmission of the burst traffic is realized.
Compared with the prior art, the invention has the remarkable advantages that:
(1) According to the method for predicting the burst flow by constructing the model according to the information such as the terminal user behavior and the environment, which is provided by the invention, the problem that the burst flow is difficult to deal with in the traditional mode of adjusting the data transmission strategy based on the packet round trip time can be solved, and the cloud edge burst flow generated by the temporary requirement of the terminal user can be better ensured not to influence the cloud edge cooperative efficiency.
(2) The burst flow flexible degradation method provided in the step 2 is different from the mode that the current control of the burst flow is mainly strategy adjustment such as data fragmentation and packetization, link congestion control and the like at the transmission bottom layer, and is directly oriented to the requirements of the end user to control the transmission data layer. The adjustment of the bottom layer transmission strategy is limited, the overlarge burst flow is oriented, the congestion of the cloud edge network still cannot be avoided by the control of the transmission strategy, but the flexible degradation of the burst flow can be within the acceptable range of a user, the burst flow among cloud edges is reduced to the greatest extent, and the stable operation of the cloud edge end system is ensured.
(3) According to the burst flow automatic caching method provided by the invention in the step 3, the information which is relatively unimportant and has low real-time requirement in the burst flow is cached for delay transmission according to the user requirement, so that the stimulation of the behavior of the burst flow on the cloud side communication network can be greatly reduced, and the stable treatment of the burst flow is realized. And the method is different from the current method for judging the importance of the data, wherein the method adopts manual work to give priority to each piece of information, and in the step (3), the importance of the information is automatically judged according to the consumption condition of the data resources by the end user, so that the priority transmission of the important information in the burst flow is realized more accurately.
Drawings
Fig. 1 is a flow chart of cloud edge burst traffic prediction for an end user according to the present invention.
Fig. 2 is a flow chart of flexible degradation and automatic caching for cloud edge burst traffic.
Detailed Description
The invention is further described with reference to the drawings and specific embodiments.
With reference to fig. 1 and fig. 2, the method for adjusting the service quality for the burst traffic under the limited network according to the invention realizes the prediction of the burst traffic of the cloud edge by analyzing and modeling the behaviors, the environments, the experience events and the cloud edge traffic conditions of the end user, and solves the problems of congestion of the cloud edge network and difficult cooperation of the cloud edge data center caused by the burst traffic according to a user demand design mechanism, and comprises the following steps:
step S1, a mechanism for predicting burst traffic between cloud edges according to the requirements of end users is designed.
In order to provide sufficient and real-time data and service support for end users, cloud edges need to cooperate and interact with each other. Under the conventional condition, the traffic interaction between cloud edges is stable and controllable, and the cloud edge limited network can execute the traffic transmission process, so that the cloud edge cooperative efficiency is ensured. However, when a user encounters an emergency, a large amount of data resources are required to be called, the cloud edge flow is rapidly increased, and when the underlying transmission strategy adjustment mechanism is difficult to reflect in time, the cloud edge network is congested. The bursty traffic L occurring between cloud edges needs to be predicted according to the end user group P and the event time T (refer to the total time from the event start node to the current node), and the above needs are to be met by constructing, training and applying the bursty traffic prediction model m= { P, T, L }.
Step S101, constructing a burst traffic prediction model m= { P, T, L }.
The end user population P may be considered as a series of end users { P } that are logging in and using the system 1 ,P 2 ...P h The more the number of users, the higher the authority of the users and the more frequent the user operation, the greater the possibility of generating burst flow between cloud edges and the more the flow in burst; the users generally have similar operations when the events of the same type are in the same event stage, and the induced cloud edge flows have similarity. In summary, the design formula is as follows:
L＝f p (P)+f T (T)，
wherein L is the predicted burst traffic, f p As the influence function of the user group P on the burst flow, f T Is the influence function of the event time T on the burst flow.
Since model M is used to predict bursty traffic, in formula l=f p (P)+f T In (T), for user group P, user group P may be represented by weighting the impact of all intrinsic properties and behavior records of all logged-in users on bursty traffic.
According to the end user group P, h end users { P { for all current logins 1 ,P 2 ...P h The characteristic of the composition, decompose P into the weight of all end users, design formula is as follows:
for the kth user P in the user group k It can be considered to be a weighted set of a series of inherent attributes and behavior records, user P k The impact on bursty traffic is equivalent to the weighting of the bursty traffic impact by the user's intrinsic properties and behavior records, and will be the kthUser P k N intrinsic properties of (1) are denoted as { g1, g2...gn }, the kth user P is taken as k The m behavioral records of (1, 2.) are denoted as { x1, x2. The influence factor of the ith intrinsic attribute in the n intrinsic attributes on the burst traffic is recorded as Q gi Kth user P k The weight of the ith intrinsic property is denoted as W gi The method comprises the steps of carrying out a first treatment on the surface of the The influence factor of the jth behavior record in the m behavior records on the burst traffic is recorded as Q xj Kth user P k The weight recorded for this jth behavior is denoted as W xj . The kth user P of the final design k The impact formula on burst traffic is as follows:
the corresponding formula of the event moment T is relatively simple, and the current time millisecond number is T now Event start time t start Then t=t now -t start 。
Step S102, training of the burst traffic prediction model m= { P, T, L }.
In the formula of step S101, W gi 、W xj Is the weight of the inherent attribute and the behavior label of the user, and the type and the weight W for describing the user label can be determined according to the inherent information such as the seat, the identity and the like of the user and the behavior actions such as the preference, the browsing record, the operation record and the like of the user information by referring to the user portrait technology gi 、W xj The expert can set the tag-burst flow influence factor Q according to the influence of different tags on the information flow gi 、Q xj Finally, calculating the weighted label type, the label weight and the label-burst flow influence factor to obtain a kth user P k Accumulating all login users at the current time point to finally obtain a user group
The burst flow L, the user group P and the event time T at certain time in the actual running process of the acquisition system are taken as sampling points to obtain a training set(training set data is distinguished by an upper-line, and the subscript of the training set refers to the sample data of the e-th time) for training the formula l=f designed in step S101 p (P)+f T (T). The training function is designed as follows,
f P (P)＝V P /(1+exp(-a 1 *(P-b 1 )))
f T (T)＝V T /(1+exp(-a 2 *(T-b 2 )))
wherein V is P For the user group P to influence the function f on the burst traffic P Maximum value of (P), V T For the influence function f of the event time T on the burst flow T Maximum value of (T), a 1 For the user group P to influence the function f on the burst traffic P (P) slope of corresponding curve, a 2 For the influence function f of the event time T on the burst flow T (T) slope of corresponding curve, b 1 For the user group P to influence the function f on the burst traffic P Center point of (P), b 2 For the influence function f of the event time T on the burst flow T Center point of (T).
Taking the influence of users on burst traffic as an example, for a training set
Parameter V P 、a 1 、b 1 Has the following meaning of V P Based on user group PMaximum value setting, a of influence of the flow rate L 1 B is set according to the median of the influence of the user group P on the burst traffic L 1 The average slope of the influence of the user group P on the burst flow L is set according to the user group P, and the burst flow L is obtained through calculation of a training set. Parameters (parameters)User population representing the e-th sample, +.>The burst traffic for the e-th sample is represented, and q represents the total number of samples.
The loss function is designed such that,
calculation of the loss function J (P) and the parameter V using gradient descent P 、a 1 、b 1 Partial derivative of (2) The inverse update function of the parameters is as follows.
Similarly, the influence of the current process T of the event on the burst flow can be also subjected to iterative calculation, and the function f is trained T (T)＝V T /(1+exp(-a 2 *(T-b 2 ) A) the corresponding parameters calculated from the training set are as follows,
wherein the parameters areEvent time representing the e-th sample, +.>The burst traffic for the e-th sample is represented, and q represents the total number of samples.
The inverse update function of the loss function and the parameters is as follows.
Step S103, according to the trained prediction model M= { P, T, L } predicting the cloud edge burst flow L of the next stage at the current time of the event.
And step S2, designing a strategy for flexibly degrading the burst traffic according to the network state.
After the cloud edge burst flow L is predicted in the step S1, comparing the cloud edge burst flow L with the cloud edge network state D, and calculating the flexible degradation proportion of the burst flow and the degradation strategies adopted by different mode data according to the comparison result and the requirement of the user on information so as to realize the balance of the requirement of the user and the consumption of network resources. When L is less than or equal to D, data degradation is not needed; and when L is more than D, monitoring whether the user receives the identification of the degraded data resource, if so, quantifying the minimum requirement of the user on the data resource, executing the degradation process, and if not, skipping the step.
Step S201, quantifying the user demand, and calculating the downgrade ratio.
In the operation process of the cloud side system, users have demands on various types of data resources with different modes on the cloud, text and message data are difficult to process so as to reduce the flow of the cloud side, but for the data resources of video, audio and pictures on the cloud, compared with the situation that the users cannot receive the needed multimode data resources, the users can quickly receive more blurred pictures, video and audio within an acceptable range, and the cloud side system is acceptable. The first step is to quantify the demands of users as much as possible and determine the degradation proportion of various data.
At this time, the cloud-edge burst traffic L > the cloud-edge network state D, and the system distinguishes the traffic L1 that cannot be compressed in L from the traffic L2 that can be compressed. The minimum transmittable successful data compression ratio K is calculated as follows:
for various data transmitted by the cloud side system, a user can set a compression threshold M in advance according to the self requirements i . Comparing compression threshold M of transmission data i And the data compression ratio K, three cases can be distinguished. First, when all transmission data are compressed threshold M i And if the data compression ratio is smaller than or equal to the data compression ratio K, the degradation is carried out according to the ratio K. Second, when all transmission data are compressed threshold M i When the compression ratio is larger than the data compression ratio K, the compression threshold M is pressed on all compressible data i And (3) compressing, selecting data resources which initiate transmission requests in the cloud edge network state D range, transmitting the data resources, and enabling the rest data to enter a cache queue for caching in the step (S3). Third, when the compression threshold M of the partial transmission data i When the data compression ratio is smaller than the data compression ratio K, the partial data L2 is compressed by a compression threshold M i After compression, the remaining traffic L-L1-L2 is compared with the remaining network states D-L1-L2 x M i Compression ratio of (2)With the remaining data pressureShrink threshold M i And carrying out the processing modes of the three modes in an iteration mode according to the comparison result until the compression ratio of all transmission data is determined. In modality three, due to compression threshold M i The part of data L2 smaller than the compression ratio K is compressed according to the threshold value, so the compression ratio is gradually increased after each iteration, the data with the undetermined compression ratio is less and less, and finally, the comparison result of all the residual transmission data compression threshold values and the data compression ratio is converged into a mode one or a mode two, and the iteration process is terminated.
Step S202, distinguishing data modes, and performing degradation through different algorithms proportionally.
After the compression ratio of each type of data is determined in S201, different compression algorithms may be adopted according to the data mode. Common compressible data resources typically include images, audio, video. For image data, a JPEG2000 algorithm is adopted in the project, and frequency components of the image are extracted mainly through a multi-resolution coding mode mainly based on wavelet conversion. The method is characterized in that a customizable video transmission acceleration method is designed for video data, key frames in the video and key contents in each frame are extracted, blurring processing is carried out on other contents in the video, and finally video data which accords with compression proportion and reserves the key contents within an acceptable range of a user is obtained. For the audio data, the scale degradation capability of the audio data is realized by a method of reducing the sampling rate, the bit depth and the channel number.
Step S3, after degrading the transmission data within the acceptable range of the user, it can be calculated in step S2 that partial data is difficult to successfully send to the opposite side under the limited network condition, the less important data in the data to be transmitted is cached first, and the cached data is sent to the current login node of the receiving user when the cloud edge burst flow L is less than the cloud edge network state D. The functions mainly comprise construction of a cache model, determination of data to be cached and establishment of a cache data retransmission process.
Step S301, constructing a transmission data cache model to form a transmission data cache library.
The invention constructs a data caching model by utilizing related concepts such as class, relation, attribute, instance and the like in the network ontology language, and generates a caching database for storing data to be cached. The key parts of the built ontology model are as follows, wherein each piece of data in the cache model needs to record information such as the type of the data, the cache content, whether the data can be degraded, whether the transmitted data is a file or a message, the data validity period, the user serial number, the message priority, the message number, the file name and the like.
Step S302, judging all the current transmission data according to the type of the data to be transmitted and the historical transmission log, and sending the data which are relatively unimportant in the current transmission data when the network bandwidth is abundant, such as buffering the data which are relatively unimportant in the current transmission data.
The method for judging the importance of data includes recording the transmission log of various data in the system, including the type of the transmitted data, the priority of the transmitted data, the sending time of the data, the receiving time of the data, the IP of the data sending node, the IP of the data receiving node, the content of the sending message or the name of the sending file, the size of the sending message or the file, and the user serial number of the received data (if the final receiving party of the data is software, the user serial number is 0). And then according to the data transmission log, the consumption condition of different types of data by various users is statistically analyzed, and the importance degree of each type of data is calculated according to the priority of the consumption data, the consumption time and the importance degree of the consumption user, wherein the calculation formula is as follows:
wherein S is y The importance of the data of the y type is obtained by weighting the importance of all the data of the y type in the history after time attenuation, S yu Weighting importance of the U-th historical record in the y-th data, U yu Z is the importance degree corresponding to the consumption user of the nth historical record in the data of the nth class yu E is the influence factor of the priority parameter of the transmission of the nth history record in the data of the y type on the importance (the higher the priority is, the more important the information is in the transmission process) -K*tyu The time influence factor of the importance of the u-th historical transmission record in the y-th type data (the closer the transmission time is to the current time, the more important the information is). And counting the importance degree of all the current transmission data, preferentially transmitting the data with high importance degree until the network resources are occupied, buffering the data with low importance degree which are not transmitted in turn, and putting the data into a buffer library.
In step S303, when the network state becomes good and the cloud edge burst traffic L is greater than the cloud edge network state D, the cached data is resent to the corresponding user. At this time, according to the cached data in the library and the importance degree corresponding to the data types, the data with low importance degree in the cache database is rearranged, the data with relatively higher importance degree in the low importance degree data is selected and sent to the user again, and the information such as the data types, the cache content, whether the data can be degraded, whether the file is valid, the data validity period, the user serial number, the message priority, the message number, the file name and the like recorded in the cache library can support the retransmission process.
In the embodiment of the invention, the sudden flow condition between cloud edges is predicted by designing a model for predicting the sudden flow of data based on users between cloud edges; when the burst traffic is predicted to far exceed the available bandwidth among cloud edges, a mechanism for calculating the optimal degradation proportion is designed, and degradation is carried out on transmission data of different modes, so that a user is ensured to receive data resources after the degradation in an acceptable range; and (3) designing a buffer retransmission mechanism, when the degraded data still is difficult to send, measuring the importance of the data, buffering the data with lower importance, and continuing to transmit when the network is better, so that the cloud-edge data resource integrated cooperative application capability is improved.
The above-described embodiment is only one applicable mode of the present invention, but the scope of the present invention is not limited thereto. The architecture and optimization algorithm of the present invention have been illustrated and described in detail in the examples, and the present invention is susceptible to any variation without departing from the principles of the technology and is intended to be covered by the scope of the claims.
Claims (5)
1. A burst traffic oriented service quality adjustment method under a limited network is characterized by comprising the following steps:
step S1, a mechanism for predicting burst traffic between cloud edges according to the requirements of end users is designed:
predicting burst flow L appearing between cloud edges according to inherent attributes and behavior records of an end user group P and event time T, constructing a burst flow prediction model, training the burst flow prediction model, and predicting cloud edge burst flow L in the next stage of the current time of an event according to the trained prediction model;
step S2, designing a strategy for flexibly degrading the burst traffic according to the network state:
after the cloud edge burst flow L is predicted, comparing the cloud edge burst flow L with the cloud edge network state D, calculating the gap between the available bandwidth and the actual interaction flow between the cloud edges, degrading the burst flow according to the gap and the mode of the flow to be transmitted, and transmitting the next-level data under the condition of meeting the basic requirement of a user so as to achieve the balance of the requirement of the user and the bandwidth limitation; when L is less than or equal to D, data degradation is not needed; when L is more than D, monitoring whether the user receives the identification of the degraded data resource, if so, quantifying the minimum requirement of the user on the data resource, executing the degradation process, and if not, skipping the step;
step S3, designing a mechanism for automatically caching burst traffic according to the network state: and counting historical data application results, comparing the importance degree of various data in the burst traffic to the user, selecting data with lower priority to buffer the data, and continuing to send the buffered data after the transmission of the burst traffic is finished, so that the smooth transmission of the burst traffic is realized.
2. The method for adjusting the quality of service for the bursty traffic under the constrained network as claimed in claim 1, wherein the calculation formula for constructing the bursty traffic prediction model in step 1 is:
L＝f p (P)+f T (T)
wherein L is the predicted burst traffic, f p As the influence function of the user group P on the burst flow, f T The method is an influence function of the event time T on the burst flow;
the training function is designed as follows,
f P (P)＝V P /(1+exp(-a 1 *(P-b 1 )))
f T (T)＝V T /(1+exp(-a 2 *(T-b 2 )))
wherein V is P For the user group P to influence the function f on the burst traffic P Maximum value of (P), V T For the influence function f of the event time T on the burst flow T Maximum value of (T), a 1 For the user group P to influence the function f on the burst traffic P (P) slope of corresponding curve, a 2 For the influence function f of the event time T on the burst flow T (T) slope of corresponding curve, b 1 For the user group P to influence the function f on the burst traffic P Center point of (P), b 2 For the influence function f of the event time T on the burst flow T A center point of (T);
the burst flow L, the user group P and the event time T at certain time in the actual running process of the acquisition system are taken as sampling points, a loss function of the user group P and a reverse updating function of parameters are designed,
wherein the method comprises the steps ofFor the user population of the e-th sample, +.>The e-th sampling burst flow, q is the total sampling times;
event time T loss function and reverse update function of parameters:
wherein the method comprises the steps ofThe event time of the e-th sample.
3. The method for adjusting the quality of service for bursty traffic in a constrained network as claimed in claim 2, wherein P is a weight of all end users, and the design formula is as follows:
kth user P k The impact formula on burst traffic is as follows:
wherein Q is gi Is the influence factor of the ith inherent attribute in n inherent attributes on burst traffic, W gi For the kth user P k Weights for the i-th intrinsic attribute; q (Q) xj The influence of the jth behavior record in the m behavior records on the burst traffic is due toA seed; w (W) xj For the kth user P k Weights for this j-th behavior record.
4. The method for adjusting the quality of service for bursty traffic in a constrained network as claimed in claim 1, wherein when L > D, the minimum transmissible successful data compression ratio K is calculated:
wherein L1 is the incompressible flow rate and L2 is the compressible flow rate;
three modes are:
first, when all transmission data are compressed threshold M i When the data compression ratio is smaller than or equal to the data compression ratio K, degrading according to the ratio K; second, when all transmission data are compressed threshold M i When the compression ratio is larger than the data compression ratio K, the compression threshold M is pressed on all compressible data i Compressing, selecting data resources which initiate transmission requests first in the cloud edge network state D range for transmission, and enabling the rest data to enter a cache queue; third, when the compression threshold M of the partial transmission data i When the data compression ratio is smaller than the data compression ratio K, the partial data L2 is compressed by a compression threshold M i After compression, the remaining traffic L '=l-L1-L2 is compared with the remaining network state D' =d-L1-L2×m i Compression ratio of (2)With residual data compression threshold M i And carrying out the processing modes of the three modes in an iteration mode according to the comparison result until the compression ratio of all transmission data is determined.
5. The method for adjusting the quality of service for bursty traffic under a restricted network according to claim 1, wherein the method for determining the importance of data is as follows:
recording a data transmission log, statistically analyzing consumption conditions of different types of data by various users according to the data transmission log, and calculating importance of each type of data, wherein the calculation formula is as follows:
wherein S is y The importance of the data of the y type is obtained by weighting the importance of all the data of the y type in the history after time attenuation, S yu Weighting importance of the U-th historical record in the y-th data, U yu Z is the importance degree corresponding to the consumption user of the nth historical record in the data of the nth class yu E, an influence factor of priority parameters of the transmission of the nth historical record in the data of the y type on the importance degree -K*tyu A time influence factor of the ith historical transmission record in the y-th data on the importance is given;
and counting the importance degree of all the current transmission data, preferentially transmitting the data with high importance degree until the network resources are occupied, buffering the data with low importance degree which are not transmitted in turn, and putting the data into a buffer library.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311815910.0A CN117560327A (en) | 2023-12-27 | 2023-12-27 | Burst traffic oriented service quality adjustment method under limited network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311815910.0A CN117560327A (en) | 2023-12-27 | 2023-12-27 | Burst traffic oriented service quality adjustment method under limited network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117560327A true CN117560327A (en) | 2024-02-13 |
Family
ID=89820740
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311815910.0A Pending CN117560327A (en) | 2023-12-27 | 2023-12-27 | Burst traffic oriented service quality adjustment method under limited network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117560327A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117938766A (en) * | 2024-03-25 | 2024-04-26 | 中国电子科技集团公司第五十四研究所 | Hierarchical transmission method for data resources under limited network |
-
2023
- 2023-12-27 CN CN202311815910.0A patent/CN117560327A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117938766A (en) * | 2024-03-25 | 2024-04-26 | 中国电子科技集团公司第五十四研究所 | Hierarchical transmission method for data resources under limited network |
| CN117938766B (en) * | 2024-03-25 | 2024-06-04 | 中国电子科技集团公司第五十四研究所 | Hierarchical transmission method for data resources under limited network |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111431941B (en) | Real-time video code rate self-adaption method based on mobile edge calculation | |
| CN102325274B (en) | Network bandwidth-adaptive video stream transmission control method | |
| US20030165150A1 (en) | Multi-threshold smoothing | |
| CN117560327A (en) | Burst traffic oriented service quality adjustment method under limited network | |
| CN102685472B (en) | Method, device and system of data transmission | |
| JP7118209B2 (en) | Congestion control method, device and system, and computer storage medium | |
| Liubogoshchev et al. | Adaptive cloud-based extended reality: Modeling and optimization | |
| CN111611129B (en) | Performance monitoring method and device of PaaS cloud platform | |
| CN112954385A (en) | Self-adaptive shunt decision method based on control theory and data driving | |
| CN110213097B (en) | Edge service supply optimization method based on dynamic resource allocation | |
| CN113132490A (en) | MQTT protocol QoS mechanism selection scheme based on reinforcement learning | |
| CN111935025B (en) | Control method, device, equipment and medium for TCP transmission performance | |
| CN115189908B (en) | Random attack survivability evaluation method based on network digital twin | |
| CN114064394A (en) | Safety monitoring method and device based on edge calculation and terminal equipment | |
| CN114389959B (en) | Network congestion control method, device, electronic equipment and storage medium | |
| CN114040257B (en) | Self-adaptive video stream transmission playing method, device, equipment and storage medium | |
| Huang et al. | Personalized QoE enhancement for adaptive video streaming: A digital twin-assisted scheme | |
| US20230060623A1 (en) | Network improvement with reinforcement learning | |
| CN115701709A (en) | Video coding method and device, computer readable medium and electronic equipment | |
| Bhattacharyya et al. | QFlow: A learning approach to high QoE video streaming at the wireless edge | |
| CN116389375A (en) | Network queue management method, device and router for live video stream | |
| CN113852817B (en) | Self-adaptive code rate transmission server, code rate determining method, transmission system and method | |
| CN115189910A (en) | Network digital twin-based deliberate attack survivability evaluation method | |
| KR100548324B1 (en) | Video data transmission bit rate control method for mobile communication system | |
| Kwon et al. | An admission control scheme for continuous media servers using caching |
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 |