CN116963174A - Congestion level prediction method, device, computer equipment and storage medium - Google Patents

Congestion level prediction method, device, computer equipment and storage medium Download PDF

Info

Publication number
CN116963174A
CN116963174A CN202311015234.9A CN202311015234A CN116963174A CN 116963174 A CN116963174 A CN 116963174A CN 202311015234 A CN202311015234 A CN 202311015234A CN 116963174 A CN116963174 A CN 116963174A
Authority
CN
China
Prior art keywords
model
sample
sub
state information
congestion level
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
Application number
CN202311015234.9A
Other languages
Chinese (zh)
Inventor
武晓鸽
戴国华
谭华
马晓亮
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
China Telecom Technology Innovation Center
China Telecom Corp Ltd
Original Assignee
China Telecom Technology Innovation Center
China Telecom Corp Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by China Telecom Technology Innovation Center, China Telecom Corp Ltd filed Critical China Telecom Technology Innovation Center
Priority to CN202311015234.9A priority Critical patent/CN116963174A/en
Publication of CN116963174A publication Critical patent/CN116963174A/en
Pending legal-status Critical Current

Links

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W28/00—Network traffic management; Network resource management
    • H04W28/02—Traffic management, e.g. flow control or congestion control
    • H04W28/0284—Traffic management, e.g. flow control or congestion control detecting congestion or overload during communication
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/14—Network analysis or design
    • H04L41/149—Network analysis or design for prediction of maintenance
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L43/00—Arrangements for monitoring or testing data switching networks
    • H04L43/08—Monitoring or testing based on specific metrics, e.g. QoS, energy consumption or environmental parameters
    • H04L43/0876—Network utilisation, e.g. volume of load or congestion level
    • H04L43/0882—Utilisation of link capacity

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Environmental & Geological Engineering (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The present application relates to the field of communications technologies, and in particular, to a congestion level prediction method, apparatus, computer device, and storage medium. The method comprises the following steps: acquiring historical state information of each historical moment in a target time window before a predicted moment of a target node; inputting each history state information into the Gaussian mixture model to obtain the membership degree of each Gaussian sub-model in the Gaussian mixture model of the target node at the prediction moment; selecting a target submodel from the Gaussian submodels according to each membership degree; and determining the congestion level of the target node at the predicted moment according to the congestion level corresponding to the target sub-model. The application can improve the congestion detection precision.

Description

Congestion level prediction method, device, computer equipment and storage medium
Technical Field
The present application relates to the field of communications technologies, and in particular, to a congestion level prediction method, apparatus, computer device, and storage medium.
Background
With the rapid development of fifth generation (The Fifth Generation, 5G) networks, vertical applications of the internet of things (Internet of Things, ioT) are receiving increasing attention in the research and industrial fields. In particular, industrial internet of things (IIoT) is considered as a powerful candidate technology to meet production automation efficiency requirements.
In IIoT systems, large-scale industrial equipment generates large amounts of real-time data that should be processed under stringent constraints. Unexpected delays/task interruptions can lead to significant safety issues and significant production losses. Traditionally, industrial data is transmitted to a centralized cloud server (Central Cloud Server, CCS) for processing. However, considering the distributed topology of large-scale IIoT devices, CCS-based processing is difficult to meet the latency requirements of time-sensitive services. To address this issue, mobile Edge computing (Mobile Edge Computing, MEC) reduces data transfer latency by allowing IIoT devices to offload processing tasks to nearby Edge Servers (ESs), which have limited computational power compared to CCS, but can shorten the response time of the tasks, thereby ensuring the security and effectiveness of real-time task-based IIoT systems.
Meanwhile, considering that mass data are generated by large-scale industrial equipment, the problem of congestion state prediction in the end network cooperation process is brought into the academic and industrial attention, and notably, considering the special structure of the IIoT end network cooperation distribution, one node is congested due to excessive burst data quantity or overload of data load, and the whole network is crashed. Thus, flexible monitoring detection of congestion is required.
However, the conventional congestion state prediction method often presents a problem of low accuracy in detecting the network state, so improvement is needed.
Disclosure of Invention
In view of the foregoing, it is desirable to provide a congestion level prediction method, apparatus, computer device, and storage medium capable of improving congestion detection accuracy.
In a first aspect, the present application provides a congestion level determination method, the method comprising:
acquiring historical state information of each historical moment in a target time window before a predicted moment of a target node;
inputting each history state information into the Gaussian mixture model to obtain the membership degree of each Gaussian sub-model in the Gaussian mixture model of the target node at the prediction moment;
selecting a target submodel from the Gaussian submodels according to each membership degree;
and determining the congestion level of the target node at the predicted moment according to the congestion level corresponding to the target sub-model.
In one embodiment, the Gaussian mixture model is constructed by:
acquiring a first training sample and a second training sample; the first training sample comprises first sample state information and sample congestion level corresponding to the first sample state information, and the second training sample comprises second sample state information;
Determining a congestion level label of second sample state information in a second training sample according to the first training sample;
training each initial sub-model according to the first training sample and the second training sample of the determined congestion level label to obtain Gaussian sub-models corresponding to the initial sub-models;
and constructing a Gaussian mixture model according to each Gaussian sub-model.
In one embodiment, training each initial sub-model according to the first training sample and the second training sample of the determined congestion level label to obtain a gaussian sub-model corresponding to each initial sub-model, including:
grouping the first training samples and the second training samples with determined congestion level labels to obtain sample groups corresponding to each congestion level;
assigning a sample set to each initial sub-model;
acquiring initialization parameters of each initial sub-model;
training each initial sub-model according to the initialization parameters and the sample set corresponding to each initial sub-model to obtain Gaussian sub-models corresponding to each initial sub-model.
In one embodiment, training each initial sub-model according to the initialization parameters and the sample set corresponding to each initial sub-model to obtain a gaussian sub-model corresponding to each initial sub-model, including:
And for each initial sub-model, adopting a maximum expected algorithm, and carrying out iterative updating on the initialization parameters of the initial sub-model according to a sample group corresponding to the initial sub-model to obtain a Gaussian sub-model corresponding to the initial sub-model.
In one embodiment, obtaining initialization parameters of each initial sub-model includes:
for each initial sub-model, determining an initial mean value of the initial sub-model according to each training sample in a sample group corresponding to the initial sub-model;
determining an initial covariance of the initial sub-model according to the initial mean vector of the initial sub-model and a sample group corresponding to the initial sub-model;
determining initial occurrence probability of the initial sub-model in the Gaussian mixture model according to the total number of samples and the number of samples of training samples in a sample group corresponding to the initial sub-model; wherein the total number of samples is the sum of the number of samples of the first training sample and the number of samples of the second training sample.
In one embodiment, determining the congestion level label of the second sample state information in the second training sample according to the first training sample includes:
for any second sample state information, determining the similarity between the second sample state information and each first sample state information;
And determining the sample congestion level corresponding to the first sample state information with the maximum similarity of the second sample state information as a congestion level label of the second sample state information.
In one embodiment, obtaining historical state information of each historical time in the target time window before the predicted time by the target node includes:
acquiring historical dimension parameters of each historical moment in a target time window before a predicted moment of a target node;
and determining the historical dimension parameters meeting the prediction performance as historical state information.
In one embodiment, the historical node dimension parameter comprises at least one of an instantaneous queue size, an average sending rate, an arrival rate, a link usage, and a buffer occupancy.
In a second aspect, the present application also provides a congestion level determination apparatus, including:
the acquisition module is used for acquiring the historical state information of each historical moment in the target time window before the predicted moment of the target node;
the membership calculation module is used for inputting each history state information into the Gaussian mixture model to obtain the membership of each Gaussian sub-model in the Gaussian mixture model of the target node at the prediction moment;
The selection module is used for selecting a target submodel from the Gaussian submodels according to each membership degree;
and the prediction module is used for determining the congestion level of the target node at the prediction moment according to the congestion level corresponding to the target sub-model.
In a third aspect, the present application also provides a computer device comprising a memory and a processor, the memory storing a computer program, the processor implementing the following steps when executing the computer program:
acquiring historical state information of each historical moment in a target time window before a predicted moment of a target node;
inputting each history state information into the Gaussian mixture model to obtain the membership degree of each Gaussian sub-model in the Gaussian mixture model of the target node at the prediction moment;
selecting a target submodel from the Gaussian submodels according to each membership degree;
and determining the congestion level of the target node at the predicted moment according to the congestion level corresponding to the target sub-model.
In a fourth aspect, the present application also provides a computer readable storage medium having stored thereon a computer program which when executed by a processor performs the steps of:
acquiring historical state information of each historical moment in a target time window before a predicted moment of a target node;
Inputting each history state information into the Gaussian mixture model to obtain the membership degree of each Gaussian sub-model in the Gaussian mixture model of the target node at the prediction moment;
selecting a target submodel from the Gaussian submodels according to each membership degree;
and determining the congestion level of the target node at the predicted moment according to the congestion level corresponding to the target sub-model.
In a fifth aspect, the application also provides a computer program product comprising a computer program which, when executed by a processor, performs the steps of:
acquiring historical state information of each historical moment in a target time window before a predicted moment of a target node;
inputting each history state information into the Gaussian mixture model to obtain the membership degree of each Gaussian sub-model in the Gaussian mixture model of the target node at the prediction moment;
selecting a target submodel from the Gaussian submodels according to each membership degree;
and determining the congestion level of the target node at the predicted moment according to the congestion level corresponding to the target sub-model.
According to the congestion level prediction method, the congestion level prediction device, the computer equipment and the storage medium, the historical state information of each historical moment in the target time window before the prediction moment is obtained, the congestion state of the prediction moment is determined according to the historical state information of the target node before the prediction moment, the correlation between the congestion state and time is fully considered in the prediction mode, the high-efficiency and high-sensitivity congestion detection can be realized through deep mining of the relation between the congestion state and the time, and the timeliness of the congestion detection is improved; by utilizing the statistical characteristics of all Gaussian sub-models in the Gaussian mixture model, the membership degree of the statistical characteristics of all historical state information corresponding to the target node belonging to all Gaussian sub-models can be determined, the target sub-model corresponding to all the historical state information corresponding to the target node is determined through all the membership degrees, and then the congestion level of the target node is indirectly determined.
Drawings
FIG. 1 is an application environment diagram of a congestion level determination method in one embodiment;
fig. 2 is a flow chart of a congestion level determination method in one embodiment;
FIG. 3 is a schematic flow diagram of constructing a Gaussian mixture model in one embodiment;
FIG. 4 is a flow diagram of determining initialization parameters in one embodiment;
fig. 5 is a block diagram of the structure of the congestion level determination apparatus in one embodiment;
fig. 6 is an internal structural diagram of a computer device in one embodiment.
Detailed Description
The present application will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present application more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the application.
With the rapid development of fifth generation (The Fifth Generation, 5G) networks, vertical applications of the internet of things (Internet of Things, ioT) are receiving increasing attention in the research and industrial fields. In particular, industrial internet of things (IIoT) is considered as a powerful candidate technology to meet production automation efficiency requirements.
In IIoT systems, large-scale industrial equipment generates large amounts of real-time data that should be processed under stringent constraints. Unexpected delays/task interruptions can lead to significant safety issues and significant production losses. In the conventional technology, industrial data is transmitted to a centralized cloud server (Central Cloud Server, CCS) for processing. However, considering the distributed topology of large-scale IIoT devices, the processing based on the central node CCS is difficult to meet the latency requirements of time sensitive services. To address this issue, mobile Edge computing (Mobile Edge Computing, MEC) reduces data transfer delay by allowing IIoT devices to offload processing tasks to nearby Edge Servers (ES), which have limited computing power compared to the central node CCS, but can shorten the response time of the tasks, thereby ensuring the security and effectiveness of real-time task-based IIoT systems.
The IIoT network intelligent manufacturing end Bian Yun collaboration scenario is schematically shown in fig. 1. The IIoT has a large number of devices, and generates real-time data and calculation tasks. Because of the limitation of the terminal equipment memory resources, the terminal needs to offload its computing tasks to the edge nodes and the central cloud nodes. Assuming that only one central node CCS, N is present in the system ES Edge nodes ESs and N EN For each IIoT terminal equipment ENs, for the network state detection unit, the node that needs to detect whether there is congestion is N EN ×(N ES +1) pieces. In this embodiment, because the service characteristics of end-network coordination are considered in this embodiment, a network state detection unit for determining a congestion level is deployed at a central node CCS to monitor, thereby assisting in traffic task scheduling of the whole network, or in EN, thereby assisting in path selection of subsequent self task transmission.
In one embodiment, as shown in fig. 2, there is provided a congestion level determining method, which is described by taking as an example that the method is applied to the central node CCS or the terminal device EN in fig. 1, including the steps of:
s201, acquiring history state information of each history time in a target time window before the predicted time of the target node.
Specifically, the total number of samples of the history state information detected by the target node at each history time is defined as T, and in consideration of time correlation, in this embodiment, a sequence with a length of Q is taken as a target time window, a sequence with a length of Q is taken as each history state information, and each history state information in the target time window is taken as a detection feature X.
Specifically, for the first node (target node in this embodiment), the detection feature X corresponding to the t-th sampling time (predicted time in this embodiment) is detected l t Can be represented by formula (1):
wherein X is l t Is a column vector of dimension Q x 1, wherein the elementsIs the state information of node l at a particular time instant (i.e., t-Q-1+q).
It can be understood that by controlling Q, the detection feature X of the target node forms a sliding window sequence based on time sequence, and the time t < Q can be regarded as the network initialization time, and the network traffic is reduced at this time, so that congestion judgment can not be performed.
S202, inputting the historical state information into the Gaussian mixture model to obtain the membership degree of each Gaussian sub-model in the Gaussian mixture model of the target node at the prediction moment.
The gaussian mixture model refers to an open source mixture model, and the gaussian mixture model includes a plurality of single gaussian models (i.e., gaussian sub-models).
Specifically, each gaussian sub-model in the high mixture model corresponds to a congestion level.
Optionally, considering the corresponding relation between the detection feature X and the congestion state, in this embodiment, it is assumed that the congestion state has L level intervals, the highest level L is that the node is in the congestion state, the lowest level 1 indicates that the node is in the idle/no-congestion state, and the first interval is defined as I (l) Where L ε {1,2, …, L }. Further, L pieces ofThe set of detection features X of the section marks is defined as i= { I, respectively (1) ,…,I (L) }. The design goal of this embodiment is to optimize the processDividing into the corresponding detection characteristic X intervals. In addition, when L>2, in this embodiment, a detection result of the network node/the current network state is given, when l=2, this embodiment may be simplified to simply determine whether the network node is congested, and by considering the multi-level network congestion state, the robustness of congestion detection may be effectively improved, thereby assisting flexible scheduling of subsequent network resources.
Specifically, a Gaussian mixture model is utilized to obtain the membership degree of each Gaussian sub-model in the Gaussian mixture model of the target node at the prediction moment, and the principle is as follows:
the present embodiment models the congestion detection problem in consideration of the statistical characteristics of the node congestion state in the entire network. Given training data set (comprising a plurality of ) Due to->Is the network status characteristics of different nodes at different moments and can be regarded as training data +.>Are independent of each other.
Assume thatGenerated from the gaussian mixture model in formula (2) (i.e. satisfying the following distribution):
wherein alpha is l Is the coefficient of Gaussian mixture model, alpha is more than or equal to 0 l Is less than or equal to 1Mu and sigma are the mean and covariance matrices, respectively, of the gaussian mixture model.
Based on equation (2), the Gaussian mixture model can be divided into L Gaussian mixture components (Gaussian Mixture Component, GMC), wherein the mean and covariance matrices of the first GMC (Gaussian sub-model) are μ, respectively l Sum sigma l ，A gaussian distribution probability density function that is a gaussian sub-model, as shown in equation (3):
wherein the gaussian distribution in equation (3) can be replaced with an arbitrary distribution to generate a specific hybrid model.
Based on the characteristics of the gaussian mixture model, it can be assumed that the training data is generated by the following process: first, based on the probability alpha l Selecting the first GMC with a mean value of mu l The covariance matrix is Σ l The method comprises the steps of carrying out a first treatment on the surface of the Second, based on the probability distribution of the first GMC, data is generated
Thereby definingFor data->The probability generated by the first GMC (i.e. the detection feature +.>Corresponding to the first congestion level interval), which is a target parameter based on the system parameter estimation of the gaussian mixture model, representing the detected characteristic data +. >Gaussian membership to the first congestion state interval. />Can be represented by formula (4):
wherein, the liquid crystal display device comprises a liquid crystal display device,from the parameter set Ω= { (α) of the gaussian mixture model 1 ,μ 1 ,Σ 1 ),(α 2 ,μ 2 ,Σ 2 ),…,(α L ,μ L ,Σ L ) A decision where Ω can be iteratively estimated by using the expectation maximization (EM, expectation Maximization) algorithm, leveraging the marked and unmarked data.
S203, selecting a target submodel from the Gaussian submodels according to each membership degree.
Specifically, the gaussian sub-model with the highest membership is determined as the target sub-model.
S204, determining the congestion level of the target node at the predicted moment according to the congestion level corresponding to the target sub-model.
Specifically, the congestion level corresponding to the target submodel is determined as the congestion level of the target node at the predicted time.
In the congestion level determining method, the historical state information of each historical moment in the target time window before the predicted moment is obtained, so that the congestion state of the predicted moment is determined according to the historical state information of the target node before the predicted moment, the correlation between the congestion state and time is fully considered in the prediction mode, and the high-efficiency and high-sensitivity congestion detection can be realized by deeply mining the relation between the congestion state and the time, so that the timeliness of the congestion detection is improved; by utilizing the statistical characteristics of all Gaussian sub-models in the Gaussian mixture model, the membership degree of the statistical characteristics of all historical state information corresponding to the target node belonging to all Gaussian sub-models can be determined, the target sub-model corresponding to all the historical state information corresponding to the target node is determined through all the membership degrees, and then the congestion level of the target node is indirectly determined.
In one embodiment, as shown in FIG. 3, the Gaussian mixture model is constructed by:
s301, a first training sample and a second training sample are obtained.
The first training sample comprises first sample state information and sample congestion level corresponding to the first sample state information, and the second training sample comprises second sample state information.
It can be appreciated that full-supervised learning requires a large amount of marked data to assist in training, and the sampling cost of full-supervised learning is higher than that of half-supervised learning, so that the full-supervised learning is more suitable for application scenarios with low sampling cost. However, in the actual IIoT communication system, since the network traffic is not fixed and a large amount of time is required to be sampled, it is generally difficult to meet the requirement of low sampling cost in a practical situation. Therefore, the application of the technology based on the full supervised learning in the IIoT communication system in the actual scene is greatly limited. Considering IIoT sampling cost, under the condition of limited cost, only a small amount of marked data, namely detection characteristics of known congestion state, can be obtained generally; while the amount of unlabeled data will be much larger than the amount of labeled data. Assume that the number of marked data is N L Number of unlabeled data is N U Wherein N is L +N U =n. Definition p L To make the marked data be the percentage of the total training data, then there isFurther, the training data is divided into marked data sets D L (first training sample) and unlabeled data set D U The two parts (second training sample) are respectively shown as the formula (5) and the formula (6):
wherein, the liquid crystal display device comprises a liquid crystal display device,for the pre-measured +.>Corresponding authentic flag,/->
S302, determining a congestion level label of second sample state information in a second training sample according to the first training sample.
Specifically, for any second sample state information, determining the similarity between the second sample state information and each first sample state information; and determining the sample congestion level corresponding to the first sample state information with the maximum similarity of the second sample state information as a congestion level label of the second sample state information.
Illustratively, the marked dataset D of equation (5) is collected L In (a)Form the first interval, where L e {1,2, …, L }. Specifically, for the first interval, collect the satisfaction +.>Is->Constructing a set of detection feature vectors(first sample status information at the congestion level /), wherein ∈>The number of elements is->
For the j th U Each not marked with a congestion level label(i.e. x ju Is j th U Second sample state information in the second training sample is not marked), the similarity between the second sample state information in the second training sample and the first sample state information under each congestion level is determined, when the sample congestion level corresponding to the first sample state information with the maximum similarity of the second sample state information is determined as the congestion level label corresponding to the second training sample, classification is performed according to the following formula (7) (namely, the congestion level label corresponding to the second sample state information is determined by the formula (7),
wherein, defineIs->Temporary index of (i), i.e.) * The congestion level label of the second sample state information, equation (7) is used to calculate the difference between the second sample state information and each first sample state information under each congestion level, and select the congestion level with the smallest difference (i.e. the highest similarity) from each difference. For all N U Personal->(second sample State information in second training sample) after classification, the number of unlabeled data contained in each section can be obtained>And will meet +.>Forms a detection vector set +.>
It will be appreciated that in this case, each interval contains both marked and unmarked detection vectors, and the set of these two types of vectors is defined as There is->Wherein the set->The number of the elements is->Wherein superscript 0 denotes an initialization parameter; thus, each interval corresponds to an initialization parameter.
And S303, training each initial sub-model according to the first training sample and the second training sample of the determined congestion level label to obtain Gaussian sub-models corresponding to each initial sub-model.
Specifically, grouping the first training samples and the second training samples with determined congestion level labels to obtain sample groups corresponding to each congestion level; assigning a sample set to each initial sub-model; acquiring initialization parameters of each initial sub-model; training each initial sub-model according to the initialization parameters and the sample set corresponding to each initial sub-model to obtain Gaussian sub-models corresponding to each initial sub-model.
Wherein the sample group corresponding to each congestion level is the vector set corresponding to each intervalOptionally, the initialization parameter of each initial sub-model is the initialization parameter corresponding to each interval.
Optionally, as shown in fig. 4, obtaining the initialization parameters of each initial sub-model includes:
s401, determining an initial mean value of each initial sub-model according to each training sample in a sample group corresponding to the initial sub-model.
Specifically, for any initial sub-model, the initial mean value of the initial sub-model is determined by the following formula (8):
s402, determining initial covariance of the initial sub-model according to the initial mean vector of the initial sub-model and a sample group corresponding to the initial sub-model.
Specifically, for any initial sub-model, the initial covariance of the initial sub-model is determined using the following equation (9):
s403, determining the initial occurrence probability of the initial sub-model in the Gaussian mixture model according to the total number of samples and the number of samples of training samples in the sample group corresponding to the initial sub-model.
Wherein, for any initial sub-model, the initial occurrence probability of the initial sub-model is determined by the following formula (10):
wherein the total number of samples N is the sum of the number of samples of the first training sample and the number of samples of the second training sample.
Training each initial sub-model according to the initialization parameters and the sample group corresponding to each initial sub-model to obtain a Gaussian sub-model corresponding to each initial sub-model, wherein the method comprises the following steps:
and for each initial sub-model, adopting a maximum expected algorithm, and carrying out iterative updating on the initialization parameters of the initial sub-model according to a sample group corresponding to the initial sub-model to obtain a Gaussian sub-model corresponding to the initial sub-model.
For any initial sub-model, based on the aforementioned initialization parameters, the next step may be to estimate the parameter Ω= { (α) using existing EM algorithms 1 ,μ 1 ,Σ 1 ),(α 2 ,μ 2 ,Σ 2 ),…,(α L ,μ L ,Σ L ) }. Specifically, the parameter defining the first GMC is Ω l ＝(α l ,μ l ,Σ l ) Order-makingThe iterative estimation process can be expressed as:
initializing: let the iteration number n=0, obtain based on equations (8) - (10)
E, step E: based on current parametersWherein L ε {1, …, L }, the Gaussian membership degree ∈>
M steps: let n=n+1, based onUpdating parameters:
repeating the step E and the step M until the following conditions are met:
where L ε {1,2, …, L }, ε is a small positive number.
S304, constructing a Gaussian mixture model according to each Gaussian sub model.
Specifically, based on the above formulas (11) to (15),
finally, outputting a parameter estimation result as a formula (16) to obtain a Gaussian mixture model:
in this embodiment, by fully considering the time correlation existing in the flow unloading and scheduling process in the end network coordination process, the detection feature is constructed, meanwhile, in order to improve the network scheduling flexibility, the embodiment establishes a multi-level network congestion state mechanism and a corresponding gaussian sub-model, constructs a gaussian mixture model of the relation between the time correlation detection feature vector and the congestion state, and fully uses unlabeled and marked data, thereby completing the semi-supervised training process of the gaussian mixture model.
In one embodiment, the present embodiment provides an alternative way to obtain the history state information of each history time in the target time window before the predicted time of the target node, that is, provides a way to refine S201. The specific implementation process can comprise the following steps: acquiring historical dimension parameters of each historical moment in a target time window before a predicted moment of a target node; and determining the historical dimension parameters meeting the prediction performance as historical state information.
The historical node dimension parameters comprise at least one of instantaneous queue size, average sending rate, arrival rate, link utilization and buffer occupancy.
Specifically, the prediction performance is a preset precision condition; in this embodiment, before a gaussian mixture model is built, prediction precision of different gaussian mixture models built by historical dimension parameters is obtained, and according to the prediction precision of the different gaussian mixture models, the historical dimension parameters corresponding to the gaussian mixture models with precision meeting a precision threshold are determined as historical state information.
In this embodiment, the problem that the system robustness is reduced is easily caused by strong dependence on a single parameter, and various characteristics of the congestion state are fully considered, so that a general congestion state detection mode is provided, and the robustness of congestion detection is improved.
Illustratively, based on the above embodiments, this embodiment provides an alternative example of congestion level determination. The specific implementation process comprises the following steps:
based on the Gaussian mixture model, a congestion detection estimation process can be performed, specifically, in actual deployment, a network state detection unit firstly collects and stores selected state information, and for the time t', a corresponding detection feature vector can be formedBringing it into formula (4), the Gaussian membership degree +.>The estimation result of the current congestion state of the node p can be expressed as +.>Wherein:
in this embodiment, the service characteristics of end network coordination are considered, so that the network state detection unit can be deployed at the central node CCS to monitor, thereby assisting in traffic task scheduling of the whole network, and can also be deployed at the terminal device EN, thereby assisting in path selection of subsequent self task transmission.
Furthermore, it should be noted that, since the checking feature is mainly composed of historical state information, the real-time requirement of the network state information acquisition process is low in this embodiment, and on the other hand, the congestion state estimation process can be regarded as a congestion state prediction process based on time-dependent historical state information, so that the subsequent link scheduling can be assisted. Meanwhile, as the multi-level network congestion state mechanism is considered, compared with the binarization traditional scheme for judging the existence of the congestion state only, the method can obtain the detection estimation result of the congestion degree of the network state, and therefore a more flexible network scheduling scheme can be provided in an auxiliary mode.
It should be understood that, although the steps in the flowcharts related to the above embodiments are sequentially shown as indicated by arrows, these steps are not necessarily sequentially performed in the order indicated by the arrows. The steps are not strictly limited to the order of execution unless explicitly recited herein, and the steps may be executed in other orders. Moreover, at least some of the steps in the flowcharts described in the above embodiments may include a plurality of steps or a plurality of stages, which are not necessarily performed at the same time, but may be performed at different times, and the order of the steps or stages is not necessarily performed sequentially, but may be performed alternately or alternately with at least some of the other steps or stages.
Based on the same inventive concept, the embodiment of the application also provides a congestion level determining device for realizing the above-mentioned congestion level determining method. The implementation of the solution provided by the apparatus is similar to the implementation described in the above method, so the specific limitations in the embodiments of one or more congestion level determination apparatuses provided below may be referred to above for the limitations of the congestion level determination method, and are not repeated here.
In one embodiment, as shown in fig. 5, there is provided a congestion level determination apparatus 1 including: an acquisition module 11, a membership calculation module 12, a selection module 13 and a prediction module 14, wherein:
an obtaining module 11, configured to obtain historical state information of each historical time in a target time window before a predicted time of a target node;
the membership calculation module 12 is configured to input each history state information into the gaussian mixture model, so as to obtain membership of each gaussian sub-model in the gaussian mixture model of the target node at the prediction moment;
a selection module 13, configured to select a target sub-model from the gaussian sub-models according to each membership degree;
and the prediction module 14 is used for determining the congestion level of the target node at the prediction moment according to the congestion level corresponding to the target sub-model.
In one embodiment, the level determination apparatus further comprises a prediction module 14, the prediction module 14 comprising:
the acquisition sub-module is used for acquiring a first training sample and a second training sample; the first training sample comprises first sample state information and sample congestion level corresponding to the first sample state information, and the second training sample comprises second sample state information;
The analysis submodule is used for determining congestion level labels of second sample state information in the second training samples according to the first training samples;
the training sub-module is used for training each initial sub-model according to the first training sample and the second training sample of the determined congestion level label to obtain Gaussian sub-models corresponding to the initial sub-models;
and the combination sub-module is used for constructing a Gaussian mixture model according to each Gaussian sub-model.
In one embodiment, the training sub-module includes:
the grouping slave module is used for grouping the first training sample and the second training sample with the determined congestion level label to obtain a sample group corresponding to each congestion level;
the matching slave module is used for distributing a sample group to each initial sub-model;
the initialization slave module is used for acquiring initialization parameters of each initial sub-model;
the training slave module is used for training each initial sub-model according to the initialization parameters and the sample group corresponding to each initial sub-model to obtain a Gaussian sub-model corresponding to each initial sub-model.
In one embodiment, the training slave module is further configured to, for each initial sub-model, use a maximum expected algorithm, and perform iterative update on an initialization parameter of the initial sub-model according to a sample set corresponding to the initial sub-model, to obtain a gaussian sub-model corresponding to the initial sub-model.
In one embodiment, the initialization slave module is further configured to determine, for each initial sub-model, an initial average value of the initial sub-model according to each training sample in the sample group corresponding to the initial sub-model;
determining an initial covariance of the initial sub-model according to the initial mean vector of the initial sub-model and a sample group corresponding to the initial sub-model;
determining initial occurrence probability of the initial sub-model in the Gaussian mixture model according to the total number of samples and the number of samples of training samples in a sample group corresponding to the initial sub-model; wherein the total number of samples is the sum of the number of samples of the first training sample and the number of samples of the second training sample.
In one embodiment, the parsing sub-module is further configured to determine, for any second sample state information, a similarity between the second sample state information and each of the first sample state information;
and determining the sample congestion level corresponding to the first sample state information with the maximum similarity of the second sample state information as a congestion level label of the second sample state information.
In one embodiment, the obtaining module 11 is configured to: acquiring historical dimension parameters of each historical moment in a target time window before a predicted moment of a target node;
And determining the historical dimension parameters meeting the prediction performance as historical state information.
In one embodiment, the historical node dimension parameters include at least one of instantaneous queue size, average sending rate, arrival rate, link usage, buffer occupancy.
The respective modules in the above congestion level determination apparatus may be implemented in whole or in part by software, hardware, or a combination thereof. The above modules may be embedded in hardware or may be independent of a processor in the computer device, or may be stored in software in a memory in the computer device, so that the processor may call and execute operations corresponding to the above modules.
In one embodiment, a computer device is provided, which may be a server, the internal structure of which may be as shown in fig. 6. The computer device includes a processor, a memory, and a network interface connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The database of the computer device is for storing data of the congestion level determination method. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program is executed by a processor to implement a congestion level determination method.
It will be appreciated by those skilled in the art that the structure shown in FIG. 6 is merely a block diagram of some of the structures associated with the present inventive arrangements and is not limiting of the computer device to which the present inventive arrangements may be applied, and that a particular computer device may include more or fewer components than shown, or may combine some of the components, or have a different arrangement of components.
In one embodiment, a computer device is provided comprising a memory and a processor, the memory having stored therein a computer program, the processor when executing the computer program performing the steps of:
acquiring historical state information of each historical moment in a target time window before a predicted moment of a target node;
inputting each history state information into the Gaussian mixture model to obtain the membership degree of each Gaussian sub-model in the Gaussian mixture model of the target node at the prediction moment;
selecting a target submodel from the Gaussian submodels according to each membership degree;
and determining the congestion level of the target node at the predicted moment according to the congestion level corresponding to the target sub-model.
In one embodiment, the processor when executing the computer program further performs the steps of: acquiring a first training sample and a second training sample; the first training sample comprises first sample state information and sample congestion level corresponding to the first sample state information, and the second training sample comprises second sample state information; determining a congestion level label of second sample state information in a second training sample according to the first training sample; training each initial sub-model according to the first training sample and the second training sample of the determined congestion level label to obtain Gaussian sub-models corresponding to the initial sub-models; and constructing a Gaussian mixture model according to each Gaussian sub-model.
In one embodiment, when the processor executes the computer program to train each initial sub-model according to the first training sample and the second training sample of the determined congestion level label to obtain the logic of the gaussian sub-model corresponding to each initial sub-model, the following steps are specifically implemented: grouping the first training samples and the second training samples with determined congestion level labels to obtain sample groups corresponding to each congestion level; assigning a sample set to each initial sub-model; acquiring initialization parameters of each initial sub-model; training each initial sub-model according to the initialization parameters and the sample set corresponding to each initial sub-model to obtain Gaussian sub-models corresponding to each initial sub-model.
In one embodiment, when the processor executes the computer program to train each initial sub-model according to the initialization parameters and the sample set corresponding to each initial sub-model to obtain the logic of the gaussian sub-model corresponding to each initial sub-model, the following steps are specifically implemented: and for each initial sub-model, adopting a maximum expected algorithm, and carrying out iterative updating on the initialization parameters of the initial sub-model according to a sample group corresponding to the initial sub-model to obtain a Gaussian sub-model corresponding to the initial sub-model.
In one embodiment, when the processor executes logic for the computer program to obtain initialization parameters for each initial sub-model, the following steps are specifically implemented: for each initial sub-model, determining an initial mean value of the initial sub-model according to each training sample in a sample group corresponding to the initial sub-model; determining an initial covariance of the initial sub-model according to the initial mean vector of the initial sub-model and a sample group corresponding to the initial sub-model; determining initial occurrence probability of the initial sub-model in the Gaussian mixture model according to the total number of samples and the number of samples of training samples in a sample group corresponding to the initial sub-model; wherein the total number of samples is the sum of the number of samples of the first training sample and the number of samples of the second training sample.
In one embodiment, when the processor executes logic for determining a congestion level label of second sample state information in the second training sample according to the first training sample, the processor specifically implements the following steps: for any second sample state information, determining the similarity between the second sample state information and each first sample state information; and determining the sample congestion level corresponding to the first sample state information with the maximum similarity of the second sample state information as a congestion level label of the second sample state information.
In one embodiment, when the processor executes logic for the computer program to obtain historical state information for each historical time within a target time window of the target node prior to the predicted time, the following steps are specifically implemented: acquiring historical dimension parameters of each historical moment in a target time window before a predicted moment of a target node; and determining the historical dimension parameters meeting the prediction performance as historical state information.
In one embodiment, the historical node dimension parameters include at least one of instantaneous queue size, average sending rate, arrival rate, link usage, buffer occupancy.
In one embodiment, a computer readable storage medium is provided having a computer program stored thereon, which when executed by a processor, performs the steps of:
acquiring historical state information of each historical moment in a target time window before a predicted moment of a target node;
inputting each history state information into the Gaussian mixture model to obtain the membership degree of each Gaussian sub-model in the Gaussian mixture model of the target node at the prediction moment;
selecting a target submodel from the Gaussian submodels according to each membership degree;
And determining the congestion level of the target node at the predicted moment according to the congestion level corresponding to the target sub-model.
In one embodiment, the computer program when executed by the processor further performs the steps of: acquiring a first training sample and a second training sample; the first training sample comprises first sample state information and sample congestion level corresponding to the first sample state information, and the second training sample comprises second sample state information; determining a congestion level label of second sample state information in a second training sample according to the first training sample; training each initial sub-model according to the first training sample and the second training sample of the determined congestion level label to obtain Gaussian sub-models corresponding to the initial sub-models; and constructing a Gaussian mixture model according to each Gaussian sub-model.
In one embodiment, the computer program trains each initial sub-model according to the first training sample and the second training sample of the determined congestion level label, and when the logic for obtaining the gaussian sub-model corresponding to each initial sub-model is executed by the processor, the following steps are specifically implemented: grouping the first training samples and the second training samples with determined congestion level labels to obtain sample groups corresponding to each congestion level; assigning a sample set to each initial sub-model; acquiring initialization parameters of each initial sub-model; training each initial sub-model according to the initialization parameters and the sample set corresponding to each initial sub-model to obtain Gaussian sub-models corresponding to each initial sub-model.
In one embodiment, the computer program trains each initial sub-model according to the initialization parameters and the sample set corresponding to each initial sub-model, and when the logic for obtaining the gaussian sub-model corresponding to each initial sub-model is executed by the processor, the following steps are specifically implemented: and for each initial sub-model, adopting a maximum expected algorithm, and carrying out iterative updating on the initialization parameters of the initial sub-model according to a sample group corresponding to the initial sub-model to obtain a Gaussian sub-model corresponding to the initial sub-model.
In one embodiment, the logic of the computer program to obtain the initialization parameters for each initial sub-model, when executed by the processor, performs the steps of: for each initial sub-model, determining an initial mean value of the initial sub-model according to each training sample in a sample group corresponding to the initial sub-model; determining an initial covariance of the initial sub-model according to the initial mean vector of the initial sub-model and a sample group corresponding to the initial sub-model; determining initial occurrence probability of the initial sub-model in the Gaussian mixture model according to the total number of samples and the number of samples of training samples in a sample group corresponding to the initial sub-model; wherein the total number of samples is the sum of the number of samples of the first training sample and the number of samples of the second training sample.
In one embodiment, the computer program, when executed by the processor, determines the congestion level label of the second sample state information in the second training sample according to the first training sample, specifically implements the steps of: for any second sample state information, determining the similarity between the second sample state information and each first sample state information; and determining the sample congestion level corresponding to the first sample state information with the maximum similarity of the second sample state information as a congestion level label of the second sample state information.
In one embodiment, the logic of the computer program to obtain historical state information for each historical time within the target time window of the target node prior to the predicted time is executed by the processor to perform the steps of: acquiring historical dimension parameters of each historical moment in a target time window before a predicted moment of a target node; and determining the historical dimension parameters meeting the prediction performance as historical state information.
In one embodiment, the historical node dimension parameters include at least one of instantaneous queue size, average sending rate, arrival rate, link usage, buffer occupancy.
In one embodiment, a computer program product is provided comprising a computer program which, when executed by a processor, performs the steps of:
acquiring historical state information of each historical moment in a target time window before a predicted moment of a target node;
inputting each history state information into the Gaussian mixture model to obtain the membership degree of each Gaussian sub-model in the Gaussian mixture model of the target node at the prediction moment;
selecting a target submodel from the Gaussian submodels according to each membership degree;
and determining the congestion level of the target node at the predicted moment according to the congestion level corresponding to the target sub-model.
In one embodiment, the computer program when executed by the processor further performs the steps of: acquiring a first training sample and a second training sample; the first training sample comprises first sample state information and sample congestion level corresponding to the first sample state information, and the second training sample comprises second sample state information; determining a congestion level label of second sample state information in a second training sample according to the first training sample; training each initial sub-model according to the first training sample and the second training sample of the determined congestion level label to obtain Gaussian sub-models corresponding to the initial sub-models; and constructing a Gaussian mixture model according to each Gaussian sub-model.
In one embodiment, the computer program trains each initial sub-model according to the first training sample and the second training sample of the determined congestion level label, and when the logic for obtaining the gaussian sub-model corresponding to each initial sub-model is executed by the processor, the following steps are specifically implemented: grouping the first training samples and the second training samples with determined congestion level labels to obtain sample groups corresponding to each congestion level; assigning a sample set to each initial sub-model; acquiring initialization parameters of each initial sub-model; training each initial sub-model according to the initialization parameters and the sample set corresponding to each initial sub-model to obtain Gaussian sub-models corresponding to each initial sub-model.
In one embodiment, the computer program trains each initial sub-model according to the initialization parameters and the sample set corresponding to each initial sub-model, and when the logic for obtaining the gaussian sub-model corresponding to each initial sub-model is executed by the processor, the following steps are specifically implemented: and for each initial sub-model, adopting a maximum expected algorithm, and carrying out iterative updating on the initialization parameters of the initial sub-model according to a sample group corresponding to the initial sub-model to obtain a Gaussian sub-model corresponding to the initial sub-model.
In one embodiment, the logic of the computer program to obtain the initialization parameters for each initial sub-model, when executed by the processor, performs the steps of: for each initial sub-model, determining an initial mean value of the initial sub-model according to each training sample in a sample group corresponding to the initial sub-model; determining an initial covariance of the initial sub-model according to the initial mean vector of the initial sub-model and a sample group corresponding to the initial sub-model; determining initial occurrence probability of the initial sub-model in the Gaussian mixture model according to the total number of samples and the number of samples of training samples in a sample group corresponding to the initial sub-model; wherein the total number of samples is the sum of the number of samples of the first training sample and the number of samples of the second training sample.
In one embodiment, the computer program, when executed by the processor, determines the congestion level label of the second sample state information in the second training sample according to the first training sample, specifically implements the steps of: for any second sample state information, determining the similarity between the second sample state information and each first sample state information; and determining the sample congestion level corresponding to the first sample state information with the maximum similarity of the second sample state information as a congestion level label of the second sample state information.
In one embodiment, the logic of the computer program to obtain historical state information for each historical time within the target time window of the target node prior to the predicted time is executed by the processor to perform the steps of: acquiring historical dimension parameters of each historical moment in a target time window before a predicted moment of a target node; and determining the historical dimension parameters meeting the prediction performance as historical state information.
In one embodiment, the historical node dimension parameters include at least one of instantaneous queue size, average sending rate, arrival rate, link usage, buffer occupancy.
The user information (including but not limited to user equipment information, user personal information, etc.) and the data (including but not limited to data for analysis, stored data, presented data, etc.) related to the present application are information and data authorized by the user or sufficiently authorized by each party.
Those skilled in the art will appreciate that implementing all or part of the above described methods may be accomplished by way of a computer program stored on a non-transitory computer readable storage medium, which when executed, may comprise the steps of the embodiments of the methods described above. Any reference to memory, database, or other medium used in embodiments provided herein may include at least one of non-volatile and volatile memory. The nonvolatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical Memory, high density embedded nonvolatile Memory, resistive random access Memory (ReRAM), magnetic random access Memory (Magnetoresistive Random Access Memory, MRAM), ferroelectric Memory (Ferroelectric Random Access Memory, FRAM), phase change Memory (Phase Change Memory, PCM), graphene Memory, and the like. Volatile memory can include random access memory (Random Access Memory, RAM) or external cache memory, and the like. By way of illustration, and not limitation, RAM can be in the form of a variety of forms, such as static random access memory (Static Random Access Memory, SRAM) or dynamic random access memory (Dynamic Random Access Memory, DRAM), and the like. The databases referred to in the embodiments provided herein may include at least one of a relational database and a non-relational database. The non-relational database may include, but is not limited to, a blockchain-based distributed database, and the like. The processor referred to in the embodiments provided in the present application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic unit, a data processing logic unit based on quantum computing, or the like, but is not limited thereto.
The technical features of the above embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.
The foregoing examples illustrate only a few embodiments of the application and are described in detail herein without thereby limiting the scope of the application. It should be noted that it will be apparent to those skilled in the art that several variations and modifications can be made without departing from the spirit of the application, which are all within the scope of the application. Accordingly, the scope of the application should be assessed as that of the appended claims.

Claims (12)

1. A method of congestion level determination, the method comprising:
acquiring historical state information of each historical moment in a target time window before a predicted moment of a target node;
inputting each history state information into a Gaussian mixture model to obtain the membership degree of each Gaussian sub-model in the Gaussian mixture model of the target node at the prediction moment;
Selecting a target submodel from the Gaussian submodels according to each membership degree;
and determining the congestion level of the target node at the predicted moment according to the congestion level corresponding to the target sub-model.
2. The method of claim 1, wherein the gaussian mixture model is constructed by:
acquiring a first training sample and a second training sample; the first training sample comprises first sample state information and a sample congestion level corresponding to the first sample state information, and the second training sample comprises second sample state information;
determining a congestion level label of second sample state information in the second training sample according to the first training sample;
training each initial sub-model according to the first training sample and the second training sample of the determined congestion level label to obtain Gaussian sub-models corresponding to each initial sub-model;
and constructing the Gaussian mixture model according to each Gaussian sub-model.
3. The method according to claim 2, wherein training each initial sub-model according to the first training sample and the second training sample of the determined congestion level label to obtain a gaussian sub-model corresponding to each initial sub-model comprises:
Grouping the first training samples and the second training samples with determined congestion level labels to obtain sample groups corresponding to each congestion level;
assigning a sample set to each initial sub-model;
acquiring initialization parameters of each initial sub-model;
training each initial sub-model according to the initialization parameters and the sample set corresponding to each initial sub-model to obtain Gaussian sub-models corresponding to each initial sub-model.
4. The method of claim 3, wherein training each initial sub-model according to the initialization parameters and the sample set corresponding to each initial sub-model to obtain the gaussian sub-model corresponding to each initial sub-model comprises:
and for each initial sub-model, adopting a maximum expected algorithm, and carrying out iterative updating on the initialization parameters of the initial sub-model according to a sample group corresponding to the initial sub-model to obtain a Gaussian sub-model corresponding to the initial sub-model.
5. A method according to claim 3, wherein said obtaining initialization parameters for each initial sub-model comprises:
for each initial sub-model, determining an initial mean value of the initial sub-model according to each training sample in a sample group corresponding to the initial sub-model;
Determining an initial covariance of the initial sub-model according to the initial mean vector of the initial sub-model and a sample group corresponding to the initial sub-model;
determining initial occurrence probability of the initial sub-model in the Gaussian mixture model according to the total number of samples and the number of samples of training samples in a sample group corresponding to the initial sub-model; wherein the total number of samples is the sum of the number of samples of the first training sample and the number of samples of the second training sample.
6. The method of claim 2, wherein determining the congestion level label of the second sample status information in the second training sample from the first training sample comprises:
for any second sample state information, determining the similarity between the second sample state information and each first sample state information;
and determining the sample congestion level corresponding to the first sample state information with the maximum similarity of the second sample state information as a congestion level label of the second sample state information.
7. The method according to claim 1, wherein the obtaining the historical state information of each historical time in the target time window before the predicted time by the target node includes:
Acquiring historical dimension parameters of each historical moment in a target time window before a predicted moment of a target node;
and determining the historical dimension parameters meeting the prediction performance as historical state information.
8. The method of claim 7, wherein the historical node dimension parameters include at least one of instantaneous queue size, average sending rate, arrival rate, link usage, and buffer occupancy.
9. A congestion level determination apparatus, the apparatus comprising:
the acquisition module is used for acquiring the historical state information of each historical moment in the target time window before the predicted moment of the target node;
the membership calculation module is used for inputting each history state information into the Gaussian mixture model to obtain membership of each Gaussian sub-model in the Gaussian mixture model of the target node at the prediction moment;
the selection module is used for selecting a target submodel from the Gaussian submodels according to each membership degree;
and the prediction module is used for determining the congestion level of the target node at the prediction moment according to the congestion level corresponding to the target sub-model.
10. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor implements the steps of the method of any one of claims 1 to 8 when the computer program is executed.
11. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the steps of the method of any of claims 1 to 8.
12. A computer program product comprising a computer program, characterized in that the computer program, when executed by a processor, implements the steps of the method of any one of claims 1 to 8.
CN202311015234.9A 2023-08-11 2023-08-11 Congestion level prediction method, device, computer equipment and storage medium Pending CN116963174A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202311015234.9A CN116963174A (en) 2023-08-11 2023-08-11 Congestion level prediction method, device, computer equipment and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202311015234.9A CN116963174A (en) 2023-08-11 2023-08-11 Congestion level prediction method, device, computer equipment and storage medium

Publications (1)

Publication Number Publication Date
CN116963174A true CN116963174A (en) 2023-10-27

Family

ID=88452976

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202311015234.9A Pending CN116963174A (en) 2023-08-11 2023-08-11 Congestion level prediction method, device, computer equipment and storage medium

Country Status (1)

Country Link
CN (1) CN116963174A (en)

Similar Documents

Publication Publication Date Title
Rakkiyappan et al. Event-triggered H∞ state estimation for semi-Markov jumping discrete-time neural networks with quantization
US11652709B2 (en) Managing computation load in a fog network
CN111431819A (en) Network traffic classification method and device based on serialized protocol flow characteristics
Dubourg et al. Metamodel-based importance sampling for the simulation of rare events
Gibberd et al. Multiple changepoint estimation in high-dimensional gaussian graphical models
Wang et al. Cooperative anomaly detection with transfer learning-based hidden Markov model in virtualized network slicing
CN111950611A (en) Big data two-classification distributed optimization method based on random gradient tracking technology
CN115562940A (en) Load energy consumption monitoring method and device, medium and electronic equipment
Gao et al. A deep learning framework with spatial-temporal attention mechanism for cellular traffic prediction
Lin et al. Distributed subgradient projection algorithm for multi-agent optimization with nonidentical constraints and switching topologies
Le et al. GCRINT: network traffic imputation using graph convolutional recurrent neural network
Mahmood et al. Mining data generated by sensor networks: a survey
Shimonishi et al. Energy optimization of distributed video processing system using genetic algorithm with bayesian attractor model
Gao et al. A data mining method using deep learning for anomaly detection in cloud computing environment
CN116963174A (en) Congestion level prediction method, device, computer equipment and storage medium
Håkansson et al. Cost-aware dual prediction scheme for reducing transmissions at IoT sensor nodes
CN109343952B (en) Bayesian network determination method, Bayesian network determination device, storage medium and electronic device
CN116484016A (en) Time sequence knowledge graph reasoning method and system based on automatic maintenance of time sequence path
CN116306030A (en) New energy prediction dynamic scene generation method considering prediction error and fluctuation distribution
D'Aronco et al. Online resource inference in network utility maximization problems
Wagle et al. Embedding alignment for unsupervised federated learning via smart data exchange
Casale et al. Fluid analysis of queueing in two-stage random environments
Pasteris et al. Data distribution and scheduling for distributed analytics tasks
CN109728958A (en) A kind of network node trusts prediction technique, device, equipment and medium
Takele et al. Resource aware long short-term memory model (RALSTMM) based on-device incremental learning for industrial Internet of Things

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