CN110166369A - A kind of electric power optical-fiber network active load equalization methods - Google Patents
A kind of electric power optical-fiber network active load equalization methods Download PDFInfo
- Publication number
- CN110166369A CN110166369A CN201910383417.3A CN201910383417A CN110166369A CN 110166369 A CN110166369 A CN 110166369A CN 201910383417 A CN201910383417 A CN 201910383417A CN 110166369 A CN110166369 A CN 110166369A
- Authority
- CN
- China
- Prior art keywords
- load
- electric power
- power optical
- business
- link
- 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
Classifications
-
- 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
- H04L41/145—Network analysis or design involving simulating, designing, planning or modelling of a network
-
- 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
- H04L41/147—Network analysis or design for predicting network behaviour
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L47/00—Traffic control in data switching networks
- H04L47/10—Flow control; Congestion control
- H04L47/12—Avoiding congestion; Recovering from congestion
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L47/00—Traffic control in data switching networks
- H04L47/10—Flow control; Congestion control
- H04L47/12—Avoiding congestion; Recovering from congestion
- H04L47/125—Avoiding congestion; Recovering from congestion by balancing the load, e.g. traffic engineering
Abstract
The invention discloses a kind of electric power optical-fiber network active load equalization methods, and the load-balancing method is the following steps are included: step 1: being predicted according to family eugenics algorithm and neural network algorithm the business of electric power optical-fiber network；Step 2: carrying out the analysis of optical fiber link load condition；Step 3: analyzed using historic load state recording of the Time Series Analysis Model to base station, neural network forecast to new business at hand when, pass through resource allocation policy optimize network load balancing.
Description
Technical field
The present invention relates to electric power technical field of optical network communication, more particularly to electric power optical-fiber network active load equilibrium side
Method.
Background technique
With the whole construction and development of national smart grid and new energy internet, electric power new business flow is also in speed
Increase, requirements at the higher level are proposed to the real-time and reliability of entire power telecom network.And different power businesses are in different transmission
Load on channel is not also identical, such as in certain hot zones, often aggregation is than other regional more users.This use
Family being unevenly distributed so that overload situations occurs on some channels in power business, and in other channels in the power communication network
There is idle state in upper thief.This user that will lead to hot zones can not may provide in time enough money due to electric power optical-fiber network
Source and meet with severe service experience.In order to improve the service quality of user, while more fully utilizing the base station of light load
Idling-resource, need to be adjusted electric power optical network service using traffic load balance technology, it is whole to improve network
Load balancing state.
Therefore, it is desirable to have a kind of electric power optical-fiber network active load equalization methods to solve problems of the prior art.
Summary of the invention
The invention discloses a kind of electric power optical-fiber network active load equalization methods, and the load-balancing method includes following
Step:
Step 1: the business of electric power optical-fiber network being predicted according to family eugenics algorithm and neural network algorithm；
Step 2: carrying out the analysis of optical fiber link load condition；
Step 3: being analyzed using historic load state recording of the Time Series Analysis Model to base station, neural network forecast arrives
New business at hand when, pass through resource allocation policy optimize network load balancing.
Preferably, the step 1 further comprises: building decision using the family eugenics algorithm and neural network algorithm
Tree-model predicts the service traffics of future time and the duration of business.
Preferably, the step 2 is according to the service traffics of the future time and the duration of business, to current all
The load of link is analyzed, and heavy duty is isolated, and light load and the merging of intermediate part load link set are marked respectively.
Preferably, the total load calculation formula of all links is as follows in the step 2:
Wherein, H is the total traffic of all links, WiFor the link in each of the links, F (Ti) it is each of the links portfolio
With time TiFunction.
Preferably, the resource allocation policy of the step 3 is to carry out grade classification to link according to load condition, according to industry
Whether business is in heavy duty link, determines whether starting load equilibrium.
The present invention analyzes the load condition around target, in advance by carrying out mobility prediction to electric power optical network service
Reasonable Resource Allocation Formula is formulated, the load balancing of network, electric power optical-fiber network active load equalization methods of the present invention are optimized
Beneficial effect include:
(1) present invention focuses on the volume forecasting of future services, can solve relevant network loading problems earlier, keep away
Exempt from link blocking occur；
(2) the method for the present invention carries out cyclic forecast by the duration time to electric power optical network service, can analyze
The business feature and development trend of overall network out；
(3) the method for the present invention comprehensively considers network overall load state change, formulates resource allocation policy in advance, realizes
Optimization to Network Load Balance meets the demand for services of hot zones user, improves network performance.
Detailed description of the invention
Fig. 1 is electric power optical network node schematic diagram.
Specific embodiment
Embodiment:
Step 1: the traffic forecast of family eugenics and neural network algorithm to electric power optical-fiber network:
The method of the present invention is calculated during the mobility prediction to user using family eugenics algorithm and neural network
Chain road influence of the state to future state before this is compared in method, research, is selected optimal portfolio input attributive character, is improved and move
Dynamic property predictablity rate.The present invention is analyzed by the mobility to portfolio, excavates rule therein, can be under link
The portfolio at one moment is made prediction.Secondly, by carrying out reasonable load balancing to link to optimize the property of whole network
Energy.Then, resource allocation policy is formulated in advance, optimizes the load balancing of network, improves the network performance of hot zones, is improved and is used
Family service experience.
In the forecast period of business, it is contemplated that neural network relatively good can handle big time scale and small time scale
The common influence of data, family eugenics can provide best flow output, so being made with neural network and family eugenics
For the main algorithm of this research.
Traditional neural network, there is " only focusing on current time " in model, and have ignored and answer last time
With, and the prediction to future time instance.Under such a scenario, family eugenics fused neural network can be borrowed come when remembeing each
The information at quarter.
Electric power optical-fiber network is conceptualized as topological structure as shown in Figure 1, and node set includes n-1 (1,2,3 ... n-1) a
Node and a destination node n, line set are the optical fiber link collection that there is syntople node to constitute.Each optical node is equal in figure
Multiple user loads are connected to, and user information is transmitted to destination node by multi-hop transmission.There are one between Fiber Node
Optical fiber link then claims two nodes to have syntople, this link belongs to optical fiber link collection Jie.Node i can cover
Node be i neighbor node integrate Jie as Ni=j | (i, j) ∈ E }.The candidate node set of node i is combined into Si, it is NiSubset.
The present invention is based on family eugenics and neural network algorithm to construct decision-tree model, to predict the stream of next business
Amount.When establishing decision-tree model, it is necessary first to input the portfolio and its characteristic in current state, objective attribute target attribute is lower a period of time
The service traffics and characteristic at quarter.In periodic statistics and prediction, the flow W of the business of current ink can be obtainednWith it is current
The duration T of businessi, then the flow W of various time points can be obtainedi{W1,W2,…,Wi-1,Wi..., in conjunction with Markovian model
The thought of type studies influence of several service conditions to predictablity rate before this.Second order Markov model considers before this two
Influence of a state to Future Positions, its predictablity rate are higher.Therefore, it under lower algorithm complexity, mentions as best one can
High predictablity rate selects 2 states of user before this as training characteristics, establishes decision-tree model.Predict future time
Service traffics Wn+1And one moment of business service traffics and after the duration, need to current all link SiLoad carry out
Analysis, isolates heavy duty, light to load, and the merging of intermediate part load link set is marked respectively.
At this point, the total load calculation formula of all links is as follows
Wherein, H is the total traffic of all links, WiFor the link in each of the links, F (Ti) it is each of the links portfolio
With time TiFunction.
That is the portfolio at current time is function related with timing node.
Time series analysis is usage history data, understands its rule of development by statistical analysis, and further to future
Development trend make prediction.ARIMA model is a kind of important analysis method in time series analysis, and precision of prediction compared with
It is high.ARIMA model includes 3 kinds of forms, the mixing auto regressive moving average of autoregression AR model, rolling average MA model and the two
Arma modeling.When using arma modeling, need to guarantee that the object to be analyzed is stable time series.If sequence is non-flat
Surely, then it needs first to carry out difference, obtains stationary sequence, otherwise can not apply the model.And " I " in ARIMA model is just represented
The stationarity of sequence.
Assuming that ytIt is a stationary time series, then corresponding p rank AR (p) model can indicate are as follows:
yt=α1yt-1+α2yt-2+…+αpyt-p+ε(t)
Wherein, p is the autoregression item of model, αi(i=1,2 ... p) be auto-regressive parameter, and ε (t) is stochastic error.y
(t) a stable time series is indicated.T indicates t-th of moment.
Corresponding q rank MA (q) model can indicate are as follows:
yt=ε (t)-θ1ε(t-1)-θ2ε(t-2)-…-θqε(t-q)
Wherein, q is the corresponding rolling average item number of model, θj(j=1,2 ..., q) is rolling average parameter, ε (t-j) table
Show that q-th of prediction is averaged the random error of item.
By AR (p) model and MA (q) models coupling, ARMA (p, q) model can be obtained, indicate are as follows:
ARMA (p, q) model can be indicated with following form:
α p (B) y (t)=θ q (B) ε (t)
α p (B)=1- α1B1-…-αpBp
α q (B)=1- θ1B1-…-θpBp
Wherein, BkLag operator is walked for k, α p (B) is a p rank autoregression multinomial, and θ q (B) is that a q rank is mobile flat
Equal multinomial.
The object of AR (p), MA (q) and ARMA (p, q) model all must be a stable time series.In the present invention,
Using the loading condition of single link as research object.It, will be this when step 1 predicts the business on the subsequent time chain road
A random sequence carried out by periodic prediction data, is set as business load amount x (t).If x (t) is a stable time sequence
Column then directly regard x (t) as y (t), carry out simulation and prediction using ARIMA model.If x (t) is the time sequence an of non-stationary
Column, need first to convert it to a stable time series.If x (t) obtains sequences y (t), y (t) after d order difference
For stationary time series.ARMA (p, q) model, the as ARIMA of x (t) (p, d, q) model so are established to y (t), indicated are as follows:
α p (B) (1-B) dx (t)=θ q (B) ε (t)
Step 3, belong to load balance optimization state, for the predictive factor in step 2, using ARIMA model to base station
Historic load state recording analyzed, when neural network forecast to new business at hand, then by working out reasonable resource
Allocation strategy optimizes the load balancing of network.
The method of the present invention comprehensively considers the load condition of whole links first, then link is carried out according to load condition etc.
Grade divides, and then considers whether business is on heavier loaded link, and decide whether to starting load balanced measure.
Since each of the links have its corresponding Business Stream thresholding, it is therefore assumed that the link circuit resource total amount of overall network is R,
Current medical resource is S, then the load condition of link are as follows:
Assuming that whole network has n link, then whole network load state are as follows:
Absolute value of the difference in order to keep the load of network more balanced, between each link and the load condition of network entirety
It is smaller, then it represents that the deviation of single link and integrality is smaller.With reference to the concept of variance in mathematics, respectively to each base station with
Network overall load seeks difference, then takes the quadratic sum of all differences, and it is averagely negative compared to network to be able to reflect each base station
The deviation situation of load.
After obtaining average load, heavy duty link is rejected, route transmission is carried out in new subnet to new business.
Finally it is noted that the above embodiments are merely illustrative of the technical solutions of the present invention, rather than its limitations.To the greatest extent
Present invention has been described in detail with reference to the aforementioned embodiments for pipe, those skilled in the art should understand that: it is still
It is possible to modify the technical solutions described in the foregoing embodiments, or part of technical characteristic is equally replaced
It changes；And these are modified or replaceed, the essence for technical solution of various embodiments of the present invention that it does not separate the essence of the corresponding technical solution
Mind and range.
Claims (5)
1. a kind of electric power optical-fiber network active load equalization methods, it is characterised in that: the load-balancing method includes following step
It is rapid:
Step 1: the business of electric power optical-fiber network being predicted according to family eugenics algorithm and neural network algorithm；
Step 2: carrying out the analysis of optical fiber link load condition；
Step 3: being analyzed using historic load state recording of the Time Series Analysis Model to base station, neural network forecast to new industry
When being engaged at hand, pass through the load balancing that resource allocation policy optimizes network.
2. electric power optical-fiber network active load equalization methods according to claim 1, it is characterised in that: the step 1 into
One step includes: that the Business Stream of decision-tree model prediction future time is built using the family eugenics algorithm and neural network algorithm
The duration of amount and business.
3. electric power optical-fiber network active load equalization methods according to claim 2, it is characterised in that: the step 2 piece
According to the service traffics of the future time and the duration of business, the load of current all links is analyzed, is isolated
Heavy duty, light load and the merging of intermediate part load link set are marked respectively.
4. electric power optical-fiber network active load equalization methods according to claim 3, it is characterised in that: in the step 2
The total load calculation formula of all links is as follows:
Wherein, H is the total traffic of all links, WiFor the link in each of the links, F (Ti) be each of the links portfolio and when
Between TiFunction.
5. electric power optical-fiber network active load equalization methods according to claim 1, it is characterised in that: the step 3
Resource allocation policy is to carry out grade classification to link according to load condition, whether is in heavy duty link according to business, determines
Whether starting load is balanced.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910383417.3A CN110166369A (en) | 2019-05-09 | 2019-05-09 | A kind of electric power optical-fiber network active load equalization methods |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910383417.3A CN110166369A (en) | 2019-05-09 | 2019-05-09 | A kind of electric power optical-fiber network active load equalization methods |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110166369A true CN110166369A (en) | 2019-08-23 |
Family
ID=67633895
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910383417.3A Pending CN110166369A (en) | 2019-05-09 | 2019-05-09 | A kind of electric power optical-fiber network active load equalization methods |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110166369A (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102711177A (en) * | 2012-04-26 | 2012-10-03 | 北京邮电大学 | Service prediction based load balancing method |
| US8953623B1 (en) * | 2011-11-23 | 2015-02-10 | Juniper Networks, Inc. | Predictive network services load balancing within a network device |
| CN104410582A (en) * | 2014-12-10 | 2015-03-11 | 国家电网公司 | Traffic balancing method for electric power communication network based on traffic prediction |
| CN108834079A (en) * | 2018-09-21 | 2018-11-16 | 北京邮电大学 | A kind of load balance optimization method based on mobility prediction in heterogeneous network |
-
2019
- 2019-05-09 CN CN201910383417.3A patent/CN110166369A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8953623B1 (en) * | 2011-11-23 | 2015-02-10 | Juniper Networks, Inc. | Predictive network services load balancing within a network device |
| CN102711177A (en) * | 2012-04-26 | 2012-10-03 | 北京邮电大学 | Service prediction based load balancing method |
| CN104410582A (en) * | 2014-12-10 | 2015-03-11 | 国家电网公司 | Traffic balancing method for electric power communication network based on traffic prediction |
| CN108834079A (en) * | 2018-09-21 | 2018-11-16 | 北京邮电大学 | A kind of load balance optimization method based on mobility prediction in heterogeneous network |
Non-Patent Citations (1)
| Title |
|---|
| 田松奇: "异构网络中面向资源分配优化的用户移动性预测算法研究", 《北京邮电大学硕士学位论文》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2021169577A1 (en) | Wireless service traffic prediction method based on weighted federated learning | |
| Alvizu et al. | Matheuristic with machine-learning-based prediction for software-defined mobile metro-core networks | |
| Le et al. | A practical model for traffic forecasting based on big data, machine-learning, and network KPIs | |
| Fan et al. | Deep learning empowered traffic offloading in intelligent software defined cellular V2X networks | |
| Guo et al. | Research on SDN/NFV network traffic management and optimization based on big data and artificial intelligence | |
| CN104602325A (en) | Analytic hierarchy process based heterogeneous wireless network selection method | |
| Fiandrino et al. | A machine-learning-based framework for optimizing the operation of future networks | |
| Pelekanou et al. | Provisioning of 5G services employing machine learning techniques | |
| CN114885388B (en) | Multi-service type self-adaptive switching judgment method combining RSS prediction | |
| Hu et al. | A study of LTE network performance based on data analytics and statistical modeling | |
| Zhong et al. | A vertical handoff decision scheme using subjective-objective weighting and grey relational analysis in cognitive heterogeneous networks | |
| Rao et al. | Network selection in heterogeneous environment: A step toward always best connected and served | |
| Manalastas et al. | Where to go next?: A realistic evaluation of AI-assisted mobility predictors for HetNets | |
| Bao et al. | Resource allocation with edge-cloud collaborative traffic prediction in integrated radio and optical networks | |
| Ramagundam | Predicting broadband network performance with ai-driven analysis | |
| CN103249050B (en) | Based on the multiple dimensioned frequency spectrum access method of business demand | |
| Hojeij et al. | Dynamic placement of O-CU and O-DU functionalities in open-ran architecture | |
| CN110166369A (en) | A kind of electric power optical-fiber network active load equalization methods | |
| Yi et al. | DMADRL: A distributed multi-agent deep reinforcement learning algorithm for cognitive offloading in dynamic MEC networks | |
| Guo et al. | Capturing spatial–temporal correlations with Attention based Graph Convolutional Network for network traffic prediction | |
| Agarwal et al. | A Low Complexity ML-Assisted Multi-knapsack-based Approach for User Association and Resource Allocation in 5G HetNets | |
| Khan et al. | Machine Learning-Based Application for Predicting 5G/B5G Service | |
| Wang et al. | Learning From Images: Proactive Caching With Parallel Convolutional Neural Networks | |
| Puspita et al. | An Intelligent Admission Control Scheme for Dynamic Slice Handover Policy in 5G Network Slicing | |
| Panchenko et al. | The method of adaptive selection of a wireless access network in a heterogeneous environment based on the theory of fuzzy sets |
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 | ||
| CB03 | Change of inventor or designer information | ||
| CB03 | Change of inventor or designer information |
Inventor after: Liu Yajing Inventor after: Zhang Yangyang Inventor after: Song Wei Inventor after: Yan Lei Inventor after: Tian Yu Inventor after: Zhao Yang Inventor after: Duan Chengyu Inventor after: Yang Chun Inventor after: Zhang Donghui Inventor after: Zhang Jiale Inventor after: Qiang Yaqian Inventor after: Lu Wenbing Inventor after: Wei Rongtao Inventor after: Luo Fang Inventor after: Wen Yifan Inventor after: Ji Yutong Inventor after: Jin Shen Inventor after: Shen Fang Inventor before: Liu Yajing Inventor before: Qiang Yaqian Inventor before: Lu Wenbing |
|
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190823 |