CN114126019B - Energy efficiency optimization-based dynamic resource mapping method and system for forward optical network - Google Patents
Energy efficiency optimization-based dynamic resource mapping method and system for forward optical network Download PDFInfo
- Publication number
- CN114126019B CN114126019B CN202111444691.0A CN202111444691A CN114126019B CN 114126019 B CN114126019 B CN 114126019B CN 202111444691 A CN202111444691 A CN 202111444691A CN 114126019 B CN114126019 B CN 114126019B
- Authority
- CN
- China
- Prior art keywords
- baseband processing
- radio frequency
- remote radio
- energy consumption
- node
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 68
- 230000003287 optical effect Effects 0.000 title claims abstract description 57
- 238000013507 mapping Methods 0.000 title claims abstract description 39
- 238000005457 optimization Methods 0.000 title claims abstract description 27
- 238000012545 processing Methods 0.000 claims abstract description 131
- 238000005265 energy consumption Methods 0.000 claims abstract description 75
- 238000012549 training Methods 0.000 claims abstract description 38
- 230000008569 process Effects 0.000 claims abstract description 35
- 230000015654 memory Effects 0.000 claims abstract description 32
- 238000013135 deep learning Methods 0.000 claims abstract description 11
- 230000007787 long-term memory Effects 0.000 claims description 18
- 238000007781 pre-processing Methods 0.000 claims description 16
- 230000006403 short-term memory Effects 0.000 claims description 15
- 238000010606 normalization Methods 0.000 claims description 7
- 230000003044 adaptive effect Effects 0.000 claims description 5
- 238000013480 data collection Methods 0.000 claims description 3
- 238000004806 packaging method and process Methods 0.000 claims description 3
- 210000004027 cell Anatomy 0.000 description 6
- 239000013598 vector Substances 0.000 description 5
- 230000008859 change Effects 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 239000011159 matrix material Substances 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 3
- 238000011161 development Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 239000013307 optical fiber Substances 0.000 description 3
- 230000000306 recurrent effect Effects 0.000 description 3
- MWRWFPQBGSZWNV-UHFFFAOYSA-N Dinitrosopentamethylenetetramine Chemical compound C1N2CN(N=O)CN1CN(N=O)C2 MWRWFPQBGSZWNV-UHFFFAOYSA-N 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 2
- 229940112112 capex Drugs 0.000 description 2
- FEBLZLNTKCEFIT-VSXGLTOVSA-N fluocinolone acetonide Chemical compound C1([C@@H](F)C2)=CC(=O)C=C[C@]1(C)[C@]1(F)[C@@H]2[C@@H]2C[C@H]3OC(C)(C)O[C@@]3(C(=O)CO)[C@@]2(C)C[C@@H]1O FEBLZLNTKCEFIT-VSXGLTOVSA-N 0.000 description 2
- 238000011478 gradient descent method Methods 0.000 description 2
- 238000007726 management method Methods 0.000 description 2
- 238000010295 mobile communication Methods 0.000 description 2
- 230000006855 networking Effects 0.000 description 2
- 210000001744 T-lymphocyte Anatomy 0.000 description 1
- 102100029469 WD repeat and HMG-box DNA-binding protein 1 Human genes 0.000 description 1
- 101710097421 WD repeat and HMG-box DNA-binding protein 1 Proteins 0.000 description 1
- 239000012141 concentrate Substances 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000013439 planning Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000013468 resource allocation Methods 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W52/00—Power management, e.g. TPC [Transmission Power Control], power saving or power classes
- H04W52/02—Power saving arrangements
- H04W52/0203—Power saving arrangements in the radio access network or backbone network of wireless communication networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B10/00—Transmission systems employing electromagnetic waves other than radio-waves, e.g. infrared, visible or ultraviolet light, or employing corpuscular radiation, e.g. quantum communication
- H04B10/25—Arrangements specific to fibre transmission
- H04B10/2575—Radio-over-fibre, e.g. radio frequency signal modulated onto an optical carrier
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W72/00—Local resource management
- H04W72/50—Allocation or scheduling criteria for wireless resources
- H04W72/52—Allocation or scheduling criteria for wireless resources based on load
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Molecular Biology (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Electromagnetism (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
- Mobile Radio Communication Systems (AREA)
Abstract
The invention relates to a method and a system for mapping dynamic resources of a front-end optical network based on energy efficiency optimization. The method comprises the following steps: collecting historical flow data of a remote radio frequency node; training a long-short-term memory network model in deep learning according to the historical flow data, and predicting the flow load condition of the next time node of each remote radio frequency node; according to the traffic load condition, aiming at minimizing the total energy consumption of the system, dynamically adjusting the connection relation of a forward optical link between a baseband processing pool of the next time node and a remote radio frequency node; the minimum total system energy consumption includes the inherent energy consumption of the baseband processing pool, the energy consumption generated in the process of the baseband processing pool executing baseband signal processing and the energy consumption generated by switching. The invention can reduce the total system energy consumption of the whole forward optical link and improve the utilization rate of the baseband processing pool resources.
Description
Technical Field
The invention relates to the technical field of mobile communication networks, in particular to a method and a system for mapping dynamic resources of a front-end optical network based on energy efficiency optimization.
Background
With the development of diversified network services and the large-scale popularization of intelligent terminal equipment, the data traffic of the mobile network is exponentially increased, and higher requirements are put forward on the performances of the 5G mobile network in various aspects. How to reduce the capital expenditure (Capital Expenditure, CAPEX) and operating expenditure (Operating Expense, OPEX) of a network while meeting the sharply increasing wireless service requirements is a matter of major concern to operators. The mobile access network architecture is developed from an integrated macro base station in early 2G age to a distributed base station networking in 3G age, and then the distributed base station networking is further developed to a centralized access network (Centralized Radio Access Network, C-RAN) architecture in 4G age, wherein the main principle of the C-RAN is to keep a far-end radio frequency node (Remote Radio Head, RRH) in a cell and establish connection with a user, and Baseband processing (BBU) units in a traditional base station are concentrated together to form a BBU pool, and the BBU pool is placed in a Central Office (CO) to perform unified scheduling on Baseband resources to be processed in the RRH. In the C-RAN, the BBU pool and the RRH carry out data transmission by utilizing an optical fiber link, and the optical fiber link has high bandwidth and high reliability, and the link is called a forward transmission (Fronthaul) optical link, so that the RRH with a simple structure can greatly reduce the cost of deployment, installation and maintenance of a network along with deployment of a dense base station. And establishing a forwarding connection between the BBU pool and the RRH through an optical fiber link. Considering the advantages in terms of reduced CAPEX and OPEX, the C-RAN has received attention from a wide range of network operators.
The front-end optical network is regarded as a novel optical access network scheme with great potential, and is also a key for realizing large-scale and industrialized deployment of the C-RAN in 5G mobile communication. However, due to the mobility of users, mobile data traffic can continuously migrate over time and space, and traffic changes in network service areas have a significant tidal effect, i.e. users tend to gather in the work area during the day, resulting in dense users in the work area, high base station load, large forward data traffic, and relatively idle base stations in populated areas. Most mobile users return to the residential area at night, and the base stations of the working area are idle. The C-RAN architecture concentrates baseband resources in a network in a BBU pool, has the capability of centralized scheduling and planning of the whole network resources, however, because the current access network structure does not have the capability of sensing network environment change, and simultaneously lacks uniform decision capability and intelligent management and control capability to schedule and allocate the network resources in time, the current optical network is still in a relatively fixed connection mode, the connection relation of a front optical link cannot be adjusted in time according to load change, and the bandwidth resources and processing capability of the BBU connected with a low-load RRH are not fully utilized, and the utilization rate of the whole resources of the BBU pool is lower. Clearly, statically configured optical networks do not fully exploit the flexibility of the C-RAN architecture and the advantages of intelligent management. With deployment of dense cells in 5G, the statically configured C-RAN can result in a huge resource waste for the BBU pool.
Disclosure of Invention
The invention aims to provide an energy efficiency optimization-based dynamic resource mapping method and system for a forward optical network, which are used for solving the problems of high total energy consumption of a system of the whole forward optical link and low utilization rate of a baseband processing pool resource.
In order to achieve the above object, the present invention provides the following solutions:
a method for mapping dynamic resources of a forward optical network based on energy efficiency optimization comprises the following steps:
Collecting historical flow data of a remote radio frequency node; the data types of the historical flow data comprise telephone service, short message service and Internet service which are received and sent; the historical flow data represents the actual flow load conditions of a plurality of historical time nodes;
Training a long-short-term memory network model in deep learning according to the historical flow data, and predicting the flow load condition of the next time node of each remote radio frequency node;
According to the traffic load condition, aiming at minimizing the total energy consumption of the system, dynamically adjusting the connection relation of a forward optical link between a baseband processing pool of the next time node and a remote radio frequency node; the minimum system total energy consumption comprises inherent energy consumption of a baseband processing pool, energy consumption generated in the process of executing baseband signal processing by the baseband processing pool and energy consumption generated by switching; the energy consumption generated by the switching comprises the energy consumption generated by the switching of the baseband processing units in the baseband processing pool between an activated state and a closed state and the energy consumption generated by the switching of the remote radio frequency node in different baseband processing units.
Optionally, the collecting historical traffic data of the remote radio frequency node further includes:
preprocessing the historical flow data to generate preprocessed historical flow data; the preprocessing comprises data normalization processing and data dimension-lifting processing.
Optionally, training a long-short-term memory network model in deep learning according to the historical traffic data, and predicting traffic load conditions of next time nodes of each remote radio frequency node specifically includes:
taking the actual flow load conditions of a plurality of historical time nodes of each remote radio frequency node as input, and taking the flow load condition of the next time node as output, training the long-term and short-term memory network model;
In each iteration process, comparing a predicted value corresponding to the predicted flow load condition with a true value corresponding to the true flow load condition, and adjusting the weight of the long-period memory network model according to the error between the predicted value and the true value until the error requirement of training is met, so as to complete the long-period memory network model after training;
And predicting the traffic load condition of the next time node of each remote radio frequency node by using the trained long-term and short-term memory network model.
Optionally, the energy consumption generated in the process of executing the baseband signal processing in the baseband processing pool is generated by fully utilizing the resource of each activated baseband processing unit through the number of the baseband processing units in the activated state in the baseband processing pool.
Optionally, the energy consumption generated by the switching is energy consumption generated by switching the baseband processing units in the baseband processing pool between an active state and a closed state and energy consumption generated by switching the remote radio frequency node in different baseband processing units.
Optionally, the dynamically adjusting method is regarded as a packaging problem by dynamically adjusting the connection relationship of the forward optical link between the baseband processing pool of the next time node and the remote radio frequency node according to the traffic load condition and with the aim of minimizing the total energy consumption of the system; the box represents the baseband processing unit, the volume of the box is the maximum flow load that each baseband processing unit can bear, the size of the article is the current moment load condition of each far-end radio frequency node, the size of the article changes with time, and the boxing operation is to adjust the connection relation between the baseband processing unit and the far-end radio frequency node.
Optionally, solving the boxing problem by using a descending order first-time adaptive resource mapping algorithm based on the minimum switching number;
the step of the descending order first adapting resource mapping algorithm based on the minimum switching number comprises the following steps:
According to the current traffic load conditions of the remote radio nodes, all the traffic load conditions of the remote radio nodes are arranged in descending order according to traffic load values;
Selecting out the execution boxing operation with the maximum flow load in the unplanned remote radio frequency node;
Judging whether a baseband processing unit connected with the current remote radio frequency node at the previous time is in an activated state or not to obtain a first judging result;
If the first judgment result is yes, judging whether the selected baseband processing unit has enough bandwidth resources to process the load flow of the current remote radio node, and obtaining a second judgment result;
if the second judgment result is yes, maintaining the connection relation between the selected baseband processing unit and the extracted far-end radio frequency node;
If the second judgment result is negative, traversing the remaining activated baseband processing units, judging whether enough bandwidth resources exist to process the extracted bandwidth resources of the remote radio frequency node, and obtaining a third judgment result;
if the third judgment result is yes, connecting a baseband processing unit with enough bandwidth resources, which is encountered for the first time, with the current remote radio frequency node in the traversal process;
if the third judging result is negative, starting a new baseband processing unit, and connecting the new baseband processing unit with the extracted remote radio frequency node;
If the first judgment result is negative, executing a third judgment process; the third judging process is to traverse the remaining activated baseband processing units and judge whether enough bandwidth resources exist to process the extracted bandwidth resources of the remote radio nodes so as to obtain a third judging result;
and updating the working states and the load conditions of all the baseband processing units.
An energy efficiency optimization-based dynamic resource mapping system of a forward optical network, comprising:
The historical flow data collection module is used for collecting historical flow data of the remote radio frequency node; the data types of the historical flow data comprise telephone service, short message service and Internet service which are received and sent; the historical flow data represents the actual flow load conditions of a plurality of historical time nodes;
The traffic load condition prediction module is used for training a long-period memory network model in deep learning according to the historical traffic data and predicting the traffic load condition of the next time node of each remote radio frequency node;
The dynamic adjustment module is used for dynamically adjusting the connection relation of the forward optical link between the baseband processing pool of the next time node and the remote radio frequency node according to the traffic load condition and with the aim of minimizing the total energy consumption of the system; the minimum system total energy consumption comprises inherent energy consumption of a baseband processing pool, energy consumption generated in the process of executing baseband signal processing by the baseband processing pool and energy consumption generated by switching; the energy consumption generated by the switching comprises the energy consumption generated by the switching of the baseband processing units in the baseband processing pool between an activated state and a closed state and the energy consumption generated by the switching of the remote radio frequency node in different baseband processing units.
Optionally, the method further comprises:
the preprocessing module is used for preprocessing the historical flow data and generating preprocessed historical flow data; the preprocessing comprises data normalization processing and data dimension-lifting processing.
Optionally, the traffic load condition prediction module specifically includes:
The training unit is used for training the long-term and short-term memory network model by taking the real traffic load conditions of a plurality of historical time nodes of each remote radio frequency node as input and the traffic load condition of the next time node as output;
The long-period memory network model building unit after training is used for comparing a predicted value corresponding to the predicted flow load condition with a true value corresponding to the true flow load condition in each iteration process, and adjusting the weight of the long-period memory network model according to the error between the predicted value and the true value until the error requirement of training is met, so as to complete the long-period memory network model after training;
And the traffic load condition prediction unit is used for predicting the traffic load condition of the next time node of each remote radio frequency node by using the trained long-short-term memory network model.
According to the specific embodiment provided by the invention, the invention discloses the following technical effects: the invention provides a method and a system for mapping dynamic resources of a front-end optical network based on energy efficiency optimization, which acquire the traffic load condition of a next time node through a long-term and short-term memory network, dynamically adjust the connection relation between a baseband processing unit and a remote radio frequency node in the front-end optical network with the aim of minimizing the total energy consumption of the system, greatly reduce the energy consumption of the whole system, improve the resource utilization rate of the front-end optical network, and meet the requirement of deployment of large-scale intelligent equipment on the high energy efficiency of the front-end optical network.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions of the prior art, the drawings that are needed in the embodiments will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
Fig. 1 is a flow chart of a dynamic resource mapping method of a front-end optical network based on energy efficiency optimization provided by the invention;
Fig. 2 is a schematic diagram of traffic prediction by using a long-short-period memory network in the method for mapping dynamic resources of a fronthaul optical network based on energy efficiency optimization provided by the invention;
Fig. 3 is a schematic flow chart of a first-time adaptive resource mapping algorithm based on a descending order of a minimum switching number in the energy efficiency optimization-based dynamic resource mapping method of the fronthaul optical network provided by the invention;
Fig. 4 is a block diagram of a dynamic resource mapping system of a front-end optical network based on energy efficiency optimization.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The invention aims to provide an energy efficiency optimization-based dynamic resource mapping method and system for a forward optical network, which can reduce the total system energy consumption of the whole forward optical link and improve the resource utilization rate of a baseband processing pool.
In order that the above-recited objects, features and advantages of the present invention will become more readily apparent, a more particular description of the invention will be rendered by reference to the appended drawings and appended detailed description.
With the development of the architecture of the C-RAN and the development of the network function virtualization (Network Functions Virtualization, NFV) technology, it is necessary to combine an artificial intelligence method with a network control technology, overall plan network resources according to the dynamic requirements of services, construct an intelligent network control policy capable of responding to environmental changes in time, improve the utilization rate of spectrum resources, and further improve the performance of the whole system. Thus, the problem can be summarized as two sub-problems: 1) Exploring the dynamic characteristics of the business demands, for example, by predicting the resource demands in the network, providing references for on-demand allocation and dynamic adjustment of network resources; 2) According to the prediction result, a strategy for dynamically adjusting and migrating the network resources is designed, and the utilization rate of the network resources is maximized.
Fig. 1 is a flowchart of a method for mapping dynamic resources of a forwarding optical network based on energy efficiency optimization, as shown in fig. 1, and the method for mapping dynamic resources of the forwarding optical network based on energy efficiency optimization includes:
Step 101: collecting historical flow data of a remote radio frequency node; the historical flow data is the embodiment of service interaction between a user and a base station in a long time, and the data types of the historical flow data comprise telephone service, short message service, internet service and other data services which are received and transmitted; the historical traffic data characterizes real traffic load conditions of a plurality of historical time nodes.
The step 101 further includes: preprocessing the historical flow data to generate preprocessed historical flow data; the preprocessing comprises data normalization processing and data dimension-lifting processing.
In practical application, because of the difference of the collected historical flow data dimension and measurement units, the range of the eigenvalue distribution is wide, and when the optimal solution is solved by using the gradient descent method, the difference of the eigenvalue range can cause the non-optimal searching direction of the gradient direction at a plurality of positions. Therefore, when the long-term memory network is used for traffic prediction, the data samples should be preprocessed first, and the feature values of different dimensions should be normalized to the same value range, so as to improve the convergence rate.
Preprocessing the collected historical traffic data includes:
Normalizing the flow data of RRH nodes in a data set by using a scaling normalization method, normalizing the value of each feature to be between 0 and 1, wherein the normalized value of the feature x can be expressed as:
the training data is normalized in the same proportion, For normalized characteristic x values, x (i) is the raw traffic load data collected by the network, max i(x(i)) and min i(x(i)) represent the maximum and minimum values, respectively.
In the training process of the long-short-term memory network, the error between the predicted value and the actual value in the iterative process needs to be judged, and the original flow load change condition is a two-dimensional time-varying sequence, so that the two-dimensional flow time-varying data in the original data set needs to be converted into a three-dimensional time-series data set. In the three-dimensional data, the input data of the long-short-period memory network is two-dimensional data, the change condition of flow data in a period of time along with time is represented, the target flow value corresponding to the two-dimensional array is the flow data of the next time node after the continuous time sequence, and each group of two-dimensional training data corresponds to one test value.
Step 102: and training a long-short-period memory network model in deep learning according to the historical flow data, and predicting the flow load condition of the next time node of each remote radio frequency node.
The step 102 specifically includes: taking the actual flow load conditions of a plurality of historical time nodes of each remote radio frequency node as input, and taking the flow load condition of the next time node as output, training the long-term and short-term memory network model; in each iteration process, comparing a predicted value corresponding to the predicted flow load condition with a true value corresponding to the true flow load condition, and adjusting the weight of the long-period memory network model according to the error between the predicted value and the true value until the error requirement of training is met, so as to complete the long-period memory network model after training; and predicting the traffic load condition of the next time node of each remote radio frequency node by using the trained long-term and short-term memory network model.
In practical application, the collected historical flow data has the characteristics of long time and large data volume.
A Long Short-Term Memory (LSTM) is a typical cyclic recurrent neural network (Recurrent Neural Network, RNN), and when a deep learning method is used to process a problem of continuous time series, the RNN is a more common model, because the RNN also considers information of a previous time node when calculating the content of the current time node, that is, takes the output of a hidden node of the previous time node t-1 as the input of the current time node when the current time node t performs data operation. However, in the RNN operation process, there is a problem of "long-term dependence", and after many time node operations, the features of the previous longer time node have been forgotten, so that when the gradient is calculated by using the error back propagation method, there is a problem that the gradient disappears, and the local optimal solution is easily trapped. The LSTM network can solve the task which can not be solved by the prior recurrent neural network RNN learning algorithm due to the unique structure of the LSTM network, and has a very good effect on predicting events with longer time intervals.
The concept of a long and short term memory network access gate (gate) is used to control the proportion of feature values passing through. As shown in fig. 2, the long-term memory network includes three gates: an Input Gate (Input Gate), an Output Gate (Output Gate), and a forget Gate (Forget Gate). In order to solve the problem of early memory forgetting encountered by RNN, the long-term and short-term memory network not only includes short-term memory values h t generated by each hidden layer, but also adds a cell state chain throughout the whole network to store long-term memory.
The first step in the long-short term memory network is to control the proportion of information forgotten in the cells of the current hidden layer by using a forgetting gate f t, which can be expressed as:
ft＝σ(Wf[ht-1,xt]+bf)
Wherein, h t-1 and x t respectively represent the output value of the previous hidden layer and the newly input information of the current hidden layer, W f and b f respectively represent the weight matrix and Bayesian vector of the forgetting gate, sigma represents the Sigmoid function, and the forgetting gate gives a value between 0 and1 to represent the proportion of the forgetting content at this time based on the contents of the two parts of information. The next step is to determine how the newly entered information is added to the cell state, which includes two aspects, the first, input gate i t, determines the method of updating the input value; second, candidate vector values C t are calculated using the tanh function, expressed as:
it＝σ(Wi[ht-1,xt]+bi)
Ct＝tanh(WC[ht-1,xt]+bc)
Wherein W i,bi and W C,bc represent the input gates and the weight matrix and bayesian vectors, respectively, when calculating candidate vector values.
Based on the previous steps, the state of the t cells is updated from C t-1 to C t at time t, denoted as:
Ct＝ft*Ct-1+it*Ct
The output gate o t is used to calculate the result of each hidden layer output, and can be expressed as:
ot＝σ(Wo[ht-1,xt]+bo)
Wherein W o and b o represent the weight matrix and bayesian vector of the output gate, respectively.
The output result is determined by the output gate and the cell state of the current hidden layer together:
ht＝ot*tanh(Ct)
Further, through training the long-term memory network, updating the weight matrix W and the Bayesian variable b, in the training process, if the output result of the output layer is different from the target result, starting to update the network weight by using error reflection, and updating the weight of each hidden layer by using a gradient descent method:
W＝W-η▽Ep(W)
where eta represents the step size of the update, The partial derivative of the output deviation to a certain hidden layer weight when the sample p is input, i.e. the weight is updated every time a new sample is input. During the training process, the weights are continually adjusted until the error is reduced to an acceptable level.
Step 103: according to the traffic load condition, aiming at minimizing the total energy consumption of the system, dynamically adjusting the connection relation of a forward optical link between a baseband processing pool of the next time node and a remote radio frequency node; the minimum system total energy consumption comprises inherent energy consumption of a baseband processing pool, energy consumption generated in the process of executing baseband signal processing by the baseband processing pool and energy consumption generated by switching; the energy consumption generated by the switching comprises the energy consumption generated by the switching of the baseband processing units in the baseband processing pool between an activated state and a closed state and the energy consumption generated by the switching of the remote radio frequency node in different baseband processing units.
The energy consumption generated in the process of executing the baseband signal processing in the baseband processing pool is generated by fully utilizing the resources of each activated baseband processing unit through the number of the baseband processing units in the activated state in the baseband processing pool.
The energy consumption generated by the switching is the energy consumption generated by the switching of the baseband processing units in the baseband processing pool between the active state and the off state and the energy consumption generated by the switching of the remote radio frequency nodes in different baseband processing units.
The invention can be interpreted as a box loading problem, the box represents the baseband processing units, the volume of the box is the maximum flow load which can be borne by each baseband processing unit, the size of the object is the current load condition of each far-end radio frequency node, the size of the object changes with time, and the box loading operation represents the adjustment of the connection relation between the baseband processing units and the far-end radio frequency nodes.
Compared with the classical boxing problem, the problem of dynamic resource mapping between the baseband processing unit and the remote radio frequency node, which is needed to be solved by the invention, is to generate the switching energy consumption as little as possible, thereby minimizing the energy consumption of the whole system. If the conventional approximate algorithm for solving the boxing problem is directly adopted to solve the problem by a First-order adaptive algorithm (First-FIT DECREASING, FFD), that is, the connection condition between the baseband processing unit of the previous time node and the remote radio frequency node is not considered to determine which baseband processing units are activated and deactivated, excessive additional energy consumption may be caused by too frequent switching. Therefore, the invention provides a descending first-adaption resource mapping algorithm (MS-FFD) based on a minimum switching number (Minimum Switches, MS) based on a descending first-adaption algorithm idea.
Referring to fig. 3, fig. 3 is a schematic flow chart of a first-time adaptive resource mapping algorithm based on a descending order of a minimum switching number in the energy efficiency optimization-based dynamic resource mapping method of the preceding optical network, which specifically includes the following steps:
S301, according to the flow prediction result of the remote radio nodes, the flow loads of all the remote radio nodes are arranged in descending order according to the numerical value.
And S302, taking out the execution boxing operation with the maximum flow load in the unplanned remote radio frequency node.
S303, judging whether the baseband processing unit connected with the remote radio frequency node at the last time is in an activated state, if so, executing step 304. If not, go to step 306.
S304, judging whether the baseband processing unit selected in S303 has enough bandwidth resources to process the load flow of the current remote radio node, if so, executing step S305, and if not, executing step 306.
And S305, maintaining the connection relation between the baseband processing unit selected in the S303 and the remote radio frequency node extracted in the S302.
S306, traversing other activated baseband processing units, judging whether enough bandwidth resources exist for processing the bandwidth resources of the remote radio frequency node extracted in S302, if yes, executing step S307, and if no, executing step S308.
S307, the baseband processing unit with enough bandwidth resource which is encountered for the first time is connected with the remote radio frequency node in the traversal process.
S308, starting a new baseband processing unit, and connecting the remote radio frequency node extracted in S302.
S309, updating the working states and the load conditions of all the baseband processing units.
The method for mapping the dynamic resources of the front-end optical network based on energy efficiency optimization greatly reduces the energy consumption in a baseband processing pool by adopting the active dynamic adjustment mode of the resources of the front-end optical network, and improves the utilization rate of the baseband resources in the front-end optical network. In addition, the long-term memory network and the short-term memory network in the deep learning are introduced into the resource mapping process of the 5G-oriented front-end optical network, so that references and ideas are provided for the industrialized application of the future artificial intelligence technology in the cloud wireless access network.
Correspondingly, the invention also provides an energy efficiency optimization-based forward optical network dynamic resource mapping system, which can realize all the flows of the energy efficiency optimization-based forward optical network dynamic resource mapping method.
Fig. 4 is a block diagram of a dynamic resource mapping system of a forwarding optical network based on energy efficiency optimization, as shown in fig. 4, and the dynamic resource mapping system of the forwarding optical network based on energy efficiency optimization includes:
A historical traffic data collection module 401, configured to collect historical traffic data of the remote radio frequency node; the data types of the historical flow data comprise telephone service, short message service and Internet service which are received and sent; the historical traffic data characterizes real traffic load conditions of a plurality of historical time nodes.
And the traffic load condition prediction module 402 is configured to train a long-short term memory network model in deep learning according to the historical traffic data, and predict the traffic load condition of a next time node of each remote radio frequency node.
A dynamic adjustment module 403, configured to dynamically adjust, according to the traffic load situation, a connection relationship between a baseband processing pool of a next time node and a forward optical link between a remote radio node with a goal of minimizing total system energy consumption; the minimum system total energy consumption comprises inherent energy consumption of a baseband processing pool, energy consumption generated in the process of executing baseband signal processing by the baseband processing pool and energy consumption generated by switching; the energy consumption generated by the switching comprises the energy consumption generated by the switching of the baseband processing units in the baseband processing pool between an activated state and a closed state and the energy consumption generated by the switching of the remote radio frequency node in different baseband processing units.
The invention also includes: the preprocessing module is used for preprocessing the historical flow data and generating preprocessed historical flow data; the preprocessing comprises data normalization processing and data dimension-lifting processing.
The traffic load condition prediction module 402 specifically includes: the training unit is used for training the long-term and short-term memory network model by taking the real traffic load conditions of a plurality of historical time nodes of each remote radio frequency node as input and the traffic load condition of the next time node as output; the long-period memory network model building unit after training is used for comparing a predicted value corresponding to the predicted flow load condition with a true value corresponding to the true flow load condition in each iteration process, and adjusting the weight of the long-period memory network model according to the error between the predicted value and the true value until the error requirement of training is met, so as to complete the long-period memory network model after training; and the traffic load condition prediction unit is used for predicting the traffic load condition of the next time node of each remote radio frequency node by using the trained long-short-term memory network model.
The energy efficiency optimization-based dynamic resource mapping system of the front-end optical network can collect long-time historical flow data of the far-end radio frequency nodes, train the long-term memory network by utilizing the affiliated historical flow data, obtain the flow prediction result of the next time node by utilizing the trained model, further dynamically adjust the connection relation between the baseband processing unit and the far-end radio frequency nodes in the front-end optical network, realize more efficient resource allocation, reduce the energy consumption of the whole system, improve the resource utilization rate of the front-end optical network and meet the access requirement of massive users.
In the present specification, each embodiment is described in a progressive manner, and each embodiment is mainly described in a different point from other embodiments, and identical and similar parts between the embodiments are all enough to refer to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant points refer to the description of the method section.
The principles and embodiments of the present invention have been described herein with reference to specific examples, the description of which is intended only to assist in understanding the methods of the present invention and the core ideas thereof; also, it is within the scope of the present invention to be modified by those of ordinary skill in the art in light of the present teachings. In view of the foregoing, this description should not be construed as limiting the invention.
Claims (8)
1. The method for mapping the dynamic resources of the forward optical network based on the energy efficiency optimization is characterized by comprising the following steps of:
Collecting historical flow data of a remote radio frequency node; the data types of the historical flow data comprise telephone service, short message service and Internet service which are received and sent; the historical flow data represents the actual flow load conditions of a plurality of historical time nodes;
Training a long-short-term memory network model in deep learning according to the historical flow data, and predicting the flow load condition of the next time node of each remote radio frequency node;
According to the traffic load condition, aiming at minimizing the total energy consumption of the system, dynamically adjusting the connection relation of a forward optical link between a baseband processing pool of the next time node and a remote radio frequency node; the total energy consumption of the system comprises inherent energy consumption of a baseband processing pool, energy consumption generated in the process of executing baseband signal processing by the baseband processing pool and energy consumption generated by switching; the energy consumption generated by the switching comprises the energy consumption generated by the switching of a baseband processing unit in the baseband processing pool between an activated state and a closed state and the energy consumption generated by the switching of a remote radio frequency node in different baseband processing units; the dynamic adjustment method for dynamically adjusting the connection relation of the forward optical link between the baseband processing pool of the next time node and the remote radio frequency node is regarded as a packaging problem according to the traffic load condition and with the aim of minimizing the total energy consumption of the system; the box represents the baseband processing unit, the volume of the box is the maximum flow load that each baseband processing unit can bear, the size of the article is the current moment load condition of each far-end radio frequency node, the size of the article changes with time, and the boxing operation is to adjust the connection relation between the baseband processing unit and the far-end radio frequency node.
2. The method for mapping dynamic resources of an optical network according to claim 1, wherein the collecting historical traffic data of a remote radio node further comprises:
preprocessing the historical flow data to generate preprocessed historical flow data; the preprocessing comprises data normalization processing and data dimension-lifting processing.
3. The method for mapping dynamic resources of an optical network according to claim 1, wherein training a long-short-term memory network model in deep learning according to the historical traffic data predicts traffic load conditions of a next time node of each remote radio frequency node, and specifically comprises:
taking the actual flow load conditions of a plurality of historical time nodes of each remote radio frequency node as input, and taking the flow load condition of the next time node as output, training the long-term and short-term memory network model;
In each iteration process, comparing a predicted value corresponding to the predicted flow load condition with a true value corresponding to the true flow load condition, and adjusting the weight of the long-period memory network model according to the error between the predicted value and the true value until the error requirement of training is met, so as to complete the long-period memory network model after training;
And predicting the traffic load condition of the next time node of each remote radio frequency node by using the trained long-term and short-term memory network model.
4. The method for mapping dynamic resources of an optical network according to claim 1, wherein the energy consumption generated during the baseband processing of the baseband processing pool is generated by using the number of baseband processing units in an active state in the baseband processing pool, so that the resources of each active baseband processing unit are fully utilized.
5. The energy efficiency optimization-based dynamic resource mapping method of the forward optical network according to claim 1, wherein the boxing problem is solved by utilizing a descending order first adaptive resource mapping algorithm based on a minimum switching number;
the step of the descending order first adapting resource mapping algorithm based on the minimum switching number comprises the following steps:
According to the current traffic load conditions of the remote radio nodes, all the traffic load conditions of the remote radio nodes are arranged in descending order according to traffic load values;
Selecting out the execution boxing operation with the maximum flow load in the unplanned remote radio frequency node;
Judging whether a baseband processing unit connected with the current remote radio frequency node at the previous time is in an activated state or not to obtain a first judging result;
If the first judgment result is yes, judging whether the selected baseband processing unit has enough bandwidth resources to process the load flow of the current remote radio node, and obtaining a second judgment result;
if the second judgment result is yes, maintaining the connection relation between the selected baseband processing unit and the extracted far-end radio frequency node;
If the second judgment result is negative, traversing the remaining activated baseband processing units, judging whether enough bandwidth resources exist to process the extracted bandwidth resources of the remote radio frequency node, and obtaining a third judgment result;
if the third judgment result is yes, connecting a baseband processing unit with enough bandwidth resources, which is encountered for the first time, with the current remote radio frequency node in the traversal process;
if the third judging result is negative, starting a new baseband processing unit, and connecting the new baseband processing unit with the extracted remote radio frequency node;
If the first judgment result is negative, executing a third judgment process; the third judging process is to traverse the remaining activated baseband processing units and judge whether enough bandwidth resources exist to process the extracted bandwidth resources of the remote radio nodes so as to obtain a third judging result;
and updating the working states and the load conditions of all the baseband processing units.
6. The utility model provides a forward optical network dynamic resource mapping system based on energy efficiency optimization which characterized in that includes:
The historical flow data collection module is used for collecting historical flow data of the remote radio frequency node; the data types of the historical flow data comprise telephone service, short message service and Internet service which are received and sent; the historical flow data represents the actual flow load conditions of a plurality of historical time nodes;
The traffic load condition prediction module is used for training a long-period memory network model in deep learning according to the historical traffic data and predicting the traffic load condition of the next time node of each remote radio frequency node;
The dynamic adjustment module is used for dynamically adjusting the connection relation of the forward optical link between the baseband processing pool of the next time node and the remote radio frequency node according to the traffic load condition and with the aim of minimizing the total energy consumption of the system; the minimum system total energy consumption comprises inherent energy consumption of a baseband processing pool, energy consumption generated in the process of executing baseband signal processing by the baseband processing pool and energy consumption generated by switching; the energy consumption generated by the switching comprises the energy consumption generated by the switching of a baseband processing unit in the baseband processing pool between an activated state and a closed state and the energy consumption generated by the switching of a remote radio frequency node in different baseband processing units; the dynamic adjustment method for dynamically adjusting the connection relation of the forward optical link between the baseband processing pool of the next time node and the remote radio frequency node is regarded as a packaging problem according to the traffic load condition and with the aim of minimizing the total energy consumption of the system; the box represents the baseband processing unit, the volume of the box is the maximum flow load that each baseband processing unit can bear, the size of the article is the current moment load condition of each far-end radio frequency node, the size of the article changes with time, and the boxing operation is to adjust the connection relation between the baseband processing unit and the far-end radio frequency node.
7. The energy efficiency optimization based dynamic resource mapping system of a forward optical network of claim 6, further comprising:
the preprocessing module is used for preprocessing the historical flow data and generating preprocessed historical flow data; the preprocessing comprises data normalization processing and data dimension-lifting processing.
8. The energy efficiency optimization-based dynamic resource mapping system of the optical network of the preceding claim 7, wherein the traffic load condition prediction module specifically comprises:
The training unit is used for training the long-term and short-term memory network model by taking the real traffic load conditions of a plurality of historical time nodes of each remote radio frequency node as input and the traffic load condition of the next time node as output;
The long-period memory network model building unit after training is used for comparing a predicted value corresponding to the predicted flow load condition with a true value corresponding to the true flow load condition in each iteration process, and adjusting the weight of the long-period memory network model according to the error between the predicted value and the true value until the error requirement of training is met, so as to complete the long-period memory network model after training;
And the traffic load condition prediction unit is used for predicting the traffic load condition of the next time node of each remote radio frequency node by using the trained long-short-term memory network model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111444691.0A CN114126019B (en) | 2021-11-30 | 2021-11-30 | Energy efficiency optimization-based dynamic resource mapping method and system for forward optical network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111444691.0A CN114126019B (en) | 2021-11-30 | 2021-11-30 | Energy efficiency optimization-based dynamic resource mapping method and system for forward optical network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114126019A CN114126019A (en) | 2022-03-01 |
| CN114126019B true CN114126019B (en) | 2024-04-23 |
Family
ID=80368721
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111444691.0A Active CN114126019B (en) | 2021-11-30 | 2021-11-30 | Energy efficiency optimization-based dynamic resource mapping method and system for forward optical network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114126019B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116390165B (en) * | 2023-03-09 | 2023-11-14 | 中南大学 | Dynamic management method and system for data-driven cellular base station |
Citations (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103945548A (en) * | 2014-04-29 | 2014-07-23 | 西安电子科技大学 | Resource distribution system and task/service scheduling method in C-RAN |
| CN105050170A (en) * | 2015-06-30 | 2015-11-11 | 东南大学 | Base station dormancy method based on flow prediction in heterogeneous network |
| CN106374996A (en) * | 2016-08-29 | 2017-02-01 | 北京邮电大学 | Optical network fault handling method and apparatus |
| CN106793045A (en) * | 2017-03-01 | 2017-05-31 | 安徽工程大学 | Cloud wirelessly networking system and its Energy dissipation method |
| WO2018093301A1 (en) * | 2016-11-16 | 2018-05-24 | Telefonaktiebolaget Lm Ericsson (Publ) | Methods and devices for adapting load on a fronthaul network |
| WO2018113797A1 (en) * | 2016-12-23 | 2018-06-28 | 中国移动通信有限公司研究院 | Front-haul transport network, data transmission method, apparatus and computer storage medium |
| CN109639470A (en) * | 2018-11-30 | 2019-04-16 | 四川安迪科技实业有限公司 | VSAT satellite communication system bandwidth distribution method based on star-like networking |
| CN109831808A (en) * | 2019-02-25 | 2019-05-31 | 重庆邮电大学 | A kind of resource allocation methods of the hybrid power supply C-RAN based on machine learning |
| WO2019137513A1 (en) * | 2018-01-15 | 2019-07-18 | Huawei Technologies Co., Ltd. | Method and apparatus for front-haul network dependent wireless network control |
| CN110839184A (en) * | 2019-10-15 | 2020-02-25 | 北京邮电大学 | Method and device for adjusting bandwidth of mobile fronthaul optical network based on flow prediction |
| CN110915251A (en) * | 2017-06-30 | 2020-03-24 | 意大利电信股份公司 | Forward-flow congestion control in mobile communication network |
| CN111586502A (en) * | 2020-03-26 | 2020-08-25 | 北京邮电大学 | Resource allocation method and system in elastic optical network |
| WO2020206705A1 (en) * | 2019-04-10 | 2020-10-15 | 山东科技大学 | Cluster node load state prediction-based job scheduling method |
| WO2020248574A1 (en) * | 2019-06-12 | 2020-12-17 | 苏州大学 | Energy consumption optimization method and system for wireless access spectrum flexible optical network |
| CN112584386A (en) * | 2020-11-17 | 2021-03-30 | 国网安徽省电力有限公司信息通信分公司 | 5G C-RAN resource prediction and allocation method and system |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10271339B2 (en) * | 2013-09-26 | 2019-04-23 | Nec Corporation | Radio base station apparatus and resource allocation method |
| EP3310123B1 (en) * | 2015-12-21 | 2019-11-06 | Huawei Technologies Co., Ltd. | Data transmission method, remote radio unit (rru) and baseband unit (bbu) |
| KR102396635B1 (en) * | 2016-02-15 | 2022-05-12 | 주식회사 에치에프알 | System for Compensating for the Latency caused by Switchover from the Fronthaul Ring Topology |
| US10965786B2 (en) * | 2018-10-31 | 2021-03-30 | At&T Intellectual Property I, L.P. | Adaptive fixed point mapping for uplink and downlink fronthaul |
| CN111130839B (en) * | 2019-11-04 | 2021-07-16 | 清华大学 | Flow demand matrix prediction method and system |
-
2021
- 2021-11-30 CN CN202111444691.0A patent/CN114126019B/en active Active
Patent Citations (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103945548A (en) * | 2014-04-29 | 2014-07-23 | 西安电子科技大学 | Resource distribution system and task/service scheduling method in C-RAN |
| CN105050170A (en) * | 2015-06-30 | 2015-11-11 | 东南大学 | Base station dormancy method based on flow prediction in heterogeneous network |
| CN106374996A (en) * | 2016-08-29 | 2017-02-01 | 北京邮电大学 | Optical network fault handling method and apparatus |
| WO2018093301A1 (en) * | 2016-11-16 | 2018-05-24 | Telefonaktiebolaget Lm Ericsson (Publ) | Methods and devices for adapting load on a fronthaul network |
| WO2018113797A1 (en) * | 2016-12-23 | 2018-06-28 | 中国移动通信有限公司研究院 | Front-haul transport network, data transmission method, apparatus and computer storage medium |
| CN106793045A (en) * | 2017-03-01 | 2017-05-31 | 安徽工程大学 | Cloud wirelessly networking system and its Energy dissipation method |
| CN110915251A (en) * | 2017-06-30 | 2020-03-24 | 意大利电信股份公司 | Forward-flow congestion control in mobile communication network |
| WO2019137513A1 (en) * | 2018-01-15 | 2019-07-18 | Huawei Technologies Co., Ltd. | Method and apparatus for front-haul network dependent wireless network control |
| CN109639470A (en) * | 2018-11-30 | 2019-04-16 | 四川安迪科技实业有限公司 | VSAT satellite communication system bandwidth distribution method based on star-like networking |
| CN109831808A (en) * | 2019-02-25 | 2019-05-31 | 重庆邮电大学 | A kind of resource allocation methods of the hybrid power supply C-RAN based on machine learning |
| WO2020206705A1 (en) * | 2019-04-10 | 2020-10-15 | 山东科技大学 | Cluster node load state prediction-based job scheduling method |
| WO2020248574A1 (en) * | 2019-06-12 | 2020-12-17 | 苏州大学 | Energy consumption optimization method and system for wireless access spectrum flexible optical network |
| CN110839184A (en) * | 2019-10-15 | 2020-02-25 | 北京邮电大学 | Method and device for adjusting bandwidth of mobile fronthaul optical network based on flow prediction |
| CN111586502A (en) * | 2020-03-26 | 2020-08-25 | 北京邮电大学 | Resource allocation method and system in elastic optical network |
| CN112584386A (en) * | 2020-11-17 | 2021-03-30 | 国网安徽省电力有限公司信息通信分公司 | 5G C-RAN resource prediction and allocation method and system |
Non-Patent Citations (6)
| Title |
|---|
| 5G前传光网络:关键问题、重要特征与发展目标;纪越峰;张佳玮;;中国科学:信息科学(第10期);全文 * |
| Bo Tian ; Qi Zhang ; Xiangjun Xin ; Qinghua Tian ; Xiangyu Wu ; Ying Tao ; Yufei Shen ; Guixing Cao ; Naijin Liu.Recursive_Neural_Network_Based_RRH_to_BBU_Resource_Allocation_in_5G_Fronthaul_Network.Recursive_Neural_Network_Based_RRH_to_BBU_Resource_Allocation_in_5G_Fronthaul_Network.2018,全文. * |
| Fu Wang ; Haipeng Yao ; Qi Zhang ; Jingjing Wang ; Ran Gao ; Dong Guo ; Mohsen Guizani.Dynamic Distributed Multi-Path Aided Load Balancing for Optical Data Center Networks.IEEE TRANSACTIONS ON NETWORK AND SERVICE MANAGEMENT.2021,全文. * |
| 一种基于C-RAN架构下资源负载预测方法;龙恳;胡渝纺;;科学技术与工程(第33期);全文 * |
| 云化移动接入网中前传链路负载自适应技术;李隆胜;毕美华;胡卫生;;聊城大学学报(自然科学版)(第02期);全文 * |
| 光通信中的新型复用形式;及睿，忻向军，张琦;新型工业化(第08期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114126019A (en) | 2022-03-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111988796B (en) | Dual-mode communication-based system and method for optimizing platform information acquisition service bandwidth | |
| CN111641681A (en) | Internet of things service unloading decision method based on edge calculation and deep reinforcement learning | |
| Piovesan et al. | Joint load control and energy sharing for renewable powered small base stations: A machine learning approach | |
| CN111901168B (en) | Network slice resource allocation method suitable for electric automobile charging and switching network | |
| CN112887142B (en) | Method for generating web application slice in virtualized wireless access edge cloud | |
| Hu et al. | Edge intelligence for real-time data analytics in an IoT-based smart metering system | |
| CN106793031A (en) | Based on the smart mobile phone energy consumption optimization method for gathering competing excellent algorithm | |
| Cicioğlu | Multi-criteria handover management using entropy‐based SAW method for SDN-based 5G small cells | |
| CN108834079A (en) | A kind of load balance optimization method based on mobility prediction in heterogeneous network | |
| CN114126019B (en) | Energy efficiency optimization-based dynamic resource mapping method and system for forward optical network | |
| CN114885388B (en) | Multi-service type self-adaptive switching judgment method combining RSS prediction | |
| Sun et al. | MEET: Mobility-enhanced edge intelligence for smart and green 6G networks | |
| CN115622603A (en) | Age minimization optimization method for auxiliary transmission information | |
| Temesgene et al. | Dynamic control of functional splits for energy harvesting virtual small cells: A distributed reinforcement learning approach | |
| Li et al. | Task computation offloading for multi-access edge computing via attention communication deep reinforcement learning | |
| Tan et al. | Uav control in smart city based on space-air-ground integrated network | |
| CN117149351A (en) | Prediction-based edge collaborative computing migration method and system | |
| CN111225398A (en) | Micro-grid wireless sensor network energy consumption optimization method based on cooperative coverage | |
| Abdullah et al. | Weather-aware fiber-wireless traffic prediction using graph convolutional networks | |
| Chang et al. | Low-latency controller load balancing strategy and offloading decision generation algorithm based on lyapunov optimization in SDN mobile edge computing environment | |
| Maksymyuk et al. | Artificial intelligence based 5G coverage design and optimization using deep generative adversarial neural networks | |
| Lin et al. | Hypergraph-based autonomous networks: adaptive resource management and dynamic resource scheduling | |
| Poulkov | Dynamic access points grouping for mobility driven user-centric wireless networks | |
| Zhang et al. | A Resource Allocation Method of Deep Reinforcement Transfer Learning in WSN | |
| Zhong et al. | Traffic prediction of converged network for smart gird based on GNN and LSTM |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |