CN115484520B - Frequency spectrum distribution network obtaining method and frequency spectrum distribution method of elastic optical network - Google Patents
Frequency spectrum distribution network obtaining method and frequency spectrum distribution method of elastic optical network Download PDFInfo
- Publication number
- CN115484520B CN115484520B CN202211359107.6A CN202211359107A CN115484520B CN 115484520 B CN115484520 B CN 115484520B CN 202211359107 A CN202211359107 A CN 202211359107A CN 115484520 B CN115484520 B CN 115484520B
- Authority
- CN
- China
- Prior art keywords
- network
- spectrum allocation
- objective function
- parameters
- function
- 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
- 238000001228 spectrum Methods 0.000 title claims abstract description 193
- 238000000034 method Methods 0.000 title claims abstract description 54
- 230000003287 optical effect Effects 0.000 title claims abstract description 49
- 238000009826 distribution Methods 0.000 title claims abstract description 38
- 238000012549 training Methods 0.000 claims abstract description 49
- 230000006870 function Effects 0.000 claims description 135
- 238000011156 evaluation Methods 0.000 claims description 11
- 230000008901 benefit Effects 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 5
- 230000008859 change Effects 0.000 abstract description 2
- 230000008569 process Effects 0.000 description 12
- 238000010586 diagram Methods 0.000 description 8
- 238000003860 storage Methods 0.000 description 8
- 238000004590 computer program Methods 0.000 description 7
- 238000012545 processing Methods 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 230000002787 reinforcement Effects 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 239000013307 optical fiber Substances 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 230000000903 blocking effect Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000000638 solvent extraction Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04Q—SELECTING
- H04Q11/00—Selecting arrangements for multiplex systems
- H04Q11/0001—Selecting arrangements for multiplex systems using optical switching
- H04Q11/0062—Network aspects
-
- 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
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L47/00—Traffic control in data switching networks
- H04L47/70—Admission control; Resource allocation
- H04L47/82—Miscellaneous aspects
- H04L47/829—Topology based
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04Q—SELECTING
- H04Q11/00—Selecting arrangements for multiplex systems
- H04Q11/0001—Selecting arrangements for multiplex systems using optical switching
- H04Q11/0062—Network aspects
- H04Q2011/0086—Network resource allocation, dimensioning or optimisation
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Signal Processing (AREA)
- Optical Communication System (AREA)
Abstract
The application discloses a spectrum allocation network obtaining method and a spectrum allocation method of an elastic optical network; the network acquisition method comprises the steps of training a first spectrum distribution network of the elastic optical network based on a first topological structure, and obtaining a second spectrum distribution network after training, a first objective function and a plurality of corresponding initial network parameters; obtaining the importance coefficient of each initial network parameter; when a second spectrum allocation network is trained on the basis of a second topological structure, a second objective function and corresponding multiple network training parameters are obtained; correcting the second objective function according to the network training parameters, the initial network parameters and the importance coefficients of all the initial network parameters to obtain a third objective function; and training the second spectrum allocation network according to the third objective function to obtain the target spectrum allocation network. According to the method and the device, when the spectrum distribution network needs to be retrained due to the change of the topological structure, the occurrence of catastrophic forgetting can be reduced, and the compatibility of the spectrum distribution network is improved.
Description
Technical Field
The present application relates to the technical field of spectrum partitioning of an elastic optical network, and in particular, to a spectrum allocation network acquisition method and a spectrum allocation method for an elastic optical network.
Background
Key network management issues for Elastic Optical Networks (EONs) are routing, modulation and spectrum allocation (RMSA), which is usually broken down into different sub-problems due to the complexity of the problem and solved by some heuristic methods. Recently, deep Reinforcement Learning (DRL) has been introduced into the spectrum allocation problem and has better performance than the traditional heuristic rule-based approach.
In the deep reinforcement learning framework, a spectrum allocation strategy is determined by a spectrum allocation network, and network parameters of the spectrum allocation network are obtained by training in an interaction process with an optical network environment. However, the topology is used as one of the factors of the optical network environment, when the topology is changed, it indicates that the optical network environment is changed, and at this time, the spectrum allocation network needs to be retrained, however, in the retraining process, the spectrum allocation network is forgotten catastrophically, which is specifically represented by forgetting network parameters obtained by previous training, so that the topology returns to a state before being changed, and the spectrum allocation network which is forgotten catastrophically cannot be compatible with the topology before being changed, so that the corresponding spectrum allocation policy cannot be accurately output.
Disclosure of Invention
The present application aims to overcome the disadvantages and deficiencies in the prior art, and provides a spectrum allocation network acquisition method and a spectrum allocation method for an elastic optical network, which can reduce the occurrence of catastrophic forgetting when a topology structure changes and the spectrum allocation network needs to be retrained, so that the retrained spectrum allocation network can accurately output a corresponding spectrum allocation strategy based on the optical network environment before and after the topology structure changes, and improve the compatibility of the spectrum allocation network.
A first aspect of an embodiment of the present application provides a method for acquiring a spectrum allocation network of an elastic optical network, including:
training a first spectrum distribution network of an elastic optical network based on a first topological structure, and obtaining a second spectrum distribution network after training, a first objective function of the second spectrum distribution network and a plurality of initial network parameters corresponding to the first objective function;
obtaining importance coefficients of the initial network parameters according to the first objective function and the initial network parameters;
when the second spectrum allocation network is trained on the basis of a second topological structure, a second objective function and a plurality of network training parameters corresponding to the second objective function are obtained;
correcting the second objective function according to the network training parameters, the initial network parameters and the importance coefficients of the initial network parameters, and determining the corrected function as a third objective function;
and adjusting the network parameters of the second spectrum allocation network according to the third objective function so as to train the second spectrum allocation network to obtain a target spectrum allocation network.
A second aspect of an embodiment of the present application provides a spectrum allocation method for a flexible optical network, including:
obtaining the target spectrum distribution network by the spectrum distribution network obtaining method of the elastic optical network;
acquiring a target traffic request, wherein the target traffic request points to the first topological structure or the second topological structure;
and inputting the target flow request into the target spectrum allocation network to obtain a spectrum allocation strategy corresponding to the target flow request.
Compared with the prior art, the method includes the steps that a first spectrum distribution network is trained on the basis of a first topological structure, a trained second spectrum distribution network, multiple initial network parameters and importance coefficients of the initial network parameters are obtained, then a second objective function and corresponding multiple network training parameters are obtained when the second spectrum distribution network is trained on the basis of a second topological structure of the elastic optical network, and then the second objective function is corrected according to the network training parameters, the initial network parameters and the importance coefficients of the initial network parameters, a third function is obtained and used for training the second spectrum distribution network, so that the target spectrum distribution network is obtained. According to the target spectrum distribution network obtained through the training in the steps, as the importance coefficients of the initial network parameters are combined in the training process, the occurrence of catastrophic forgetting can be reduced, so that the target spectrum distribution network can accurately output a corresponding spectrum distribution strategy based on the first topological structure and the second topological structure, and the compatibility of the spectrum distribution network is improved.
Drawings
FIG. 1 is a diagram of a flexible optical network according to an embodiment of the present application is also disclosed.
Fig. 2 is a flowchart of step S4 of a spectrum allocation network acquiring method of an elastic optical network according to an embodiment of the present application.
Fig. 3 is a flowchart of a spectrum allocation method for a flexible optical network according to an embodiment of the present application.
Detailed Description
To make the objects, technical solutions and advantages of the present application more clear, embodiments of the present application will be described in detail in one embodiment with reference to the accompanying drawings.
It should be understood that the embodiments described are only some embodiments of the present application, and not all embodiments. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments in the present application without any creative effort belong to the protection scope of the embodiments in the present application.
When the following description refers to the accompanying drawings, like numbers in different drawings represent the same or similar elements unless otherwise indicated. In the description of the present application, it is to be understood that the terms "first," "second," "third," and the like are used solely to distinguish one from another and are not necessarily used to describe a particular order or sequence, nor are they to be construed as indicating or implying relative importance. The specific meaning of the above terms in the present application can be understood by those of ordinary skill in the art as appropriate. As used in this application and the appended claims, the singular forms "a", "an", and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. The word "if"/"if" as used herein may be interpreted as "at ...when" or "when ...when" or "in response to a determination".
Further, in the description of the present application, "a plurality" means two or more unless otherwise specified. "and/or" describes the association relationship of the associated objects, meaning that there may be three relationships, e.g., a and/or B, which may mean: a exists alone, A and B exist simultaneously, and B exists alone. The character "/" generally indicates that the former and latter associated objects are in an "or" relationship.
Please refer to fig. 1, which is a flowchart illustrating a spectrum allocation network acquiring method of an elastic optical network according to an embodiment of the present application, including:
s100: training a first spectrum distribution network based on a first topological structure of an elastic optical network to obtain a second spectrum distribution network after training, a first objective function of the second spectrum distribution network and a plurality of initial network parameters corresponding to the first objective function.
The elastic optical network is a wide area network, a metropolitan area network or a newly-built large-scale local area network which uses optical fibers as main transmission media, the optical fibers are connected through nodes, and a plurality of adjacent nodes are connected to form a link of the optical network. A frequency slot is a unit of data stored and transmitted in a link, and the common bandwidth of a single frequency slot is 12.5GHz.
The topology structure refers to the mutual positions and mutual connection layout conditions of all nodes in an optical network, and three common basic topology structures include a bus type, a ring type and a star type, and each topology structure can also be divided into different types of topology structures based on different node numbers and connection conditions among the nodes. The first topology is a kind of topology, and the number of nodes and the layout of the interconnections between the nodes are not limited.
The first spectrum allocation network is a neural network to be subjected to deep reinforcement learning and used for outputting a spectrum allocation strategy, the second spectrum allocation network is a neural network obtained by learning training based on the first topological structure, and when a traffic request is received, a source node, a destination node and a current candidate path meeting the spectrum requirements of the traffic request can be obtained based on the first topological structure. The starting node of the current candidate path corresponds to a source node of the flow request, the destination node corresponds to a destination node of the flow request, and a link included in the current candidate path needs to meet the spectrum requirement.
The first objective function is a function for indicating an adjustment direction of the spectrum allocation network based on the first topology structure to implement training of the first spectrum allocation network, thereby obtaining the second spectrum allocation network. The initial network parameters refer to optimal network parameters of the second spectrum allocation network.
S200: and obtaining the importance coefficient of each initial network parameter according to the first objective function and each initial network parameter.
The importance coefficients of the initial network parameters represent the importance of the initial network parameters in the second spectrum allocation network. Optionally, the importance coefficients of the initial network parameters may be obtained by performing second derivative calculation on the first objective function and the initial network parameters, or may be approximated to the square of the first derivative of the objective function to obtain the importance coefficients of the initial network parameters.
S300: and acquiring a second objective function and a plurality of network training parameters corresponding to the second objective function when the second spectrum allocation network is trained based on a second topological structure.
The second topological structure is a topological structure with the number of nodes, or the mutual positions of all the nodes, or the mutual connection relationship of all the nodes different from that of the first topological structure.
The second objective function is a function for indicating the adjustment direction of the spectrum allocation network based on the second topological structure, and the network training parameter refers to a network parameter to be trained and learned in the second spectrum allocation network.
S400: and correcting the second objective function according to the network training parameters, the initial network parameters and the importance coefficients of the initial network parameters, and determining the corrected function as a third objective function.
The third objective function is a function indicating an adjustment direction of the first spectrum allocation network based on the second topology and a negative influence on the initial network parameter during training.
S500: and adjusting the network parameters of the second spectrum allocation network according to the third objective function so as to train the second spectrum allocation network to obtain a target spectrum allocation network.
Compared with the prior art, in this embodiment, a first spectrum allocation network is trained based on a first topological structure to obtain a trained second spectrum allocation network, multiple initial network parameters and importance coefficients of the initial network parameters, a second objective function and corresponding multiple network training parameters are obtained when the second spectrum allocation network is trained based on a second topological structure of the elastic optical network, and the second objective function is corrected according to the network training parameters, the initial network parameters and the importance coefficients of the initial network parameters to obtain a third function and is used for training the second spectrum allocation network to obtain a target spectrum allocation network. According to the target spectrum distribution network obtained through the training in the steps, as the importance coefficients of the initial network parameters are combined in the training process, the occurrence of catastrophic forgetting can be reduced, so that the target spectrum distribution network can accurately output a corresponding spectrum distribution strategy based on the first topological structure and the second topological structure, and the compatibility of the spectrum distribution network is improved.
In one possible embodiment, the S100: the method comprises the steps of training a first spectrum allocation network based on a first topological structure of an elastic optical network, and obtaining a second spectrum allocation network after training, a first objective function of the second spectrum allocation network and a plurality of initial network parameters corresponding to the first objective function, wherein the steps comprise:
s101: based on a first topological structure of the elastic optical network, a first environment state when a first sample flow request is received and a second environment state when a preset number of second sample flow requests are received are obtained.
The first environment state comprises a plurality of first candidate paths corresponding to the first sample traffic request; the receiving time of the second sample traffic request is later than that of the first sample traffic request, and the second environment state includes a plurality of second candidate paths corresponding to the second sample traffic request.
S102: and inputting the first sample traffic request, the first environment state, the second sample traffic request and the second environment state into the first spectrum allocation network to obtain a plurality of spectrum allocation strategies and instant rewards corresponding to the spectrum allocation strategies.
The instant incentive indicates whether the corresponding spectrum allocation strategy can be executed or is blocked, for example, if the spectrum allocation strategy can be smoothly executed by the elastic optical network, or the spectrum allocation strategy is executed by the elastic optical network without being blocked, the corresponding instant incentive value is a positive value, and if the spectrum allocation strategy cannot be smoothly executed by the elastic optical network, or the spectrum allocation strategy is executed by the elastic optical network without being blocked, the corresponding instant incentive value is a negative value.
S103: and inputting each frequency spectrum allocation strategy into a pre-constructed frequency spectrum allocation evaluation network to obtain a plurality of corresponding value functions.
The cost function is the impact of the corresponding instantaneous reward on the future spectrum allocation situation of the resilient optical network.
S104: acquiring an advantage function according to the instantaneous rewards corresponding to the spectrum allocation strategies and the value functions; the dominance function is used to indicate a network parameter adjustment direction of the first spectrum allocation network and the spectrum allocation evaluation network.
Wherein the merit function is obtained by the following formula:
wherein,
is the merit function;
a total number of first and second sample traffic requests;
is a preset discount factor;
to receiveAn instantaneous reward of the corresponding sample traffic request;
a received first sample traffic request;
the received last second sample flow request;
is the cost function.
The spectrum allocation evaluation network is a neural network used for predicting the value of a spectrum allocation strategy output by the first spectrum allocation network, the spectrum allocation evaluation network works together with the first spectrum allocation network, and the spectrum allocation evaluation network can transmit a value function corresponding to the spectrum allocation strategy to the first spectrum allocation network so as to improve the learning accuracy of the first spectrum allocation network.
Specifically, the network parameters of the spectrum allocation evaluation network may be adjusted by the following formula:
wherein,
network parameters of the network are evaluated for the spectrum allocation,
network parameters of the network are evaluated for the adjusted spectrum allocation,
the learning rate of the network is evaluated for the spectrum allocation,
is the partial derivative.
Wherein,
the parameter expression symbols are used for distinguishing the network parameters of the spectrum allocation evaluation network before and after adjustment, so that the formula for adjusting the network parameters of the spectrum allocation evaluation network is convenient to understand. In practical tests and applications, each acquisition
Will all be
Is assigned to
And the frequency spectrum distribution evaluation network can update the parameters and realize corresponding functions in time.
S105: and obtaining the first objective function and each initial network parameter according to the spectrum allocation strategy and the advantage function.
Obtaining a first objective function by the following formula:
wherein,
in order to be said first objective function,
allocating a strategy for the frequency spectrum;
is the merit function;
entropy of the policy distribution corresponding to the first topology;
the strength of the entropy regularization term.
In this embodiment, the dominance function is obtained by obtaining the environmental state, link topology information, the corresponding spectrum allocation policy, and the corresponding instantaneous reward when the elastic optical network receives the first sample traffic request and the plurality of second sample traffic requests, and then the network parameter of the first spectrum allocation line network is adjusted according to the dominance function to obtain the second spectrum allocation network, so that the elastic optical network executes the spectrum allocation policy based on the second spectrum allocation network, thereby reducing the blocking probability when receiving other subsequent traffic requests.
In one possible embodiment, the S200: obtaining importance coefficients of the initial network parameters according to the first objective function and the initial network parameters, wherein the importance coefficients comprise:
obtaining importance coefficients of the initial network parameters by the following formula:
wherein,
for each of the importance coefficients of the initial network parameters,
the calculation coefficient is a preset calculation coefficient;
in order to be said first objective function,
for each of said initial network parameters.
In this embodiment, the importance coefficients of the initial network parameters can be calculated by the above formula.
Referring to fig. 2, in one possible embodiment, the S400: modifying the second objective function according to the network training parameters, the initial network parameters and the importance coefficients of the initial network parameters, and determining the modified function as a third objective function, including:
s401: obtaining a negative influence function according to each network training parameter, each initial network parameter and the importance coefficient of each initial network parameter; the negative influence function is the negative influence on the second spectrum allocation network when each network training parameter replaces each initial network parameter.
Since the data of each initial network parameter may become each network training parameter in the learning process, and this process may decrease the accuracy of the second spectrum allocation network outputting the spectrum allocation policy based on the first topology structure, for the second spectrum allocation network, this process is negatively affected, and this negative effect may also be affected by the importance coefficients of each initial network parameter, and therefore, according to each network training parameter, each initial network parameter, and the importance coefficient of each initial network parameter, the larger the obtained negative effect function is, the more serious the negative effect is.
Wherein, the step of obtaining a negative influence function according to each item of the network training parameters, each item of the initial network parameters, and the importance coefficients of each item of the initial network parameters includes:
obtaining the negative impact function by the following formula:
wherein,
is the negative influence function;
is the elastic coefficient of the negative influence function;
an importance coefficient for each of the initial network parameters;
training parameters for each of the networks;
for each of said initial network parameters.
S402: and correcting the second objective function according to the negative influence function to obtain a third objective function.
Wherein the step of correcting the second objective function according to the negative influence function to obtain the third objective function comprises:
obtaining the third objective function by the following formula:
wherein,
in order to be said third objective function,
is the second objective function;
is the negative impact function.
In this embodiment, the second objective function is modified according to the negative impact function, so that the third objective function combines with the negative impact function, and the third objective function can reduce the occurrence of catastrophic forgetting in the process of adjusting the network parameters of the second spectrum allocation network.
In one possible embodiment, the S500: adjusting network parameters of the second spectrum allocation network according to the third objective function to train the second spectrum allocation network to obtain a target spectrum allocation network, including:
obtaining a network parameter for adjusting the second spectrum allocation network by:
wherein,
allocating network parameters of a network for the adjusted second spectrum;
allocating network parameters of the network to be adjusted for the second spectrum,
allocating a learning rate of a network for the second spectrum;
is a cleft operator;
is the third objective function.
Wherein,
the parameter is simply represented by a symbol for distinguishing the network parameters of the second spectrum allocation network before and after the adjustment, so as to facilitate understanding of the formula for adjusting the network parameters of the second spectrum allocation network. In practical tests and applications, each acquisition
Will all be
Is assigned to
And the second spectrum allocation network can update the parameters and realize corresponding functions in time.
In this embodiment, the network parameters of the second spectrum allocation network are adjusted through the third objective function to train the second spectrum allocation network, so that the occurrence of catastrophic forgetting caused by training can be reduced.
Referring to fig. 3, an embodiment of the present application further provides a spectrum allocation method for a flexible optical network, including:
s1: the target spectrum distribution network is obtained by the spectrum distribution network obtaining method of the elastic optical network.
S2: and acquiring a target traffic request, wherein the target traffic request points to the first topological structure or the second topological structure.
The target flow request refers to a currently received flow request, the flow request includes a source node, a destination node and a spectrum demand, when the elastic optical network of the to-be-adjusted partition receives the target flow request, a plurality of current candidate paths are obtained according to the source node, the destination node and the spectrum demand and the spectrum availability distribution of the current environment state of the elastic optical network, and specifically, a plurality of candidate paths between the source node and the destination node on the topology are obtained through a shortest path algorithm (dijkstra algorithm) according to the source node and the destination node. The spectrum demand is the number of spectrum slots required by the corresponding traffic request, where a spectrum slot is a unit for storing and transmitting data in a link. The target traffic request obtained in this embodiment is directed to the first topology or the second topology, that is, a traffic request for smoothly performing spectrum allocation based on the first topology or the second topology.
S3: and inputting the target flow request into the target spectrum allocation network to obtain a spectrum allocation strategy corresponding to the target flow request.
Compared with the prior art, the embodiment inputs the target flow request into the target spectrum allocation network obtained by the spectrum allocation network obtaining method of the elastic optical network, so as to obtain the spectrum allocation strategy which can be smoothly performed based on the first topological structure or the second topological structure, and improve the compatibility of the spectrum allocation method of the elastic optical network.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart block or blocks and/or flowchart block or blocks. These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
In a typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.
The memory may include forms of volatile memory in a computer readable medium, random Access Memory (RAM) and/or non-volatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM). The memory is an example of a computer-readable medium.
Computer-readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static Random Access Memory (SRAM), dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), digital Versatile Disks (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium, which can be used to store information that can be accessed by a computing device. As defined herein, a computer readable medium does not include a transitory computer readable medium such as a modulated data signal and a carrier wave.
It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrases "comprising a," "...," or "comprising" does not exclude the presence of additional identical elements in the process, method, article, or apparatus comprising the element.
The above are merely examples of the present application and are not intended to limit the present application. Various modifications and changes may occur to those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.
Claims (10)
1. A method for acquiring a spectrum allocation network of an elastic optical network is characterized by comprising the following steps:
training a first spectrum distribution network based on a first topological structure of an elastic optical network to obtain a second spectrum distribution network after training, a first objective function of the second spectrum distribution network and a plurality of initial network parameters corresponding to the first objective function;
obtaining importance coefficients of the initial network parameters according to the first objective function and the initial network parameters;
when the second spectrum allocation network is trained based on a second topological structure, a second objective function and a plurality of network training parameters corresponding to the second objective function are obtained;
correcting the second objective function according to the network training parameters, the initial network parameters and the importance coefficients of the initial network parameters, and determining the corrected function as a third objective function;
and adjusting the network parameters of the second spectrum allocation network according to the third objective function so as to train the second spectrum allocation network to obtain a target spectrum allocation network.
2. The method according to claim 1, wherein the step of modifying the second objective function according to the network training parameters, the initial network parameters, and importance coefficients of the initial network parameters, and determining the modified function as a third objective function includes:
obtaining a negative influence function according to each network training parameter, each initial network parameter and the importance coefficient of each initial network parameter; the negative influence function is the negative influence on the second spectrum allocation network when each network training parameter replaces each initial network parameter;
and correcting the second objective function according to the negative influence function to obtain a third objective function.
3. The method according to claim 2, wherein the step of obtaining a negative impact function according to the network training parameters, the initial network parameters, and the importance coefficients of the initial network parameters includes:
obtaining the negative impact function by the following formula:
wherein,
is the negative impact function;
is the elastic coefficient of the negative influence function;
an importance coefficient for each of the initial network parameters;
training parameters for each of the networks;
for each of said initial network parameters.
4. The method according to claim 2, wherein the step of modifying the second objective function according to the negative impact function to obtain the third objective function includes:
obtaining the third objective function by the following formula:
wherein,
in order to be said third objective function,
is the second objective function;
is the negative impact function.
5. The method according to claim 1, wherein the step of obtaining the importance coefficients of the initial network parameters according to the first objective function and the initial network parameters comprises:
obtaining importance coefficients of the initial network parameters by the following formula:
wherein,
for each of the importance coefficients of the initial network parameters,
the calculation coefficient is a preset calculation coefficient;
in order to be said first objective function,
for each of said initial network parameters.
6. The method according to claim 1, wherein the adjusting the network parameter of the second spectrum allocation network according to the third objective function to train the second spectrum allocation network to obtain the target spectrum allocation network comprises:
obtaining a network parameter for adjusting the second spectrum allocation network by:
wherein,
allocating network parameters of a network for the adjusted second spectrum;
network for distributing the second spectrumThe network parameters of the network to be adjusted,
allocating a learning rate of a network for the second spectrum;
is a cleft operator;
is the third objective function;
a received first sample traffic request;
the total number of first and second sample traffic requests.
7. The method according to claim 1, wherein the step of training the first spectrum allocation network based on the first topology structure of the elastic optical network to obtain the trained second spectrum allocation network, the first objective function of the second spectrum allocation network, and a plurality of initial network parameters corresponding to the first objective function includes:
acquiring a first environment state when a first sample flow request is received and a second environment state when a preset number of second sample flow requests are received based on a first topological structure of the elastic optical network;
inputting the first sample traffic request, the first environment state, the second sample traffic request and the second environment state into the first spectrum allocation network to obtain a plurality of spectrum allocation strategies and instant rewards corresponding to the spectrum allocation strategies;
inputting each frequency spectrum allocation strategy into a pre-constructed frequency spectrum allocation evaluation network to obtain a plurality of corresponding value functions;
obtaining an advantage function according to the instantaneous rewards corresponding to the spectrum allocation strategies and the value functions; the dominance function is used for indicating the network parameter adjustment direction of the first spectrum allocation network and the spectrum allocation evaluation network;
and obtaining the first objective function and each initial network parameter according to the spectrum allocation strategy and the advantage function.
8. The method according to claim 7, wherein the step of obtaining an advantage function according to the instantaneous reward corresponding to each spectrum allocation policy and each cost function includes:
the merit function is obtained by the following formula:
wherein,
is the merit function;
a total number of first and second sample traffic requests;
a preset discount factor;
an instant reward for receiving a corresponding sample traffic request;
a received first sample traffic request;
the received last second sample flow request;
is the cost function.
9. The method according to claim 7, wherein the step of obtaining the first objective function and each of the initial network parameters according to the spectrum allocation policy and the dominance function includes:
wherein,
in order to be said first objective function,
allocating a policy for the spectrum;
is the merit function;
entropy of the policy distribution corresponding to the first topology;
the strength of the entropy regularization term.
10. A method for allocating spectrum of a flexible optical network, comprising:
obtaining the target spectrum allocation network by the method for acquiring a spectrum allocation network of an elastic optical network according to any one of claims 1 to 9;
acquiring a target traffic request, wherein the target traffic request points to the first topological structure or the second topological structure;
and inputting the target flow request into the target spectrum allocation network to obtain a spectrum allocation strategy corresponding to the target flow request.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211359107.6A CN115484520B (en) | 2022-11-02 | 2022-11-02 | Frequency spectrum distribution network obtaining method and frequency spectrum distribution method of elastic optical network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211359107.6A CN115484520B (en) | 2022-11-02 | 2022-11-02 | Frequency spectrum distribution network obtaining method and frequency spectrum distribution method of elastic optical network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115484520A CN115484520A (en) | 2022-12-16 |
| CN115484520B true CN115484520B (en) | 2023-04-18 |
Family
ID=84395259
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211359107.6A Active CN115484520B (en) | 2022-11-02 | 2022-11-02 | Frequency spectrum distribution network obtaining method and frequency spectrum distribution method of elastic optical network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115484520B (en) |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110365589B (en) * | 2019-07-30 | 2021-09-28 | 国网福建省电力有限公司 | Electric power optical transmission route and frequency spectrum allocation method based on elastic optical network |
| CN111586502B (en) * | 2020-03-26 | 2021-06-15 | 北京邮电大学 | Resource allocation method and system in elastic optical network |
| CN114584871B (en) * | 2022-04-28 | 2022-08-05 | 华南师范大学 | Spectrum allocation method, device, storage medium and equipment of elastic optical network |
| CN115225989A (en) * | 2022-06-01 | 2022-10-21 | 北京邮电大学 | Routing and resource allocation method and device in elastic optical network |
-
2022
- 2022-11-02 CN CN202211359107.6A patent/CN115484520B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN115484520A (en) | 2022-12-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11410046B2 (en) | Learning-based service migration in mobile edge computing | |
| CN110264326B (en) | Method, device and equipment for identifying abnormal account set and risk account set | |
| CN104052811A (en) | Service scheduling method and device and system | |
| EP2962427B1 (en) | Method and system to represent the impact of load variation on service outage over multiple links | |
| US20150327135A1 (en) | Apparatus and method for dynamic hybrid routing in sdn networks to avoid congestion and balance loads under changing traffic load | |
| WO2017215378A1 (en) | Software-defined network, node, path calculation method and device, and storage medium | |
| Rkhami et al. | Learn to improve: A novel deep reinforcement learning approach for beyond 5G network slicing | |
| CN114072820A (en) | Executing machine learning models | |
| CN114650249A (en) | Algorithm model and path determination method, electronic device, SDN controller and medium | |
| CN109617806B (en) | Data traffic scheduling method and device | |
| CN105634974A (en) | Route determining method and apparatus in software-defined networking | |
| CN107168777A (en) | The dispatching method and device of resource in distributed system | |
| CN114302407A (en) | Network decision method and device, electronic equipment and storage medium | |
| JP7436747B2 (en) | OTN network resource optimization method and apparatus, computing device and storage medium | |
| CN116541106A (en) | Computing task unloading method, computing device and storage medium | |
| CN115484520B (en) | Frequency spectrum distribution network obtaining method and frequency spectrum distribution method of elastic optical network | |
| CN113965837B (en) | Optical network dynamic spectrum partitioning method, device, storage medium and computer equipment | |
| CN115809684A (en) | Network model heterogeneous deployment method, device, equipment and storage medium | |
| Chaurasia et al. | Trust computation in VANET cloud | |
| CN113708982A (en) | Service function chain deployment method and system based on group learning | |
| JP7428831B2 (en) | OTN network resource optimization method and apparatus, computing device and storage medium | |
| CN115988364A (en) | Spectrum allocation network acquisition method of optical network, storage medium and computer device | |
| CN116932190A (en) | Method for determining mapping relation and operation of calculation map, many-core system and readable medium | |
| CN116980420B (en) | Cluster communication method, system, device, equipment and medium | |
| CN111279765B (en) | Method, system and computer program product for identifying endpoints in a network |
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 |