CN112383846B - Cloud-fog elastic optical network-oriented spectrum resource allocation method for advance reservation request - Google Patents
Cloud-fog elastic optical network-oriented spectrum resource allocation method for advance reservation request Download PDFInfo
- Publication number
- CN112383846B CN112383846B CN202011271498.7A CN202011271498A CN112383846B CN 112383846 B CN112383846 B CN 112383846B CN 202011271498 A CN202011271498 A CN 202011271498A CN 112383846 B CN112383846 B CN 112383846B
- Authority
- CN
- China
- Prior art keywords
- time
- service request
- spectrum
- path
- cloud
- 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 98
- 238000000034 method Methods 0.000 title claims abstract description 28
- 230000003287 optical effect Effects 0.000 title claims abstract description 21
- 238000013468 resource allocation Methods 0.000 title claims abstract description 10
- 239000011159 matrix material Substances 0.000 claims abstract description 35
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 21
- 230000009471 action Effects 0.000 claims abstract description 15
- 238000012545 processing Methods 0.000 claims abstract description 10
- 230000002787 reinforcement Effects 0.000 claims abstract description 8
- 230000003595 spectral effect Effects 0.000 claims description 23
- 238000004364 calculation method Methods 0.000 claims description 12
- 230000007246 mechanism Effects 0.000 claims description 4
- 230000007613 environmental effect Effects 0.000 claims description 3
- 235000008694 Humulus lupulus Nutrition 0.000 claims description 2
- 150000001875 compounds Chemical class 0.000 claims description 2
- 238000011478 gradient descent method Methods 0.000 claims description 2
- 230000002194 synthesizing effect Effects 0.000 claims description 2
- 230000002123 temporal effect Effects 0.000 claims description 2
- 239000004576 sand Substances 0.000 claims 1
- 230000000903 blocking effect Effects 0.000 description 4
- 238000011160 research Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 238000013467 fragmentation Methods 0.000 description 3
- 238000006062 fragmentation reaction Methods 0.000 description 3
- 238000013508 migration Methods 0.000 description 3
- 230000005012 migration Effects 0.000 description 3
- 230000003068 static effect Effects 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 230000003111 delayed effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 239000000758 substrate Substances 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
- 238000012360 testing method Methods 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
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L45/00—Routing or path finding of packets in data switching networks
- H04L45/12—Shortest path evaluation
-
- 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
- H04Q2011/0073—Provisions for forwarding or routing, e.g. lookup tables
-
- 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)
- Signal Processing (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention discloses a spectrum resource allocation method for a cloud-fog elastic optical network advance reservation request, which comprises the following steps: calculating k shortest candidate paths of the service request by using a shortest path algorithm; dividing time slices and frequency spectrum slots based on frequency spectrum resources of a link, acquiring a path resource matrix according to the state of each time frequency spectrum unit, and acquiring the number of the time slices and the frequency spectrum slots required by processing a service request according to the path resource matrix; confirming the action distributed by the service request by using a reinforcement learning algorithm, and acquiring rewards according to the action; confirming whether the distribution scheme is effective according to the reward, and if the distribution scheme is effective, recording the distribution scheme; the allocation scheme comprises the starting time of service request scheduling, the shortest candidate path, and the number of time slices and frequency spectrum slots required for processing the service request; and traversing k shortest candidate paths in sequence and selecting the distribution scheme generating the maximum reward. The invention has good robustness and can improve the utilization rate of frequency spectrum resources to the maximum extent.
Description
Technical Field
The invention relates to the technical field of elastic optical networks and cloud-fog communication, in particular to a spectrum resource allocation method for a cloud-fog elastic optical network advance reservation request.
Background
With the rapid development of 5G communication, internet of things (IoT) and virtual reality technologies, traditional cloud computing cannot meet its needs with high latency and huge energy consumption. Edge computing is a good complement to cloud computing, being closer to the device and with lower latency, and the cooperation of cloud computing and edge computing can fuse their advantages and provide higher quality of service. Meanwhile, as the bandwidth requirements of service requests are more and more diversified, new requirements are provided for the network to have the capability of flexibly providing frequency spectrums.
Elastic Optical Networks (EONs) are the underlying networks that are expected to carry flexible requests between cloud computing and edge computing. Based on the OFDM technology, the substrate spectrum resources are cut into independent spectrum time slots, each spectrum time slot usually occupies 6.25GHz or 12.5GHz, and a plurality of spectrum time slots can be efficiently and flexibly provided for arriving requests. Therefore, the application of Elastic Optical Networks (EON) makes cloud edge computing and 5G technologies better able to improve quality of life.
There are often many service requests for mass data migration or mass data backup between cloud-edge data centers, and these mass data migration or backup service requests do not need to be responded to immediately, and they always have a certain deadline. These service requests are completed before the expiration date, e.g., 8 am the next day. Therefore, these requests are also referred to as Advance Reservation (AR) requests. Due to the introduction of the time domain, these requests can be delayed appropriately to relieve network resource pressure and avoid network congestion. For allocating an advance reservation request, not only the spectrum domain resources but also the time domain should be considered. The request may be successfully allocated if both time resources and spectrum resources meet the requirements.
Routing and Spectrum Allocation (RSA) problems have been a hot problem in EON. Although many researches have researched the problem of large-capacity data transmission in some aspects and most of the researches propose the traditional heuristic RSA algorithm, in static RSA and dynamic RSA, the traditional heuristic RSA algorithm cannot be continuously optimized and is limited by scalability, and the technical problems that service requests cannot be reasonably distributed and processed and the blocking rate is high exist.
Disclosure of Invention
The invention provides a spectrum resource allocation method facing a cloud-fog elastic optical network advance reservation request, which solves the problem of spectrum resource allocation of cross-data center transmission services such as data backup, application data synchronization and virtual machine migration in the existing Internet of things.
S1, for a service requestK shortest candidate paths of the service request r are calculated by using a shortest path algorithm, wherein,representing the number of services carried by the service request r, s representing a source node, d representing a destination node, taAnd tdRespectively representing the arrival time and the deadline of the service request r;
s2, dividing time slices and frequency spectrum slots based on frequency spectrum resources of each link, obtaining a path resource matrix corresponding to the shortest candidate path in the step S1 according to the state of each time frequency spectrum unit, and obtaining the number n of the time slices needed for processing the service request r according to the path resource matrixtAnd the number n of spectral slotsf;
S3, the number n of time slices obtained from the step S2tAnd the number n of spectral slotsfConfirming the action A allocated to the service request R in the path resource matrix obtained in the step S2 by using a reinforcement learning algorithm, and acquiring a corresponding reward R according to the action A;
s4, according to the reward R obtained in the step S3, whether the distribution scheme under the shortest candidate route is effective is confirmed, if so, the distribution scheme under the shortest candidate route and the corresponding reward R are recorded, and then the step S5 is executed, and if not, the step S5 is directly executed;
the allocation scheme includes a start time t of scheduling of a service request rsThe shortest candidate path, the number of time slices n required for processing the service request rtAnd the number n of spectral slotsf;
And S5, traversing the k shortest candidate paths in sequence according to the method of the steps S2-S4, and selecting the distribution scheme generating the maximum reward R as the distribution scheme of the service request R.
The invention has the beneficial effects that: for an incoming advance reservation request, the invention firstly finds k shortest candidate paths by using a shortest path method, traverses each candidate path and calculates available spectrum resources corresponding to each candidate path; different service time and the number of frequency spectrum slots can be allocated to each service request, then the optimal allocation scheme is selected by utilizing the deep neural network, meanwhile, a reward is obtained for each allocation scheme, and the optimal allocation scheme is decided according to the reward; the method has good robustness, can select a proper routing path for all the services of the advance reservation requests and allocate the optimal service time and spectrum resources for each advance reservation request, thereby maximizing the utilization rate of the spectrum resources and reducing the blocking rate and the initial time delay of the service requests.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
FIG. 1 is a schematic flow chart of the present invention.
Fig. 2 is a schematic diagram of the synthesis of the environmental state S.
Fig. 3 is a flow chart of the DQN algorithm.
Fig. 4 is a schematic diagram of a cluster.
FIG. 5 shows the time-frequency spectrum continuity TFcA statistical representation of the parameters in (1).
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to fig. 1 to 5 in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all embodiments. All other embodiments, which can be obtained by a person skilled in the art without inventive effort based on the embodiments of the present invention, are within the scope of the present invention.
A spectrum resource allocation method for a pre-reservation request in a cloud-cloud elastic optical network, as shown in fig. 1, includes the following steps:
s1, for a service requestK shortest candidate paths of the service request r are calculated by using a shortest path algorithm, wherein,representing the number of services carried by the service request r, s representing the source node, d representing the destination node, taAnd tdRespectively representing the arrival time and the deadline of the service request r;
the service requestFor reserving the request service in advance, each shortest candidate path is composed of one or more links.
S2, based on each linkThe time slice and the frequency spectrum slot are divided into spectrum resources, the path resource matrix corresponding to the shortest candidate path in the step S1 is respectively obtained according to the state of each time spectrum unit, and the number n of the time slices required for processing the service request r is obtained according to the path resource matrixtAnd the number n of spectral slotsfThe method comprises the following steps:
s21, dividing the link on the shortest candidate path into time slices and frequency spectrum slots, establishing time frequency spectrum units based on the time slices and the frequency spectrum slots, and respectively confirming the state of each time frequency spectrum unit on the link;
the expression of the state of the time spectrum unit is as follows:
wherein S is(t,f)Representing a time-spectral unit u(t,f)State of (1), time spectrum unit u(t,f)Is composed of the t-th time slice and the f-th frequency spectrum slot.
S22, confirming the link resource matrix of the link according to the state of each time spectrum unit on the link obtained in the step S21;
the expression of the link resource matrix is:
in the formula of UlA link resource matrix representing the link/,representing a time-spectrum unit u on a link l(T,F)T represents the number of time slices on link l; f denotes the number of spectrum slots on link i.
S23, confirming the link resource matrix of each link on the shortest candidate path according to the methods of the step S21 and the step S22, and confirming the path resource matrix of the shortest candidate path according to the link resource matrix;
the expression of the path resource matrix is as follows:
in the formula of UPA path resource matrix representing the shortest candidate path P, L representing all link sets comprised by the shortest candidate path P,representing the time-spectrum unit u on all links in the shortest candidate path P(T,F)The state of (1).
The path resource matrix represents the state of each time spectrum unit in the shortest candidate path, and the available spectrum resources in the shortest candidate path can be quickly identified according to the path resource matrix.
S24, calculating the service duration time Deltat required by the service request r according to the path resource matrix, and calculating the number n of the time slices required by the service request r according to the service duration time DeltattAnd the number n of spectral slotsf;
The number of time slices ntThe calculation formula of (2) is as follows:
wherein τ represents the size of a time slice, and Δ t represents the service duration of the service request r;
the service duration Δ t is obtained by processing the service request r by respectively trying different start times and using available spectrum resources in the path resource matrix according to the following constraint conditions:
max△t=td-ta;
ta≤ts≤td;
τ≤△t≤td-ts;
in the formula, tsRepresents the starting time of the scheduling of the service request r;
the calculation formula of the service duration time Δ t is as follows:
△t=te-ts;
in the formula, teRepresenting the end time of the service request r scheduling;
the number n of spectrum slotsfThe calculation formula of (2) is as follows:
in the formula, FslotRepresenting the capacity of a spectrum bin, GB representing the guard bandwidth [. ]]Indicating that the whole is taken.
In this embodiment, the capacity F of one spectrum slotslotAt 12.5GHZ, the size of a time slice τ is one hour.
S3, the number n of time slices obtained from the step S2tAnd the number n of spectral slotsfConfirming the action A allocated to the service request R in the path resource matrix obtained in the step S2 by using a reinforcement learning algorithm, and acquiring a corresponding reward R according to the action A;
in this embodiment, the reinforcement learning algorithm is a DQN algorithm, and step S4 includes the following steps:
s31, as shown in FIG. 2, establishing a resource environment according to the path resource matrix established in step S2, and the number n of time slices required by the service request rtAnd the number n of spectral slotsfAnd establishing a request environment corresponding to the resource environment, and synthesizing the resource environment and the request environment to obtain an environment state S.
And S32, inputting the environment state S obtained in the step S31 into the evaluate network of the DQN algorithm to obtain an action A, wherein the action A represents the position of the service request r to be distributed in the path resource matrix.
And S33, judging and calculating the reward R corresponding to the position according to the reward mechanism.
The reward mechanism of the reward R is as follows:
in the formula, SRU represents a spectrum resource utilization value, and TSAE represents a time spectrum allocation efficiency; the smaller the spectrum resource utilization value SRU, the better, indicating that more resources may be left for subsequent requests, and therefore, the smaller the SRU,the larger, i.e. the more awards R; the larger the time-spectrum allocation efficiency TSAE, the better, indicating less spectrum fragmentation in the path resource matrix, i.e., more available resources.
The calculation formula of the frequency spectrum resource utilization value SRU is as follows:
SRU=(te-ts)×nt×h(r);
where h (r) represents the number of route hops from source node s to destination node d;
the calculation formula of the time spectrum allocation efficiency TSAE is as follows:
TSAE=Cs×Ri×TFc;
in the formula, CsDenotes the size of the cluster, RiIndicating resource idleness, TFcRepresents temporal spectral continuity;
the calculation of the time spectrum allocation efficiency TSAE comprehensively considers two factors of a cluster and a resource idleness degree on the basis of the time spectrum continuity, so that the spectrum fragmentation can be reduced, and the spectrum resources are utilized to the maximum extent.
The cluster is formed by connecting the position distributed by the service request r and the distributed time spectrum units around the cluster, and the size C of the clustersI.e. the number of time-spectrum units in the cluster; resource idleness degree RiRepresenting the fraction of time spectrum units that are free in the path resource matrix. As shown in fig. 4, if the allocated location of the service request is available block 1, cluster 1 is formed, and the size C of cluster 1s64; if the service request is allocated the available block 2, cluster 2 is formed, and the size C of cluster 2s17; since in available block 1 and available block 2The number of time spectrum units is the same, so the resource idleness R in both conditionsiSame as Ri=0.32。
The time-frequency spectrum continuity TFcThe calculation formula of (2) is as follows:
in the formula (I), the compound is shown in the specification,andrepresenting the number of available spectral blocks, num, along the time axis and the spectral axis, respectively2uIndicating the number of two consecutive spectral units (along the time axis and the spectral axis, respectively).
Time-frequency spectrum continuity TFcRepresents the situation of spectral fragmentation in the path resource matrix, as shown in fig. 5, the corresponding TF in fig. 5c=1.08。
S4, according to the reward R obtained in the step S3, whether the distribution scheme under the shortest candidate route is effective is confirmed, if so, the distribution scheme under the shortest candidate route and the corresponding reward R are recorded, and then the step S5 is executed, and if not, the step S5 is directly executed; the allocation scheme includes a start time t of scheduling of a service request rsThe shortest candidate path, the number of time slices n required for processing the service request rtAnd the number n of spectral slotsf。
Whether the position allocated in step S3 is occupied can be determined according to the sign of the reward R, and if the reward R is positive, the allocation scheme is valid, and if the reward R is negative, the allocation scheme is invalid.
Preferably, after recording the distribution scheme under the shortest candidate path, the environment state S is synchronously updated according to the action a to obtain a new environment state S_And the experience (S, A, R, S)_) Storing the updated network parameters into an experience pool of the evaluate network, and then judging whether the set updated network parameters are reachedIf yes, updating the network parameters, otherwise, directly executing the step S5.
As shown in fig. 3, the DQN algorithm includes two networks, namely an evaluate network and a target network, respectively, and the evaluate network is used to calculate an estimated Q value, denoted as Qevaluate(ii) a the target network is used for calculating the actual Q value, which is marked as Qtarget. As shown in fig. 3, according to the set time for updating the network parameters, the evaluate network and the target network extract part of experience (S, a, R, S) from the experience pool at intervals_) The evaluate network obtains Q according to the environment state Sevaluate(S, A), the target network according to the new environment state S_To obtain Qtarget(S_,A_) Then calculating a loss function from the two Q values, wherein A_Indicating according to the new environmental state S-The estimated new action.
The loss function is Qevaluate(S, A) and Qtarget(S_,A_) The specific formula of the mean square error L is as follows:
L=E((Qtarget(S_,A_)-Qevaluate(S,A))2);
the evaluate network updates the network parameters by adopting a gradient descent method, and the target network copies the updated parameters of the evaluate network, which is the prior art and is not described in this embodiment again.
And S5, traversing the k shortest candidate paths in sequence according to the method of the steps S2-S4, and then taking the allocation scheme generating the maximum reward R as the spectrum resource allocation scheme of the service request R by the elastic optical network.
The invention firstly establishes a two-dimensional resource model of frequency domain and time domain facing the service of the advance reservation request, carries out interaction with the environment through reinforcement learning, scores the frequency spectrum resource allocation scheme to optimize the allocation of frequency spectrum resources, and then updates the state of the corresponding time frequency spectrum unit according to the determined allocation scheme to prepare for the arrival of the next service request.
Since Deep Reinforcement Learning (DRL) shows the potential for successful Learning strategies for combinatorial and distributed problems, the present invention relies on obtaining feedback and rewards from the environment, and the DQN algorithm can learn the optimization strategy step by step, and is therefore well suited for decision-making problems. In the research of the static Spectrum Allocation strategy of the advance reservation request service, the optimal solution of the computing Resource of the Integer Linear Programming (ILP), the DRDA method and three traditional heuristic algorithms are compared, the performance of the DRDA in the static RSA problem is tested, and the simulation result shows that the performance of the DRDA method is very close to the optimal solution of the Resource computed by the ILP. In the research of dynamic Spectrum Allocation strategy, the invention provides a Time Spectrum Allocation Efficiency (TSAE) measurement standard for measuring the available resource state in an elastic optical network, a DQN algorithm Allocation scheme is adopted for scoring, simulation test and large-scale network experiment are adopted for comparing DRDA with three traditional heuristic algorithms from three aspects of average TSAE, request blocking rate and average initial delay, and the result shows that the DRDA method has good robustness, and compared with other three heuristic algorithms, the DRDA method keeps lower initial delay while obtaining the lowest blocking rate.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.
Claims (10)
1. A spectrum resource allocation method for a cloud-fog elastic optical network advance reservation request is characterized by comprising the following steps:
s1, for a service requestK shortest candidate paths of the service request r are calculated by using a shortest path algorithm, wherein,representing the number of services carried by the service request r, s representing a source node, d representing a destination node, taAnd tdRespectively representing the arrival time and the deadline of the service request r;
s2, dividing time slices and frequency spectrum slots based on frequency spectrum resources of each link, obtaining a path resource matrix corresponding to the shortest candidate path in the step S1 according to the state of each time frequency spectrum unit, and obtaining the number n of the time slices needed for processing the service request r according to the path resource matrixtAnd the number n of spectral slotsf;
S3, the number n of time slices obtained from the step S2tAnd the number n of spectral slotsfConfirming the action A allocated to the service request R in the path resource matrix obtained in the step S2 by using a reinforcement learning algorithm, and acquiring a corresponding reward R according to the action A;
s4, according to the reward R obtained in the step S3, whether the distribution scheme under the shortest candidate route is effective is confirmed, if so, the distribution scheme under the shortest candidate route and the corresponding reward R are recorded, and then the step S5 is executed, and if not, the step S5 is directly executed;
the allocation scheme includes a start time t of scheduling of a service request rsThe shortest candidate path, the number of time slices n required for processing the service request rtAnd the number n of spectral slotsf;
S5, according to the method of steps S2-S4, traversing k shortest candidate paths in sequence, and selecting the distribution scheme generating the maximum reward R as the distribution scheme of the service request R.
2. The method for allocating spectrum resources for the cloud-fog-oriented optical network advance reservation request according to claim 1, wherein the step S2 includes the following steps:
s21, dividing the link on the shortest candidate path into time slices and frequency spectrum slots, establishing time frequency spectrum units based on the time slices and the frequency spectrum slots, and respectively confirming the state of each time frequency spectrum unit on the link;
s22, confirming the link resource matrix of the link according to the state of each time spectrum unit on the link obtained in the step S21;
s23, confirming the link resource matrix of each link on the shortest candidate path according to the methods of the step S21 and the step S22, and confirming the path resource matrix of the shortest candidate path according to the link resource matrix;
s24, calculating the service duration time Deltat required by the service request r according to the path resource matrix, and calculating the number n of the time slices required by the service request r according to the service duration time DeltattAnd the number n of spectral slotsf。
3. The method for allocating spectrum resources for cloud-fog-oriented optical network reservation request in advance as claimed in claim 2, wherein in step S24, the time slice number n istThe calculation formula of (2) is as follows:
wherein τ represents the size of a time slice, and Δ t represents the service duration of the service request r;
the number n of spectrum slotsfThe calculation formula of (2) is as follows:
in the formula, FslotRepresenting the capacity of a spectrum bin, GB representing the guard bandwidth [. ]]Indicating that the whole is taken.
4. The method for allocating spectrum resources for the cloud-fog elastic optical network reservation request in advance as claimed in claim 2 or 3, wherein the service duration Δ t is obtained by processing the service request r by respectively trying different start times according to the following constraint conditions:
max△t=td-ta;
ta≤ts≤td;
τ≤△t≤td-ts;
in the formula, tsThe starting time of scheduling of a service request r is represented, and tau represents the size of a time slice;
the calculation formula of the service duration time Δ t is as follows:
△t=te-ts;
in the formula, teIndicating the end time of the service request r schedule.
5. The method for allocating spectrum resources for the cloud-fog-oriented optical network advance reservation request according to claim 1, wherein the step S3 includes the following steps:
s31, establishing resource environment according to the path resource matrix established in the step S2, and the number n of the time slices needed by the service request rtAnd the number n of spectral slotsfEstablishing a request environment corresponding to the resource environment, and synthesizing the resource environment and the request environment to obtain an environment state S;
s32, inputting the environmental state S obtained in the step S31 into an evaluate network of a reinforcement learning algorithm to obtain an action A, wherein the action A represents the position of a service request r to be distributed in the path resource matrix;
and S33, judging and calculating the reward R corresponding to the position according to the reward mechanism.
6. The method for allocating spectrum resources for cloud-fog-oriented elastic optical network reservation request in advance as claimed in claim 5, wherein in step S33, the reward mechanism is:
in the formula, SRU represents a spectrum resource utilization value, and TSAE represents a time spectrum allocation efficiency.
7. The method for allocating the spectrum resources for the cloud-fog elastic optical network reservation request in advance as claimed in claim 6, wherein the formula for calculating the spectrum resource utilization value SRU is as follows:
SRU=(te-ts)×nt×h(r);
where h (r) represents the number of route hops from source node s to destination node d, tsAnd teRespectively representing the start time and the end time of the scheduling of the service request r.
8. The method for allocating the spectrum resources for the cloud-fog elastic optical network reservation request in advance according to claim 6, wherein the calculation formula of the time-spectrum allocation efficiency TSAE is as follows:
TSAE=Cs×Ri×TFc;
in the formula, CsDenotes the size of the cluster, RiIndicating resource idleness, TFcRepresenting the temporal spectral continuity.
9. The method for allocating spectrum resources for cloud-fog-oriented optical network reservation request in advance as claimed in claim 8, wherein the time-spectrum continuity (TF)cThe calculation formula of (2) is as follows:
10. The cloud-fog oriented elastic optical network advance reservation request of claim 1The method for allocating spectrum resources is characterized in that, in step S4, after recording the allocation scheme and the corresponding reward R under the shortest candidate route, the environment state S is synchronously updated according to the action a to obtain a new environment state S_And the experience (S, A, R, S)_) Storing the network parameter into an experience pool, judging whether the set time for updating the network parameter is reached, if so, updating the network parameter by using a gradient descent method, otherwise, directly executing the step S5.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011271498.7A CN112383846B (en) | 2020-11-13 | 2020-11-13 | Cloud-fog elastic optical network-oriented spectrum resource allocation method for advance reservation request |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011271498.7A CN112383846B (en) | 2020-11-13 | 2020-11-13 | Cloud-fog elastic optical network-oriented spectrum resource allocation method for advance reservation request |
Publications (2)
Publication Number | Publication Date |
---|---|
CN112383846A CN112383846A (en) | 2021-02-19 |
CN112383846B true CN112383846B (en) | 2022-06-21 |
Family
ID=74582388
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202011271498.7A Active CN112383846B (en) | 2020-11-13 | 2020-11-13 | Cloud-fog elastic optical network-oriented spectrum resource allocation method for advance reservation request |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112383846B (en) |
Families Citing this family (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114338661B (en) * | 2021-08-27 | 2024-05-03 | 南京曦光信息科技研究院有限公司 | Distributed edge data center system based on optical packet switching and application |
CN113438173B (en) * | 2021-08-30 | 2021-11-23 | 华南师范大学 | Routing and spectrum allocation method, device, storage medium and electronic equipment |
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 |
CN116320843B (en) * | 2023-04-24 | 2023-07-25 | 华南师范大学 | Queue request mobilization method and device for elastic optical network |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2013045521A1 (en) * | 2011-09-30 | 2013-04-04 | Telefonica, S.A. | A system and a method to perform spectrum allocation in an optical network |
CN105634990A (en) * | 2014-11-27 | 2016-06-01 | 中兴通讯股份有限公司 | Resource reservation method, device and processor based on time spectrum continuity |
CN111246320A (en) * | 2020-01-08 | 2020-06-05 | 郑州大学 | Deep reinforcement learning flow dispersion method in cloud-fog elastic optical network |
CN111865800A (en) * | 2020-07-07 | 2020-10-30 | 烽火通信科技股份有限公司 | Routing frequency spectrum allocation method and device suitable for elastic optical network |
-
2020
- 2020-11-13 CN CN202011271498.7A patent/CN112383846B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2013045521A1 (en) * | 2011-09-30 | 2013-04-04 | Telefonica, S.A. | A system and a method to perform spectrum allocation in an optical network |
CN105634990A (en) * | 2014-11-27 | 2016-06-01 | 中兴通讯股份有限公司 | Resource reservation method, device and processor based on time spectrum continuity |
CN111246320A (en) * | 2020-01-08 | 2020-06-05 | 郑州大学 | Deep reinforcement learning flow dispersion method in cloud-fog elastic optical network |
CN111865800A (en) * | 2020-07-07 | 2020-10-30 | 烽火通信科技股份有限公司 | Routing frequency spectrum allocation method and device suitable for elastic optical network |
Non-Patent Citations (2)
Title |
---|
Dynamic Virtual Network Embedding with Distance Adaptive RSA Over SDM-Based Elastic Optical Networks;Lijie Chen,等;《2018 Conference on Lasers and Electro-Optics Pacific Rim (CLEO-PR)》;20190429;全文 * |
刘焕淋 ; 等.基于频谱感知的业务分割-合并的弹性光网络资源分配策略.《电子与信息学报》.2016,全文. * |
Also Published As
Publication number | Publication date |
---|---|
CN112383846A (en) | 2021-02-19 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN112383846B (en) | Cloud-fog elastic optical network-oriented spectrum resource allocation method for advance reservation request | |
EP3090517B1 (en) | Inter-domain sdn traffic engineering | |
JP5324637B2 (en) | Dynamic flowlet scheduling system, flow scheduling method, and flow scheduling program | |
WO2017148101A1 (en) | Systems and methods for performing traffic engineering through network slices | |
CN112822050B (en) | Method and apparatus for deploying network slices | |
US11689961B2 (en) | Systems and methods for distribution of application logic in digital networks | |
CN113692021B (en) | Intelligent resource allocation method for 5G network slice based on affinity | |
CN113708972A (en) | Service function chain deployment method and device, electronic equipment and storage medium | |
Gvozdiev et al. | On low-latency-capable topologies, and their impact on the design of intra-domain routing | |
WO2017082185A1 (en) | Resource allocating device and resource allocating method | |
US10983828B2 (en) | Method, apparatus and computer program product for scheduling dedicated processing resources | |
Zhu et al. | Drl-based deadline-driven advance reservation allocation in eons for cloud–edge computing | |
CN116137593A (en) | Virtual network function migration method for digital twin auxiliary dynamic resource demand prediction | |
CN110933000A (en) | Distributed data multi-stage aggregation method, device, server and storage medium | |
Ren et al. | End-to-end network SLA quality assurance for C-RAN: a closed-loop management method based on digital twin network | |
CN117436547A (en) | Cluster division method for dynamic federation learning, terminal and edge server | |
WO2022166348A1 (en) | Routing method, routing apparatus, controller and computer-readable storage medium | |
Liu et al. | Joint jobs scheduling and routing for metro-scaled micro datacenters over elastic optical networks | |
CN110417682B (en) | Periodic scheduling method in high-performance network | |
Yang et al. | Cross-layer self-similar coflow scheduling for machine learning clusters | |
Li et al. | Energy-efficient offloading based on hybrid bio-inspired algorithm for edge–cloud integrated computation | |
JP2016039531A (en) | Virtual network allocation method and device | |
JP6039592B2 (en) | Scheduler and program used in a network for performing dynamic bandwidth allocation | |
Tsukuda et al. | Reducing Inconsistency between Software-Defined Networking Controllers | |
Vaezi et al. | Analytical performance modeling of elastic optical links with aligned spectrum allocation |
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 |