Elastic optical network resource allocation method and device for balancing transmission path energy consumption and path length

Abstract

The invention provides an elastic optical network resource allocation method and equipment for balancing transmission path energy consumption and path length. The method comprises the following steps: allocating a target modulation format for a service request for a target elastic optical network and determining network resource demand data; distributing a target wave band for each initial transmission path, and determining the initial transmission path meeting the network resource demand data as an available transmission path; determining the available transmission paths of which the generalized signal-to-noise ratios are equal to or greater than a threshold value as candidate transmission paths; and determining a target transmission path and allocating network resources according to the resource utilization rates of the C wave band and the L wave band of the target elastic optical network, the service bandwidth type corresponding to the service request and the routing feature vector of each candidate transmission path based on a depth deterministic strategy gradient model. According to the invention, the network energy consumption can be effectively reduced in the optical transmission process, the frequency spectrum fragmentation rate and the service blocking rate can be reduced in the resource allocation process, and the service transmission quality can be ensured.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of elastic optical networks, and in particular to an elastic optical network resource allocation method and device for balancing energy consumption and path length of a transmission path. BACKGROUND

[0002] Elastic Optical Networks (EONs) significantly improve spectrum efficiency and resource utilization through flexible spectrum allocation mechanisms, and have become an important technical development direction in the field of optical communications. As a hot issue in the research field, optical energy saving has always been concerned by enterprises and scholars. However, current researches focus a lot on C+L band optical devices and optical transmission technology. The research on the optimization design of C+L band elastic optical networks mainly concentrates on network capacity evaluation, network control technology and optical channel quality modeling. However, there are still very few researches on energy consumption in the transmission process of C+L band and the optimization of resources in the transmission process. Therefore, with the urgent need to build a green optical network and the diversification of network services, designing a resource allocation algorithm with the main optimization purpose of energy saving and consumption reduction will be the focus of future work in combination with new flexible optical network technologies.

[0003] In recent years, some research results on elastic optical network routing optimization methods combined with deep learning methods have emerged. In previous research, someone proposed a routing mechanism algorithm considering energy consumption based on a deep deterministic policy gradient (DDPG) model in the C+L band elastic optical network scenario. The NSFNET network was selected for simulation, with a single time slot of 12.5GHz, and 458 frequency slots in the C band and the L band. Traffic services were randomly generated within a certain range and were preferentially allocated to the C band, followed by the L band. The available modulation formats included BPSK, QPSK, 8QAM and 16QAM.

[0004] However, the existing elastic optical network routing optimization method does not consider the size of the service bandwidth demand in the band selection, has poor flexibility in modulation format setting, has low resource utilization rate in the L band, and has unreasonable allocation order, which can lead to the inability to balance the energy consumption and path length of the service transmission path (i.e., routing), and the inability to effectively guarantee the service transmission quality. SUMMARY

[0005] In view of this, the embodiments of the present application provide an elastic optical network resource allocation method and device for balancing the energy consumption and path length of a transmission path, so as to eliminate or improve one or more defects in the prior art.

[0006] An aspect of the present application provides a method for balancing energy consumption and path length of transmission paths in an elastic optical network resource allocation method, comprising:

[0007] allocating a target modulation format for a current service request for a target elastic optical network and determining network resource requirement data corresponding to the target modulation format;

[0008] allocating a target band for each of the initial transmission paths according to the path band type of each of the initial transmission paths corresponding to the service request, and determining the initial transmission path whose current available network resources of the target band satisfy the network resource requirement data as an available transmission path; wherein the target band is a C band or an L band;

[0009] determining a general signal-to-noise ratio value corresponding to each of the available transmission paths, and determining the available transmission path whose general signal-to-noise ratio value is equal to or greater than a signal-to-noise ratio threshold as a candidate transmission path;

[0010] based on a deep deterministic policy gradient model for balancing energy consumption and path length of transmission paths, selecting one of the candidate transmission paths as a target transmission path according to the current C band resource utilization, the current L band resource utilization, the service bandwidth type corresponding to the service request, and the routing feature vector corresponding to each of the candidate transmission paths, and allocating network resources for the target transmission path for processing the service request.

[0011] In some embodiments of the present application, the method for allocating a target modulation format for a current service request for a target elastic optical network and determining network resource requirement data corresponding to the target modulation format comprises:

[0012] obtaining each candidate modulation format specified by the current service request for the target elastic optical network;

[0013] determining the number of frequency slots required for each of the candidate modulation formats for the service request;

[0014] determining whether there are multiple candidate modulation formats with the same number of frequency slots in each of the candidate modulation formats;

[0015] if not, selecting the highest order modulation format in each of the candidate modulation formats as the target modulation format for the service request;

[0016] determining the number of frequency slots corresponding to the target modulation format as the network resource requirement data corresponding to the target modulation format.

[0017] In some embodiments of the present application, before determining the number of frequency slots corresponding to the target modulation format, the method further comprises:

[0018] If it is determined that there are multiple candidate modulation formats with the same number of frequency slots in the candidate modulation formats, then among the number of frequency slots corresponding to each of the candidate modulation formats, it is determined whether the number of frequency slots corresponding to the multiple candidate modulation formats with the same number of frequency slots is the minimum value.

[0019] If yes, then the modulation format with the lowest order is selected from the candidate modulation formats as the target modulation format of the service request.

[0020] If no, then the modulation format with the highest order is selected from the candidate modulation formats as the target modulation format of the service request.

[0021] In some embodiments of the present application, the path waveband type includes: only C waveband, only C+L waveband, and mixed waveband; wherein the mixed waveband includes C waveband and C+L waveband.

[0022] Correspondingly, the method of allocating a target waveband to each of the initial transmission paths according to the path waveband type of each of the initial transmission paths corresponding to the service request, and determining the initial transmission path whose current available network resources of the target waveband satisfy the network resource requirement data as an available transmission path, comprises:

[0023] If the path waveband type of the initial transmission path corresponding to the service request is the only C+L waveband, then it is determined whether the available network resources of the L waveband corresponding to the initial transmission path satisfy the network resource requirement data.

[0024] If yes, then the target waveband of the initial transmission path is configured as the L waveband and the initial transmission path is determined as an available transmission path.

[0025] In some embodiments of the present application, the method of allocating a target waveband to each of the initial transmission paths according to the path waveband type of each of the initial transmission paths corresponding to the service request, and determining the initial transmission path whose current available network resources of the target waveband satisfy the network resource requirement data as an available transmission path, further comprises:

[0026] If the path waveband type of the initial transmission path corresponding to the service request is the only C waveband or the mixed waveband, then the target waveband of the initial transmission path is configured as the C waveband.

[0027] if the path waveband type of the initial transmission path corresponding to the service request is the C+L waveband only, and it is determined that the available network resources of the L waveband corresponding to the initial transmission path do not meet the network resource requirement data, the target waveband of the initial transmission path is configured as the C waveband;

[0028] In the initial transmission path with the target waveband being the C waveband, the initial transmission path with the available network resources of the C waveband meeting the network resource requirement data is determined as an available transmission path.

[0029] In some embodiments of the present application, the deep deterministic policy gradient model for balancing energy consumption and path length of transmission paths determines one of the candidate transmission paths as a target transmission path according to the current C waveband resource utilization, the current L waveband resource utilization of the target elastic optical network, the service bandwidth type corresponding to the service request, and the routing feature vector corresponding to each of the candidate transmission paths, and allocates network resources for the target transmission path to process the service request, comprising:

[0030] The current C waveband resource utilization, the current L waveband resource utilization of the target elastic optical network, and the service bandwidth type corresponding to the service request are taken as a current network state vector of the target elastic optical network, and the network state vector and the routing feature vector corresponding to each of the candidate transmission paths are taken as a current state vector;

[0031] The state vector is input into the deep deterministic policy gradient model for balancing energy consumption and path length of transmission paths, so that the deep deterministic policy gradient model performs a plurality of iteration rounds to obtain optimal continuous action parameters; wherein in each iteration round, an actor network of the deep deterministic policy gradient model generates continuous action parameters for balancing energy consumption and path length of transmission paths according to the state vector corresponding to the current iteration round, and a critic network in the deep deterministic policy gradient model outputs a Q value for the continuous action parameters according to the state vector, the continuous action parameters, and a reward function, so that the actor network determines a state vector corresponding to the next iteration round according to the Q value;

[0032] One of the candidate transmission paths is determined as a target transmission path based on the optimal continuous action parameters, and network resources are allocated for the target transmission path to process the service request.

[0033] In some embodiments of the present application, the routing feature vector corresponding to the candidate transmission path includes the normalized energy consumption, the normalized path length, and the generalized signal-to-noise ratio value of the candidate transmission path.

[0034] Correspondingly, the selecting the target transmission path from the candidate transmission paths based on the optimal continuous action parameter comprises:

[0035] determining a comprehensive cost score of each of the candidate transmission paths based on the optimal continuous action parameter, the normalized energy consumption and the normalized path length of each of the candidate transmission paths;

[0036] selecting the candidate transmission path with the minimum comprehensive cost score as the target transmission path from the candidate transmission paths.

[0037] In some embodiments of the present application, the reward function comprises a cost penalty parameter group, a success reward parameter group and a failure penalty parameter group;

[0038] the cost penalty parameter group is calculated by the sum of a first parameter and a second parameter, the first parameter is calculated by the product of a first weight and the normalized energy consumption of the candidate transmission path, and the second parameter is calculated by the product of a second weight and the normalized path length of the candidate transmission path;

[0039] the success reward parameter group is calculated by a preset allocated success reward threshold and an allocated success indication function;

[0040] the failure penalty parameter group is calculated by a preset allocated failure penalty threshold and an allocated failure indication function.

[0041] In some embodiments of the present application, the sum of the first weight and the second weight is equal to 1; and the values of the first weight and the second weight are adjusted in real time according to the C-band resource utilization of the target elastic optical network.

[0042] Another aspect of the present application provides an elastic optical network resource allocation device for balancing transmission path energy consumption and path length, comprising:

[0043] a modulation format allocation module configured to allocate a target modulation format for a service request for a target elastic optical network and determine network resource requirement data corresponding to the target modulation format;

[0044] a band screening module configured to allocate a target band for each of the initial transmission paths according to the path band type of each of the initial transmission paths corresponding to the service request, and determine an available transmission path by determining that the initial transmission path whose available network resources of the target band currently satisfy the network resource requirement data; wherein the target band is a C-band or an L-band.

[0045] a GSNR screening module configured to determine a general signal-to-noise ratio (GSNR) value corresponding to each of the available transmission paths, and determine the available transmission paths with a GSNR value equal to or greater than a signal-to-noise ratio (SNR) threshold as candidate transmission paths;

[0046] a DDPG screening module configured to determine a target transmission path from the candidate transmission paths based on a deep deterministic policy gradient (DDPG) model for balancing energy consumption and path length of the transmission path, according to a current C-band resource utilization, a current L-band resource utilization, a service bandwidth type corresponding to the service request, and a routing feature vector corresponding to each of the candidate transmission paths, and allocate network resources for the target transmission path to process the service request.

[0047] A third aspect of the present application provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method for balancing energy consumption and path length of the transmission path in the elastic optical network.

[0048] A fourth aspect of the present application provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program is executable by a processor to implement the method for balancing energy consumption and path length of the transmission path in the elastic optical network.

[0049] A fifth aspect of the present application provides a computer program product comprising a computer program, wherein the computer program is executable by a processor to implement the method for balancing energy consumption and path length of the transmission path in the elastic optical network.

[0050] The application provides a method for allocating resources of a flexible optical network for balancing energy consumption and path length of a transmission path, which comprises the following steps: allocating a target modulation format for a service request of a target flexible optical network and determining network resource requirement data corresponding to the target modulation format; allocating a target waveband for each initial transmission path according to a path waveband type of each initial transmission path corresponding to the service request, and determining an initial transmission path whose current available network resources of the target waveband satisfy the network resource requirement data as an available transmission path; wherein the target waveband is a C waveband or an L waveband; determining a generalized signal-to-noise ratio value corresponding to each available transmission path, and determining the available transmission path whose generalized signal-to-noise ratio value is equal to or greater than a signal-to-noise ratio threshold as a candidate transmission path; based on a deep deterministic policy gradient model for balancing energy consumption and path length of a transmission path, selecting one of the candidate transmission paths as a target transmission path according to a C waveband resource utilization rate, an L waveband resource utilization rate of the target flexible optical network, a service bandwidth type corresponding to the service request and a routing feature vector corresponding to each candidate transmission path, and allocating network resources for the target transmission path for processing the service request; from the perspective of energy saving and consumption reduction, the energy consumption and path length can be fully considered, the service path can be selected according to the service path energy consumption and path length, and the most suitable path for the service is given; thus, the network energy consumption can be effectively reduced in the optical transmission process, the spectrum fragmentation rate and the service blocking rate can be reduced in the resource allocation process, and the service transmission quality can be ensured; the low-energy operation of the entire network can be ensured in the case of multiple services, the network transmission capacity can be improved, and the carbon footprint can be reduced; not only a new idea for energy consumption optimization of a C+L waveband flexible optical network is provided, but also important theoretical and technical support for the construction of a future green optical network is provided.

[0051] Additional advantages, objects, and features of the application will be set forth in part by the description that follows, and will become apparent to those skilled in the art upon examination of the following figures and detailed description thereof or can be learned by practice of the application. The advantages of the application will be realized and attained by the structure particularly pointed out in the written description and claims hereof as well as the appended drawings.

[0052] It will be appreciated by those skilled in the art that the objectives and advantages of the application can be realized and attained by the embodiments particularly pointed out in the written description and claims hereof as well as the appended drawings. BRIEF DESCRIPTION OF DRAWINGS

[0053] The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this application, illustrate embodiments of the application and together with the description serve to explain the principles of the application. The components in the drawings are not necessarily to scale, emphasis instead being placed upon illustrating the principles of the application. For purposes of clarity and a consistent approach, portions of the drawings may have been exaggerated to show details of the application that would otherwise be obscure, i.e., made larger with respect to other components, where appropriate. In the drawings:

[0054] Figure 1 Flowchart of the method for allocating resources in a flexible optical network to balance energy consumption and path length of transmission paths according to an embodiment of the application.

[0055] Figure 2 Flowchart of step 100 of the method for allocating resources in a flexible optical network to balance energy consumption and path length of transmission paths according to an embodiment of the application.

[0056] Figure 3 Diagram of three different types of paths in a C-C+L hybrid optical network according to an embodiment of the application.

[0057] Figure 4 Flowchart of step 200 of the method for allocating resources in a flexible optical network to balance energy consumption and path length of transmission paths according to an embodiment of the application.

[0058] Figure 5 Flowchart of the complete execution of steps 100 and 200 according to an embodiment of the application.

[0059] Figure 6 Flowchart of step 400 of the method for allocating resources in a flexible optical network to balance energy consumption and path length of transmission paths according to an embodiment of the application.

[0060] Figure 7 Flowchart of step 400 of the method for allocating resources in a flexible optical network to balance energy consumption and path length of transmission paths according to an embodiment of the application.

[0061] Figure 8 Principle diagram of the DDPG algorithm according to an application example of the application.

[0062] Figure 9 Diagram of the spectrum allocation algorithm according to an application example of the application.

[0063] Figure 10 Structure diagram of the device for allocating resources in a flexible optical network to balance energy consumption and path length of transmission paths according to an embodiment of the application. DETAILED DESCRIPTION

[0064] For the purposes of the present application, the technical solutions and advantages thereof, further detailed explanations will be given below in conjunction with the embodiments and the drawings. Herein, the illustrative embodiments of the present application and the explanations thereof are used to explain the present application, but are not intended to limit the present application.

[0065] It should also be noted that, in order to avoid obscuring the present application due to unnecessary details, only structures and / or processing steps closely related to the solutions according to the present application are shown in the drawings, and other details not closely related to the present application are omitted.

[0066] It should be emphasized that the term "comprises / comprising" as used herein is intended to mean that features, elements, steps or components are present, but does not exclude the presence or addition of one or more other features, elements, steps or components.

[0067] It should also be noted that, unless otherwise specified, the term "connected" as used herein can not only mean direct connection, but also indirect connection in the presence of an intermediate.

[0068] Hereinafter, embodiments of the present application will be described with reference to the accompanying drawings. In the drawings, the same reference numerals represent the same or similar parts, or the same or similar steps.

[0069] In today's digital era, with the wide popularity of 5G technology, the deep development of artificial intelligence (AI), and the continuous innovation of augmented reality and virtual reality (AR / VR) technology, various intelligent devices and application scenarios are emerging, and people's dependence on data is increasing day by day. Data traffic is rising at an alarming rate, showing an explosive growth trend. Traditional communication networks have been difficult to meet the growing demand for high-performance computing and new services in terms of transmission rate, capacity and delay. Optical networks, with their high speed, large capacity and low delay, have become a key technology to solve the bottleneck of traditional communication, providing strong support for efficient data transmission in modern society.

[0070] With the rapid growth of high-speed, multi-granularity business needs, the fixed spectrum grid mode adopted by traditional wavelength division multiplexing (WDM) networks gradually exposes the defects of low spectrum utilization and insufficient flexibility. Elastic optical networks (EONs) significantly improve spectrum efficiency and resource utilization through flexible spectrum allocation mechanisms, becoming an important technical development direction in the field of optical communication.

[0071] To accommodate the rapidly growing traffic, long-term capacity expansion of optical backbone networks is necessary. Network operators are therefore constantly seeking new capacity expansion techniques and exploring new expansion solutions to increase the bandwidth capacity of optical networks. As an emerging technology, multi-band optical networks significantly improve the transmission capacity and spectral efficiency of optical fibers by expanding the available spectral range (such as C-band, L-band, S-band, etc.), and have become a research hotspot in the field of optical communications. This application focuses on the intelligent routing algorithm research of C+L band flexible optical networks.

[0072] With the rapid growth of network traffic and traffic, the energy consumption problem of the information communication technology (ICT) industry is increasingly prominent, and the carbon emissions are also surprisingly large, especially in the core optical network that performs high-speed and massive transmission. In fact, energy conservation and environmental protection are important issues and global consensus in today's society, and energy consumption is closely related to sustainable development goals, and all walks of life are working hard to reduce energy consumption to slow down climate change and reduce carbon footprint. As one of the important infrastructures of information communication technology, the improvement of energy efficiency and sustainability of optical networks is crucial to the sustainable development of the whole society. In recent years, the focus of telecommunications research has shifted to network design that optimizes resources. As the Internet's demand for energy grows, researchers not only need to focus on optimizing network design, but also need to consider the energy consumption problem of the network.

[0073] In summary, optical energy saving as a hot issue in the research field has always been the focus of attention of enterprises and scholars, however, the current research focuses a lot on the research of C+L band optical devices and optical transmission technology, and the research on the optimal design of C+L band flexible optical networks mainly focuses on network capacity evaluation, network control technology and optical channel quality modeling. There are still very few studies on energy consumption problems and resource optimization problems in the transmission process considering C+L transmission process. Therefore, with the urgent need to build a green optical network and the diversification of network business development, designing a resource allocation algorithm with energy saving as the main optimization purpose combined with new flexible optical network technologies will be the focus of future work.

[0074] Among them, the routing mechanism algorithm considering energy consumption based on DDPG in the scenario of C+L band flexible optical network is as follows:

[0075] (1) Prefer to select C-band for service allocation, and select L-band for resource allocation when all modulation formats of all candidate service paths in C-band do not meet the service allocation requirements.

[0076] (2) Select the modulation format from the highest to the lowest, and calculate multiple working routes using KSP algorithm after selecting the modulation format, and calculate the GSNR of all candidate routing paths under the selected modulation format.

[0077] (3) For the routes meeting the GSNR threshold, use the DDPG algorithm to select the route with the minimum energy consumption and allocate resources. If the estimated GSNR in the considered frequency band is lower than the threshold, select a modulation format with lower GSNR requirement. If none of the modulation formats in the selected frequency band meets the GSNR threshold requirement, select another transmission frequency band and re-evaluate the GSNR.

[0078] (4) If neither the C band nor the L band meets the allocation requirements of the service path allocation, the service path allocation fails.

[0079] However, the above algorithm has the following problems:

[0080] (1) The size of the service bandwidth requirement is not considered in the selection of the frequency band. The new request allocates resources in the order of the C band to the L band and in the order of the FF. The C band is strictly prioritized, which leads to a rapid rise in the utilization rate of the C band under high load. When the utilization rate exceeds 70%, new requests may be forced to downgrade the modulation format or be blocked due to insufficient resources, while the L band resources are not fully utilized, causing imbalance in network resource allocation.

[0081] (2) The modulation format is mechanically selected from high order to low order (such as 16QAM, 8QAM, QPSK to BPSK), which may still force the use of high-order modulation in a high-noise or interference environment, leading to an increase in the bit error rate and retransmission rate, which in turn increases the overall energy consumption, which is contrary to the energy-saving goal.

[0082] (3) The utilization rate of the L band resource is low. The L band is only used after the C band is exhausted, which leads to a long-term low load state of the L band. The load is not shared by the L band, which reduces the overall capacity and energy efficiency of the network, especially in the C / L band coordination scenario, which wastes the advantage of spectrum expansion.

[0083] (4) The order of L band resource allocation is not reasonable. The traditional first-fit (FF) allocation method is used in the L band, and continuous resources are not reserved for large services in priority. This may lead to the need for multiple searches or forced splitting of large services into multiple segments for transmission in the L band, increasing processing delay and device energy consumption.

[0084] (5) Only the algorithm design in the L band full upgrade scenario is considered. In fact, during the migration of the C band to the C+L transmission band, operators need to deploy new L band ready erbium-doped fiber amplifiers (EDFA) and reconfigurable optical add-drop multiplexers (ROADM). Therefore, through partial device upgrade and optical fiber link amplification, the C band is gradually expanded to the C+L band, which is a more economical and feasible solution. However, the L band partial upgrade scenario is not considered in this method.

[0085] To solve the problems of the prior art, such as not considering the bandwidth requirement of a service, poor flexibility of modulation format setting, low utilization of L-band resources, and unreasonable allocation sequence, embodiments of the present application provide a method for allocating resources of an elastic optical network for balancing energy consumption and path length, a device for allocating resources of an elastic optical network for balancing energy consumption and path length, an entity, a computer readable storage medium, and a computer program product for executing the method for allocating resources of an elastic optical network for balancing energy consumption and path length. It is necessary to focus on the research on resource optimization considering energy consumption in C+L bands. In this work, the routing, modulation level and spectrum assignment (RMLSA) problem considering energy consumption and balancing of routing length in C+L band elastic optical networks (EONs) is studied, and a deep deterministic policy gradient (DDPG) intelligent routing algorithm based on deep learning is proposed to solve the RMLSA problem considering energy consumption and balancing of routing length.

[0086] The embodiments are described in detail as follows.

[0087] Based on this, the embodiments of the present application provide a method for allocating resources of an elastic optical network for balancing energy consumption and path length, which can be implemented by a device for allocating resources of an elastic optical network for balancing energy consumption and path length, as shown in Figure 1 The method for allocating resources of an elastic optical network for balancing energy consumption and path length specifically includes the following contents.

[0088] Step 100: allocating a target modulation format for a service request for a target elastic optical network and determining network resource requirement data corresponding to the target modulation format.

[0089] The target elastic optical network refers to the elastic optical network currently designated by the service request, i.e., a C+L multi-band elastic optical network (EONs). In one or more embodiments of the present application, the elastic optical network can refer to an elastic optical network operating in C band to C+L band, which can also be referred to as a C-C+L hybrid optical network.

[0090] The C+L band is a general term for the combination of traditional C band (1525-1565 nm) and L band (1565-1625 nm) in the field of optical communication, aiming to significantly improve the fiber transmission capacity and distance by expanding the available spectrum bandwidth.

[0091] The service request can include a plurality of modulation formats specified to select one of the plurality of modulation formats as a target modulation format for the service request.

[0092] The network resource requirement data can include a number of frequency slots required by the modulation format for the service request.

[0093] Step 200: According to the path wavelength type of each initial transmission path corresponding to the service request, a target wavelength is allocated to each initial transmission path, and the initial transmission path whose current available network resource of the target wavelength satisfies the network resource requirement data is determined as an available transmission path; wherein the target wavelength is a C wavelength or an L wavelength.

[0094] In step 200, each initial transmission path corresponding to the service request can be calculated as an initial transmission path by a KSP algorithm based on the service request and the current network topology of the target elastic optical network. KSP algorithm (K-Shortest Paths) is a general algorithm for solving the first K shortest paths between two points in a network topology, and has important applications in path planning, network optimization and communication routing.

[0095] In one or more embodiments of the present application, the transmission path can be referred to as a route, and the path length (i.e. route length) refers to the number of route hops.

[0096] Step 300: Determine the generalized signal-to-noise ratio value corresponding to each of the available transmission paths, and determine the available transmission path whose generalized signal-to-noise ratio value is equal to or greater than a signal-to-noise ratio threshold as a candidate transmission path.

[0097] Step 400: Based on a deep deterministic policy gradient model for balancing the energy consumption and path length of the transmission path, according to the current C wavelength resource utilization, L wavelength resource utilization of the target elastic optical network, the service bandwidth type corresponding to the service request, and the route feature vector corresponding to each of the candidate transmission paths, select one of the candidate transmission paths as a target transmission path, and allocate network resources to the target transmission path for processing the service request.

[0098] The core idea of the deep deterministic policy gradient model DDPG is to use the Actor-Critic architecture to learn the optimal policy. The actor is responsible for outputting a deterministic policy, i.e. outputting an action a given a state s. The critic is responsible for evaluating the goodness of the actor's policy, and estimates the value of taking action a in state s by learning a value function Q(s, a, w).

[0099] The deep deterministic policy gradient model for balancing the energy consumption and path length of a transmission path refers to a deep deterministic policy gradient model that simultaneously considers the energy consumption (i.e., the current C-band resource utilization, L-band resource utilization of the target elastic optical network, and the service bandwidth type corresponding to the service request) and the path length (i.e., the routing feature vector corresponding to each of the candidate transmission paths). The action a in the deep deterministic policy gradient model is a continuous action parameter for balancing the energy consumption and path length of a transmission path.

[0100] As can be seen from the above description, the elastic optical network resource allocation method for balancing the energy consumption and path length of a transmission path provided by the embodiments of the present application can fully consider the energy consumption and path length problem from the perspective of energy saving and consumption reduction, can select a path for a service according to the energy consumption and path length of the service path, and give the most suitable path for the service. Furthermore, the network energy consumption can be effectively reduced in the optical transmission process, the spectrum fragmentation rate and service blocking rate can be reduced in the resource allocation process, and the service transmission quality can be guaranteed. The low-energy operation of the entire network can be ensured in the case of multiple services, the network transmission capacity can be improved, and the carbon footprint can be reduced. Not only does the method provide a new idea for the energy consumption optimization of a C+L-band elastic optical network, but also provides important theoretical and technical support for the construction of a future green optical network.

[0101] In order to further improve the effectiveness and flexibility of the modulation format allocation in the elastic optical network resource allocation for balancing the energy consumption and path length of a transmission path, in the elastic optical network resource allocation method for balancing the energy consumption and path length of a transmission path provided by the embodiments of the present application, referring to Figure 2 , the step 100 in the elastic optical network resource allocation method for balancing the energy consumption and path length of a transmission path specifically includes the following contents:

[0102] Step 110: obtaining each candidate modulation format specified by a service request for a target elastic optical network;

[0103] Step 120: determining the number of frequency slots required by each of the candidate modulation formats for the service request;

[0104] Step 130: determining whether there are multiple candidate modulation formats with the same number of frequency slots in each of the candidate modulation formats. If not, step 140 is performed.

[0105] Step 140: selecting the highest order modulation format in each of the candidate modulation formats as the target modulation format of the service request.

[0106] Step 150: determining the number of frequency slots corresponding to the target modulation format as the network resource requirement data corresponding to the target modulation format.

[0107] In an example, in a C+L band optical network, common modulation formats include on-off keying (OOK), binary phase shift keying (BPSK), quadrature phase shift keying (QPSK), 16-quadrature amplitude modulation (16QAM), etc.

[0108] In the present application, when selecting a modulation format, for each modulation format, the number of required frequency slots is first calculated, and then two cases are considered: in the first case, the number of required frequency slots is the same for different modulation formats, and the number of required frequency slots is the least among all available modulation formats, a low-order modulation format is selected. For example, the available modulation formats set by a request are QPSK, 8QAM and 16QAM, and the number of required frequency slots is 2, 1 and 1 respectively, then among the two modulation formats of 8QAM and 16QAM with the same number of required frequency slots, QPSK is selected, i.e. the lowest order modulation format is selected. Compared with high-order modulation, low-order modulation has lower requirement on signal-to-noise ratio under the same channel condition. Taking QPSK and 16-QAM as examples, QPSK can ensure better bit error rate performance at a lower signal-to-noise ratio, while 16-QAM needs a higher signal-to-noise ratio to improve the signal-to-noise ratio tolerance to achieve the same bit error rate performance.

[0109] In the second case related to modulation selection, the number of required frequency slots is different for the available modulation formats, or the number of required frequency slots is the same for several modulation formats but not the least, then a high-order modulation is selected to achieve the minimum frequency slot occupation, for example, the available modulation format set of a request includes BPSK, QPSK and 8QAM, and the number of required frequency slots is 3, 3 and 2 respectively. At this time, 8QAM is selected in the present application.

[0110] In order to further improve the flexibility and applicability of modulation format allocation in the elastic optical network resource allocation for balancing the energy consumption and path length of transmission paths, in an embodiment of the present application, a method for balancing the energy consumption and path length of transmission paths in an elastic optical network resource allocation method is provided, referring to Figure 2 Before step 150 in step 100 in the method for balancing the energy consumption and path length of transmission paths in an elastic optical network resource allocation method, the following contents are further included:

[0111] If it is known from the judgment in step 130 that there are multiple candidate modulation formats with the same number of frequency slots in each of the candidate modulation formats, step 141 is performed;

[0112] Step 141: among the number of frequency slots corresponding to each of the candidate modulation formats, it is judged whether the number of frequency slots corresponding to the multiple candidate modulation formats with the same number of frequency slots is the minimum value; if yes, step 142 is performed; if no, step 140 is performed;

[0113] Step 142: selecting the lowest order modulation format as the target modulation format of the service request from the candidate modulation formats.

[0114] To further improve the reliability and effectiveness of the band selection in the elastic optical network resource allocation for balancing the energy consumption and path length of transmission path, in an embodiment of the present application, a method for elastic optical network resource allocation for balancing the energy consumption and path length of transmission path is provided, wherein the path band types include: only C band, only C+L band and mixed band; wherein the mixed band includes C band and C+L band.

[0115] Specifically, when selecting the C+L band, the transmission loss and bandwidth demand need to be considered comprehensively, and in the C-C+L mixed optical network, there can be three different types of paths for different requests, such as Figure 3 As shown, including all C band links, part of C band links and all C+L band links.

[0116] In the C-C+L mixed optical network, there can be three different types of paths for three different services, including service allocation in all C band links r1, service allocation in part of C band links r3 and service allocation in all C+L band links r2, for r1 and r3, due to the unavailability of part of L band, all services will be allocated in C band, and the service allocation is performed in a first fitting manner; for r3, all links on the selected path are deployed with C+L band, in order to avoid request blocking due to insufficient C band resources, the request r3 preferentially selects L band resources, large bandwidth services are allocated resources in a second fitting manner, and small bandwidth services are allocated resources in a first fitting manner, when the L band does not meet the service resource allocation demand, C band resources are selected for service allocation. This is to avoid excessive consumption of C band resources on the selected C-C+L mixed path or all links with C band, which can cause request blocking.

[0117] Based on this, referring to Figure 4 , the step 200 in the method for elastic optical network resource allocation for balancing the energy consumption and path length of transmission path specifically contains the following contents:

[0118] Step 210: if the path band type of the initial transmission path corresponding to the service request is the only C+L band, it is judged whether the available network resources of the L band corresponding to the initial transmission path meet the network resource demand data; if yes, step 220 is executed.

[0119] Step 220: configuring the target band of the initial transmission path as L band and determining the initial transmission path as an available transmission path.

[0120] To further improve the reliability and effectiveness of spectrum allocation in the elastic optical network resource allocation for balancing the energy consumption and path length of transmission path, in an embodiment of the present application, a method for elastic optical network resource allocation for balancing the energy consumption and path length of transmission path is provided, referring to Figure 4 Step 200 in the method for elastic optical network resource allocation for balancing the energy consumption and path length of transmission path further comprises the following contents:

[0121] Step 230: If the path waveband type of the initial transmission path corresponding to the service request is the C waveband only or the mixed waveband, the target waveband of the initial transmission path is configured as the C waveband.

[0122] Step 240: If the path waveband type of the initial transmission path corresponding to the service request is the C+L waveband only, and it is known from step 210 that the available network resources of the L waveband corresponding to the initial transmission path do not meet the network resource requirement data, step 250 is performed.

[0123] Step 250: The target waveband of the initial transmission path is configured as the C waveband.

[0124] Step 260: In the initial transmission path with the C waveband as the target waveband, the initial transmission path with the available network resources of the C waveband meeting the network resource requirement data is determined as the available transmission path.

[0125] In an example, Figure 5 A complete execution example of the foregoing steps 100 and 200 is provided, in which the network resources can be abbreviated as resources.

[0126] To further effectively reduce the network energy consumption in the optical transmission process, reduce the spectrum fragmentation rate and service blocking rate in the resource allocation process, and ensure the service transmission quality, in an embodiment of the present application, a method for elastic optical network resource allocation for balancing the energy consumption and path length of transmission path is provided, referring to Figure 6 Step 400 in the method for elastic optical network resource allocation for balancing the energy consumption and path length of transmission path comprises the following contents:

[0127] Step 410: The current C waveband resource utilization, L waveband resource utilization of the target elastic optical network, and the service bandwidth type corresponding to the service request are taken as the current network state vector of the target elastic optical network, and the network state vector and the respective routing feature vector corresponding to each candidate transmission path are taken as the current state vector.

[0128] Step 420: inputting the state vector into a deep deterministic policy gradient model for balancing energy consumption and path length of transmission paths, so that the deep deterministic policy gradient model performs a plurality of iteration rounds to obtain an optimal continuous action parameter; wherein in each iteration round, an actor network of the deep deterministic policy gradient model generates a continuous action parameter for balancing energy consumption and path length of transmission paths according to the state vector corresponding to the current iteration round, and a critic network in the deep deterministic policy gradient model outputs a Q value for the continuous action parameter according to the state vector, the continuous action parameter and a reward function, so that the actor network determines a state vector corresponding to a next iteration round according to the Q value.

[0129] Step 430: selecting one of the candidate transmission paths as a target transmission path based on the optimal continuous action parameter, and allocating network resources for the target transmission path for processing the service request.

[0130] In order to further improve the effectiveness and reliability of target transmission path selection, in an embodiment of the present application, a method for allocating resources of an elastic optical network for balancing energy consumption and path length of transmission paths is provided, wherein the route feature vector corresponding to the candidate transmission path comprises: normalized energy consumption, normalized path length and the generalized signal-to-noise ratio value of the candidate transmission path.

[0131] Correspondingly, referring to Figure 7 , step 430 of the method for allocating resources of an elastic optical network for balancing energy consumption and path length of transmission paths specifically comprises the following contents:

[0132] Step 431: determining a comprehensive cost score corresponding to each of the candidate transmission paths based on the optimal continuous action parameter, the normalized energy consumption and the normalized path length corresponding to each of the candidate transmission paths.

[0133] Step 432: selecting the candidate transmission path with the minimum comprehensive cost score as the target transmission path from among the candidate transmission paths.

[0134] In order to further improve the effectiveness and reliability of the application of the penalty function, in an embodiment of the present application, a method for allocating resources of an elastic optical network for balancing energy consumption and path length of transmission paths is provided, wherein the reward function comprises a cost penalty parameter group, a success reward parameter group and a failure penalty parameter group.

[0135] The cost penalty parameter group is calculated by the sum of a first parameter and a second parameter, the first parameter is calculated by the product of a first weight and the normalized energy consumption of the candidate transmission path, and the second parameter is calculated by the product of a second weight and the normalized path length of the candidate transmission path.

[0136] The success reward parameter group is calculated by a preset allocation success reward threshold and an allocation success indication function.

[0137] The failure penalty parameter group is calculated by a preset allocation failure penalty threshold and an allocation failure indication function.

[0138] The sum of the first weight and the second weight is equal to 1, and the values of the first weight and the second weight are adjusted in real time according to the C-band resource utilization of the target elastic network.

[0139] In order to further illustrate the above-mentioned embodiments, the application further provides an application example of an elastic optical network resource allocation method for balancing transmission path energy consumption and path length, that is, an energy consumption and path balancing intelligent resource allocation optimization algorithm based on DDPG deep reinforcement learning. In the research of C+L multi-band elastic optical networks (EONs), the application example studies the routing, modulation format and spectrum allocation problem considering energy consumption and routing length. In order to balance the energy consumption and length of service routing, the application example proposes a resource allocation optimization algorithm based on DDPG to efficiently solve this optimization problem. The algorithm is applied to the network scenario of L-band partial upgrade, and is optimized in the problems of waveband, modulation format selection and spectrum allocation, and considers stimulated Raman scattering (SRS) in signal-to-noise ratio analysis. The results show that, compared with the traditional RMLSA algorithm, the optimization resource allocation algorithm based on DDPG can effectively reduce the network energy consumption in the optical transmission process, ensure a low spectrum fragmentation rate and service blocking rate in the resource allocation process, and ensure the service transmission quality. The RMLSA optimization method based on the DDPG algorithm proposed in this study not only provides a new idea for the energy consumption optimization of C+L band optical networks, but also provides important theoretical and technical support for the construction of future green optical networks.

[0140] Firstly, DDPG is described. The core idea of DDPG is to use the Actor-Critic architecture to learn the optimal strategy. The actor (Actor) is responsible for outputting a deterministic strategy, that is, outputting an action a given a state s. The critic (Critic) is responsible for evaluating the goodness of the actor's strategy, and learns a value function Q(s, a, w) to estimate the value of taking action a in state s. The principle of DDPG algorithm is as shown in Figure 8

[0141] ​The DDPG algorithm integrates four neural networks, which are responsible for approximating the Q-value function and the policy. The critic target network plays a key role in the entire algorithm system, and it is mainly responsible for approximating the Q-value function Q w' (S t+1 ,π θ' (S t+1 )) of the next state-action pair. In this process, the next action value π θ' (S t+1 ) used to estimate the Q-value function of the next moment is provided by the actor target network. Based on the operation results of the two target networks, the target value of the Q-value function under the current state can be further derived:

[0142] y i =r i +γQ w' (S I+1 ,π θ' (S I+1 )) (1-1)

[0143] To evaluate the performance of the current policy, the Critic network is trained to output the Q-value function Q w (S t ,a t ) under the given state and action. In the DDPG algorithm, in order to enable the agent to explore the environment more comprehensively and improve the adaptability and generalization ability of the algorithm, a Gaussian noise function is introduced at the behavior policy level, so that the agent is no longer limited to the given policy when executing actions, but has a certain degree of randomness, thus having the opportunity to explore more different state spaces and avoiding falling into local optimal solutions. The target definition of the Critic network is:

[0144] y i -Q w (S i ,a i ) (1-2)

[0145] The parameter update of the Critic network is realized by minimizing a loss value, which is usually calculated in the form of mean square error. When updating the Critic network, the loss function aims to measure the difference between the predicted Q-value and the actual Q-value, and optimizes the network parameters by minimizing this difference. The loss function for Critic network update is:

[0146]

[0147] where a i =π θ (Si )+ε, where ε represents the exploration noise on the action policy.

[0148] In the DDPG algorithm, the Actor target network is used to generate the optimal action policy for the next state, while the Actor training network is responsible for outputting the policy action under the current state. At the same time, the Critic network evaluates the Q value of the state-action pair, providing feedback signals for the Actor training network. Based on this feedback, the Actor network updates its parameters using the Policy Gradient method, gradually optimizing the policy output to maximize the long-term cumulative reward:

[0149]

[0150] DDPG uses a soft update mechanism to update the parameters w' and θ' of the target network. In each learning process, the parameters of the target network are only updated from the online network with a small proportion, rather than directly copied. This gradual updating method effectively avoids parameter mutation and significantly improves the stability of training:

[0151] w'←ξw+(1-ξ)w' (1-5)

[0152] θ' =←ξθ+(1-ξ)θ' (1-6)

[0153] DDPG combines the characteristics of value function-based and policy-based reinforcement learning methods. This hybrid architecture not only enables DDPG to handle continuous action space problems, but also introduces action noise to give the algorithm some exploration ability, thereby achieving better policy learning in complex environments.

[0154] Based on this, the application example of the present application starts from the perspective of energy saving and consumption reduction, and proposes a routing algorithm based on the reinforcement learning deep deterministic policy gradient algorithm (DDPG) for minimum energy consumption and shortest path balance. It can fully consider the problems of energy consumption and path length, and can select the path of the business according to the energy consumption and path length of the business path, and give the most suitable path of the business. Realize in the case of multiple businesses, ensure the low energy consumption operation of the whole network, improve the network transmission capacity, and reduce the carbon footprint.

[0155] Single objective optimization such as shortest path can lead to high energy consumption, for example, some short paths may pass through high load nodes (such as long distance links requiring optical amplifiers), requiring more power to maintain signal quality, which in turn increases overall energy consumption. For example: path A (3 hops, energy consumption = 150 W) and path B (5 hops, energy consumption = 100 W). Conversely, optimizing only energy consumption may choose long paths to bypass many nodes, resulting in increased delay or more resource occupation, affecting other services. The demand in the actual network is usually multi-dimensional. For example, real-time services may pay more attention to delay (path length), while environmentally friendly or cost-sensitive scenarios value energy consumption more. The dynamic trade-off capability of DDPG can adapt to different scenarios, improving flexibility.

[0156] The core idea is to model routing and resource allocation as a continuous action space optimization problem, with the Actor network of DDPG outputting actions (selecting routing and resource allocation strategies) and the Critic network evaluating the value of actions (combining energy consumption and path length). The algorithm framework is as follows:

[0157] 1. State space definition (State Space): encode network state and candidate route features into state vectors.

[0158]

[0159] where, E i : normalized energy consumption of the i-th route:

[0160]

[0161] L i : normalized path length of the i-th route:

[0162]

[0163] G i : GSNR value of the i-th route (threshold needs to be met).

[0164] U C ,U L : utilization rate of C band and L band (0-1);

[0165] B: service bandwidth type (0: small bandwidth <50G, 1: large bandwidth >50G).

[0166] 2. Action space design (Action space): output continuous action parameter α∈[0,1], used to balance energy consumption and path length, comprehensive cost:

[0167] i=α·E i +(1-α)L i

[0168] Select the route with the minimum comprehensive cost:

[0169] a t = argmin i (αE i +(1-α)L i )

[0170] 3. Reward Function: The reward function needs to optimize energy consumption, path length and service success rate at the same time:

[0171]

[0172] Where, β1, β2: weight coefficients of energy consumption and path length (can be dynamically adjusted);

[0173] γ: allocation success reward threshold (+10);

[0174] η: allocation failure penalty threshold (-5);

[0175] Indicator function (1 for success, otherwise 0).

[0176] 4. DDPG network structure:

[0177] (1) Actor network (policy function):

[0178] Input: state vector s t ;

[0179] Output: continuous action parameter α ∈ [0, 1];

[0180] Network design:

[0181] α = σ (W2·ReLU (W1·s t +b1) +b2)

[0182] Where, W1, W2: weight matrix;

[0183] σ: Sigmoid function (restricts output range).

[0184] (2) Critic network (value function)

[0185] Input: state s t + action α;

[0186] Output: Q value estimate Q(s t , α);

[0187] Network design:

[0188] Q = W4·ReLU (W3·[st ; a] + b3) + b4

[0189] where, [s t ; a]: concatenation vector of state and action.

[0190] 5. Training procedure:

[0191] (1) Experience replay storage: store transition samples (s t , a t , r t , s t+1 ) into replay buffer (ReplayBuffer).

[0192] (2) Network update:

[0193] Critic loss function:

[0194] where, Q': target Critic network.

[0195] μ': target Actor network.

[0196] Actor gradient update:

[0197]

[0198] (3) Target network update:

[0199] Soft update parameters (τ « 1):

[0200] θ μ' ← τθ μ + (1 - τ)θ μ' θ Q‘ ← τθ Q + (1 - τ)θ Q’

[0201] 6. Dynamic weight adjustment strategy:

[0202] Adjust the weight coefficients of energy consumption and path length according to network load:

[0203]

[0204] High load (Uc>70%): pay more attention to reducing energy consumption (increase).

[0205] Low load: balance the optimization of energy consumption and path length, tend to short path, and improve business response speed;

[0206] 7. Routing logic:

[0207] For the candidate route set that meets the GSNR threshold:

[0208] (1) Normalized energy consumption and path length:

[0209]

[0210] (2) Calculate the comprehensive cost:

[0211] C i = aE i + (1-a)L i

[0212] By dynamically learning the weight parameter a e [0, 1] by DDPG, the following goals are achieved:

[0213] When the network load is high, a→1 (focus on energy consumption).

[0214] When the service delay is sensitive, a→0 (focus on path).

[0215] (3) Select the optimal route:

[0216]

[0217] In order to verify the performance of the proposed DDPG-EMC algorithm 2, the application example is simulated and analyzed under the NSFNET network, and the slot width is set to 12.5GHz, of which 1 slot is used as a protection band between any two optical paths. The total number of available frequency slots in the C band and the L band is 358 and 558 respectively, and the maximum distance of each span is set to 80 kilometers.

[0218] Since the scenario of the application example is C-C+L hybrid EONs, but how to upgrade the network to this scenario is not the focus of the application example, the application example simply upgrades the link with higher utilization to C+L band. Specifically, the shortest path between all node pairs in the network is calculated by Dijkstra algorithm or DSK algorithm, and the Dijkstra algorithm is a classic greedy algorithm for solving single-source shortest path problems in weighted directed or undirected graphs. First, calculate the number of uses of each link, denoted by N. Sort all links in descending order of N value. Then, according to the upgrading requirements and certain principles, select the links that need to be upgraded. The application example determines the link to be upgraded according to the size of N value. The number of upgraded links refers to the product of the number of network links and the link upgrade rate. The application example upgrades the link according to a link upgrade rate of 70%.

[0219] Specifically, the energy consumption and path balanced intelligent resource allocation optimization algorithm based on DDPG deep reinforcement learning provided by the application example is implemented as algorithm 1 shown in Table 1.

[0220] Table 1

[0221]

[0222]

[0223] The algorithm 1 flow includes: using the KSP algorithm to calculate a plurality of initial transmission paths (i.e., working routes) {R n} and then selecting a modulation format for the service, the modulation format selection method being as shown in the foregoing, see Figure 9 For the selected modulation format, the number of required frequency slots is calculated, and the service wavelength band selection and spectrum allocation method selection are completed according to the algorithm 2, and the available number of frequency slots is searched, the available modulation format is selected, and the same as in the L wavelength full upgrade network scenario, the available transmission path {R m} that meets the resource allocation condition for the selected target modulation format is calculated, and the generalized signal-to-noise ratio (GSNR) value is calculated for the candidate transmission path {R m'} that meets the GSNR threshold, and the application example uses the DDPG algorithm to select the target transmission path Rselect with the minimum energy consumption and the shortest path and allocate resources, otherwise, the above steps are repeated according to the selection principle by using other modulation formats in turn. If any modulation format in the selected frequency band of the service does not meet the GSNR threshold requirement, the service allocation fails.

[0224] The algorithm 2 is as shown in Table 2.

[0225] Table 2

[0226]

[0227]

[0228] The algorithm 2 is expressed as: in the running environment of the C-C+L hybrid optical network, when three different types of services are involved, three different types of path planning results can occur, for r1 and r3, all services will be allocated in the C wavelength band due to the unavailability of part of the L wavelength band, and the available frequency slots are searched in the first fitting manner; for r2, all links on the selected path are deployed with the C+L wavelength band, in order to avoid request blocking due to insufficient C wavelength band resources, the request r2 preferentially selects L wavelength band resources, large-bandwidth services search available frequency slots in the second fitting manner, and small-bandwidth services search available frequency slots in the first fitting manner, when the L wavelength band does not meet the service resource allocation requirement, C wavelength band resources are selected for service allocation, small-bandwidth services search available frequency slots in the second fitting manner, and large-bandwidth services search available frequency slots in the first fitting manner.

[0229] The application example considers applying the algorithm in the L-band partially upgraded network scene. Compared with the traditional RMLSA algorithm, the proposed intelligent routing optimization algorithm based on DDPG can effectively reduce the network energy consumption in the optical transmission process, ensure a low spectrum fragmentation rate and service blocking rate in the resource allocation process, and ensure the service transmission quality.

[0230] By using the proposed intelligent routing optimization algorithm based on DDPG for balancing energy consumption and path, the path selection of the service is performed according to the service path energy consumption and path length, the most suitable path of the service is given, and the low-energy operation of the entire network is ensured. Compared with the traditional RMLSA algorithm, the algorithm adopts a dynamic adjustment of the waveband allocation and spectrum allocation mode (combination of pre-fitting / after-fitting), and optimizes the modulation format selection mode, thereby reducing the service blocking rate and the spectrum fragmentation rate. In the selection of the modulation format, the number of frequency slots required by different modulation formats is the same, and the low-order modulation format is selected when the number of frequency slots required in all available modulation formats is the least. In the environment with a relatively low signal-to-noise ratio, the transmission performance is maintained, the bit error rate is relatively low, and the robustness is relatively strong.

[0231] From the software aspect, the application also provides a device for balancing the transmission path energy consumption and path length of an elastic optical network resource allocation method, which is used to execute all or part of the device for balancing the transmission path energy consumption and path length of the elastic optical network resource allocation method, as shown in Figure 10 , the device for balancing the transmission path energy consumption and path length of the elastic optical network resource allocation method specifically includes the following contents:

[0232] The modulation format allocation module 10 is configured to allocate a target modulation format for the service request of the target elastic optical network and determine the network resource requirement data corresponding to the target modulation format.

[0233] The waveband screening module 20 is configured to allocate a target waveband to each of the initial transmission paths according to the path waveband type of each of the initial transmission paths corresponding to the service request, determine the initial transmission path whose available network resources of the target waveband currently satisfy the network resource requirement data as an available transmission path; and the target waveband is a C waveband or an L waveband.

[0234] The GSNR screening module 30 is configured to determine the generalized signal-to-noise ratio value corresponding to each of the available transmission paths, and determine the available transmission path with the generalized signal-to-noise ratio value equal to or greater than a signal-to-noise ratio threshold value as a candidate transmission path.

[0235] The DDPG screening module 40 is configured to determine a target transmission path from the candidate transmission paths based on a deep deterministic policy gradient model for balancing energy consumption and path length of the transmission path, and allocate network resources for the target transmission path to process the service request according to the current C-band resource utilization, the L-band resource utilization, the service bandwidth type of the service request, and the routing feature vector of each candidate transmission path.

[0236] The embodiments of the elastic optical network resource allocation device for balancing energy consumption and path length of a transmission path provided in the present application can be used to execute the processing procedures of the embodiments of the elastic optical network resource allocation method for balancing energy consumption and path length of a transmission path described above, and the functions thereof will not be repeated here. Please refer to the detailed description of the embodiments of the elastic optical network resource allocation method for balancing energy consumption and path length of a transmission path described above.

[0237] The part of the elastic optical network resource allocation device for balancing energy consumption and path length of a transmission path for performing the elastic optical network resource allocation can be completed in a server or a client device. Specifically, it can be selected according to the processing capability of the client device, the use scenario of the user, and the like. The present application does not limit this. If all operations are completed in the client device, the client device can further include a processor for specific processing of the elastic optical network resource allocation for balancing energy consumption and path length of a transmission path.

[0238] The client device described above can have a communication module (i.e., a communication unit) that can be communicatively connected to a remote server to realize data transmission with the server. The server can include a server of a task scheduling center side, and can also include a server of an intermediate platform in other implementation scenarios, such as a server of a third-party server platform that is communicatively connected to the server of the task scheduling center. The server can include a single computer device, or a server cluster composed of multiple servers, or a distributed server structure.

[0239] The server and the client device can communicate using any suitable network protocol, including network protocols that have not yet been developed as of the filing date of this application. The network protocol can include, for example, a TCP / IP protocol, a UDP / IP protocol, an HTTP protocol, an HTTPS protocol, and the like. Of course, the network protocol can also include, for example, a RPC protocol (Remote Procedure Call Protocol), a REST protocol (Representational State Transfer), and the like used on top of the above-mentioned protocols.

[0240] From the above description, it can be seen that the elastic optical network resource allocation device for balancing the transmission path energy consumption and path length provided by the embodiments of the present application can fully consider the energy consumption and path length from the perspective of energy saving and consumption reduction, can select the path of the service according to the service path energy consumption and path length, and give the most suitable path for the service; further, it can effectively reduce the network energy consumption in the optical transmission process, can reduce the spectrum fragmentation rate and service blocking rate in the process of guaranteeing resource allocation, and can guarantee the service transmission quality; can ensure low energy consumption operation of the entire network, improve the network transmission capacity, and reduce the carbon footprint in the case of multiple services; not only provides a new idea for energy consumption optimization of C+L band elastic optical network, but also provides important theoretical and technical support for future green optical network construction.

[0241] The embodiments of the present application also provide an electronic device, which can include a processor, a memory, a receiver and a transmitter, the processor being configured to execute the elastic optical network resource allocation method for balancing the transmission path energy consumption and path length mentioned in the above embodiments, wherein the processor and the memory can be connected through a bus or other means. The receiver can be connected with the processor and the memory through wired or wireless means.

[0242] The processor can be a central processing unit (CPU). The processor can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, or a combination thereof.

[0243] The memory, as a non-transitory computer readable storage medium, can be configured to store non-transitory software programs, non-transitory computer executable programs and modules, such as the program instructions / modules of the method for balancing energy consumption and path length of a transmission path in an elastic optical network.

[0244] The memory can include a program storage area and a data storage area, where the program storage area can store an operating system and at least one application required by a function; and the data storage area can store data created by the processor and the like. In addition, the memory can include a high-speed random access memory, and can also include a non-transitory memory, such as at least one magnetic disk storage device, a flash memory device, or other non-transitory solid-state memory device. In some embodiments, the memory can optionally include a memory disposed remotely from the processor, and these remote memories can be connected to the processor through a network. Examples of the network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof.

[0245] The one or more modules are stored in the memory and, when executed by the processor, perform the method for balancing energy consumption and path length of a transmission path in an elastic optical network in the embodiments.

[0246] In some embodiments of the present application, a user equipment can include a processor, a memory, and a transceiver which can include a receiver and a transmitter, the processor, the memory, the receiver and the transmitter can be connected through a bus system, the memory is configured to store computer instructions, and the processor is configured to execute the computer instructions stored in the memory to control the transceiver to transceive signals.

[0247] As an implementation manner, the functions of the receiver and the transmitter in the present application can be implemented by a transceiver circuit or a transceiver dedicated chip, and the processor can be implemented by a dedicated processing chip, a processing circuit or a general-purpose chip.

[0248] As another implementation manner, the server provided by the embodiments of the present application can be implemented by using a general-purpose computer. That is, program codes for implementing the functions of the processor, the receiver and the transmitter are stored in the memory, and the general-purpose processor implements the functions of the processor, the receiver and the transmitter by executing the codes in the memory.

[0249] The embodiments of the present application further provide a computer readable storage medium, which stores a computer program. The computer program is executed by a processor to implement the steps of the method for balancing energy consumption and path length of transmission paths in a flexible optical network resource allocation method.

[0250] The embodiments of the present application further provide a computer program product, which comprises a computer program. The computer program is executed by a processor to implement the steps of the method for balancing energy consumption and path length of transmission paths in a flexible optical network resource allocation method.

[0251] Those skilled in the art should understand that all the exemplary components, systems and methods described in connection with the embodiments disclosed herein can be implemented or realized in hardware, software or a combination thereof. The particular implementations shown are merely exemplary and are not intended to limit the scope of the application, which is defined by the appended claims. Stated differently, specific implementations can be realized in hardware, software, or a combination thereof, depending on the particular application and design constraints. Those skilled in the art can realize the described functionality by using different methods, and the described implementations are not limited to any particular method. When realized in hardware, the functionality can be realized in, for example, an electronic circuit, an application specific integrated circuit (ASIC), a suitable firmware, a plug-in, a functional card, etc. When realized in software, the elements of the application are the program or code segments to perform the necessary tasks. The program or code segments can be stored in a machine readable medium, or transmitted by a carrier wave in a data signal over a transmission medium or communication link.

[0252] It is to be understood that the application is not limited to the particular configurations and processes described herein and shown in the figures. Detailed descriptions of known methods are omitted for the sake of brevity. In the above embodiments, several specific steps are described and shown as examples. However, the method processes of the application are not limited to the specific steps described and shown, and one skilled in the art can make various changes, modifications and additions, or change the order of the steps, after understanding the spirit of the application.

[0253] In the present application, the features described and / or illustrated for one embodiment can be used in the same way or in a similar way in one or more other embodiments, and / or in combination with or instead of features of other embodiments.

[0254] The above descriptions are only the preferred embodiments of the present application, and are not intended to limit the present application. The embodiments of the present application can be variously changed and modified by those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the scope of protection of the present application.

Claims

1. A method for allocating resources in an elastic optical network for balancing transmission path energy consumption and path length, characterized in that: include: Allocating a target modulation format for a current service request for a target elastic optical network and determining network resource demand data corresponding to the target modulation format; Allocating a target band to each of the initial transmission paths corresponding to the service request according to the path band type of each of the initial transmission paths, and determining the initial transmission path whose current available network resources in the target band meet the network resource demand data as an available transmission path; wherein the target band is C-band or L-band; Determine a generalized signal-to-noise ratio value corresponding to each of the available transmission paths, and determine the available transmission paths whose generalized signal-to-noise ratio values ​​are equal to or greater than a signal-to-noise ratio threshold as candidate transmission paths; Based on a deep deterministic policy gradient model for balancing the energy consumption and path length of transmission paths, one of the candidate transmission paths is selected as the target transmission path according to the current C-band resource utilization and L-band resource utilization of the target elastic optical network, the service bandwidth type corresponding to the service request, and the routing feature vector corresponding to each of the candidate transmission paths, and network resources are allocated to the target transmission path for processing the service request.

2. The elastic optical network resource allocation method for balancing transmission path energy consumption and path length according to claim 1, characterized in that: The allocating a target modulation format for a current service request for a target elastic optical network and determining network resource demand data corresponding to the target modulation format includes: Obtaining each candidate modulation format specified by a current service request for a target elastic optical network; determining the number of frequency slots required by each of the candidate modulation formats for the service request; Determining whether there are multiple candidate modulation formats with the same number of frequency slots among the candidate modulation formats; If not, selecting the highest-order modulation format from among the candidate modulation formats as the target modulation format for the service request; The number of frequency slots corresponding to the target modulation format is determined as network resource requirement data corresponding to the target modulation format.

3. The elastic optical network resource allocation method for balancing transmission path energy consumption and path length according to claim 2, characterized in that: Before determining the number of frequency slots corresponding to the target modulation format as the network resource requirement data corresponding to the target modulation format, the method further includes: If it is determined that there are multiple candidate modulation formats with the same number of frequency slots among the candidate modulation formats, determining whether the number of frequency slots corresponding to the multiple candidate modulation formats with the same number of frequency slots is a minimum value among the numbers of frequency slots corresponding to the respective candidate modulation formats; If yes, selecting the lowest-order modulation format among the candidate modulation formats as the target modulation format for the service request; If not, a highest-order modulation format is selected from each candidate modulation format as the target modulation format of the service request.

4. The elastic optical network resource allocation method for balancing transmission path energy consumption and path length according to claim 1, characterized in that: The path band types include: C band only, C+L band only, and mixed band; wherein the mixed band includes C band and C+L band; Correspondingly, allocating a target band to each of the initial transmission paths corresponding to the service request according to the path band type of each initial transmission path, and determining the initial transmission path whose current available network resources of the target band meet the network resource demand data as an available transmission path includes: If the path band type of the initial transmission path corresponding to the service request is the C+L band only, determining whether the available network resources of the L band corresponding to the initial transmission path meet the network resource requirement data; If so, the target band of the initial transmission path is configured as the L band and the initial transmission path is determined as an available transmission path.

5. The elastic optical network resource allocation method for balancing transmission path energy consumption and path length according to claim 4, characterized in that: The method further comprises: allocating a target band to each of the initial transmission paths according to the path band type of each initial transmission path corresponding to the service request, and determining the initial transmission path whose current available network resources of the target band meet the network resource demand data as an available transmission path. If the path band type of the initial transmission path corresponding to the service request is the C-band only or the mixed band, configuring the target band of the initial transmission path to be the C-band; If the path band type of the initial transmission path corresponding to the service request is the C+L band only, and it is determined that the available network resources of the L band corresponding to the initial transmission path do not meet the network resource requirement data, then configuring the target band of the initial transmission path to be the C band; In the initial transmission path where the target band is the C band, the initial transmission path where the available network resources of the C band meet the network resource requirement data is determined as an available transmission path.

6. The elastic optical network resource allocation method for balancing transmission path energy consumption and path length according to claim 1, characterized in that: The deep deterministic policy gradient model based on balancing energy consumption and path length of transmission paths selects one of the candidate transmission paths as a target transmission path according to the current C-band resource utilization and L-band resource utilization of the target elastic optical network, the service bandwidth type corresponding to the service request, and the routing feature vector corresponding to each of the candidate transmission paths, and allocates network resources to the target transmission path for processing the service request, including: The current C-band resource utilization rate and L-band resource utilization rate of the target elastic optical network and the service bandwidth type corresponding to the service request are used as the current network state vector of the target elastic optical network, and the network state vector and the routing feature vectors corresponding to each of the candidate transmission paths are used as the current state vector; Inputting the state vector into a deep deterministic policy gradient model for balancing energy consumption and path length of a transmission path, so that the deep deterministic policy gradient model executes multiple iterations to obtain optimal continuous action parameters; wherein, in each iteration, an actor network of the deep deterministic policy gradient model generates continuous action parameters for balancing energy consumption and path length of the transmission path according to the state vector corresponding to the current iteration, and causes a critic network in the deep deterministic policy gradient model to output a Q value for the continuous action parameter according to the state vector, the continuous action parameter, and a reward function, so that the actor network determines the state vector corresponding to the next iteration according to the Q value; One of the candidate transmission paths is selected as a target transmission path based on the optimal continuous action parameter, and network resources are allocated to the target transmission path for processing the service request.

7. The elastic optical network resource allocation method for balancing transmission path energy consumption and path length according to claim 6, characterized in that: The routing feature vector corresponding to the candidate transmission path includes: the normalized energy consumption, the normalized path length and the generalized signal-to-noise ratio value of the candidate transmission path; Correspondingly, selecting one of the candidate transmission paths as the target transmission path based on the optimal continuous action parameter includes: Determining a comprehensive cost score corresponding to each of the candidate transmission paths based on the optimal continuous action parameter, the normalized energy consumption and the normalized path length corresponding to each of the candidate transmission paths; Among the candidate transmission paths, the candidate transmission path with the smallest comprehensive cost score is selected as the target transmission path.

8. The elastic optical network resource allocation method for balancing transmission path energy consumption and path length according to claim 7, characterized in that: The reward function includes a cost penalty parameter group, a success reward parameter group and a failure penalty parameter group; The cost penalty parameter group is calculated by adding a first parameter and a second parameter, wherein the first parameter is calculated by multiplying a first weight and a normalized energy consumption of the candidate transmission path; The second parameter is calculated by multiplying the second weight by the normalized path length of the candidate transmission path; The success reward parameter group is calculated by a preset allocation success reward threshold and an allocation success indication function; The failure penalty parameter group is calculated by a preset allocation failure penalty threshold and an allocation failure indication function.

9. The elastic optical network resource allocation method for balancing transmission path energy consumption and path length according to claim 8, characterized in that: The sum of the first weight and the second weight is equal to 1; and the values ​​of the first weight and the second weight are adjusted in real time according to the C-band resource utilization of the target elastic network.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the elastic optical network resource allocation method for balancing transmission path energy consumption and path length as described in any one of claims 1 to 9 is implemented.

Images