CN115442216A - Network slice fault self-healing method, device, equipment and computer program product - Google Patents

Network slice fault self-healing method, device, equipment and computer program product Download PDF

Info

Publication number
CN115442216A
CN115442216A CN202110628791.2A CN202110628791A CN115442216A CN 115442216 A CN115442216 A CN 115442216A CN 202110628791 A CN202110628791 A CN 202110628791A CN 115442216 A CN115442216 A CN 115442216A
Authority
CN
China
Prior art keywords
self
action
healing
slice
fault
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202110628791.2A
Other languages
Chinese (zh)
Other versions
CN115442216B (en
Inventor
邢彪
丁东
冯杭生
陈嫦娇
陈向荣
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
China Mobile Communications Group Co Ltd
China Mobile Group Zhejiang Co Ltd
Original Assignee
China Mobile Communications Group Co Ltd
China Mobile Group Zhejiang Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by China Mobile Communications Group Co Ltd, China Mobile Group Zhejiang Co Ltd filed Critical China Mobile Communications Group Co Ltd
Priority to CN202110628791.2A priority Critical patent/CN115442216B/en
Publication of CN115442216A publication Critical patent/CN115442216A/en
Application granted granted Critical
Publication of CN115442216B publication Critical patent/CN115442216B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/06—Management of faults, events, alarms or notifications
    • H04L41/0654—Management of faults, events, alarms or notifications using network fault recovery
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L43/00—Arrangements for monitoring or testing data switching networks
    • H04L43/08—Monitoring or testing based on specific metrics, e.g. QoS, energy consumption or environmental parameters
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W24/00—Supervisory, monitoring or testing arrangements
    • H04W24/04—Arrangements for maintaining operational condition

Abstract

The invention discloses a network slice fault self-healing method, a device, equipment and a computer program product, wherein the method comprises the following steps: acquiring multidimensional index data of a network slice, and detecting the multidimensional index data; when abnormal data are detected, generating a slice self-healing action and a slice fault action according to the multi-dimensional index data and judging the slice self-healing action and the slice fault action so as to determine behavior scores of the generated slice self-healing action and the generated slice fault action; and performing iterative optimization on the generated slice self-healing action and the generated slice fault action according to the behavior score so as to determine a target self-healing action capable of enabling the network slice fault to be self-healed and perform fault repair on the network slice. According to the invention, the double intelligent bodies respectively generate the slice self-healing action and the slice fault action to form countermeasures, so that the generated slice self-healing action is forced to continuously improve the self-fault-healing capability, the performance of a fault self-healing strategy is improved, and the fault self-healing recovery capability of the network slice is further improved.

Description

Network slice fault self-healing method, device, equipment and computer program product
Technical Field
The present invention relates to the field of communications technologies, and in particular, to a network slice fault self-healing method, apparatus, device, and computer program product.
Background
The traditional network slice fault self-healing method needs to rely on the experience of technical personnel to manually set a self-healing strategy, the self-healing strategy is mainly used for simply setting threshold values for various index data to realize triggering of slice fault self-healing actions, when the self-healing strategy is realized by relying on manual setting rules, the fault self-healing efficiency is low, errors are easy to occur, and the strategy cannot be timely adjusted when a network slice changes. In the existing slice fault self-healing process, firstly the fault type and the fault influence range need to be analyzed, then a proper fault processing process is determined, fault repair or fault self-healing is realized based on a artificially defined strategy, the self-healing process flow is long, more time is spent on fault analysis and positioning, the service influence time is long, and the user experience is influenced, so that the fault self-healing performance of the existing network slice fault self-healing strategy which depends on the manual setting rule is poor.
Disclosure of Invention
The invention mainly aims to provide a network slice fault self-healing method, a network slice fault self-healing device, a network slice fault self-healing equipment and a computer program product, and aims to solve the technical problem that the existing network slice fault self-healing strategy relying on manual setting rules is poor in fault self-healing performance.
In addition, in order to achieve the above object, the present invention further provides a network slice fault self-healing method, which includes the following steps:
acquiring multidimensional index data of the network slice, and detecting the acquired multidimensional index data;
when abnormal index data exist in the multidimensional index data of the network slice, generating a slice self-healing action and a slice fault action according to the abnormal index data;
judging the generated slice self-healing action and slice fault action to determine the behavior scores of the slice self-healing action and the slice fault action;
and performing iterative optimization on the generated slice self-healing action and the generated slice fault action according to the behavior score to determine a target self-healing action, and performing fault repair on the network slice by using the target self-healing action.
Optionally, the step of generating a slice self-healing action and a slice fault action according to the multidimensional index data includes:
inputting the multidimensional index data into a preset target self-healing model, wherein the target self-healing model is obtained by utilizing historical multidimensional index data of a network slice to carry out iterative training on a preset self-healing model to be trained, and the target self-healing model comprises an action generator;
and generating a slice self-healing action and a slice fault action by utilizing an action generator of the target self-healing model according to the multi-dimensional index data.
Optionally, the action generator includes a self-healing action generator and a fault action generator, and the step of generating a slice self-healing action and a slice fault action by using the action generator of the target self-healing model according to the multidimensional index data includes:
determining a current full information state of the network slice according to the multidimensional index data, wherein the full information state comprises a first state observable by the self-healing action generator and a second state observable by the fault action generator;
inputting a first state in the full information state into the self-healing action generator, and generating a slice self-healing action by using the self-healing action generator;
and inputting a second state in the full information state into the fault action generator, and generating a slice fault action by using the fault action generator.
Optionally, the target self-healing model further includes a self-healing action evaluator and a fault action evaluator, the generated slice self-healing action and slice fault action are evaluated to determine the slice self-healing action and the behavior score of the slice fault action, including:
inputting a first state, the slice self-healing action and the slice fault action in the full information state into a self-healing action evaluator in the target self-healing model to evaluate the slice self-healing action and determine a behavior score of the slice self-healing action;
determining a fault damage radius according to the slice fault action;
and inputting a second state in the full information state, the fault damage radius, the slice self-healing action and the slice fault action into a fault action evaluation device in the target self-healing model so as to evaluate the slice fault action and determine the behavior score of the slice fault action.
Optionally, the step of performing iterative optimization on the generated slice self-healing action and slice fault action according to the behavior score to determine a target self-healing action, and performing fault repair on the network slice by using the target self-healing action includes:
feeding back the behavior score of the slice self-healing action to the self-healing action generator, feeding back the behavior score of the slice fault action to the fault action generator, returning and executing the step of generating the slice self-healing action by using the self-healing action generator in the target self-healing model, and generating the slice fault action by using the fault action generator in the target self-healing model so as to iteratively optimize the slice self-healing action and the slice fault action until the behavior score of the slice self-healing action and the behavior score of the slice fault action meet preset conditions to obtain the target self-healing action;
and switching the state of the network slice according to the target self-healing action so as to switch the network slice from the current full information state to the target full information state corresponding to the target self-healing action, and repairing the fault of the network slice.
Optionally, after the step of performing iterative optimization on the generated slice self-healing action and slice fault action according to the behavior score to determine a target self-healing action, and performing fault repairing on the network slice by using the target self-healing action, the method includes:
inputting the target self-healing action into a preset reward function to obtain a reward value corresponding to the target self-healing action;
and generating an experience replay data set according to the reward value, and updating the model parameters of the target self-healing model according to the experience replay data set.
Optionally, before the step of generating a slice self-healing action and a slice fault action according to the target index data, the method further includes:
acquiring historical index data of the network slice and preprocessing the historical index data to obtain sample data;
obtaining model architecture parameters, and establishing a basic self-healing model according to the model architecture parameters;
and performing iterative training on the basic self-healing model by using the sample data to obtain a target self-healing model.
In addition, in order to achieve the above object, the present invention further provides a network slice fault self-healing device, including:
the data detection module is used for acquiring the multidimensional index data of the network slice and detecting the acquired multidimensional index data;
the action generation module is used for generating a slice self-healing action and a slice fault action according to the multi-dimensional index data when abnormal data exist in the multi-dimensional index data of the network slice, and judging the generated slice self-healing action and the generated slice fault action so as to determine behavior scores of the slice self-healing action and the slice fault action;
and the double-agent countermeasure module is used for performing iterative optimization on the generated slice self-healing action and the generated slice fault action according to the behavior score so as to determine a target self-healing action and performing fault repair on the network slice by utilizing the target self-healing action.
In addition, to achieve the above object, the present invention further provides a network slice fault self-healing device, including: the self-healing network slice fault processing system comprises a memory, a processor and a self-healing network slice fault program which is stored on the memory and can run on the processor, wherein when the self-healing network slice fault program is executed by the processor, the steps of the self-healing network slice fault processing method are realized.
Furthermore, to achieve the above object, the present invention further provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, the computer program implements the steps of the network slice fault self-healing method as described above.
The embodiment of the invention provides a network slice fault self-healing method, a device, equipment and a computer program product. Compared with the existing self-healing strategy fault self-healing method for the network slice, the method has the advantages that the self-healing strategy fault self-healing performance is poor, in the embodiment of the invention, the multidimensional index data of the network slice are obtained, and the obtained multidimensional index data are detected; when abnormal data exist in the multi-dimensional index data of the network slice, generating a slice self-healing action and a slice fault action according to the multi-dimensional index data, and judging the generated slice self-healing action and the generated slice fault action to determine behavior scores of the slice self-healing action and the slice fault action; and performing iterative optimization on the generated slice self-healing action and the generated slice fault action according to the behavior score to determine a target self-healing action, and performing fault repair on the network slice by using the target self-healing action. When the fact that most of network slices are abnormal in index data is detected, faults are actively manufactured through generating slice fault actions and are resisted with the generated slice self-healing actions, the generated slice self-healing actions are forced to improve the fault repairing capacity of the self, the fault self-healing performance of a network slice self-healing strategy is improved, the fault problems in the network slices can be timely identified and repaired, the fault self-healing is recovered before the network slice faults are interrupted, serious consequences are avoided, and the self-healing recovery capacity of the network slices is improved.
Drawings
Fig. 1 is a schematic diagram of a hardware structure of an embodiment of a network slice fault self-healing device according to an embodiment of the present invention;
fig. 2 is a schematic flow chart of a network slice fault self-healing method according to a first embodiment of the present invention;
fig. 3 is a schematic diagram of a countermeasure process of an action generator in a second embodiment of the network slice fault self-healing method according to the present invention;
fig. 4 is a schematic diagram of a neural network hierarchical architecture of an action generator and an action evaluator in a third embodiment of the network slice fault self-healing method according to the present invention;
fig. 5 is a schematic functional module diagram of a network slice fault self-healing device according to an embodiment of the present invention.
The implementation, functional features and advantages of the present invention will be further described with reference to the accompanying drawings.
Detailed Description
It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
In the following description, suffixes such as "module", "component", or "unit" used to denote elements are used only for facilitating the explanation of the present invention, and have no specific meaning in itself. Thus, "module", "component" or "unit" may be used mixedly.
The main solution of the embodiment of the invention is as follows: the method comprises the steps of obtaining multidimensional index data of a network slice, and detecting the obtained multidimensional index data; when abnormal data exist in the multi-dimensional index data of the network slice, generating a slice self-healing action and a slice fault action according to the multi-dimensional index data, and judging the generated slice self-healing action and the generated slice fault action to determine behavior scores of the slice self-healing action and the slice fault action; and performing iterative optimization on the generated slice self-healing action and the generated slice fault action according to the behavior score to determine a target self-healing action, and performing fault repair on the network slice by using the target self-healing action. Through initiatively making the trouble, make the two intelligent agent that generates section self-healing action and generate section fault action fight against, improve the fault self-healing performance of network section self-healing strategy, in time discern and repair the trouble problem in the network section, make the fault self-healing recover before the network section trouble causes the interrupt, avoid causing serious consequence, and then improved the self-healing recovery ability of network section.
The embodiment of the invention relates to the following main technical terms:
network Slice (NS): network slicing refers to customizing different logical networks according to different service requirements on a physical or virtual network infrastructure. The network slice may be a complete end-to-end network including the terminal device, the access network, the transport network, the core network and the application server, and may provide a complete communication service with certain network capabilities. The network slice may also be any combination of terminal device, access network, transport network, core network and application server.
Chaos Engineering (Chaos Engineering): the chaos engineering is a technical means which can ensure the availability of the system and improve the elastic capability of the technical architecture, aims to kill faults in swaddling, namely identifies the faults before the faults cause interruption, actively identifies and repairs the fault problems by actively manufacturing the faults and testing the behavior of the system under various pressures, and avoids causing serious consequences.
NSMF: the Network Slice Management Function (Network Slice Management Function) is responsible for receiving Network Slice requirements, managing life cycle, performance, faults and the like of the Network slices, arranging the composition of the Network slices, decomposing the Network Slice requirements into the requirements of each Network Slice subnet or Network Function, and sending Network Slice subnet Management requests to each NSSMF.
NSSMF: a Network Slice Subnet Management Function (Network Slice Subnet Management Function) receives a Network Slice Subnet deployment request issued from the NSMF, manages the Network Slice Subnet, arranges the composition of the Network Slice Subnet, maps the SLA (Service Level Agreement) request of the Network Slice Subnet to a QoS (Quality of Service) request of the Network Service, and issues the deployment request of the Network Service.
The embodiment of the invention considers that in the existing related scheme, the self-healing strategy of the network slice fault depends on the manually set rule, the self-healing recovery process is long in time consumption and easy to make mistakes, and when the slice fault changes, more time is used for fault analysis and positioning, and the self-healing strategy cannot be adjusted in time. In order to solve the problems, chaotic engineering is introduced into a self-healing strategy of a network slice, and the fault problem is actively identified and repaired by actively manufacturing the fault and testing the behavior of the system under various pressures, so that serious consequences are avoided. However, the existing network slice fault self-healing method based on chaotic engineering mainly depends on artificial fault manufacturing, and the rule of manufacturing faults is easy to learn by a self-healing system, so that the generalization capability and the self-healing recovery capability of a self-healing strategy are reduced. Therefore, the problem of poor fault self-healing performance generally exists in the conventional network slice fault self-healing strategy.
Therefore, the embodiment of the invention provides a solution, when the multi-dimensional index data of the network slice is detected to be abnormal, the fault is actively manufactured by generating the slice fault action and forms an countermeasure with the generated slice self-healing action, so that the generated slice self-healing action is forced to continuously improve the fault repairing capability of the self, the performance of the fault self-healing strategy is further improved, the fault is identified and repaired before the network slice fault is interrupted, the fault self-healing recovery capability of the network slice is improved, and the serious consequences are avoided.
Specifically, referring to fig. 1, fig. 1 is a schematic diagram of functional modules of a terminal device belonging to a network slice fault self-healing apparatus according to an embodiment of the present invention, where the terminal device (also called a terminal or device) may be a PC, or may be a mobile terminal device having a data processing function, such as a smart phone, a tablet computer, and a portable computer.
As shown in fig. 1, the terminal device may include: a processor 1001, such as a CPU, a network interface 1004, a user interface 1003, a memory 1005, a communication bus 1002. Wherein a communication bus 1002 is used to enable connective communication between these components. The user interface 1003 may include a Display (Display), an input unit such as a Keyboard (Keyboard), and the optional user interface 1003 may also include a standard wired interface, a wireless interface. The network interface 1004 may optionally include a standard wired interface, a wireless interface (e.g., WI-FI interface). The memory 1005 may be a high-speed RAM memory or a non-volatile memory (e.g., a magnetic disk memory). The memory 1005 may alternatively be a storage device separate from the processor 1001.
Optionally, the terminal may further include a camera, a Radio Frequency (RF) circuit, a sensor, an audio circuit, a WiFi module, and the like. Such as light sensors, motion sensors, and other sensors. Specifically, the light sensor may include an ambient light sensor that may adjust the brightness of the display screen according to the brightness of ambient light, and a proximity sensor that may turn off the display screen and/or the backlight when the mobile terminal is moved to the ear. As one of the motion sensors, the gravity acceleration sensor can detect the magnitude of acceleration in each direction (generally, three axes), detect the magnitude and direction of gravity when the mobile terminal is stationary, and can be used for applications (such as horizontal and vertical screen switching, related games, magnetometer attitude calibration), vibration recognition related functions (such as pedometer and tapping) and the like for recognizing the attitude of the mobile terminal; of course, the mobile terminal may also be configured with other sensors such as a gyroscope, a barometer, a hygrometer, a thermometer, and an infrared sensor, which are not described herein again.
Those skilled in the art will appreciate that the terminal structure shown in fig. 1 is not intended to be limiting and may include more or fewer components than those shown, or some components may be combined, or a different arrangement of components.
As shown in fig. 1, a memory 1005, which is a computer program product, may include an operating system, a network communication module, a user interface module, and a network slice fault self-healing program.
In the terminal shown in fig. 1, the network interface 1004 is mainly used for connecting to a backend server and performing data communication with the backend server; the user interface 1003 is mainly used for connecting a client (user side) and performing data communication with the client; and the processor 1001 may be configured to invoke a network slice fault self-healing program stored in the memory 1005, which when executed by the processor, implements the operations in the network slice fault self-healing method provided by the embodiments described below.
Based on the hardware structure of the equipment, the embodiment of the network slice fault self-healing method is provided.
Referring to fig. 2, in a first embodiment of the network slice fault self-healing method according to the present invention, the network slice fault self-healing method includes:
step S10, obtaining multidimensional index data of the network slice, and detecting the obtained multidimensional index data;
with the development of communication technology, especially the popularization and application of the 5th-generation (5G) technology, the concept of network slicing is introduced to cope with the difference of the demands of different communication services on network performance. In order to ensure the continuity of network services, a network slice needs to have a certain self-healing recovery capability, and a fault is self-healed before the network slice fault causes network interruption.
The invention provides a network slice fault self-healing method applicable to 5G network slices, which is based on reinforcement learning and chaotic engineering of double intelligent bodies, generates fault actions and self-healing actions by actively manufacturing faults, improves the self-healing generalization capability of slices by utilizing the countermeasure between the double intelligent bodies of the fault actions and the self-healing actions, and forces the generated self-healing actions to continuously improve the self-healing capability of the slices, thereby improving the self-healing recovery capability of the network slices. Specifically, multi-dimensional index data of the network slice is obtained first, and the obtained multi-dimensional index data is detected to determine whether an abnormality exists in the network slice. It can be known that, in a network environment where the network slice is located, the NSMF is generally provided with an NSSMF and is responsible for managing the network slice, and is mainly responsible for monitoring multidimensional index data of the network slice, the network slice generally includes a terminal device, an access network, a transmission network, a core network, and an application server, and the access network, the transmission network, and the core network in the network slice include a plurality of network sub-slices managed by the NSSMF, such as a wireless access network sub-slice, a transmission network sub-slice, a core network sub-slice, and the like. In this embodiment, when obtaining multi-dimensional index data of a network slice, mainly obtaining KPI (Key Performance Indicator or target quantization management Indicator) data of a network sub-slice from the NSMF, where the KPI of the network sub-slice includes KPI of a wireless access network sub-slice, a transmission network sub-slice, a core network sub-slice, and the like, where the KPI of the wireless access network sub-slice includes transmission delay of a wireless access network, average throughput of uplink/downlink users, average throughput of uplink/downlink cells, average occupancy of CPUs, number of online users, qoS flow establishment success rate, call establishment success rate, and the like; the KPI of the transmission network sub-slice comprises transmission network transmission time delay, bandwidth utilization rate, packet loss rate, data transmission quantity, bit error rate and the like; the KPI of the core network sub-slice comprises core network transmission delay, virtualized storage resource utilization rate, virtualized network resource utilization rate, virtualized computing resource utilization rate, the number of error codes, request success rate and the like. The acquired KPI data are detected, and when abnormal KPI data are detected, for example, a certain KPI value obviously exceeds a normal range, it indicates that a fault exists in the network slice or a potential fault exists, and the fault needs to be identified and automatically repaired before the service interruption is caused by the fault, so that the fault of the network slice is self-healed, and serious consequences such as downtime are avoided.
Step 20, when abnormal data are detected in the multi-dimensional index data of the network slice, generating a slice self-healing action and a slice fault action according to the multi-dimensional index data, and judging the generated slice self-healing action and the generated slice fault action to determine behavior scores of the slice self-healing action and the slice fault action;
in this embodiment, the network slice fault self-healing method is provided with a self-healing model, and when it is detected that most of the acquired index data has abnormal data, the model can generate a slice self-healing action and a slice fault action according to the acquired multidimensional index data, and then evaluate the generated fault action and self-healing action, thereby determining behavior scores of the generated slice self-healing action and the generated slice fault action. When generating a slice fault action, the model is converged toward a direction with a high behavior score of the slice fault action, and further a fault action which makes the network slice difficult to repair is generated. And then evaluating the generated self-healing action and the fault action of the slice, and according to the behavior scores of the self-healing action and the fault action of the slice, enabling the self-healing action and the fault action of the slice to resist against each other, so that the self-healing action of the generated slice is forced to continuously improve the fault repairing capability of the self-healing action.
The refinement of the step S20 comprises the steps A1-A2:
step A1, inputting the multidimensional index data into a preset target self-healing model, wherein the target self-healing model is obtained by performing iterative training on the preset self-healing model to be trained by using historical multidimensional index data of a network slice, and the target self-healing model comprises an action generator;
and A2, generating a slice self-healing action and a slice fault action by utilizing an action generator of the target self-healing model according to the multi-dimensional index data.
Further, when generating a slice self-healing action and a slice failure action, preprocessing may be performed on the acquired multidimensional index data, where the preprocessing includes but is not limited to normalization processing, taking normalization processing as an example, the acquired index data is scaled in proportion and uniformly mapped to a smaller interval, that is, the data is scaled to a value between a given minimum value and a given maximum value, usually between 0 and 1, to obtain target index data, and subsequently, when processing the obtained target index data, the convergence speed of the data and the precision of data processing may be improved. And inputting the multidimensional index data obtained through preprocessing into a preset target self-healing model, wherein the target self-healing model is obtained by performing iterative training on a basic self-healing model to be trained by using historical multidimensional index data of network slices, the model comprises a motion generator of double intelligent bodies, and a slice self-healing motion and a slice fault motion can be respectively generated according to the input multidimensional index data.
Further, the action generator comprises a self-healing action generator and a fault action generator, and the step A2 of generating the slice self-healing action and the slice fault action by using the action generator of the target self-healing model according to the multidimensional index data comprises the steps A21-A23:
step A21, determining a current full information state of the network slice according to the multidimensional index data, wherein the full information state comprises a first state observable by the self-healing action generator and a second state observable by the fault action generator;
step A22, inputting a first state in the full information state into the self-healing action generator, and generating a slice self-healing action by using the self-healing action generator;
step a23, inputting a second state of the full information states into the failure action generator, and generating a slice failure action by using the failure action generator.
In this embodiment, the action generator of the target self-healing model includes a self-healing action generator (actor 1) and a fault action generator (actor 2), and when a slice self-healing action and a slice fault action are generated by using the action generator of the preset target self-healing model, a current full information state of a network slice is first determined according to input target index data, where the full information state includes a first state(s) that the actor1 can observe 1 ) And a second state(s) observable by actor2 2 ). And then uses the state s of the network slice that the operator 1 can observe according to 1 Generating a slice self-healing action, and using the status s of the network slice that can be observed by the operator 2 2 A slice failure action is generated.
And S30, performing iterative optimization on the generated slice self-healing action and the generated slice fault action according to the behavior score to determine a target self-healing action, and performing fault repair on the network slice by using the target self-healing action.
In the target self-healing model, different action generators generate corresponding actions according to partial states of network slices which can be observed by the action generators, the generated slice self-healing actions and the slice fault actions are evaluated, behavior scores of the generated actions are determined, iterative optimization is carried out on the generated slice self-healing actions and the generated slice fault actions according to the behavior scores, the self-healing action generators are forced to generate the slice self-healing actions with higher fault repairing capacity, and then the target self-healing actions which can enable the network slice faults to self-heal are determined. And carrying out fault repair on the network slice by using the target self-healing action so as to enable the fault of the network slice to be self-healed. The method for repairing the network slice includes, but is not limited to, switching the state of the network slice according to the determined target self-healing action.
In the embodiment, the multidimensional index data of the network slice is obtained, and the obtained multidimensional index data is detected; when abnormal data exist in the multi-dimensional index data of the network slice, generating a slice self-healing action and a slice fault action according to the multi-dimensional index data, and judging the generated slice self-healing action and the generated slice fault action to determine behavior scores of the slice self-healing action and the slice fault action; and performing iterative optimization on the generated slice self-healing action and the generated slice fault action according to the behavior score to determine a target self-healing action, and performing fault repair on the network slice by using the target self-healing action. When the fact that most of network slices are abnormal in index data is detected, faults are actively manufactured through generating slice fault actions and are resisted with the generated slice self-healing actions, the generated slice self-healing actions are forced to improve the fault repairing capacity of the self, the fault self-healing performance of a network slice self-healing strategy is improved, the fault problems in the network slices can be timely identified and repaired, the fault self-healing is recovered before the network slice faults are interrupted, serious consequences are avoided, and the self-healing recovery capacity of the network slices is improved.
Further, on the basis of the above embodiment of the present invention, a second embodiment of the network slice fault self-healing method of the present invention is provided.
The present embodiment is a step of refining step S20 in the first embodiment, and the present embodiment is different from the above embodiments of the present invention in that the target self-healing model in the present embodiment further includes a self-healing action evaluator and a fault action evaluator, and in step S20 in the above embodiments, the generated slice self-healing action and slice fault action are evaluated to determine refinement of behavior scores of the slice self-healing action and the slice fault action: comprising steps B1-B3:
step B1, inputting a first state in the full information state, the slice self-healing action and the slice fault action into a self-healing action evaluator in the target self-healing model to evaluate the slice self-healing action and determine a behavior score of the slice self-healing action;
b2, determining a fault damage radius according to the slice fault action;
and B3, inputting the second state in the full information state, the fault damage radius, the slice self-healing action and the slice fault action into a fault action evaluation device in the target self-healing model so as to evaluate the slice fault action and determine the behavior score of the slice fault action.
Based on the above embodiments, in the present embodiment, when the generated slice self-healing action and the slice failure action are evaluated, the full information state of the network slice, the generated slice self-healing action, the slice failure action, and the like are input into the evaluator of the target self-healing model for evaluation, and the behavior scores of the generated slice self-healing action and the slice failure action are further determined. Specifically, the target self-healing model further comprises a self-healing action evaluator and a fault action evaluator, and the generated slice self-healing action (a) 1 ) Self-healing action generator actor1 can observe the state s of network slice 1 And the generated slice fault action (a) 2 ) Inputting the data into a self-healing action evaluator (critic 1) to evaluate the self-healing action of the slice to obtain a self-healing action a of the slice 1 Behavior score of Q 1 (s 1 ,a 1 ,a 2 ). Determining fault damage radius from generated slice fault actions(c 2 ) The state s of the network slice that can be observed by the fault action generator operator 2 2 Failure radius of failure c 2 Slice failure action a 2 And self-healing action of slicing a 1 Inputting the data into a fault action evaluator (critic 2) to evaluate the slice fault action to obtain a slice fault action a 2 Behavior score of Q 2 (s 2 ,c 2 ,a 2 ,a 1 )。
Further, the refinement of step S30 in the above embodiment further includes steps C1-C2:
step C1, feeding back a behavior score of the slice self-healing action to the self-healing action generator, feeding back a behavior score of the slice fault action to the fault action generator, returning and executing the step of generating the slice self-healing action by using the self-healing action generator in the target self-healing model, and generating the slice fault action by using the fault action generator in the target self-healing model so as to optimize and update the slice self-healing action and the slice fault action until the behavior score of the slice self-healing action and the behavior score of the slice fault action meet preset conditions, thereby obtaining the target self-healing action;
and step C2, switching the network slice from the current full information state to a target full information state corresponding to the target self-healing action so as to repair the network slice.
Further, after the behavior scores of the slice self-healing action and the slice fault action are determined, the target self-healing action is determined according to the behavior scores, so that when fault repair is performed on the network slice, the behavior scores of the slice self-healing action and the slice fault action are required to be fed back to the corresponding action generators actor1 and actor2 respectively, the steps of generating the slice self-healing action by using the self-healing action generator in the target self-healing model and generating the slice fault action by using the fault action generator in the target self-healing model are returned and executed, and new slice self-healing action and new slice fault action are generated. And then evaluating the newly generated self-healing action and fault action of the slice, and continuously optimizing and updating the generated self-healing action and fault action of the slice through a DNQ (Deep Q-Learning) cycle structure to form the countermeasure of the actor1 and actor2 double intelligent bodies and improve the stability and the convergence of the self-healing model. And determining the target self-healing action until the generated slice self-healing action and the generated slice fault action meet preset conditions, wherein the preset conditions can be that the behavior score of the slice self-healing action is highest, and the fault generated by the slice fault action can be repaired.
Referring to fig. 3, fig. 3 is a schematic diagram of a countermeasure process of a slice self-healing action and a slice failure action generated by the target self-healing model in this embodiment, and in fig. 3, a network slice outputs a full information state s all Then, the activator 1 and the activator 2 can only obtain partial state information which can be observed by the activator 1 and the activator 2, while the critic1 and the critic2 can obtain the full information state and simultaneously obtain the strategy actions taken by the two agents, namely the activator 1 and the activator 2. That is, although different operators cannot acquire all information and cannot acquire the policies of other operators, the evaluator corresponding to each operator can observe all information and instruct the corresponding operator to optimize the policy of the operator. Specifically, the behavior scores obtained by the judgers critic1 and critic2, i.e., the value Q 1 And Q 2 And respectively feeding back to the action generators actor1 and actor2, determining whether the currently generated action is the optimal strategy or not by the action generators actor1 and actor2 according to the determined action strategy corresponding to the current action score, and if not, generating a new action to optimize and update the own strategy. The determined behavior policies of the action generators, namely DPG (Deterministic Policy Gradient), of the action generators, action 1 and action 2 enable the behavior of each step to directly obtain a determined value through the function μ, where the DPG is shown in the following formula 1:
a t ＝μ(s t |θ μ ) (1)
in the formula 1, a t The action selected for time t, s t Is the state of the environment at time t, θ μ For weight value, the function mu is the optimal behavior strategy, and the training can obtain a deterministic optimal behavior strategy function and a deterministic strategy gradientThe formula is shown in the following formula 2:
Figure BDA0003102219590000131
wherein the content of the first and second substances,
Figure BDA0003102219590000141
the Gradient is expressed, it should be noted that the Deterministic strategy in the present invention is based on the DDPG (Deep Deterministic Policy Gradient) of DPG, and the Deep learning neural network is fused with the strategy learning method of DPG, that is, both the cost function and the strategy function are expressed by the neural network. Compared with DPG, the DDPG simulates a strategy function mu and a strategy function Q, namely a strategy network and a Q network, and then trains the neural network by using a deep learning method to obtain the action generators actor1 and actor2 of the target self-healing model to select and generate action deterministic strategy gradients.
Further, the deterministic strategy gradient shown in the above equations 1-2 is only one preferred strategy of the present embodiment, and is used for illustrating the embodiments of the present invention, and is not used to limit the present invention. In this embodiment, the generated slice self-healing action and the slice fault action are optimized and updated according to the behavior score obtained by the evaluator, and after the target self-healing action is determined, the network slice is switched from the current full information state to the target full information state according to the target full information state of the network slice corresponding to the target self-healing action, so as to complete fault repair of the network slice. The switching of the network slice full information state includes, for example, wireless network sub-slice: switching to a standby service channel; transmission of mesh sub-slices: switching to a standby logical port; core net slicing: switching to a standby virtual machine, etc.
In the embodiment, a slice self-healing action and a slice fault action are generated by controlling double intelligent bodies, the generated slice self-healing action and the generated slice fault action are judged by utilizing a judging device, based on a deterministic strategy gradient, the generated slice self-healing action and the generated slice fault action are optimized and updated according to judged action scores, so that the actions of the double intelligent bodies of an action generator generate counteractions, the self-healing action generator is forced to generate the self-healing action with higher fault repairing capability, an optimal action strategy, namely a target self-healing action, is further determined, the generalization capability of a self-healing strategy is improved, the state of a network slice is switched according to the determined target self-healing action, and the fault self-healing capability of the network slice is improved.
Based on the foregoing embodiment, a third embodiment of the network slice fault self-healing method according to the present invention is provided, where in the foregoing embodiment, the step after step S30 further includes steps D1-D2:
step D1, inputting the target self-healing action into a preset reward function to obtain a reward value corresponding to the target self-healing action;
and D2, generating an empirical replay data set according to the reward value, and updating the model parameters of the target self-healing model according to the empirical replay data set.
Based on the foregoing embodiment, in this embodiment, after the fault recovery is performed on the network slice according to the determined target self-healing action, the target self-healing action is input into a preset reward function, so as to obtain a reward value corresponding to the target self-healing action, an experience replay data set is generated according to the reward value, and a model parameter of the target self-healing model is updated by using the generated experience replay data set.
Specifically, the action generator and the evaluator of the target self-healing model both include an estimation network and a target network, and generate empirical replay data such as (s, a, r, s '), where s is the full information state of the network slice before the target self-healing action is switched, s' is the full information state of the network slice after the target self-healing action is switched, and a = (a =) (a 1 ，a 2 ) The actions performed by the actor1 and the actor2 in the full information state of the network slice, r represents the reward value which is the profit obtained from the environment after the actor performs the actions, and the action, the reward and the result (s, a, r, s') of the next state in each state of the network slice are recorded in the generated experience playback data set. When the experience playback data set has oneAnd the fixed storage capacity can cover the data stored firstly when the data is fully stored.
When the model parameters are updated by using the empirical playback data set, the estimation network is mainly updated, the parameters of the target network can be obtained by tracking the parameters of the estimation network, and when the estimation network is updated, the action generator operator network mu (s theta) is initialized firstly μ ) And a critic network Q (s, a | θ) Q ) The initialized weight values are respectively theta μ And theta Q Then, the target network Q' = Q (s, a | θ) is initialized Q ) And μ' = μ (s | θ) μ ) Randomly selecting N pieces of experience playback data from the experience playback data set, and updating the parameters of the judger according to an objective function shown in the following formula 3:
y i ＝r i +γQ'(s i+1 ,μ'(s i+1 |θ μ′ |θ Q′ )) (3)
wherein, y i Representing the target network, r i Representative of a prize value, θ μ' And theta Q' Representing the target weight value, gamma represents the discount factor, and the evaluator parameters are updated by minimizing the loss function shown in the following equation 4:
Figure BDA0003102219590000151
in equation 4, L is a loss function, in this embodiment, the Q value is obtained based on the square loss of the real and estimated Q values, the estimated Q value is obtained from the current state s and the action a input of the action estimation network output to the estimation network, the real Q value is obtained by adding the real reward r and the discount value of the Q value, and the real Q value is obtained by inputting the state s 'at the next time and the action a' obtained by the action real network to the real network. The evaluator parameters are updated by minimizing the loss function. For updating the action generator actor, based on the deterministic strategy gradient shown in the formula 1, the actor network is updated by the deterministic strategy gradient, and the actor aims to obtain an action with a high Q value as much as possibleTherefore, the loss of the operator can be simply understood as the larger the obtained feedback Q value is, the smaller the loss is, and the smaller the obtained feedback Q value is, the larger the loss is, and therefore, the parameter gradient of the operator itself is used
Figure BDA0003102219590000161
Action gradient in combination with an evaluator
Figure BDA0003102219590000162
Gradient updating method for operator network according to gradient updating mode shown in the following formula 5
Figure BDA0003102219590000163
The updating is performed so that the operator updates the self-parameter in a direction in which a relatively large Q value is more likely to be obtained.
Figure BDA0003102219590000164
Finally, by tracking and learning the parameter update of the estimated network, the target network parameters are updated according to the following formula 6, wherein the formula 6 is a weight update mode of the critic and the actor target networks:
θ′ i ←τθ i +(1-τ)θ′ i (6)
in this embodiment, τ is set to a number very close to 1, so that the parameter θ of the target network does not fluctuate greatly, thereby improving the stability of the model.
Before step S20, further comprising steps E1-E3:
e1, acquiring historical index data of the network slice and preprocessing the historical index data to obtain sample data;
step E2, obtaining model architecture parameters, and establishing a basic self-healing model according to the model architecture parameters;
and E3, performing iterative training on the basic self-healing model by using the sample data to obtain a target self-healing model.
Further, based on the above-described embodiment, the model needs to be pre-trained before generating the slice self-healing action and the slice failure action from the processed target index data. When the model is pre-trained, firstly, historical multi-dimensional index data of a network slice is obtained and is pre-processed to obtain sample data of the model pre-training, then a strategy function, an evaluation function and model architecture parameters are obtained, a self-healing model of a foundation is built, and then the sample data is used for carrying out iterative training on the foundation model to obtain a target self-healing model. The architecture parameters of the model comprise the number of the layers of the neural network of the model, the number of the neurons arranged in each layer of the neural network and the like.
Based on the target self-healing model in the above embodiment, the basic self-healing model built in this embodiment is a dual-agent DDPG model, and includes two action generators (actors) and two criterics (criterics), where each actor or criteric includes two neural networks with the same structure, namely, a target network (target _ net) and an estimation network (eval _ net), and only the parameter update frequencies are different. Taking the above actor1 and critic1 as examples, the built basic self-healing model is shown in fig. 4, fig. 4 is a schematic diagram of a neural network hierarchy architecture of a self-healing action generator and a self-healing action evaluator of the built basic self-healing model to be trained, and in fig. 4, an input layer of the actor1 is used for inputting a state s of a current slice network 1 The hidden layer comprises 3 full connection layers (sense), and 256 neurons and 128 neurons are respectively arranged in the hidden layer. A discard layer is introduced after each fully connected layer to effectively avoid model overfitting, neurons are discarded in the discard layer with probability p and other neurons are retained with probability q =1-p, and the discard probability p =0.2 is set in the embodiment, that is, 20% of neurons are randomly ignored to be disabled. The output layer is a fully connected layer (sense): and 3 neurons are set, and self-healing actions of the network subsections are output, wherein the self-healing actions comprise a wireless network subsection (switching to a standby service channel), a transmission network subsection (switching to a standby logic port) and a core network subsection (switching to a standby virtual machine).
The evaluator critic1 is respectively provided with two input layers, one input layer 1 is used for receiving the full information state of the network slice in the latest T time period, and the other input layer 2 is used for receiving the generated slice self-healing action and the slice fault actionSetting two full-connection layers (Dense) under an input layer 1, respectively setting 256 and 128 neurons, setting 1 full-connection layer (Dense) under an input layer 2, setting 16 neurons, merging the action and the state through a merging layer (merge), finally setting a full-connection layer containing 128 neurons and an output layer containing only 1 neuron, and finally outputting and evaluating Q selected by the action 1 (s 1 ,a 1 ,a 2 ) The value is obtained.
After a neural network hierarchical structure of a basic self-healing model is built, the built and self-healing model is trained by using acquired sample data, two agent operators are respectively trained during training, and specifically, a multi-dimensional KPI (kernel-based Key Performance indicator) is cut into historical network slices including a wireless access network sub-slice, a transmission network sub-slice and a core network sub-slice 1 ) The self-healing operation corresponding to the generated network slice is output after being input to an operator 1 composed of a fully connected neural network. The state KPI(s) of the network slice is then sliced 1 ) And the generated slice self-healing action (a) 1 ) Slice failure operation (a) 2 ) Inputting the signal into a critic1 which is composed of a multi-branch fully-connected neural network, and outputting a Q value Q for evaluating the action selection 1 (s 1 ,a 1 ,a 2 ). Will Q 1 The value is fed back to actor1, and actor1 is according to Q 1 Self-healing actions which can reduce the business influence to the minimum degree are selected according to the value, and the converged operator 1 model weight can be used as a slice self-healing action generator.
Then slicing KPIs(s) in multiple dimensions for the historical network including radio access network sub-slices, transmission network sub-slices, core network sub-slices KPIs 2 ) And failure radius of rupture (c) 2 ) Inputting the data into an actor2 composed of a fully connected neural network, and outputting the generated slice fault generation operation (a) 2 ). Slicing a network into states KPI(s) 2 ) Failure radius of failure (c) 2 ) And the self-healing action of the generated slice (a) 1 ) Slice failure operation (a) 2 ) Inputting the data into a critic2 consisting of a multi-branch fully-connected neural network, and outputting a Q value Q for evaluating the selection of the action 2 (s 2 ,c 2 a 2 ,a 1 ). Will Q 2 Value is fed back to action generator, operator 2 is according to Q 2 The value is selected to select the slice fault action which can make the slice hard to self-heal to the maximum extent under a certain fault damage radius. And training the converged operator 2 model weight to be used as a slice fault action generator. In the model pre-training process, the experience replay data (s, a, r, s') of all stages in the training process are stored to form an experience replay data set which is used for a basic data set for parameter updating in the model operation process, and the iterative training of the built basic self-healing model is completed by using sample data to obtain the target self-healing model.
In the embodiment, iterative training is performed on the built basic self-healing model by using historical multi-dimensional index data of the network slice to obtain the target self-healing model, and after fault repair of the network slice is completed according to the target self-healing action, an experience playback data set is generated by using the target self-healing action, so that model parameters are updated, and the self-healing performance of the target self-healing model is improved.
In addition, referring to fig. 5, an embodiment of the present invention further provides a network slice fault self-healing device, where the network slice fault self-healing device includes:
the data detection module 10 is configured to acquire multidimensional index data of the network slice and detect the acquired multidimensional index data;
the action generating module 20 is configured to, when it is detected that abnormal data exists in the multidimensional index data of the network slice, generate a slice self-healing action and a slice fault action according to the multidimensional index data, and evaluate the generated slice self-healing action and the generated slice fault action to determine behavior scores of the slice self-healing action and the slice fault action;
and the double-agent countermeasure module 30 is used for performing iterative optimization on the generated slice self-healing action and the generated slice fault action according to the behavior score so as to determine a target self-healing action, and performing fault repair on the network slice by using the target self-healing action.
Optionally, the action generating module 20 is further configured to:
inputting the multidimensional index data into a preset target self-healing model, wherein the target self-healing model is obtained by performing iterative training on the preset self-healing model to be trained by using historical multidimensional index data of a network slice, and the target self-healing model comprises an action generator;
and generating a slice self-healing action and a slice fault action by utilizing an action generator of the target self-healing model according to the target index data.
Optionally, the action generating module 20 is further configured to:
determining a current full information state of the network slice according to the multidimensional index data, wherein the full information state comprises a first state observable by the self-healing action generator and a second state observable by the fault action generator;
inputting a first state in the full information state into the self-healing action generator, and generating a slice self-healing action by using the self-healing action generator;
and inputting a second state in the full information state into the fault action generator, and generating a slice fault action by using the fault action generator.
Optionally, the action generating module 20 is further configured to:
inputting a first state, the slice self-healing action and the slice fault action in the full information state into a self-healing action evaluator in the target self-healing model to evaluate the slice self-healing action and determine a behavior score of the slice self-healing action;
determining a fault damage radius according to the slice fault action;
and inputting a second state in the full information state, the fault damage radius, the slice self-healing action and the slice fault action into a fault action evaluation device in the target self-healing model so as to evaluate the slice fault action and determine the behavior score of the slice fault action.
Optionally, the dual agent countermeasure module 30 is further configured to:
feeding back the behavior score of the slice self-healing action to the self-healing action generator, feeding back the behavior score of the slice fault action to the fault action generator, returning and executing the step of generating the slice self-healing action by using the self-healing action generator in the target self-healing model, and generating the slice fault action by using the fault action generator in the target self-healing model so as to optimize and update the slice self-healing action and the slice fault action until the behavior score of the slice self-healing action and the behavior score of the slice fault action meet preset conditions to obtain the target self-healing action;
and switching the network slice from the current full information state to a target full information state corresponding to the target self-healing action so as to repair the network slice.
Optionally, the network slice fault self-healing device further includes a self-healing policy updating module, configured to:
inputting the target self-healing action into a preset reward function to obtain a reward value corresponding to the target self-healing action;
and generating an experience replay data set according to the reward value, and updating the model parameters of the target self-healing model according to the experience replay data set.
Optionally, the network slice fault self-healing device further includes a model training module, configured to:
acquiring historical index data of the network slice and preprocessing the historical index data to obtain sample data;
obtaining model architecture parameters, and establishing a basic self-healing model according to the model architecture parameters;
and performing iterative training on the basic self-healing model by using the sample data to obtain a target self-healing model.
In addition, an embodiment of the present invention further provides a computer program product, where the computer program product includes a computer program, and when the computer program is executed by a processor, the computer program implements the operations in the network slice fault self-healing method provided in the foregoing embodiment.
The embodiments of the device and the computer program product of the present invention can refer to the embodiments of the network slice fault self-healing method of the present invention, and are not described herein again.
It is noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity/action/object from another entity/action/object without necessarily requiring or implying any actual such relationship or order between such entities/actions/objects; the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or system. Without further limitation, an element defined by the phrase "comprising a … …" does not exclude the presence of another identical element in a process, method, article, or system that comprises the element.
For the apparatus embodiment, since it is substantially similar to the method embodiment, it is described relatively simply, and reference may be made to some descriptions of the method embodiment for relevant points. The above-described apparatus embodiments are merely illustrative, in that elements described as separate components may or may not be physically separate. Some or all of the modules can be selected according to actual needs to achieve the purpose of the scheme of the invention. One of ordinary skill in the art can understand and implement it without inventive effort.
The above-mentioned serial numbers of the embodiments of the present invention are only for description, and do not represent the advantages and disadvantages of the embodiments.
Through the above description of the embodiments, those skilled in the art will clearly understand that the method of the above embodiments can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware, but in many cases, the former is a better implementation manner. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium (e.g., ROM/RAM, magnetic disk, optical disk) as described above, and includes instructions for enabling a terminal device (e.g., a mobile phone, a computer, a server, an air conditioner, or a network device) to execute the network slice fault self-healing method according to the embodiments of the present invention.
The above description is only a preferred embodiment of the present invention, and not intended to limit the scope of the present invention, and all modifications of equivalent structures and equivalent processes, which are made by using the contents of the present specification and the accompanying drawings, or directly or indirectly applied to other related technical fields, are included in the scope of the present invention.

Claims (10)

1. A network slice fault self-healing method is characterized by comprising the following steps:
acquiring multidimensional index data of the network slice, and detecting the acquired multidimensional index data;
when abnormal data exist in the multi-dimensional index data of the network slice, generating a slice self-healing action and a slice fault action according to the multi-dimensional index data, and judging the generated slice self-healing action and the generated slice fault action to determine behavior scores of the slice self-healing action and the slice fault action;
and performing iterative optimization on the generated slice self-healing action and the generated slice fault action according to the behavior score to determine a target self-healing action, and performing fault repair on the network slice by using the target self-healing action.
2. The network slice fault self-healing method according to claim 1, wherein the step of generating a slice self-healing action and a slice fault action according to the multidimensional index data includes:
inputting the multidimensional index data into a preset target self-healing model, wherein the target self-healing model is obtained by performing iterative training on the preset self-healing model to be trained by using historical multidimensional index data of a network slice, and the target self-healing model comprises an action generator;
and generating a slice self-healing action and a slice fault action by utilizing an action generator of the target self-healing model according to the multi-dimensional index data.
3. The network slicing fault self-healing method according to claim 2, wherein the action generator comprises a self-healing action generator and a fault action generator, and the step of generating slicing self-healing actions and slicing fault actions by the action generator of the target self-healing model according to the multidimensional index data comprises:
determining a current full information state of the network slice according to the multidimensional index data, wherein the full information state comprises a first state observable by the self-healing action generator and a second state observable by the fault action generator;
inputting a first state in the full information state into the self-healing action generator, and generating a slice self-healing action by using the self-healing action generator;
and inputting a second state in the full information state into the fault action generator, and generating a slice fault action by using the fault action generator.
4. The network slice fault self-healing method according to claim 3, wherein the target self-healing model further comprises a self-healing action evaluator and a fault action evaluator, and the evaluating the generated slice self-healing action and slice fault action to determine the behavior scores of the slice self-healing action and the slice fault action comprises:
inputting a first state, the slice self-healing action and the slice fault action in the full information state into a self-healing action evaluator in the target self-healing model to evaluate the slice self-healing action and determine a behavior score of the slice self-healing action;
determining a fault damage radius according to the slice fault action;
and inputting a second state in the full information state, the fault damage radius, the slice self-healing action and the slice fault action into a fault action evaluation device in the target self-healing model so as to evaluate the slice fault action and determine the behavior score of the slice fault action.
5. The network slice fault self-healing method according to claim 4, wherein the step of iteratively optimizing the generated slice self-healing action and slice fault action according to the behavior score to determine a target self-healing action, and performing fault healing on the network slice using the target self-healing action comprises:
feeding back the behavior score of the slice self-healing action to the self-healing action generator, feeding back the behavior score of the slice fault action to the fault action generator, returning and executing the step of generating the slice self-healing action by using the self-healing action generator in the target self-healing model, and generating the slice fault action by using the fault action generator in the target self-healing model, so as to perform iterative optimization on the slice self-healing action and the slice fault action until the behavior score of the slice self-healing action and the behavior score of the slice fault action meet preset conditions, thereby obtaining the target self-healing action;
and switching the state of the network slice according to the target self-healing action so as to switch the network slice from the current full information state to the target full information state corresponding to the target self-healing action, and repairing the fault of the network slice.
6. The method according to claim 1, wherein the step of iteratively optimizing the generated slice self-healing action and slice fault action according to the behavior score to determine a target self-healing action, and performing fault repair on the network slice using the target self-healing action comprises:
inputting the target self-healing action into a preset reward function to obtain a reward value corresponding to the target self-healing action;
and generating an experience replay data set according to the reward value, and updating the model parameters of the target self-healing model according to the experience replay data set.
7. A network slice fault self-healing method according to claim 1, wherein before the step of generating a slice self-healing action and a slice fault action according to the target index data, the method further comprises:
acquiring historical index data of the network slice and preprocessing the historical index data to obtain sample data;
obtaining model architecture parameters, and establishing a basic self-healing model according to the model architecture parameters;
and performing iterative training on the basic self-healing model by using the sample data to obtain a target self-healing model.
8. The utility model provides a network section fault self-healing device which characterized in that, network section fault self-healing device includes:
the data detection module is used for acquiring the multidimensional index data of the network slice and detecting the acquired multidimensional index data;
the action generation module is used for generating a slice self-healing action and a slice fault action according to the multi-dimensional index data when abnormal data exist in the multi-dimensional index data of the network slice, and judging the generated slice self-healing action and the generated slice fault action so as to determine behavior scores of the slice self-healing action and the slice fault action;
and the double-agent countermeasure module is used for performing iterative optimization on the generated slice self-healing action and the generated slice fault action according to the behavior score so as to determine a target self-healing action and performing fault repair on the network slice by utilizing the target self-healing action.
9. The utility model provides a network section fault self-healing equipment which characterized in that, network section fault self-healing equipment includes: a memory, a processor, and a network slice fault self-healing program stored on the memory and executable on the processor, the network slice fault self-healing program when executed by the processor implementing the steps of the network slice fault self-healing method according to any of claims 1 to 7.
10. A computer program product, characterized in that the computer program product comprises a computer program which, when executed by a processor, implements the network slice fault self-healing method according to any of claims 1-7.
CN202110628791.2A 2021-06-04 2021-06-04 Network slice fault self-healing method, device, equipment and computer storage medium Active CN115442216B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110628791.2A CN115442216B (en) 2021-06-04 2021-06-04 Network slice fault self-healing method, device, equipment and computer storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110628791.2A CN115442216B (en) 2021-06-04 2021-06-04 Network slice fault self-healing method, device, equipment and computer storage medium

Publications (2)

Publication Number Publication Date
CN115442216A true CN115442216A (en) 2022-12-06
CN115442216B CN115442216B (en) 2023-09-05

Family

ID=84271986

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110628791.2A Active CN115442216B (en) 2021-06-04 2021-06-04 Network slice fault self-healing method, device, equipment and computer storage medium

Country Status (1)

Country Link
CN (1) CN115442216B (en)

Citations (15)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107872339A (en) * 2016-09-27 2018-04-03 中兴通讯股份有限公司 O＆M implementation method and device, Virtual Networking System in a kind of virtualization network
CN108337110A (en) * 2018-01-02 2018-07-27 中兴通讯股份有限公司 A kind of virtual resource management method and device, computer readable storage medium
CN108632065A (en) * 2017-03-20 2018-10-09 华为技术有限公司 Manage method, the network equipment and the system of network slice example
CN108933686A (en) * 2017-05-25 2018-12-04 中兴通讯股份有限公司 A kind of network dicing method, computer equipment and storage medium
CN109067579A (en) * 2018-08-01 2018-12-21 重庆邮电大学 Towards the 5G network slice topology design of bottom layer node failure and reliable mapping method
CN109768875A (en) * 2017-11-10 2019-05-17 华为技术有限公司 Policy management method, device, equipment and the system of network slice
CN110602717A (en) * 2019-09-18 2019-12-20 四川长虹电器股份有限公司 Disaster recovery processing method for 5G network slice
WO2020087948A1 (en) * 2018-10-31 2020-05-07 中兴通讯股份有限公司 Network slice template generation method, apparatus and device, and storage medium
CN111555910A (en) * 2020-04-22 2020-08-18 北京思特奇信息技术股份有限公司 Self-healing and self-operation maintenance method and system based on network slice
CN111865665A (en) * 2020-06-23 2020-10-30 广州衡昊数据科技有限公司 Network equipment fault self-healing method and device
CN112313906A (en) * 2018-06-22 2021-02-02 华为技术有限公司 Data analysis management, configuration specification and process, provisioning, and service-based architecture
CN112365366A (en) * 2020-11-12 2021-02-12 广东电网有限责任公司 Micro-grid management method and system based on intelligent 5G slice
CN112534778A (en) * 2018-08-06 2021-03-19 苹果公司 Management data analysis KPI for 5G network traffic and resources
CN112748864A (en) * 2019-10-31 2021-05-04 伊姆西Ip控股有限责任公司 Method for allocating storage disk, electronic device and computer program product
CN112821560A (en) * 2021-01-26 2021-05-18 珠海博威电气股份有限公司 Distributed self-healing protection method for overhead line distribution network

Patent Citations (15)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107872339A (en) * 2016-09-27 2018-04-03 中兴通讯股份有限公司 O＆M implementation method and device, Virtual Networking System in a kind of virtualization network
CN108632065A (en) * 2017-03-20 2018-10-09 华为技术有限公司 Manage method, the network equipment and the system of network slice example
CN108933686A (en) * 2017-05-25 2018-12-04 中兴通讯股份有限公司 A kind of network dicing method, computer equipment and storage medium
CN109768875A (en) * 2017-11-10 2019-05-17 华为技术有限公司 Policy management method, device, equipment and the system of network slice
CN108337110A (en) * 2018-01-02 2018-07-27 中兴通讯股份有限公司 A kind of virtual resource management method and device, computer readable storage medium
CN112313906A (en) * 2018-06-22 2021-02-02 华为技术有限公司 Data analysis management, configuration specification and process, provisioning, and service-based architecture
CN109067579A (en) * 2018-08-01 2018-12-21 重庆邮电大学 Towards the 5G network slice topology design of bottom layer node failure and reliable mapping method
CN112534778A (en) * 2018-08-06 2021-03-19 苹果公司 Management data analysis KPI for 5G network traffic and resources
WO2020087948A1 (en) * 2018-10-31 2020-05-07 中兴通讯股份有限公司 Network slice template generation method, apparatus and device, and storage medium
CN110602717A (en) * 2019-09-18 2019-12-20 四川长虹电器股份有限公司 Disaster recovery processing method for 5G network slice
CN112748864A (en) * 2019-10-31 2021-05-04 伊姆西Ip控股有限责任公司 Method for allocating storage disk, electronic device and computer program product
CN111555910A (en) * 2020-04-22 2020-08-18 北京思特奇信息技术股份有限公司 Self-healing and self-operation maintenance method and system based on network slice
CN111865665A (en) * 2020-06-23 2020-10-30 广州衡昊数据科技有限公司 Network equipment fault self-healing method and device
CN112365366A (en) * 2020-11-12 2021-02-12 广东电网有限责任公司 Micro-grid management method and system based on intelligent 5G slice
CN112821560A (en) * 2021-01-26 2021-05-18 珠海博威电气股份有限公司 Distributed self-healing protection method for overhead line distribution network

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
"Modeling system based on fuzzy dynamic Bayesian network for fault diagnosis and reliability prediction", 《2015 ANNUAL RELIABILITY AND MAINTAINABILITY SYMPOSIUM (RAMS)》 *
廖皓: "网络切片场景下服务功能链故障检测与诊断算法研究", 《中国优秀硕士论文电子期刊网》 *
李千目,赵学龙,张宏,刘凤玉: "一种基于自愈策略的网络故障诊断系统", 仪器仪表学报, no. 1 *

Also Published As

Publication number Publication date
CN115442216B (en) 2023-09-05

Similar Documents

Publication Publication Date Title
CN110598802B (en) Memory detection model training method, memory detection method and device
US20200193313A1 (en) Interpretability-based machine learning adjustment during production
US11487493B2 (en) Automatic display unit backup during failures of one more display units through the utilization of graphic user interface objects defined for control transfer and reversion after resolution of the failures
US11704597B2 (en) Techniques to generate network simulation scenarios
KR102087959B1 (en) Artificial intelligence operations system of telecommunication network, and operating method thereof
US11500370B2 (en) System for predictive maintenance using generative adversarial networks for failure prediction
CN113486584B (en) Method and device for predicting equipment failure, computer equipment and computer readable storage medium
CN109743286A (en) A kind of IP type mark method and apparatus based on figure convolutional neural networks
CN116418653A (en) Fault positioning method and device based on multi-index root cause positioning algorithm
US20170141949A1 (en) Method and apparatus for processing alarm information in cloud computing
WO2023093431A1 (en) Model training method and apparatus, and device, storage medium and program product
CN107122464A (en) A kind of aid decision-making system and method
CN115442216B (en) Network slice fault self-healing method, device, equipment and computer storage medium
CN113409096B (en) Target object identification method and device, computer equipment and storage medium
Banerjee et al. On detection of manipulative cognitive functions in cognitive autonomous networks
US11810011B2 (en) Generation of predictive model metrics using intermediate operations
CN112764957A (en) Application fault delimiting method and device
CN109902486A (en) Electronic device, abnormal user processing strategie Intelligent Decision-making Method and storage medium
CN109436980A (en) The condition detection method and system of elevator components
WO2023186091A1 (en) Sample determination method, apparatus and device
CN116450808B (en) Data processing method and device and storage medium
CN112202686B (en) Adaptive access identification method for differential flow control and terminal equipment
US20220277326A1 (en) Performance and quality improvements for a market research platform
CN103699937B (en) A kind of method and apparatus for forecasting software project characteristic
CN116010808A (en) Base station detection method, base station detection device, electronic equipment and computer storage medium

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant