A first node, a third node, a fourth node for processing parameters for configuring a node in a communication network and a method performed thereby

Abstract

A method for processing parameters for configuring a second node (112) performed by a first node (111). The first node (111) determines (207) parameters for configuring the second node (112). The determination (207) is based on an analysis performed using a generative adversarial network, the analysis comprising iteratively performing: i) generating a set of parameters estimated to be associated with a performance of a first node group (121) above a first threshold, and ii) distinguishing the generated set of parameters from a first set of parameters observed to be associated with a performance of the node group (121) above the first threshold to obtain a score for each parameter. The score indicates the degree to which the generated parameter differs from the first set of parameters. The determined parameter has a score resulting from the distinguishing that is below another threshold. The first node (111) also outputs an indication including the determined parameter.

Description

Technical Field

[0001] The present disclosure generally relates to a first node for processing parameters for configuring a second node in a communication network and methods performed thereby. The present disclosure also generally relates to a third node for processing parameters for configuring a second node in a communication network and methods performed thereby. The present disclosure further generally relates to a fourth node for processing parameters for configuring a second node in a communication network and methods performed thereby. Background Art

[0002] The computer system may include one or more nodes. A node may include one or more processors (which together with the computer program code can perform different functions and actions), a memory, a receiving port, and a sending port. A node may be, for example, a server. A node may be included in a communication network.

[0003] Some types of nodes within a communication network may be wireless devices, such as stations (STAs), user equipment (UEs), mobile terminals, wireless terminals, terminals, and / or mobile stations (MSs). A wireless device is enabled to communicate wirelessly in a cellular communication network or wireless communication network (sometimes also referred to as a cellular radio system, a cellular system, or a cellular network). Communication may be performed, for example, between two wireless devices, between a wireless device and a regular phone, and / or between a wireless device and a server via a radio access network (RAN) and one or more core networks that may be included in a telecommunications network. A wireless device may also be referred to as a mobile phone, a cell phone, a laptop computer, or a tablet computer with wireless capabilities, to mention only some other examples. A wireless device in this context may be, for example, a portable, pocket-storable, handheld, computer-included, or vehicle-mounted mobile device that is enabled to transmit voice and / or data via a RAN with another entity (such as another terminal or server).

[0004] A communication network may cover a geographical area that may be divided into cell areas, each of which is served by another type of node, a network node or a transmission point (TP), for example an access node, such as a base station (BS), for example a radio base station (RBS), which may sometimes be referred to as, for example, an evolved Node B ("eNB"), "eNodeB", "Node B", "B Node", or BTS (Base Transceiver Station), depending on the technology and terminology used. Based on transmission power and thus also based on cell size, base stations may have different classes, such as, for example, wide area base stations, medium range base stations, local area base stations, and home base stations. A cell is a geographical area in which radio coverage is provided by a base station at a base station site. One base station located at a base station site may serve one or more cells. Further, each base station may support one or more communication technologies. Any cell may also be referred to as a node. In this document, in the case where a base station manages more than one cell, a node may generally refer to one base station when managing one of the cells.

[0005] In 3rd Generation Partnership Project (3GPP) Long Term Evolution (LTE), base stations (which may be referred to as eNodeBs or even eNBs) may be directly connected to one or more core networks. All data transmission in LTE is controlled by radio base stations.

[0006] Standardization organization 3GPP is currently in the process of specifying a new radio interface known as NR or 5G-UTRA and a fifth generation (5G) packet core network (which may be referred to as Next Generation Core Network, abbreviated as NG-CN, NGC, 5G CN or 5GC).

[0007] The 3GPP system including a 5G access network (AN), a 5G core network, and a UE may be referred to as a 5G system.

[0008] During operation in a communication network, the performance of a node (e.g., a cell) may vary. It may be optimal, or it may be degraded. The performance of a node may be understood as a measure of the execution of multiple tasks. The performance of a node may be measured by one or more indicators, metrics, or parameters. By comparing the observed values ​​of these parameters in the node with the corresponding thresholds of each parameter, it may be determined whether the performance in the node is optimal or degraded. Typically, the performance of nodes (such as cells) operating at various bandwidths (e.g., 700, 1900, and 2100 Hz) may be degraded due to large traffic, interference, coverage issues, weather factors, and various other reasons. Performance degradation may be measured according to various call event parameters (such as mute, garbled, soft drop, hard drop, access failure, etc.). Measurements of these parameters may be obtained from each node every day. The degradation of these event parameters in a consistent manner may be understood as indicating poor performance of the node. For example, if the cell is ranked higher, it is expected that the cell may experience a lower percentage of mute, garbled, soft drop, hard drop, access failure events, while if the cell is ranked lower, it may behave the opposite.

[0009] Poor performance of a node can result in a waste of processing, energy, and time-frequency resources, leading to poor performance of the network and a poor user experience. Summary of the invention

[0010] The degradation of node (e.g., cell) performance has an important relationship with the change of the configuration management (CM) characteristics of the node. Typically, in existing methods, depending on the generation of the network, the degraded performance of the node can be manually addressed during the installation of the base station when the problem has already occurred, that is, after the fact or in retrospect, or during the roll-out, in an attempt to achieve correctness by design. Both approaches can be problematic because in the first approach, the problem has already occurred, i.e., the degradation of performance, while in the second approach, once the base station is deployed, the characteristics that may have been available may no longer be maintained due to environmental issues (such as, for example, high temperature or poor maintenance). Therefore, both approaches may still have a negative impact on the performance of the nodes in the communication network.

[0011] It is an object of embodiments herein to improve the process of configuring the performance of a node in a communication network.A particular object of embodiments herein is to improve the process of configuring the parameters of a node in a communication network.

[0012] Embodiments herein provide methods that enable fine-tuning of deployments in a manner that mitigates and even eliminates poor performance.

[0013] Embodiments herein describe an automated correction method to restore a poorly performing node to its correct state. Certain embodiments herein may describe an automated method to remotely tune CM parameters so that their configuration may best suit their changing traffic scenarios.

[0014] As a step toward achieving an automatically correct network, nodes that may need correction may first be identified. Embodiments herein may then recommend appropriate CM parameter changes that may be needed in that node to achieve improved performance. Thus, embodiments herein may be understood to address two problems: a) how to rank nodes based on performance, and b) what CM parameters may need to be changed in order to correct a poorly performing node.

[0015] Embodiments herein may rely on an autoencoder approach to rank nodes based on their historical performance data. Based on daily rankings, poorly performing nodes with a higher chance of needing correction in configuration management parameters may be shortlisted, since typically node performance is degraded due to misconfiguration of these CM parameters while intending to adjust with traffic that varies throughout the day.

[0016] Further in particular, the embodiments herein rely on an unsupervised learning second generation neural network based technology autoencoder to model the behavior of the node. The autoencoder can be understood as a method based on a deep neural network (DNN) to learn the probability distribution of node performance and detect deviations of daily node performance from their derived empirical distribution. In this method, it will be observed that those nodes that can be identified as deviating from normal behavior tend to have the best performance or the worst performance, because deviations can occur on both sides of normal behavior.

[0017] Alternative approaches such as using a similarity or distance metric (e.g., a Euclidean distance metric) to identify configuration settings for a given traffic condition will not be able to identify CM settings for a given traffic condition very effectively because the distribution of CM parameters for a given traffic setting is not learned. That is, by using a similarity or distance metric, only the closest situations in historical data that are similar to the current situation, such as traffic and events, can be determined, and CM parameters for the identified similar situations can be identified. However, the best solution for the current situation may not be determined. In addition, there is no obvious way to decide which similarity metric may need to be employed.

[0018] According to a first aspect of the embodiments herein, an object is achieved by a method performed by a first node. The method is used to process parameters for configuring a second node. The first node operates in a communication network. The first node determines one or more parameters for configuring the second node for operation of delivering a service in the communication network. The determination is based on an analysis performed by a generative adversarial network (GAN). The GAN is managed by the first node. The analysis includes iteratively performing: i) generating a parameter set estimated to be associated with the performance of a first node group operating in the communication network above a first threshold; and ii) distinguishing the generated parameter set and the first parameter set observed to be associated with the performance of the first node group above the first threshold to obtain a corresponding score for each parameter. The corresponding score indicates the degree to which the generated parameter set is different from the first parameter set. The determined one or more parameters have a corresponding score generated by the discrimination below a second threshold. The first node also outputs an indication including the determined one or more parameters.

[0019] According to a second aspect of the embodiments herein, the purpose is achieved by a method performed by a third node. The method is used to process parameters for configuring a second node. The third node is an autoencoder operating in a communication network. The third node obtains a first parameter set that is observed to be associated with the performance of a first node group operating in the communication network that is higher than a first threshold. The first parameter set is a parameter for configuring the second node for operation of delivering a service in the communication network. Obtaining includes the following i-iv. i) Training a first data model with corresponding data of nodes included in the node group, the corresponding data of the node indicating the corresponding performance of the node over a period of time. The node group includes a first node group. ii) Obtaining a reconstruction of the corresponding error of the trained first data model for each node in the node group. iii) Sort the nodes included in the node group based on their corresponding reconstruction errors. iv) Selecting a first node group from the node group based on their corresponding reconstruction errors being higher than a first threshold. The third node also provides a first indication indicating the obtained first parameter set to the first node operating in the communication network.

[0020] According to a third aspect of the embodiments herein, the object is achieved by a method performed by a fourth node. The method is used to process parameters for configuring a second node. The fourth node operates in a communication network. The fourth node obtains a fixed-length vector encoding a plurality of traffic conditions observed to be associated with the performance of a first node group operating in the communication network. The obtaining includes: training a second data model with corresponding sequential traffic data of nodes included in the first node group to encode the corresponding sequential traffic data into a fixed-length vector. The fourth node provides a second indication indicating the obtained fixed-length vector encoding the plurality of traffic conditions to the first node operating in the communication network.

[0021] According to a fourth aspect of the embodiments herein, the purpose is achieved by a first node for processing parameters for configuring a second node. The first node is configured to operate in a communication network. The first node is also configured to determine one or more parameters for configuring the second node for operation of delivering a service in the communication network. The determination is configured to be based on an analysis configured to be performed by a GAN, which is configured to be managed by the first node. The analysis is configured to include iteratively performing i-ii. i) Generate a parameter set estimated to be associated with the performance of a first node group configured to operate in the communication network above a first threshold. ii) Distinguish the parameter set configured to be generated and the first parameter set observed to be associated with the performance of the first node group above the first threshold to obtain a corresponding score for each parameter. The corresponding score is configured to indicate the degree to which the generated parameter set is different from the first parameter set. The one or more parameters configured to be determined are configured to have a corresponding score generated by the discrimination below a second threshold. The first node is also configured to output an indication including the one or more parameters configured to be determined.

[0022] According to a fifth aspect of the embodiments herein, the purpose is achieved by a third node for processing parameters for configuring a second node. The third node is configured as an autoencoder configured to operate in a communication network. The third node is also configured to obtain a first parameter set configured to be observed to be associated with the performance of a first node group configured to operate in the communication network above a first threshold. The first parameter set is configured to be a parameter for configuring the second node for operation of delivering services in the communication network. Obtaining is configured to include i-iv. i) training a first data model with corresponding data of nodes configured to be included in the node group, the corresponding data of the node indicating the corresponding performance of the node over a period of time. The node group is configured to include a first node group. ii) obtaining a reconstruction of the corresponding error of the first data model configured to be trained for each node in the node group. iii) sorting the nodes configured to be included in the node group based on their corresponding reconstruction errors. In addition, iv) selecting a first node group from the node group based on their corresponding errors configured to be reconstructed being above a first threshold. The third node is further configured to provide, to the first node configured to operate in the communication network, a first indication configured to indicate a first set of parameters configured to be obtained.

[0023] According to a sixth aspect of the embodiments herein, the purpose is achieved by a fourth node for processing parameters configuring a second node. The fourth node is configured to operate in a communication network. The fourth node is also configured to obtain a fixed-length vector encoding a plurality of traffic conditions, which are configured to be observed to be associated with the performance of a first node group configured to operate in the communication network. The acquisition is configured to include: training a second data model with corresponding sequential traffic data of nodes configured to be included in the first node group to encode the corresponding sequential traffic data into a fixed-length vector. The fourth node is configured to provide a second indication indicating a fixed-length vector configured to be obtained to encode a plurality of traffic conditions to the first node configured to operate in the communication network.

[0024] By determining one or more parameters for configuring the second node, the first node is enabled to automatically correct the configuration of the second node using the determined one or more parameters so that the performance of the second node (e.g., given similar conditions, such as traffic) can be corrected to be as similar as possible to the performance observed in the first node group (i.e., the best performing node in the node group).

[0025] By training the first data model with corresponding data of the nodes included in the node group, the third node can be enabled to predict, or ultimately the first node can be enabled to predict, for any given node, what the expected performance may be, that is, the expected value of each of the modeled events, and therefore, detect any performance that may deviate from the expected performance, as explained below.

[0026] By obtaining the first parameter set, the third node can be enabled to provide a first indication to the first node, thereby enabling the first node to determine one or more parameters using the first parameter set as an input to the GAN. By training the first data model with the corresponding data of the nodes included in the node group, the third node can enable the first node to predict what the expected performance may be for any given node, that is, the expected value of each of the modeled events, and therefore, detect any performance that may deviate from the expected performance. By obtaining a reconstruction of the corresponding error of the trained first data model for each node in the node group, the third node can be enabled to determine which nodes in the node group deviate from normal behavior. By sorting or scoring the nodes included in the node group based on their corresponding reconstruction errors, the third node can be enabled to determine which nodes in the node group deviate the most from normal behavior. By selecting the first node group from the node group, the nodes with the best performance included in the node group can be identified, which then enables their parameter settings to be evaluated so that the first parameter set can be identified, and given similar situations, such as similar traffic conditions, suitable configurations of parameters can be recommended to other nodes to optimize, improve or correct their performance in the communication network.

[0027] By obtaining the fixed length vector, the fourth node can be enabled to provide the second indication, and in turn, the first node can be enabled to determine one or more parameters for configuring the second node for operation, and to do so based on different traffic conditions. In other words, the fourth node can enable the first node to determine which parameters may be most suitable for configuring the second node to obtain good performance given specific traffic conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] According to the following description, examples of embodiments herein are described in more detail with reference to the accompanying drawings.

[0029] Figure 1 are schematic diagrams showing two non-limiting examples of communication networks in a) and b), respectively, according to embodiments herein.

[0030] Figure 2 is a flow chart depicting an embodiment of a method in a first node according to embodiments herein.

[0031] Figure 3 is a flow chart depicting an embodiment of a method in a third node according to embodiments herein.

[0032] Figure 4 is a flow chart depicting an embodiment of a method in a fourth node according to embodiments herein.

[0033] Figure 5is a flow chart depicting a non-limiting example of a method in a communication network according to embodiments herein.

[0034] Figure 6 is a schematic block diagram illustrating two non-limiting examples a) and b) of a first node according to embodiments herein.

[0035] Figure 7 is a schematic block diagram illustrating two non-limiting examples a) and b) of a third node according to embodiments herein.

[0036] Figure 8 is a schematic block diagram illustrating two non-limiting examples a) and b) of a fourth node according to embodiments herein. DETAILED DESCRIPTION

[0037] Several embodiments are included herein that address limitations of existing methods. The embodiments herein may be understood as being directed to improving the performance of nodes (such as cells) in a communication network. As an overview, the embodiments herein may be understood as involving methods for ranking nodes and recommending CM parameter changes.

[0038] Embodiments will now be described more fully below with reference to the accompanying drawings in which examples are shown. In this section, embodiments herein are described by exemplary embodiments. It should be noted that these embodiments are not mutually exclusive. Components from one embodiment or example may be assumed to exist in another embodiment or example by default, and it will be apparent to those skilled in the art how those components may be used in other exemplary embodiments.

[0039] Figure 1 Two non-limiting examples of communication networks 100 in which embodiments herein may be implemented are depicted in panels "a" and "b," respectively. In some example implementations, such as in Figure 1 As depicted in the non-limiting example of a), the communication network 100 may be a computer network. The communication network 100 may support a 5G system or a system with similar functionality. In other example implementations, such as in Figure 1 b), the communication network 100 can be implemented in a telecommunications network 105 (sometimes also referred to as a cellular radio system, a cellular network, or a wireless communication system). In some examples, the telecommunications network 105 can include network nodes that can serve receiving nodes (such as wireless devices) with a serving beam.

[0040] In some examples, telecommunications network 105 may be, for example, a network such as a 5G system, a 5G network, or a next generation network. The telecommunications network 105 may also support other technologies, such as a Long Term Evolution (LTE) network, e.g., LTE frequency division duplex (FDD), LTE time division duplex (TDD), LTE half-duplex frequency division duplex (HD-FDD), LTE operating in an unlicensed band, Wideband Code Division Multiple Access (WCDMA), Universal Terrestrial Radio Access (UTRA) TDD, GSM / Enhanced Data rates for GSM Evolution (EDGE) Radio Access Network (GERAN) network, Ultra Mobile Broadband (UMB), EDGE network, a network including any combination of radio access technologies (RATs) (such as, for example, multi-standard radio (MSR) base stations, multi-RAT base stations, etc.), any third generation partnership project (3GPP) cellular network, a wireless local area network (WLAN) or WiFi network, Worldwide Interoperability for Microwave Access (WiMax), a low power short range network based on IEEE802.15.4 (such as IPv6 over low power wireless personal area network (6LowPAN)), Zigbee, Z-Wave, Bluetooth Low Energy (BLE), or any cellular network or system.

[0041] The communication network 100 includes a plurality of nodes. Figure 1 , wherein the first node 111, the second node 112, the third node 113, and the fourth node 114 are depicted. Each of the first node 111, the second node 112, the third node 113, and the fourth node 114 can be understood as a first computer system, a second computer system, a first computer system, and a fourth computer system, respectively.

[0042] Any one of the first node 111, the second node 112, the third node 113, and the fourth node 114 may be a network node. In a specific example, the first node 111, the third node 113, and the fourth node 114 may be core network nodes.

[0043] The first node 111 may be understood as a node having the capability of including or managing a generative adversarial network (GAN), such as a first core network node.

[0044] The third node 113 can be understood as an autoencoder or a node having the ability to manage an autoencoder, such as a second core network node. That is, a node capable of executing a deep learning model such as an autoencoder.

[0045] The fourth node 114 can be understood as a long short-term memory autoencoder or a node capable of managing a long short-term memory autoencoder, such as a third core network node. That is, a node capable of executing a deep learning model such as a long short-term memory autoencoder.

[0046] The second node 112 may be a radio network node in a radio access network (RAN) of the telecommunication network 105. The telecommunication network 105 may cover a geographical area, which in some embodiments may be divided into cell areas, wherein each cell area may be served by a radio network node, although one radio network node may serve one or several cells. Figure 1 In the example of b, the second node 112 serves the cell 120. The second node 112 may be, for example, a gNodeB. That is, a transmission point such as a radio base station (e.g., eNodeB) or a home node B, a home eNode B, or any other network node capable of serving a wireless device (such as a user equipment or a machine type node in the communication network 100). Based on the transmission power and thus also based on the cell size, the second node 112 may have different categories, such as, for example, a macro eNodeB, a home eNodeB, or a micro base station. In some examples, the radio network node may serve the receiving node with a service beam. The second node 112 may support one or several communication technologies, and its name may depend on the technology and terminology used. In a typical example of an embodiment herein, any reference to the second node 112 may be understood to refer to the second node 112 that manages the cell 120. This may be expressed as the second node 112 may refer to the cell 120, rather than to different cells that may be managed by the same radio network node.

[0047] The communication network 100 may include other nodes, such as the second node 112. In particular, the communication network 100 includes a first node group 121. The first node group 121 may be included in a node group 123 together with the second node group 122. Any node in the node group 123 may be understood as a corresponding computer system, respectively. Any node in the node group 123 may be understood as a corresponding network node, respectively. More particularly, any node in the node group 123 may be understood as a corresponding radio network node, respectively, as described above. In a typical example, in the case of beamforming transmission, any node in the node group 123 may be understood as a corresponding cell (similar to the cell 120) or a corresponding beam, respectively. In such an example, any node in the node group 123 may be a cell or a beam, and one or more nodes in the node group 123 may be managed by the same network node. In other examples herein, any node in the node group 123 may be understood as a corresponding non-radio network node, respectively.

[0048] Although in Figure 1, the node group 123 is shown as including five nodes, of which three nodes are included in the first node group 121, as indicated by the cells with solid lines, and two nodes are included in the second node group 122, as indicated by the cells with dashed lines, which can be understood as being for illustration purposes only. The number of nodes included in each of the node group 123, the first node group 121, and the second node group 122 may vary and is not limited to Figure 1 The specific examples depicted.

[0049] Any radio network node that may be included in the communication network 100 may be directly connected to one or more core networks.

[0050] In some examples, such as Figure 1 b), any one of the first node 111, the third node 113, and the fourth node 114 may be implemented as, for example, a standalone server in a host computer in the cloud 130. In some examples, any one of the first node 111, the second node 112, the third node 113, the fourth node 114, and the nodes in the node group 123 may be a distributed node or a distributed server, wherein some of their respective functions are implemented locally, for example, by a client manager, and some of their functions are implemented in the cloud 130, for example, by a server manager. However, in other examples, any one of the first node 111, the second node 112, the third node 113, the fourth node 114, and the nodes in the node group 123 may also be implemented as a processing resource in a server cluster. Any one of the first node 111, the third node 113, and the fourth node 114 may be under the ownership or control of a service provider, or may be operated by or on behalf of the service provider.

[0051] Any of the first node 111, the second node 112, the third node 113, the fourth node 114, and the nodes in the node group 123 may be enabled to communicate wirelessly in the communication network 100, and in some specific examples, may be able to support beamforming transmissions. The communication may be performed, for example, via a RAN and possibly one or more core networks included within the communication network 100.

[0052] Typically, the second node 112 and the nodes in the node group 123 may not exist in the communication network 100 at the same time. The nodes in the node group 123 may exist in the communication network 100 for a period of time, and the second node 112 may exist at a different later period of time. Alternatively, the second node 112 may be included in the node group 123.

[0053] The communication network 100 may include wireless devices. Figure 1b) a wireless device 140 is depicted. The wireless device 140 may also be referred to as, for example, a UE, a mobile terminal, a wireless terminal and / or a mobile station, a mobile phone, a cellular phone, or a laptop computer with wireless capabilities, or a user terminal equipment (CPE), to mention only some further examples. The wireless device 140 in this context may be, for example, a portable, pocket-storable, handheld, computer-included, or vehicle-mounted mobile device that enables voice and / or data to be communicated via the RAN with another entity such as a server, a laptop, a personal digital assistant (PDA), or a tablet computer (sometimes referred to as a tablet computer with wireless capabilities, or simply a tablet), a machine-to-machine (M2M) device, a device equipped with a wireless interface such as a printer or a file storage device, a modem, a laptop embedded device (LEE), a laptop mounted device (LME), a USB dongle, or any other radio network element capable of communicating over a radio link in the communication network 100. Any wireless device may be wireless, i.e., it may be enabled to communicate wirelessly in the communication network 100, and in some specific examples, may be capable of supporting beamforming transmissions. The communication may be performed, for example, between two devices, between a device and a radio network node and / or between a device and a server.The communication may be performed, for example, via the RAN and possibly one or more core networks respectively comprised within the communication network 100.

[0054] The first node 111 is configured to communicate with the third node 113 through a first link 151 (e.g., a radio link, an infrared link, or a wired link) within the communication network 100. The first node 111 is configured to communicate with the fourth node 114 through a second link 152 (e.g., a radio link, an infrared link, or a wired link) within the communication network 100. The second node 112 can communicate with the third node 113 through a third link 153 (e.g., a radio link, an infrared link, or a wired link). Any one of the nodes in the node group 123 can communicate with the third node 113 through a corresponding fourth link 154 (e.g., a radio link or an infrared link). The second node 112 can communicate with the wireless device 140 through a fifth link 155 (e.g., a radio link or an infrared link).

[0055] Any of the first link 151, the second link 152, the third link 153, the respective fourth link 154 and the fifth link 155 may be a direct link or include one or more links, for example via one or more other network nodes, a radio network node or a core network node.

[0056] Any one of the first link 151, the second link 152, the third link 153, the fourth link 154, and the fifth link 155 may be a direct link, or it may be connected via one or more computer systems or one or more core networks (in the communication network 100). Figure 1 The intermediate network may be one of a public, private, or host network or a combination of more than one of a public, private, or host network; if any, the intermediate network may be a backbone network or the Internet, which is Figure 1 Not shown.

[0057] Generally, the use of "first", "second", "third", "fourth", "fifth", etc. herein may be understood as any manner of referring to different elements or entities, and may be understood not to give a cumulative or chronological character to the nouns they modify.

[0058] Now refer to Figure 2 The depicted flowchart describes an embodiment of a method performed by the first node 111. The method may be understood as processing parameters for configuring the second node 112. The first node 111 may be understood as operating in a communication network 100. In some examples, the communication network 100 may be a fifth generation (5G) network.

[0059] The method may include the actions described below. In some embodiments, some actions may be performed. In some embodiments, all actions may be performed. Where applicable, one or more embodiments may be combined. All possible combinations are not described to simplify the description. It should be noted that the examples herein are not mutually exclusive. Components from one example may be assumed to exist in another embodiment by default, and it will be apparent to those skilled in the art how those components may be used in other examples. Figure 2 Optional actions are indicated by dashed boxes.

[0060] Action 201

[0061] The ultimate goal of the methods described herein may be understood to enable the communication network 100 to perform automatic correction of nodes (such as the second node 112) included in the communication network 100 that may perform suboptimally or poorly, and recommend appropriate parameter changes in such nodes so that they can improve their performance. The parameters discussed herein may be understood to refer to configuration management (CM) parameters. That is, any set of configurable parameters in a node (e.g., a cell) that may directly or indirectly affect the performance of the node (e.g., a cell).

[0062] In order to be able to identify which parameter changes may be suitable for improving the performance of the second node 112 or other similar nodes, the first node 111 may first determine which nodes may perform well according to the standard based on historical data from other nodes (e.g., node group 123). Then, the first node 111 can determine which parameters (i.e., for example, which configuration management (CM) parameters) may be responsible for the good performance of these nodes. Node group 123 can be considered as a node group whose historical performance data may have been collected in the communication network 100 for a period of time, during which period of time, node group 123 may have been recorded under multiple traffic conditions. In this way, the applicability of different parameter settings to different traffic conditions can also be analyzed. It can be noted that the nodes included in node group 123 may not always be constant. For example, historical data may be considered for analysis for a specific time period. In this time period, when node A may perform well, node A may be included in node group 123 for a certain time, and when it does not perform well, node A may not be in the group.

[0063] The performance of the node group 123 may be measured, for example, by evaluating a plurality of call events (e.g., silence, garbled characters, soft drop, hard drop, access failure, etc.). For each node in the node group 123, a data set may be recorded at regular intervals, the data set including call events, traffic conditions (such as the number of user connections, key performance indicators (KPIs) related to data transmission, etc.), and corresponding CM parameter settings for the node group 123 (e.g., a cell group). The recorded data set may be considered to have the following structure: Among them, the event data of node c at time t is given by is given by, and the flow and CM data are similarly given by and Given.

[0064] In the node group 123, for a given criterion (e.g., a first threshold), the nodes included in the first node group 121 can be considered as nodes with good performance (i.e., performance above the first threshold). The second node group 122 can be considered as nodes with normal or poor performance (i.e., performance below the first threshold). The first threshold can be a single value. For example, a single score can be calculated as a composite value based on multiple events that are a single indication of performance. Then, the first threshold can be selected as the value of the score. In a typical embodiment, the first threshold can be a value of a deviation or difference from what can be considered as a "normal" performance of a node. That is, the first threshold can typically be the value of a reconstruction error, as will be explained later. The technician will appreciate that the first threshold can take other forms.

[0065] It is also understood that the second node 112 itself may be included in the node group 123 as long as historical data for performing the analysis has been collected for the second node 112. That is, the historical data of the second node 112 may be used for the analysis so as to perform automatic correction of the parameter configuration of the second node 110 in a future time period. However, in a typical example, the second node 112 may not be included in the node group 123 for historical data analysis.

[0066] The analysis of the historical data may be performed by the third node 113, which may be included in the first node 111, or co-located with the first node 111, or may be an independent node, such as in Figure 1 As depicted in the non-limiting examples of.

[0067] According to the foregoing, in this action 201, the first node 111 may obtain a first indication from a third node 113 operating in the communication network 100. The first indication may indicate a first set of parameters observed to be associated with performance of the first node group 121 operating in the communication network 100 above a first threshold. In other words, in this action 201, the first indication may point to parameters corresponding to performance of a well-behaved node in the node group 123.

[0068] In some embodiments, where the third node 113 may be a node independent from the first node 111 , the obtaining in action 201 may be performed by, for example, receiving via the first link 151 .

[0069] In some embodiments, the third node 113 may be an autoencoder managed by the first node 111 .

[0070] In such an embodiment where the third node 113 may be an autoencoder managed by the first node 111 (e.g., co-located with or included in the first node 111), obtaining 201 the first parameter set may further include the following actions 202-205:

[0071] Action 202

[0072] In this action 202 , the first node 111 may train the first data model via the third node 113 with the corresponding data of the nodes included in the node group 123 , the corresponding data of the nodes indicating the corresponding performance of the nodes over a period of time. As mentioned above, the node group includes the first node group 121 .

[0073] The data model can be understood as a computer data model, for example, a machine learning model. The first data model can be, for example, an autoencoder model based on deep learning, etc.

[0074] The respective data of the nodes included in the node group 123 indicating the respective performance of the nodes over a period of time may be understood as data about one or more events respectively collected for the nodes included in the node group 123 .

[0075] Training the first data model may be understood as an autoencoder that may be able to learn the normal behavior of the node. Additionally, the first node 111 via the third node 113 may also consider how the node should behave given a specific event based on a laboratory deployment, or if possible when the node may have been deployed for the first time. This information may then be used as input in the training of the first data model in this action 202 in order to better understand any possible deviations from normal behavior.

[0076] The first data model may be understood as a mathematical function that best fits the normal performance of the nodes in the node group 123. All events may be input variables of the first data model.

[0077] By training the first data model in this action 202 with corresponding data of the nodes included in the node group 123, the first node 111 is enabled to predict what the expected performance may be for any given node, i.e., the expected value of each of the modeled events, and therefore, detect any performance that may deviate from the expected performance, as explained below.

[0078] Action 203

[0079] In this action 203, the first node 111 can obtain a reconstruction of the corresponding error of the trained first data model for each node in the node group 123 via the third node 113. The corresponding error can be understood as a reconstruction error of the autoencoder. The autoencoder can be understood as learning what should be a normal behavior (e.g., in terms of events) for a given set of events for a node, which normal behavior can provide a reconstruction error when fed to the autoencoder. The higher the reconstruction error, the higher the deviation from normal behavior. The autoencoder can be understood as directly outputting an error. Therefore, the corresponding error can be an indicator of a deviation from normal behavior.

[0080] By obtaining a reconstruction of the corresponding error of the trained first data model for each node in the node group 123 in this action 203, the first node 111 can determine which nodes of the node group 123 deviate from normal behavior and by how much, as explained below.

[0081] Action 204

[0082] In this action 204, the first node 111, via the third node 113, may rank or score the nodes included in the node group 123 based on their respective reconstruction errors. The ranking thus obtained may then be used to divide the collected data D into a tuple in which the nodes in the node group 123 may perform well and a tuple in which the nodes may perform poorly, i.e., D=D good ∪D bad The collected data D nodes may perform well, that is, D good It can be understood as indicating that the parameters as configured in these nodes can be set correctly for example for given traffic conditions. The reason behind this is that the normal behavior of the nodes in the node group 123 modeled by the third node 113 with higher reconstruction errors can be understood as being in order to better deviate from the normal behavior, i.e. the behavior with the best observed performance.

[0083] By sorting or scoring the nodes included in the node group 123 based on their corresponding reconstruction errors in this action 204, the first node 111 can determine which nodes in the node group 123 deviate the most from normal behavior, that is, which nodes in the node group 123 have the highest corresponding reconstruction errors.

[0084] Action 205

[0085] In this action 205 , the first node 111 , via the third node 113 , may select the first node group 121 from the node group 123 based on their respective reconstruction errors being above a first threshold.

[0086] By selecting the first node group 121 from the node group 123 in this action 205, the nodes with the best performance included in the node group 123 can be identified, which then enables their parameter settings to be evaluated so that the first parameter set mentioned in action 201 can be identified and suitable parameter configurations can be recommended to other nodes given similar circumstances (e.g. similar traffic conditions) to optimize, improve or correct their performance in the communication network 100, as will be explained later.

[0087] The obtained first parameter set may be used later in action 207 , in particular in the generation and identification in an analysis which will be explained in detail later in action 207 .

[0088] Action 206

[0089] As previously mentioned, the historical data collected for the node group 123, ie, the corresponding data of the nodes included in the node group 123 indicating the corresponding performance of the nodes over a period of time, may include traffic conditions associated with the observed performance data.

[0090] In this action 206, the first node 111 may obtain a second indication from the fourth node 114 operating in the communication network 100. The second indication may indicate a fixed-length vector encoding a plurality of traffic conditions observed to be associated with the performance of the first node group 121 (i.e., the node with the best performance in the node group 123) operating in the communication network 100. The obtained fixed-length vector may be used in action 207, in particular in the generation and identification in the analysis to be explained in detail later in action 207.

[0091] In some embodiments, where the fourth node 114 may be a node independent from the first node 111 , the obtaining in action 206 may be performed by receiving via the second link 152 , for example.

[0092] The traffic conditions of nodes (such as cells) can be understood to change over time. Therefore, the nature of these data can be understood to be continuous in nature, and therefore the data can be modeled using a long short-term memory (LSTM) encoder-decoder architecture similar to the architecture used for machine translation. The LSTM encoder (which uses the input as the traffic conditions) can output an encoding vector that can represent the traffic conditions observed over a period of time. The encoder vector can be understood to represent the traffic time series for the nodes included in the first node group 121. Therefore, in some embodiments, the fourth node 114 can manage the LSTM autoencoder.

[0093] In some embodiments, the fourth node 114 may be managed by the first node 111 and may manage the LSTM autoencoder.

[0094] In further specific embodiments, a plurality of traffic conditions may vary over time, and the fourth node 114 may be managed by the first node 111, and the LSTM autoencoder may be managed to generate a fixed length vector. In some such embodiments, the obtaining in the action 206 may include training a second data model with corresponding sequential traffic data of nodes included in the first node 121 group to encode the corresponding sequential traffic data into a fixed length vector.

[0095] The second data model can be an LSTM autoencoder model. Therefore, the LSTM autoencoder model can be used to encode the time-varying traffic into a fixed-length vector: ET t .

[0096] By obtaining the second indication in this action 206, the first node 111 may be enabled to determine one or more parameters for configuring the second node 112 for operation of delivering services in the communication network 100, and based on different traffic conditions, as explained in the next action 207. In other words, the first node 111 may be enabled to determine which parameters may be most suitable for configuring the second node 112 under given specific traffic conditions in order to obtain good performance.

[0097] Action 207

[0098] In this action 207, the first node 111 determines one or more parameters for configuring the second node 112 for operation of delivering a service in the communication network 100. A service may be understood as a function that may be used by a device operating in a communication network, such as radio coverage. In other words, in this action 207, the first node 111 applies to the second node 112 learned knowledge about the parameters that may be best suited to result in the best performance of the first node group 121, for example, for given similar conditions (such as traffic), as will be explained later. It is understood that the second node 112 is merely an illustrative example and the methods described herein may be applied to any node.

[0099] In this action 207, determining one or more parameters is based on an analysis performed by a generative adversarial network (GAN) managed by the first node 111. The analysis includes iteratively performing the following actions: i) generating a set of parameters estimated to be associated with the performance of the first node group 121 operating in the communication network 100 above a first threshold, and ii) distinguishing the generated parameter set from the first parameter set observed to be associated with the performance of the first node group 121 above the first threshold as obtained in action 201 to obtain a corresponding score for each parameter. The corresponding score indicates the degree to which the generated parameter set is different from the first parameter set. In other words, in this action 207, the first node 111 can provide a score about how close the generated parameter set can be to the data actually collected or recorded from the first node group 121 in action 201 via the discriminator learning.

[0100] Generating the parameter set in action i) may be understood as, for example, obtaining CM parameters for given traffic conditions, which may be necessary for good performance of the node. It may be understood that in action 201, real data may be obtained for training. Since there may be infinite possible scenarios, it is impossible for the real data to maintain all possible scenarios. The goal of generating the parameter set may be understood as "filling the gap". That is, for example, given the currently observed traffic conditions, CM parameters are recommended for improving the performance of a given node.

[0101] In some embodiments, the GAN may be a conditional GAN, that is, a GAN that learns to sample / generate points from a distribution conditioned on certain features or attributes.

[0102] In such embodiments where the GAN may be a conditional GAN, performing the analysis may further include adjusting based on a plurality of traffic conditions. In such embodiments, the determined one or more parameters may include a plurality of sets of one or more parameters. Each of the plurality of sets of one or more parameters may be determined for a different traffic condition in the plurality of traffic conditions. That is, for example, a first set of one or more parameters may be determined for when traffic is high, and a second set of one or more parameters may be determined for when traffic is low.

[0103] In some embodiments, the determination in action 207 may include a first stage and a second stage. The first stage may include performing generation and discrimination by: i. training a GAN with a first set of parameters observed to be associated with the performance of the first node group 121 above a first threshold and a plurality of traffic conditions to obtain a data model of parameters based on the plurality of traffic conditions, e.g., an overall model or a third data model. In other words, the first stage may include training a conditional GAN ​​for, for example, CM parameters of nodes that may have been identified as being within a particular time period (i.e., the data D in actions 201-205); good The above) performs well, and these parameters can be adjusted on multiple traffic conditions represented by the encoding vector obtained in action 206.

[0104] The second stage may include: ii) executing a data model, i.e., a third data model, by inputting the first observed set of traffic conditions as a first fixed-length vector, and iii) determining one or more parameters corresponding to the first observed set of traffic conditions based on the obtained data model, i.e., the third data model.

[0105] In other words, the first stage may include training the generator portion of the conditional GAN ​​to generate ideal CM parameters for given traffic conditions. The generator may be trained to learn P(C|ET), i.e., the distribution of CM parameters C observed over a period of time given some traffic conditions (which may have been encoded as ET).

[0106] During the prediction or second phase, the first node 111 may generate several parameter outputs from the generator for a given traffic condition until the discriminator can pass the generated parameters with an acceptable score. These parameters may then be recommended as settings for a given node (such as the second node 112) for a given traffic condition. Thus, the one or more parameters determined in this action 207 have corresponding scores generated by the discrimination that are below a second threshold. In other words, the one or more parameters determined are the parameters that are most similar to the observed parameters (i.e., the first parameter set).

[0107] To briefly summarize what may be required for action 207: first, a data model for predicting parameters based on the configuration parameters of the best performance node, such as under given specific traffic conditions, is established through iteration until the predicted data obtained using the data model cannot be distinguished from the real observed data. Then, the established model is used to output the parameters to be used under given real traffic conditions to obtain the best performance of the second node 112.

[0108] By determining one or more parameters for configuring the second node 112 in this action 207, the first node 111 is able to automatically correct the configuration of the second node 112 using the determined one or more parameters so that, for example, the performance of the second node 112 under given similar conditions (such as traffic) can be corrected to be as similar as possible to the performance observed in the first node group 121 (i.e., the best performing node in the node group 123).

[0109] Action 208

[0110] In this action 208, the first node 111 may output an indication, such as a third indication, comprising the determined one or more parameters. Outputting may be understood as providing or presenting internally, such as to another node within the first node 111, or by sending the third indication to another node located externally.

[0111] By outputting a third indication indicating the determined one or more parameters in this action 208, the first node 111 can enable an automated method to remotely tune the CM parameters of the second node 112 so that their configuration can best adapt to its changing traffic scenarios to prevent degradation of the performance of the second node 112 or even improve its performance.

[0112] Now refer to Figure 3 The depicted flowchart describes an embodiment of a method performed by the third node 113. The method may be understood as being for processing parameters configuring the second node 112. The third node 113 is an autoencoder operating in the communication network 100.

[0113] The method may include the actions described below. In some embodiments, some actions may be performed. In some embodiments, all actions may be performed. Where applicable, one or more embodiments may be combined. All possible combinations are not described to simplify the description. It should be noted that the embodiments herein are not mutually exclusive. Components from one example may be assumed to exist in another embodiment by default, and it will be apparent to those skilled in the art how those components may be used in other examples. Figure 3 Optional actions are indicated with dashed boxes. The detailed description of some of the following corresponds to the same reference numbers provided above with respect to the actions described for the first node 111 and will therefore not be repeated here. For example, the parameter may be a CM parameter.

[0114] The description of the following actions may be understood to correspond to actions 202-205, which are described as being performed by the first node 111 in an embodiment where the first node 111 manages the third node 113. Any further details of the method may be found in the description provided for actions 202-205.

[0115] Action 301

[0116] In this action 301, the third node 113 obtains a first set of parameters observed to be associated with performance of the first node group 121 operating in the communication network 100 above a first threshold. The first set of parameters are parameters configuring the second node 112 for operation of delivering services in the communication network 100.

[0117] Obtaining in this action 301 may be understood as determining, calculating or deriving.

[0118] Obtaining the first parameter set in this action 301 may further include the following actions 302-305:

[0119] Action 302

[0120] In this action 302, the third node 113 trains a first data model with corresponding data of nodes included in the node group 123, and the corresponding data of the node indicates the corresponding performance of the node over a period of time. As mentioned above, the node group includes the first node group 121. The first data model can be, for example, an autoencoder model based on deep learning.

[0121] By training the first data model with corresponding data of nodes included in the node group 123 in this action 302, the third node 113 can be enabled to predict or ultimately the first node can be enabled to predict what the expected performance may be for any given node, i.e., the expected value of each of the modeled events, and thereby detect any performance that may deviate from the expected performance, as explained below.

[0122] Action 303

[0123] In this action 303 , the third node 113 obtains, for each node in the node group 123 , a reconstruction of the corresponding error of the trained first data model.

[0124] By obtaining a reconstruction of the corresponding error of the trained first data model for each node in the node group 123 in this action 203, the third node 113 can determine which nodes in the node group 123 deviate from normal behavior and by how much, as explained below.

[0125] Action 304

[0126] In this action 304 , the first node 111 , via the third node 113 , ranks or scores the nodes included in the node group 123 based on their respective reconstruction errors.

[0127] By sorting or scoring the nodes included in the node group 123 based on their corresponding reconstruction errors in this action 304, the third node 113 can determine which nodes in the node group 123 deviate the most from normal behavior, that is, which nodes in the node group 123 have the highest corresponding reconstruction errors.

[0128] Action 305

[0129] In this action 305 , the third node 113 selects the first node group 121 from the node group 123 based on their respective reconstruction errors being above a first threshold.

[0130] By selecting the first node group 121 from the node group 123 in this action 205, the nodes with the best performance included in the node group 123 can be identified, which then enables their parameter settings to be evaluated so that the first parameter set mentioned in action 301 can be identified, and suitable parameter configurations given similar circumstances (e.g. similar traffic conditions) can be recommended to other nodes to optimize, improve or correct their performance in the communication network 100, as will be explained later.

[0131] Action 306

[0132] In this action 306, the third node 113 provides a first indication indicating the obtained first parameter set to the first node 111 operating in the communication network 100. In other words, in this action 306, the first indication may point to parameters corresponding to the performance of the well-behaved nodes in the node group 123.

[0133] The first indication may be, for example, a cell list, a message, etc.

[0134] In some embodiments, the first parameter set may include a first plurality of sets having a first one or more parameters, each set in the first plurality of sets having the first one or more parameters corresponding to a different flow condition in the plurality of flow conditions.

[0135] The first parameter set may then be used by the first node 111 as an input to the GAN to determine one or more parameters. Thus, by providing the first indication in this action 306, the third node 113 may enable the first node 111 to determine one or more parameters to configure the second node 112 for operation as previously explained in action 207 so that the second node 111 obtains good performance. In some embodiments, this determination may be enabled for different traffic conditions.

[0136] Now refer to Figure 4 The depicted flowchart describes an embodiment of a method performed by the fourth node 114. The method may be understood as being for processing parameters for configuring the second node 112. The fourth node 114 operates in the communication network 100.

[0137] The method includes the actions described below. Where applicable, one or more embodiments may be combined. All possible combinations are not described to simplify the description. It should be noted that the embodiments herein are not mutually exclusive. Components from an example may be assumed to exist in another embodiment by default, and it will be apparent to those skilled in the art how those components may be used in other examples.

[0138] , the detailed description of some of the following corresponds to the same reference numerals provided above with respect to the actions described for the first node 111, and thus will not be repeated here. For example, the parameter may be a CM parameter.

[0139] The description of the following actions may be understood to correspond to action 206, which in an embodiment where the first node 111 manages the fourth node 114 is described as being performed by the first node 111. Any further details of the method may be found in the description provided for action 206.

[0140] Action 401

[0141] As previously mentioned, historical data collected for node group 123 (ie, corresponding data of nodes included in node group 123 indicating corresponding performance of the nodes over a period of time) may include traffic conditions associated with observed performance data.

[0142] In this action 206, the fourth node 114 obtains a fixed-length vector encoding a plurality of traffic conditions observed to be associated with the performance of the first node group 121 (i.e., the node with the best performance in the node group 123) operating in the communication network 100. The obtained fixed-length vector can be used in the generation and identification in the analysis, as previously described.

[0143] The obtaining in this action 401 includes training a second data model with corresponding sequential traffic data of nodes included in the first node group 121 to encode the corresponding sequential traffic data into a fixed-length vector.

[0144] In some embodiments, a plurality of traffic conditions may vary over time, and the fourth node 114 may manage an LSTM autoencoder to obtain a fixed-length vector.

[0145] The second data model can be an LSTM autoencoder model. Therefore, the LSTM autoencoder model can be used to encode the time-varying traffic into a fixed-length vector: ET t .

[0146] The fixed length vector may then be used by the first node 111 as an input to the GAN to determine one or more parameters. Obtaining the fixed length vector in this action 401 may enable the fourth node 114 to provide a second indication in the next action 402 and enable the first node 111 to determine one or more parameters for configuring the second node 112 for operation of delivering services in the communication network 100 as explained in action 207, and based on different traffic conditions. In other words, the fourth node 114 may enable the first node 111 to determine which parameters may be most suitable for configuring the second node 112 under given specific traffic conditions in order to obtain good performance.

[0147] Action 402

[0148] In this action 402, the fourth node 114 provides a second indication to the first node 111 operating in the communication network 100. The second indication indicates the obtained fixed length vector encoding the plurality of traffic conditions.

[0149] By providing the second indication in this action 402, the fourth node 114 enables the first node 111 to determine one or more parameters for configuring the second node 112 for operation of delivering services in the communication network 100, as explained in action 207, and based on different traffic conditions. In other words, the fourth node 114 may enable the first node 111 to determine which parameters may be most suitable for configuring the second node 112 given certain traffic conditions in order to obtain good performance.

[0150] Figure 5 is a schematic diagram depicting a non-limiting example of how the first node 111 , the third node 113 , and the fourth node 114 interact in the communication network 100 to perform the method described above.

[0151] The method may start with a data set as previously described, which includes call events, traffic conditions, and parameter settings for a node group 123 (eg, a cell set) recorded at regular intervals. Among them, the event data of cell c at time t is given by The flow and parameter data are similarly given by and Given. In step 1, the third node 113 (autoencoder) trains a first computer model in action 302, obtains a reconstruction of the corresponding error in action 303, and uses the reconstruction error as a first threshold in action 304 to sort the nodes (here are cells) and identify the cells that perform well, selecting these cells in action 305. Then, according to action 301, it obtains a first parameter set. That is, configuration management data of nodes with ideal CM parameter settings for given traffic conditions that result in good performance and / or ranking. In action 306, the third node 113 provides a first indication to the first node 111, and the first node 111 uses the first indication as an input to the discriminator according to action 201. In step 2, the fourth node 114 manages the LSTM autoencoder according to action 401 to obtain a fixed length vector that encodes multiple traffic conditions that vary over time as ET. Then, according to action 402, the fourth node 114 provides a second indication to the first node 111 as an input to the discriminator. In step 3, the first node 111 determines one or more parameters by training the GAN with a first parameter set observed to be associated with the performance of the first node group 121 above a first threshold and a plurality of traffic conditions to obtain a data model of parameters based on the plurality of traffic conditions. In other words, in this example, the first node 111 can be understood as a data model of parameters based on the plurality of traffic conditions within a specific time period (i.e., the data D in step 1). good(above) A conditional GAN ​​is trained for the CM parameters of cells identified as being well-performing. The GAN may be adjusted on the traffic conditions represented by the encoded vectors obtained in step 2. At step 4, the generator portion of the conditional GAN ​​is trained according to action 207 to generate ideal parameters for given traffic conditions. The generator is trained to learn P(C|ET), i.e., the distribution of CM parameters C observed over a period of time given some traffic conditions (which have been encoded as ET). At step 5, according to action 207, the discriminator portion of the conditional GAN ​​learns to provide a score on how close the generated CM parameters are to those observed in the training data. During the prediction phase, the first node 111 generates several CM parameter outputs from the generator for a given traffic situation until the discriminator passes it with an acceptable score. Then, according to action 208, these CM parameters are recommended to be set for a given cell for a given traffic situation.

[0152] As a general overview, the embodiments herein may be understood as providing a machine learning driven method for automatically detecting poorly performing nodes (e.g., cells) on a daily level, and enabling any necessary corrections to be performed on the CM parameters in the node to improve its performance. The second node 112 is used herein as an example of such a node. However, it will be appreciated that the same node may perform poorly during a particular time period and perform well during a different time period, for example based on circumstances such as traffic. The embodiments herein may be understood as enabling a mechanism to monitor key performance metrics of nodes, and to identify them when the performance of these nodes may be degraded by studying statistical parameters of events recorded from each node. In other words, the embodiments herein may be understood as providing a method for better diagnosing potential poorly performing nodes in advance based on historical trends of events obtained from the nodes, and estimating which CM parameters may change performance, and then taking any necessary actions so that degradation of the performance of the node can be prevented.

[0153] One advantage of embodiments herein is that ranking nodes (eg, cells) based on recent historical performance enables identification of poorly performing cells in an automated manner, allowing instantaneous necessary corrections to be performed to improve their performance.

[0154] Embodiments herein enable identification of CM parameters that may be responsible for poor performance of a node, and enable automatic correction of the CM parameters to restore the performance of the node to its optimal level.

[0155] Another advantage of the embodiments herein is that in the automatic correction method, the OPEX cost of the site may be reduced or minimized.

[0156] Another advantage of embodiments herein is that they may enable providing a better Quality of Experience (QoE) to users.

[0157] Yet another advantage of embodiments herein is that they may enable ensuring that the customer experience is not severely impacted.

[0158] Furthermore, as an added advantage, embodiments herein may be understood to enable fewer visits to sites to be performed, as remote monitoring may be understood to enable a significant amount of productive time of field engineers to be saved.

[0159] Figure 6 In panels a) and b), respectively, the first node 111 may include a method for performing the above Figure 2 and / or Figure 5 Two different examples of arrangements of the described method actions. The first node 111 is for processing parameters for configuring the second node 112. The first node 111 is configured to operate in the communication network 100.

[0160] Some of the detailed descriptions below correspond to the same reference numbers provided above with respect to the actions described for the first network node 111 and will therefore not be repeated here. For example, the parameter may be a CM parameter. Figure 6 Optional units are indicated by dashed lines.

[0161] In some embodiments, the first node 111 may include Figure 6 a The following arrangement depicted.

[0162] The first node 111 is configured to determine one or more parameters for configuring the operation of the second node 112 for delivering a service in the communication network 100, for example by means of a determination unit 601 within the first node 111. The determination is configured to be based on an analysis configured to be performed by a GAN, which is configured to be managed by the first node 111. The analysis is configured to include iteratively performing: i) generating a set of parameters estimated to be associated with a performance of the first node group 121 configured to operate in the communication network 100 above a first threshold, and ii) distinguishing the set of parameters configured to be generated and the first set of parameters observed to be associated with a performance of the first node group 121 above the first threshold, to obtain a corresponding score for each parameter. The corresponding score is configured to indicate the degree to which the generated set of parameters differs from the first set of parameters. The one or more parameters configured to be determined are configured to have a corresponding score resulting from the distinguishing that is below a second threshold.

[0163] The first node 111 is further configured, for example by means of an output unit 602 within the first node 111 , which is configured to output an indication comprising one or more parameters configured to be determined.

[0164] In some embodiments, the GAN may be configured as a conditional GAN. In such embodiments, performing the analysis may be further configured to include adjusting based on a plurality of traffic conditions. The one or more parameters configured to be determined may be configured to include a plurality of sets of one or more parameters. Each of the plurality of sets of one or more parameters may be configured to be determined for a different traffic condition in the plurality of traffic conditions.

[0165] In some embodiments, the determination may be configured to include a first stage and a second stage. The first stage may be configured to include performing generation and discrimination by: i) training a GAN. The training of the GAN is configured to utilize a first parameter set and a plurality of traffic conditions configured to be observed to be associated with the performance of the first node group 121 above a first threshold. This is to obtain a data model of parameters based on a plurality of traffic conditions. The second stage may be configured to include: ii) executing a data model by inputting a first observed set of traffic conditions as a first fixed-length vector, and iii) determining one or more parameters corresponding to the first observed set of traffic conditions based on the data model configured to be obtained.

[0166] In some embodiments, the third node 113 may also be configured to obtain a first indication from the third node 113 configured to operate in the communication network 100, for example by means of an obtaining unit 603 within the first node 111. The first indication is configured to indicate a first parameter set configured to be observed to be associated with the performance of the first node group 121 above a first threshold. The first parameter set configured to be obtained is configured to be used in generating and distinguishing in the analysis.

[0167] In some embodiments in which the third node 113 may be an autoencoder configured to be managed by the first node 111, the obtaining of the first parameter set may also be configured to include the following. i) Training, for example, by means of a training unit 604 within the first node 111, the training unit 604 being configured to train a first data model with corresponding data of nodes configured to be included in the node group 123, the corresponding data of the nodes indicating the corresponding performance of the nodes over a period of time. The node group 123 is configured to include the first node group 121. ii) Obtaining, for example, by means of an obtaining unit 603 within the first node 111, the obtaining unit 603 being configured to obtain, for each node in the node group 123, a reconstruction of the corresponding error of the first data model configured to be trained. ii) Sorting, for example, by means of a sorting unit 605 within the first node 111, the sorting unit 605 being configured to sort the nodes configured to be included in the node group 123 based on their corresponding reconstruction errors. Furthermore, iv) selection is performed, for example by means of a selection unit 605 within the first node 111, which is configured to select the first node group 121 from the node group 123 based on their respective errors configured to be reconstructed being above a first threshold.

[0168] In some embodiments, the first node 111 may also be configured to obtain a second indication from a fourth node 114 configured to operate in the communication network 100, for example by means of an obtaining unit 603 within the first node 111. The second indication is configured to indicate a fixed-length vector encoding a plurality of traffic conditions configured to be observed to be associated with the performance of the first node group 121 configured to operate in the communication network 100. The fixed-length vector configured to be obtained may be configured to be used in generation and identification in the analysis.

[0169] In some embodiments, a plurality of traffic conditions may vary over time. In some such embodiments, the fourth node 114 may be configured to be managed by the first node 111 and may be configured to manage an LSTM autoencoder to generate a fixed length vector. Obtaining may be configured to include training a second data model with corresponding sequential traffic data of nodes configured to be included in the first node group 121 to encode the corresponding sequential traffic data into a fixed length vector.

[0170] The embodiments herein may be implemented by one or more processors (such as Figure 6The processor 607 in the first node 111 depicted in a) is implemented together with the computer program code for performing the functions and actions of the embodiments herein. The processor 607 can be understood as a hardware component in this article. The program code mentioned above can also be provided as a computer program product, for example in the form of a data carrier carrying the computer program code for performing the embodiments herein when loaded into the first node 111. However, it is feasible to utilize other data carriers (such as memory sticks). In addition, the computer program code can be provided as a pure program code on a server and downloaded to the first node 111.

[0171] The first node 111 may further comprise a memory 608 comprising one or more memory units. The memory 608 is arranged to store the obtained information, store data, configurations, schedules and applications etc. to perform the methods herein when executed in the first node 111.

[0172] In some embodiments, the first node 111 may receive information from, for example, the second node 112, the third node 113, the fourth node 114, and / or the node group 123 via the receiving port 609. In some examples, the receiving port 609 may, for example, be connected to one or more antennas in the first node 111. In other embodiments, the first node 111 may receive information from another structure in the communication network 100 via the receiving port 609. Since the receiving port 609 may communicate with the processor 607, the receiving port 609 may send the received information to the processor 607. The receiving port 609 may also be configured to receive other information.

[0173] The processor 607 in the first node 111 may also be configured to transmit or send information to, for example, the second node 112 , the third node 113 , the fourth node 114 and / or the node group 123 via a transmission port 610 , which may be in communication with the processor 607 and the memory 608 .

[0174] about Figure 6 Any of the described units 601-606 may be a processor 607 of the first node 111, or an application running on such a processor.

[0175] Those skilled in the art will also understand that Figure 6Any of the described units 601-606 may refer to a combination of analog and digital circuits, and / or one or more processors configured with software and / or firmware stored, for example, in a memory, which when executed by one or more processors, such as processor 607, performs as described above. One or more of these processors and other digital hardware may be included in a single application specific integrated circuit (ASIC), or multiple processors and various digital hardware may be distributed among multiple separate components, whether individually packaged or assembled into a system on a chip (SoC).

[0176] Therefore, the methods according to the embodiments described herein for the first node 111 can be implemented respectively by means of a computer program 611 product, which includes instructions, i.e., software code portions, which when executed on at least one processor 607 cause at least one processor 607 to perform actions described herein as being performed by the first node 111. The computer program 611 product can be stored on a computer-readable storage medium 612. The computer-readable storage medium 612 on which the computer program 611 is stored may include instructions, which when executed on at least one processor 607 cause at least one processor 607 to perform actions described herein as being performed by the first node 111. In some embodiments, the computer-readable storage medium 612 may be a non-transitory computer-readable storage medium (such as a memory stick), or stored in a cloud space. In other embodiments, the computer program 611 product may be stored on a carrier containing a computer program, wherein the carrier is one of the following: an electronic signal, an optical signal, a radio signal, or a computer-readable storage medium 612, as described above.

[0177] The first node 111 may include an interface unit that facilitates communication between the first node 111 and other nodes or devices (e.g., the second node 112, the third node 113, the fourth node 114, and / or the node group 123). In some specific examples, the interface may, for example, include a transceiver configured to send and receive radio signals over an air interface according to a suitable standard.

[0178] In other embodiments, the first node 111 may include Figure 6 b. The first node 111 may include processing circuitry 607 (e.g., one or more processors, such as processor 607 in the first node 111) and memory 608. The first node 111 may also include radio circuitry 613, which may include, for example, a receive port 609 and a transmit port 610. The processing circuitry 607 may be configured or operable to communicate with the user about the Figure 6 A similar approach is performed according to Figure 2 and / or Figure 5 The radio circuit 613 may be configured to establish and maintain a wireless connection with at least, for example, the second node 112, the third node 113, the fourth node 114 and / or the node group 123. A circuit may be understood herein as a hardware component.

[0179] Therefore, embodiments herein also relate to a first node 111 operable to operate in a communication network 100. The first node 111 may include a processing circuit 607 and a memory 608, the memory 608 containing instructions executable by the processing circuit 607, whereby the first node 111 may also be operable to perform the instructions described herein with respect to the first node 111, for example, in Figure 2 and / or Figure 5 The actions described in .

[0180] Figure 7 In panels a) and b) respectively, the third node 113 may include a method for performing the above Figure 3 and / or Figure 5 Two different examples of arrangements of the described method actions. The third node 113 is for processing parameters for configuring the second node 112. The third node 113 is configured as an autoencoder configured to operate in the communication network 100.

[0181] Some of the detailed descriptions below correspond to the same reference numbers provided above with respect to the actions described for the first node 111, and therefore will not be repeated here. For example, the parameter may be a CM parameter. Figure 7 Optional units are indicated by dashed lines.

[0182] In some embodiments, the third node 113 may include Figure 7 a The following arrangement depicted.

[0183] The third node 113 is configured to obtain, for example, by means of an obtaining unit 701 within the third node 113, the obtaining unit 701 being configured to obtain a first set of parameters configured to be observed to be associated with performance of the group of first nodes 121 configured to operate in the communication network 100 above a first threshold. The first set of parameters is configured to configure the second node 112 for operation of delivering services in the communication network 100. The obtaining is configured to include the following. i) training, for example, by means of a training unit 702 within the first node 113, the training unit 702 being configured to train a first data model using corresponding data of nodes configured to be included in the node group 123, the corresponding data of the nodes indicating the corresponding performance of the node over a period of time. The node group 123 is configured to include the group of first nodes 121. ii) obtaining, for example, by means of an obtaining unit 701 within the first node 111, the obtaining unit 701 being configured to obtain a reconstruction of a corresponding error of the first data model for each node in the node group 123. ii) sorting, for example, by means of a sorting unit 703 in the first node 111, the sorting unit 703 being configured to sort the nodes configured to be included in the node group 123 based on their corresponding reconstruction errors. In addition, iv) selecting, for example, by means of a selection unit 704 in the first node 111, the selection unit 704 being configured to select the first node group 121 from the node group 123 based on their corresponding errors configured to be reconstructed being higher than a first threshold.

[0184] The third node 113 is further configured to provide, for example by means of a providing unit 705 within the third node 114, a first indication configured to indicate a first parameter set configured to be obtained to the first node 111 configured to operate in the communication network 100.

[0185] In some embodiments, the first parameter set can be configured to include a first plurality of sets of first one or more parameters. Each set of the first plurality of sets of one or more parameters can be configured to correspond to a different flow condition of the plurality of flow conditions.

[0186] The embodiments herein may be implemented by one or more processors (such as Figure 7The processor 706 in the third node 113 depicted in a) is implemented together with the computer program code for executing the functions and actions of the embodiments herein. The processor 706 can be understood as a hardware component in this article. The program code mentioned above can also be provided as a computer program product, for example in the form of a data carrier carrying the computer program code for executing the embodiments herein when loaded into the third node 113. However, it is feasible to use other data carriers (such as memory sticks). In addition, the computer program code can be provided as a pure program code on a server and downloaded to the third node 113.

[0187] The third node 113 may further comprise a memory 707 comprising one or more memory units. The memory 707 is arranged to store the obtained information, store data, configurations, schedules and applications etc. to perform the methods herein when executed in the third node 113.

[0188] In some embodiments, the third node 113 may receive information from, for example, the first node 111, the second node 112, the fourth node 114, and / or the node group 123 via the receiving port 708. In some examples, the receiving port 708 may be connected to, for example, one or more antennas in the third node 113. In other embodiments, the third node 113 may receive information from another structure in the communication network 100 via the receiving port 708. Since the receiving port 708 may communicate with the processor 706, the receiving port 708 may send the received information to the processor 706. The receiving port 708 may also be configured to receive other information.

[0189] The processor 706 in the third node 113 may also be configured to transmit or send information to, for example, the first node 111 , the second node 112 , the fourth node 114 and / or the node group 123 via a transmission port 709 , which may communicate with the processor 706 and the memory 707 .

[0190] about Figure 7 Any of the described units 701-705 may be a processor 706 of the third node 113, or an application running on such a processor.

[0191] Those skilled in the art will also understand that Figure 7Any of the described units 701-705 may refer to a combination of analog and digital circuits, and / or one or more processors configured with software and / or firmware stored, for example, in a memory, which when executed by one or more processors, such as processor 706, performs as described above. One or more of these processors and other digital hardware may be included in a single application specific integrated circuit (ASIC), or multiple processors and various digital hardware may be distributed among multiple separate components, whether packaged separately or assembled into a system on a chip (SoC).

[0192] Therefore, the methods according to the embodiments described herein for the third node 113 can be implemented respectively by means of a computer program 710 product, which includes instructions, i.e., software code portions, which when executed on at least one processor 706 cause at least one processor 706 to perform actions described herein as being performed by the third node 113. The computer program 710 product may be stored on a computer-readable storage medium 711. The computer-readable storage medium 711 on which the computer program 710 is stored may include instructions, which when executed on at least one processor 706 cause at least one processor 706 to perform actions described herein as being performed by the third node 113. In some embodiments, the computer-readable storage medium 711 may be a non-transitory computer-readable storage medium (such as a memory stick), or stored in a cloud space. In other embodiments, the computer program 710 product may be stored on a carrier containing a computer program, wherein the carrier is one of the following: an electronic signal, an optical signal, a radio signal, or a computer-readable storage medium 711, as described above.

[0193] The third node 113 may include an interface unit that facilitates communication between the third node 113 and other nodes or devices (e.g., the first node 111 and / or the third node 113). In some specific examples, the interface may, for example, include a transceiver configured to send and receive radio signals over an air interface according to a suitable standard.

[0194] In other embodiments, the third node 113 may include Figure 7 b. The third node 113 may include processing circuitry 706 (e.g., one or more processors, such as processor 706 in the third node 113) and memory 707. The third node 113 may also include radio circuitry 712, which may include, for example, a receive port 708 and a transmit port 709. The processing circuitry 706 may be configured or operable to employ a method related to Figure 7 A similar approach as described above is performed according to Figure 3 and / or Figure 5 The radio circuit 712 may be configured to establish and maintain a wireless connection with at least, for example, the first node 111, the second node 112, the fourth node 114 and / or the node group 123. A circuit may be understood herein as a hardware component.

[0195] Therefore, the embodiments herein also relate to a third node 113 operable to operate in the communication network 100. The third node 113 may include a processing circuit 706 and a memory 707, wherein the memory 707 contains instructions executable by the processing circuit 706, whereby the third node 113 may also be operable to perform the operations described herein, for example, Figure 3 and / or Figure 5 The actions described with respect to the third node 113 in .

[0196] Figure 8 In panels a) and b), respectively, the fourth node 114 may include a method for performing the above-mentioned Figure 4 and / or Figure 5 Two different examples of arrangements of the described method actions. The fourth node 114 is for processing parameters for configuring the second node 112. The fourth node 114 is configured to operate in the communication network 100.

[0197] , some of the detailed descriptions below correspond to the same reference numbers provided above with respect to the actions described for the first node 111, and therefore will not be repeated here. For example, the parameter may be a CM parameter. Figure 8 Optional units are indicated by dashed lines.

[0198] In some embodiments, the fourth node 114 may include Figure 8 a The following arrangement depicted.

[0199] The fourth node 114 is configured to obtain, for example, by means of a receiving unit 801 within the fourth node 114, a fixed-length vector encoding a plurality of traffic conditions configured to be observed to be associated with the performance of the first node group 121 configured to operate in the communication network 100. The obtaining is configured to include training a second data model with corresponding sequential traffic data of nodes configured to be included in the first node group 121 to encode the corresponding sequential traffic data into a fixed-length vector.

[0200] The fourth node 114 is configured to provide, for example by means of a providing unit 802 within the fourth node 114, a second indication to the first node 111 operating in the communication network, the second indication indicating a fixed length vector configured to be obtained encoding a plurality of traffic conditions.

[0201] In some embodiments, the plurality of traffic conditions may vary over time.In some such embodiments, the fourth node 114 may be configured to manage an LSTM autoencoder to obtain a fixed length vector.

[0202] The embodiments herein may be implemented by one or more processors (such as Figure 8 The processor 803 in the fourth node 114 depicted in a) is implemented together with the computer program code for executing the functions and actions of the embodiments herein. The processor 803 can be understood as a hardware component in this article. The program code mentioned above can also be provided as a computer program product, for example in the form of a data carrier carrying the computer program code for executing the embodiments herein when loaded into the fourth node 114. However, it is feasible to utilize other data carriers (such as memory sticks). In addition, the computer program code can be provided as a pure program code on a server and downloaded to the fourth node 114.

[0203] The fourth node 114 may further include a memory 804 including one or more memory units. The memory 804 is arranged to store the obtained information, store data, configurations, schedules and applications, etc. to perform the methods herein when executed in the fourth node 114.

[0204] In some embodiments, the fourth node 114 may receive information from, for example, the first node 111, the second node 112, the third node 113, and / or the node group 123 via the receiving port 805. In some examples, the receiving port 805 may, for example, be connected to one or more antennas in the fourth node 114. In other embodiments, the fourth node 114 may receive information from another structure in the communication network 100 via the receiving port 805. Since the receiving port 805 may communicate with the processor 803, the receiving port 805 may send the received information to the processor 803. The receiving port 805 may also be configured to receive other information.

[0205] The processor 803 in the fourth node 114 may also be configured to transmit or send information to, for example, the first node 111 , the second node 112 , the third node 113 and / or the node group 123 via a transmission port 806 , which may communicate with the processor 803 and the memory 804 .

[0206] about Figure 8 Any of the described units 801-802 may be a processor 803 of the fourth node 114, or an application running on such a processor.

[0207] Those skilled in the art will also understand that Figure 8The described units 801-802 may refer to a combination of analog and digital circuits, and / or one or more processors configured with software and / or firmware stored, for example, in a memory, which when executed by one or more processors, such as processor 803, performs as described above. One or more of these processors and other digital hardware may be included in a single application specific integrated circuit (ASIC), or multiple processors and various digital hardware may be distributed among multiple separate components, whether packaged separately or assembled into a system on a chip (SoC).

[0208] Therefore, the methods according to the embodiments described herein for the fourth node 114 can be implemented respectively by means of a computer program 807 product, which includes instructions, i.e., software code portions, which, when executed on at least one processor 803, cause at least one processor 803 to perform actions described herein as being performed by the fourth node 114. The computer program 807 product can be stored on a computer-readable storage medium 808. The computer-readable storage medium 808 on which the computer program 807 is stored may include instructions, which, when executed on at least one processor 803, cause at least one processor 803 to perform actions described herein as being performed by the fourth node 114. In some embodiments, the computer-readable storage medium 808 may be a non-transitory computer-readable storage medium (such as a memory stick), or stored in a cloud space. In other embodiments, the computer program 807 product may be stored on a carrier containing a computer program, wherein the carrier is one of the following: an electronic signal, an optical signal, a radio signal, or a computer-readable storage medium 808, as described above.

[0209] The fourth node 114 may include an interface unit that facilitates communication between the fourth node 114 and other nodes or devices (e.g., the first node 111, the second node 112, the third node 113, and / or the node group 123). In some specific examples, the interface may, for example, include a transceiver configured to send and receive radio signals over an air interface according to a suitable standard.

[0210] In other embodiments, the fourth node 114 may include Figure 8 b. The fourth node 114 may include a processing circuit 803 (e.g., one or more processors, such as the processor 803 in the fourth node 114) and a memory 804. The fourth node 114 may also include a radio circuit 809, which may include, for example, a receiving port 805 and a transmitting port 806. The processing circuit 803 may be configured or operable to employ a method related to Figure 8 A similar approach as described above is performed according to Figure 4 and / or Figure 5 The radio circuit 809 may be configured to establish and maintain a wireless connection with at least, for example, the first node 111, the second node 112, the third node 113 and / or the node group 123. A circuit may be understood herein as a hardware component.

[0211] Therefore, the embodiments herein also relate to a third node 113 operable to operate in the communication network 100. The third node 113 may include a processing circuit 803 and a memory 804, wherein the memory 804 contains instructions executable by the processing circuit 803, whereby the third node 113 may also be operable to perform the instructions described herein, for example, Figure 4 and / or Figure 5 The actions described with respect to the third node 113 in .

[0212] When the word “includes” or “comprising” is used, it should be interpreted as non-limiting, meaning “consisting at least of…”

[0213] The embodiments herein are not limited to the preferred embodiments described above. Various alternatives, modifications and equivalents may be used. Therefore, the above embodiments should not be understood to limit the scope of the present invention.

[0214] As used herein, the expression "in some embodiments" has been used to indicate that features of a described embodiment can be combined with any other embodiments or examples disclosed herein.

[0215] As used herein, the expression "in some examples" has been used to indicate that features of the described examples can be combined with any other embodiments or examples disclosed herein.

[0216] As used herein, the expression "at least one of:" followed by a list of alternatives separated by commas and wherein the last alternative precedes an "and" term may be understood to mean that only one of the list of alternatives may be applicable, more than one of the list of alternatives may be applicable, or all of the list of alternatives may be applicable. This expression may be understood to be equivalent to the expression "at least one of:" followed by a list of alternatives separated by commas and wherein the last alternative precedes an "or" term.

Claims

1. A method performed by a first node (111), the method for processing parameters for configuring a second node (112), the first node (111) operating in a communication network (100), the method comprising: - determining (207) one or more parameters for configuring the second node (112) for operation of delivering a service in the communication network (100), the determination (207) being based on an analysis performed by a generative adversarial network (GAN) managed by the first node (111), the analysis comprising iteratively performing: i. generating a set of parameters estimated to be associated with the performance of a first group of nodes (121) operating in said communication network (100) above a first threshold, and ii. distinguishing the generated parameter set from a first parameter set observed to be associated with the performance of the first node group (121) above the first threshold to obtain a respective score for each parameter, wherein the respective score indicates the degree to which the generated parameter set differs from the first parameter set, and wherein the determined one or more parameters have a corresponding score resulting from the discrimination that is below a second threshold, and - output (208) includes an indication of the determined one or more parameters; wherein the GAN is a conditional GAN, and wherein performing the analysis further comprises adjusting based on a plurality of flow conditions, and wherein the determined one or more parameters comprises a plurality of sets of one or more parameters, each set of the plurality of sets of one or more parameters being determined for a different flow condition of the plurality of flow conditions; Also includes: - obtaining (201) a first indication from a third node (113) operating in said communication network (100), the first indication being indicative of said first set of parameters observed to be associated with said performance of said first group of nodes (121) being above said first threshold, and wherein said obtained first set of parameters is used for said generating and said discriminating in said analyzing; - obtaining (206) a second indication from a fourth node (114) operating in the communication network (100), the second indication indicating a fixed-length vector encoding the plurality of traffic conditions observed to be associated with the performance of the first group of nodes (121) operating in the communication network (100), and wherein the obtained fixed-length vector is used for the generation and the identification in the analysis.

2. The method according to claim 1, wherein: The determining (207) includes a first stage and a second stage, wherein the first stage includes performing the generating and the distinguishing by: i. training the GAN with the first set of parameters observed to be associated with the performance of the first group of nodes (121) above the first threshold and the plurality of traffic conditions to obtain a data model of the parameters based on the plurality of traffic conditions, And wherein, the second stage includes: ii. executing the data model by inputting a first observed set of flow conditions as a first fixed length vector, and iii. Based on the obtained data model, determine the one or more parameters corresponding to the first observed set of flow conditions.

3. The method according to claim 1, wherein: The third node (113) is an autoencoder managed by the first node (111), and wherein obtaining (201) the first set of parameters further comprises: training (202) a first data model using corresponding data of nodes included in a node group (123), the corresponding data of the nodes indicating corresponding performance of the nodes over a period of time, the node group (123) including the first node group (121), obtaining (203) for each node in the set of nodes (123) a reconstruction of the corresponding error of the trained first data model, sorting (204) the nodes included in the node group (123) based on their respective reconstruction errors, and The first group of nodes (121) is selected (205) from the group of nodes (123) based on their respective reconstruction errors being above the first threshold.

4. The method according to claim 1, wherein: The multiple traffic conditions change over time, and wherein the fourth node (114) is managed by the first node (111), and the fourth node (114) manages a long short-term memory autoencoder to generate the fixed-length vector, and wherein the obtaining (206) includes: training a second data model with corresponding sequential traffic data of the nodes included in the first node group (121) to encode the corresponding sequential traffic data into the fixed-length vector.

5. A method performed by a third node (113) for processing parameters for configuring a second node (112), the third node (113) being an autoencoder operating in a communication network (100), the method comprising: - obtaining (301) a first set of parameters observed to be associated with a performance of a first set of nodes (121) operating in the communication network (100) above a first threshold, the first set of parameters being parameters for configuring the operation of a second node (112) for delivering a service in the communication network (100), the obtaining (301) comprising: training (302) a first data model using corresponding data of nodes included in a node group (123), the corresponding data of the nodes indicating corresponding performance of the nodes over a period of time, the node group (123) including the first node group (121), obtaining (303) for each node in the set of nodes (123) a reconstruction of the corresponding error of the trained first data model, sorting (304) the nodes included in the node group (123) based on their respective reconstruction errors, and selecting (305) the first group of nodes (121) from the group of nodes (123) based on their respective reconstruction errors being above the first threshold, and - providing (306) a first indication indicating the obtained first set of parameters to a first node (111) operating in the communication network (100); Wherein, the first node (111) executes the method according to claim 1.

6. The method according to claim 5, wherein: The first parameter set includes a first plurality of sets of first one or more parameters, each set of the first plurality of sets of first one or more parameters corresponding to a different flow condition of a plurality of flow conditions.

7. A method performed by a fourth node (114), the method for processing parameters for configuring a second node (112), the fourth node (114) operating in a communication network (100), the method comprising: - obtaining (401) a fixed length vector encoding a plurality of traffic conditions observed to be associated with performance of a first group of nodes (121) operating in the communication network (100), the obtaining (401) comprising: training a second data model with corresponding sequential traffic data of nodes included in the first group of nodes (121) to encode the corresponding sequential traffic data into the fixed length vector, and - providing (402) to a first node (111) operating in the communication network (100) a second indication indicating the obtained fixed length vector encoding the plurality of traffic conditions; Wherein, the first node (111) executes the method according to claim 1.

8. The method according to claim 7, wherein: The plurality of traffic conditions vary over time, and wherein the fourth node (114) manages a long short-term memory autoencoder to obtain the fixed-length vector.

9. A first node (111) for processing parameters for configuring a second node (112), the first node (111) being configured to operate in a communication network (100), the first node (111) being further configured to: - determining one or more parameters for configuring the operation of said second node (112) for delivering a service in said communication network (100), wherein The determining is configured to be based on an analysis configured to be performed by a generative adversarial network (GAN) configured to be managed by the first node (111), the analysis configured to include iteratively performing: i. generating a set of parameters estimated to be associated with the performance of a first group of nodes (121) configured to operate in said communication network (100) above a first threshold, and ii. distinguishing between the parameter set configured to be generated and a first parameter set observed to be associated with the performance of the first node group (121) above the first threshold, to obtain a respective score for each parameter, wherein the respective score is configured to indicate the degree to which the generated parameter set differs from the first parameter set, and wherein the one or more parameters configured to be determined are configured to have a corresponding score resulting from the discriminating being below a second threshold, and - outputting an indication comprising the one or more parameters configured to be determined; wherein the GAN is configured as a conditional GAN, and wherein performing the analysis is further configured to include adjusting based on a plurality of traffic conditions, and wherein the one or more parameters configured to be determined are configured to include a plurality of sets of one or more parameters, each of the plurality of sets of one or more parameters being configured to be determined for a different traffic condition of the plurality of traffic conditions; The first node (111) is further configured to: - obtaining a first indication from a third node (113) configured to operate in said communication network (100), said first indication being configured to indicate said first set of parameters configured to be observed to be associated with said performance of said first group of nodes (121) being above said first threshold, and wherein said first set of parameters configured to be obtained is configured to be used for said generating and said discerning in said analyzing; - Obtaining a second indication from a fourth node (114) configured to operate in the communication network (100), the second indication being configured to indicate a fixed-length vector encoding the plurality of traffic conditions configured to be observed to be associated with the performance of the first group of nodes (121) configured to operate in the communication network (100), and wherein the fixed-length vector configured to be obtained is configured to be used for the generation and the identification in the analysis.

10. The first node (111) according to claim 9, wherein: The determination is configured to include a first stage and a second stage, wherein the first stage is configured to include: The generation and identification are performed by the following operations: i. training the GAN with the first set of parameters configured to be observed to be associated with the performance of the first group of nodes (121) above the first threshold and the plurality of traffic conditions to obtain a data model of the parameters based on the plurality of traffic conditions, And wherein the second stage is configured to include: ii. executing the data model by inputting a first observed set of flow conditions as a first fixed length vector, and iii. Based on the data model configured to be obtained, determining the one or more parameters corresponding to the first observed set of flow conditions.

11. The first node (111) according to claim 9, wherein: The third node (113) is an autoencoder configured to be managed by the first node (111), and wherein obtaining the first parameter set is further configured to include: training a first data model using corresponding data of nodes configured to be included in a node group (123), the corresponding data of the nodes indicating corresponding performance of the nodes over a period of time, the node group (123) being configured to include the first node group (121), obtaining, for each node in the set of nodes (123), a reconstruction of the corresponding error of the first data model configured to be trained, ordering the nodes configured to be included in the node group (123) based on their respective reconstruction errors, and The first group of nodes (121) is selected from the group of nodes (123) based on their respective errors being configured to be reconstructed being above the first threshold.

12. The first node (111) according to claim 9, wherein: The multiple traffic conditions vary over time, and wherein the fourth node (114) is configured to be managed by the first node (111) and is configured to manage a long short-term memory autoencoder to generate the fixed-length vector, and wherein the obtaining is configured to include training a second data model with corresponding sequential traffic data of the nodes configured to be included in the first node group (121) to encode the corresponding sequential traffic data into the fixed-length vector.

13. A third node (113) for processing parameters for configuring a second node (112), the third node (113) being configured as an autoencoder, the autoencoder being configured to operate in a communication network (100), the third node (113) being further configured to: - obtaining a first set of parameters configured to be observed to be associated with performance of a first set of nodes (121) configured to operate in the communication network (100) above a first threshold, the first set of parameters being configured to be parameters for configuring a second node (112) for operation of delivering a service in the communication network (100), wherein Get is configured to include: training a first data model using corresponding data of nodes configured to be included in a node group (123), the corresponding data of the nodes indicating corresponding performance of the nodes over a period of time, the node group (123) being configured to include the first node group (121), obtaining, for each node in the set of nodes (123), a reconstruction of the corresponding error of the first data model configured to be trained, sorting the nodes configured to be included in the node group (123) based on their respective reconstruction errors, and selecting the first group of nodes (121) from the group of nodes (123) based on their respective errors being configured to be reconstructed being above the first threshold, and - providing a first indication to a first node (111) configured to operate in the communication network (100) according to claim 9, the first indication being configured to indicate the first set of parameters configured to be obtained.

14. The third node (113) according to claim 13, wherein: The first parameter set is configured to include a first plurality of sets of first one or more parameters, each set of the first plurality of sets of first one or more parameters being configured to correspond to a different flow condition of a plurality of flow conditions.

15. A fourth node (114) for processing parameters for configuring a second node (112), the fourth node (114) being configured to operate in a communication network (100), the fourth node (114) being further configured to: - obtaining a fixed length vector encoding a plurality of traffic conditions configured to be observed to be associated with performance of a first group of nodes (121) configured to operate in the communication network (100), wherein The method is configured to include: training a second data model with corresponding sequential traffic data of nodes configured to be included in the first node group (121) to encode the corresponding sequential traffic data into the fixed-length vector, and - providing a second indication to a first node (111) configured to operate in the communication network (100) according to claim 9, the second indication being indicative of the fixed length vector configured to be obtained encoding the plurality of traffic conditions.

16. The fourth node (114) according to claim 15, wherein: The plurality of traffic conditions vary over time, and wherein the fourth node (114) is configured to manage a long short-term memory autoencoder to obtain the fixed-length vector.

Images