[go: up one dir, main page]

WO2025074431A1 - Method and system of predicting worst affected cells or base stations in a network - Google Patents

Method and system of predicting worst affected cells or base stations in a network Download PDF

Info

Publication number
WO2025074431A1
WO2025074431A1 PCT/IN2024/051995 IN2024051995W WO2025074431A1 WO 2025074431 A1 WO2025074431 A1 WO 2025074431A1 IN 2024051995 W IN2024051995 W IN 2024051995W WO 2025074431 A1 WO2025074431 A1 WO 2025074431A1
Authority
WO
WIPO (PCT)
Prior art keywords
data
network
base stations
processor
features
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
PCT/IN2024/051995
Other languages
French (fr)
Inventor
Aayush Bhatnagar
Ankit Murarka
Jugal Kishore
Chandra GANVEER
Sanjana Chaudhary
Gourav Gurbani
Yogesh Kumar
Avinash Kushwaha
Dharmendra Kumar Vishwakarma
Sajal Soni
Niharika PATNAM
Shubham Ingle
Harsh Poddar
Sanket KUMTHEKAR
Mohit Bhanwria
Shashank Bhushan
Vinay Gayki
Aniket KHADE
Durgesh KUMAR
Zenith KUMAR
Gaurav Kumar
Manasvi Rajani
Kishan Sahu
Sunil Meena
Supriya Kaushik DE
Kumar Debashish
Mehul Tilala
Satish Narayan
Rahul Kumar
Harshita GARG
Kunal Telgote
Ralph LOBO
Girish DANGE
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Jio Platforms Ltd
Original Assignee
Jio Platforms Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Jio Platforms Ltd filed Critical Jio Platforms Ltd
Publication of WO2025074431A1 publication Critical patent/WO2025074431A1/en
Pending legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W24/00—Supervisory, monitoring or testing arrangements
    • H04W24/02—Arrangements for optimising operational condition
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W24/00—Supervisory, monitoring or testing arrangements
    • H04W24/08—Testing, supervising or monitoring using real traffic

Definitions

  • the present invention relates to the field of network monitoring and analysis, and more particularly relates to a system and a method of predicting worst affected cells or base stations in a network.
  • One or more embodiments of the present disclosure provide a method and system of predicting worst affected cells or base stations in a network.
  • the method includes pre-processing the data for data definition, data normalization, or data cleaning.
  • the method includes assessing performance of the one or more machine learning models using the testing dataset.
  • the method includes identifying areas or cells at risk of the call releases due to one of a poor signal strength, congestion, and resource unavailability in the network.
  • the system of predicting the worst affected cells or the base stations in the network includes a probing unit configured to collect data from the base stations and aggregate the data.
  • the system further includes a pre-processing unit configured to extract features from the data and split the data into a training dataset and a testing dataset.
  • the system further includes an anomaly detection unit configured to detect unusual patterns or relationships between the features in the data.
  • the system further includes a prediction unit configured to assess the unusual patterns or relationships detected for predicting the worst affected cells or the base stations in the network.
  • a non-transitory computer-readable medium having stored thereon computer-readable instructions is disclosed.
  • the computer- readable instructions are executed by a processor.
  • the processor is configured to collect data from the base stations and aggregate the data. Further, the processor is configured to extract features from the data and split the data into a training dataset and a testing dataset. Further, the processor is configured to detect unusual patterns or relationships between the features in the data. Further, the processor is configured to assess the unusual patterns or relationships detected for predicting worst affected cells or base stations in the network.
  • FIG. 3 is an exemplary block diagram of an architecture implemented in the system of the FIG. 2, according to one or more embodiments of the present invention
  • FIG. 5 is a schematic representation of a method of predicting the worst affected cells or the base stations in the network, according to one or more embodiments of the present invention.
  • the worst affected cells refer to the cells or the base stations experiencing performance degradation or ability to serve users effectively.
  • the cells facing challenges like network congestion, interference, hardware failure, and high user density are termed as the worst affected cells.
  • the UE 110 includes, but not limited to, a first UE 110a, a second UE 110b, and a third UE 110c, and should nowhere be construed as limiting the scope of the present disclosure.
  • the UE 110 may include a plurality of UEs as per the requirement.
  • each of the first UE 110a, the second UE 110b, and the third UE 110c, will hereinafter be collectively and individually referred to as the “User Equipment (UE) 110”.
  • UE User Equipment
  • the UE 110 is one of, but not limited to, any electrical, electronic, electro-mechanical or an equipment and a combination of one or more of the above devices such as a smartphone, virtual reality (VR) devices, augmented reality (AR) devices, laptop, a general-purpose computer, desktop, personal digital assistant, tablet computer, mainframe computer, or any other computing device.
  • a smartphone virtual reality (VR) devices
  • AR augmented reality
  • laptop a general-purpose computer
  • desktop personal digital assistant
  • tablet computer tablet computer
  • mainframe computer or any other computing device.
  • the environment 100 includes the server 115 accessible via the network 105.
  • the server 115 may include, by way of example but not limitation, one or more of a standalone server, a server blade, a server rack, a bank of servers, a server farm, hardware supporting a part of a cloud service or system, a home server, hardware running a virtualized server, one or more processors executing code to function as a server, one or more machines performing server-side functionality as described herein, at least a portion of any of the above, some combination thereof.
  • the entity may include, but is not limited to, a vendor, a network operator, a company, an organization, a university, a lab facility, a business enterprise side, a defense facility side, or any other facility that provides service.
  • the network 105 includes, by way of example but not limitation, one or more of a wireless network, a wired network, an internet, an intranet, a public network, a private network, a packet-switched network, a circuit-switched network, an ad hoc network, an infrastructure network, a Public-Switched Telephone Network (PSTN), a cable network, a cellular network, a satellite network, a fiber optic network, or some combination thereof.
  • PSTN Public-Switched Telephone Network
  • the network 105 may also include, by way of example but not limitation, one or more of a wireless network, a wired network, an internet, an intranet, a public network, a private network, a packet-switched network, a circuit- switched network, an ad hoc network, an infrastructure network, a Public-Switched Telephone Network (PSTN), a cable network, a cellular network, a satellite network, a fiber optic network, a VOIP or some combination thereof.
  • PSTN Public-Switched Telephone Network
  • FIG. 2 is an exemplary block diagram of the system 120 of predicting the worst affected cells or the base stations in the network 105, according to one or more embodiments of the present invention.
  • the processor 205 includes one or more modules.
  • the one or more modules/units includes, but not limited to, a probing unit 225, a pre-processing unit 230, a model training unit 235, an anomaly detection unit 240, and a prediction unit 245 communicably coupled to each other of monitoring the network 105.
  • the probing unit 225 is configured to collect data from the base stations and aggregate the data.
  • the base stations are referred as Next- Generation Node B (gNBs). Further the base stations or the gNBs can be used interchangeably without limiting the scope of the invention.
  • the base stations are configured to continuously transmit the data to the probing unit 225 on periodic basis. For example, the probing unit 225 receives the data from the one base stations in every one hour.
  • the probing unit 225 collects the data from the base stations via an interface.
  • the interface includes at least one of, but not limited to, one or more Application Programming Interfaces (APIs) which are used by the probing unit 225 for collecting the data.
  • the one or more APIs are sets of rules and protocols that allow different entities to communicate with each other.
  • the one or more APIs define the methods and data formats that entities can use to request and exchange information, enabling integration and functionality across various platforms.
  • the data includes, but is not limited to, signal strength metrics, traffic data, network load metrics, mobility patterns, and Quality of Service metrics (QoS).
  • the signal strength metrics are measurements of received signal strength indicator and reference signal received power to assess network coverage and quality.
  • the traffic data includes information on data usage patterns such as, but not limited to, download and upload speeds, the volume of data transmitted, and the types of applications being used, the network load metrics encompass data on the number of active users, resource utilization, and load balancing across the gNBs.
  • the mobility patterns include information on user movement and handover events as devices transition between different the gNBs or cells.
  • the QoS metrics consist of data on latency, jitter, packet loss, and throughput, which are used to evaluate the performance of different services.
  • the features include one of timestamps, call parameters, geographic coordinates, and network load metrics.
  • the call parameters are specific metrics related to voice or video calls in the network 105.
  • the call parameters include, but are not limited to, all setup time, call drop rate, and voice quality metrics.
  • the geographic coordinates are specific latitude and longitude values that pinpoint the physical location of the user or the network node, the geographic coordinates include, but not limited to, location-based services and network optimization.
  • the network load metrics are measurements that provide insight into the utilization of the network resources.
  • the network load metrics may include, but not limited to, active users, resources utilization, data throughput, and latency.
  • the testing dataset is a separate subset of the collected data used to evaluate the performance of the trained model.
  • the testing dataset helps assess how well the model generalizes to new, unseen data.
  • the examples of testing dataset include, but not limited to, new user movement data and new QoS measurements.
  • the model predicts outcomes based on the features in the testing dataset, and the results are compared to the actual labels to measure accuracy, precision, recall, and other performance metrics.
  • the cell or the gNBs is identified as the worst affected cells or the gNBs.
  • the one or more machine learning models may be trained to predict network 105 performance issues, allowing them to recognize patterns associated with worst affected the gNBs based on the historical data and enabling proactive measures to mitigate issues before escalating, ultimately supporting operational efficiency by allowing network operators to prioritize resources and interventions for the cells or the gNBs that require the most attention.
  • the trained model is deployed in the network management system to enable real-time predictions and proactive management of the network 105.
  • the training data is provided to the one or more models by the model training unit 235.
  • the one or more models generates outputs based on the provided training data.
  • the anomaly detection unit 240 assesses performance of the one or more machine learning models.
  • the assessing refers to the process of evaluating, measuring, or estimating the quality, value, or performance of the one or more machine learning models.
  • the model with the superior performance as compared to the other models is selected by the anomaly detection unit 240 to detect the anomaly.
  • the anomaly detection unit 240 Upon training the one or more machine learning models using the dataset from the model training unit 235, the anomaly detection unit 240 is configured to detect unusual patterns or relationships between the historical data or real-time data.
  • the anomaly detection unit 240 identifies unusual patterns in the network data by collecting metrics from sources like the gNBs and user devices, including signal strength and latency.
  • the networks data generates the information and metrics within the network environment.
  • the anomaly detection unit 240 extracts the feature and establishes a standard for normal behavior using historical data.
  • the anomaly detection unit 240 then applies algorithms such as, but not limited to, statistical methods, isolation forests, and clustering techniques to detect deviations in real-time.
  • the anomaly detection unit 240 Upon detecting an anomaly, the anomaly detection unit 240 generates alerts for network operators, detailing the issue and affected the gNBs.
  • the detection process involves collecting metrics and data from various sources, extracting relevant features, and establishing the standard for normal behavior using the historical data. Algorithms such as, but not limited to, statistical methods, isolation forests, and clustering techniques are then applied to analyze real-time data against the standard for normal behaviour, identifying significant deviations that may indicate anomalies. Furthermore, a feedback mechanism allows operators to validate the anomalies, refining the detection algorithms and improving future accuracy to maintain network performance and reliability.
  • the process of evaluating the performance metrics of the anomaly detection unit 240 which helps to identify strengths and weaknesses, allowing for adjustments to improve accuracy such as, bit not limited to, refining feature selection or retraining with additional data.
  • the example of the strength includes, but not limited to, real-time monitoring, predictive maintenance, enhanced user experience, and automated decision making.
  • the examples of the weaknesses include, but not limited to, false positives, dependency on historical data, and complexity of implementation.
  • the anomaly detection unit 240 select the best-performing model for deployment or tune hyperparameters to enhance performance, ensuring that the models reliably detect unusual patterns in the network 105 and contribute to effective network management and operational efficiency.
  • the prediction unit 245 Upon detecting the unusual patterns or relationships between the features in the data, the prediction unit 245 is configured to assess the unusual patterns or relationships detected for predicting the worst affected cells or the gNBs in the network 105 that are likely to experience call releases.
  • the prediction unit 245 can predict call releases, enabling operators to take proactive measures like reallocating resources or optimizing load balancing.
  • the capability of the prediction unit 245 helps maintain service quality and enhances user experience within the network 105.
  • the threshold is set based on assessing the unusual patterns or relationships between the features in the data. In one scenario at a peak time the number of the number of the users connected with the base station are increasing.
  • the prediction unit 245 identifies areas or cells at risk of the call releases due to one of a poor signal strength, congestion, and resource unavailability in the network 105.
  • the prediction unit 245 monitors signal strength metrics collected from user devices and the gNBs. If the signal strength falls below the predefined threshold, then the prediction unit 245 identifies areas or cells at risk of the call releases due to at least one of the poor signal strengths and indicates that users may struggle to maintain the stable connection, leading to potential call drops.
  • the predefined threshold represents the minimum signal strength required for reliable communication in the network 105 when the predefined threshold is exceeded, users may expect the high- quality connection.
  • the prediction unit 245 evaluates the availability of network resources, such as bandwidth and processing power. When the availability of network resources is insufficient to handle the current load, calls may be released because the network cannot maintain the required quality of service. Overloading of the gNBs or temporary unavailability of certain resources due to the maintenance or other issues can lead to such situations.
  • the one or more models is retrained based on updated data received from the base stations due to which the one or more models are getting advances and leads to increase in the accuracy of the prediction.
  • the updated data is collected from the base stations on periodic basis.
  • the updated data is fed to the one or more models by the model training unit 235, so that based on the updated data, the model is trained again.
  • the data integration 320 undergoes aggregation, where data from the gNBs is combined and aligned correctly.
  • the aggregation process may include, but not limited to, merging data, performing initial consistency checks, and preparing for the pre-processing.
  • the data integration 320 ensures that the data is consistent and ready for the one or more machine learning operations, where the data collected from the gNBs is aggregated for subsequent analysis.
  • the data is split into the training dataset and the testing dataset.
  • the data ensures that the one or more machine learning models are trained on one portion of the data while another portion is reserved for evaluating the performance of the models.
  • the training dataset is used to build the predictive model, and the testing dataset is used to evaluate the accuracy of the predictions.
  • the model training 330 uses the one or more machine learning models to train on the training dataset. Specifically, the model training 330 involves analyzing the data to extract patterns, trends, and relationships from the features, such as, but not limited to, call parameters, geographic coordinates, and network load metrics. During the model training 330 phase, the one or more machine learning model learns from the data to identify underlying patterns that may later be used to make predictions or detect anomalies. The choice of the one or more machine learning algorithms applied at the stage depends on the complexity and volume of the data. [0070] After the model is trained, the prediction module 335 is responsible for detecting unusual patterns or relationships between the features in the data.
  • Both the training results and predictions generated by the one or more machine learning models are stored in the database 220.
  • the database 220 holds historical data and can be used to re-train models over time or for further analysis.
  • the database 220 ensures that there is the persistent record of the predictions, which can be accessed for future comparisons or investigations.
  • the anomaly detected results are sending and predictions to the data consumers 310.
  • the data consumers 310 could be network operators or automation systems that monitor network performance and take proactive actions based on the predictions. For example, the data consumers may use this information to adjust network resources, troubleshoot poor signal areas, or mitigate congestion before leads to significant service degradation.
  • the processing hub interface 340 serves as the central communication link between various components within the architecture 300.
  • the processing hub interface 340 role is to manage the flow of the data between modules such as the data integration 320, the data pre-processing 325, the model training 330, and the prediction module 335.
  • the processing hub interface 340 ensures that data is transferred efficiently between the processing hub 315, enabling seamless interaction.
  • FIG. 4 is an exemplary flowchart diagram of predicting the worst affected cells or the gNBs in the network 105, according to one or more embodiments of the present invention.
  • the process begins with the collection and aggregation of the raw data from the gNBs in the network 105.
  • the data includes, but is not limited to, call release reasons, timestamps, call parameters, network load metrics, and other relevant information.
  • the purpose of this step is to gather essential information that will later be used for training and prediction.
  • the training dataset is used to train the one or more machine learning models.
  • the prediction module 335 identifies the patterns, relationships, and connections between variables in the data which includes, but not limited to, factors contributing to call releases or network congestion. The objective is to develop the prediction module 335 capable of recognizing patterns linked to potential network issues, thereby enabling early detection and response to problems.
  • the trained model upon evaluating against the testing dataset to assess the accuracy and performance, produces insights by identifying cells or the gNBs that are at the highest risk of facing issues, such as call releases. These insights are based on the unusual patterns or anomalies detected during the training and testing phases. The objective is to predict the worst-affected areas in the network 105 and highlight potential problems before they result in disruptions.
  • the one or more machine learning model continuously adapts based on new data and changing network conditions to ensure the one or more machine learning model remains effective in predicting network issues as conditions evolve over time.
  • FIG. 5 is a schematic representation of a method 500 of predicting worst affected cells or the gNBs in the network 105, according to one or more embodiments of the present invention.
  • the method 500 is described with the embodiments as illustrated in FIG. 2 and should nowhere be construed as limiting the scope of the present disclosure.
  • the method 500 includes the step of extracting features from the data.
  • the features include one of the timestamps, call parameters, geographic coordinates, and network load metrics.
  • the present invention further discloses a non-transitory computer-readable medium having stored thereon computer-readable instructions.
  • the computer-readable instructions are executed by the processor 205.
  • the processor 205 is configured to collect data from the base stations and aggregate the data. Further, the processor 205 is configured to extract features from the data and split the data into the training dataset and the testing dataset. Further, the processor 205 is configured to detect unusual patterns or relationships between the features in the data. Further, the processor 205 is configured to assess the unusual patterns or relationships detected for predicting worst affected cells or the base stations in the network 105 that are likely to experience call releases.
  • the present disclosure includes technical advancement by improving predictive accuracy with advanced the machine learning models and diverse data sources, such as call parameters and geographic coordinates.
  • Automated anomaly detection and targeted risk identification address potential issues like poor signal strength and congestion.
  • the invention robust the data pre-processing ensures high-quality inputs for reliable predictions, while systematic performance assessments facilitate continuous improvement.
  • the scalable architecture optimizes resource allocation, reducing costs and boosting user satisfaction, and computer-readable instructions allow for easy integration across platforms, promoting widespread adoption and overall efficiency.
  • the present invention offers multiple advantages over the prior art and the above listed are a few examples to emphasize on some of the advantageous features.
  • the listed advantages are to be read in a non-limiting manner.

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Data Exchanges In Wide-Area Networks (AREA)

Abstract

The present disclosure relates to a system (120) and a method (500) of predicting worst affected cells or base stations in a network (105) The method (500) includes the step of collecting data from base stations and aggregating the data. The method (500) includes the step of extracting features from the data. The method (500) further includes the step of splitting the data into a training dataset and a testing dataset. The method (500) further includes the step of training the training dataset using one or more machine learning models. The method (500) further includes the step of detecting unusual patterns or relationships between the features in the data. Further the method (500) includes the step of assessing the unusual patterns or relationships detected for predicting worst affected cells or the base stations in the network (105).

Description

METHOD AND SYSTEM OF PREDICTING WORST AFFECTED CELLS OR BASE STATIONS IN A NETWORK
FIELD OF THE INVENTION
[0001] The present invention relates to the field of network monitoring and analysis, and more particularly relates to a system and a method of predicting worst affected cells or base stations in a network.
BACKGROUND OF THE INVENTION
[0002] With the increase in number of users, the network service provisions have been implementing to up-gradations to enhance the service quality so as to keep pace with such high demand. With advancement of technology, there is a demand for the telecommunication service to induce up to date features into the scope of provision so as to enhance user experience and implement advanced monitoring mechanisms. There are regular data analyses to observe issues beforehand for which many data collection as well as assessment practices are implemented in a network.
[0003] Probing units, a crucial component within telecommunications networks, are tasked with the responsibility of data analytics, providing valuable insights into network performance, and particularly focusing on Raw Call Release Reason data. One of its key functions is to collect and analyse this raw data to pinpoint Next-Generation Node B with the highest count of released calls, which are essentially the network nodes facing the most call drops. The gNBs having independent network function is responsible for radio communication with UEs in its coverage area.
[0004] However, in traditional systems, identifying these problematic gNBs is a manual process that necessitates the analysis of the Call Release Reason (CRR) data for the gNodeB. Only after this analysis, the probing unit notifies the telecom operators to take corrective actions aimed at reducing call drops. The delay in addressing network issues may result in customer dissatisfaction and service disruptions.
[0005] There is a need for a mechanism, more specifically a system and method thereof to predict and flag specific network areas or cells susceptible to call drops due to issues like weak signal strength, network congestion, or resource constraints. SUMMARY OF THE INVENTION
[0006] One or more embodiments of the present disclosure provide a method and system of predicting worst affected cells or base stations in a network.
[0007] In one aspect of the present invention, the method of predicting the worst affected cells or base stations in the network is disclosed. The method includes the step of collecting data from the base stations and aggregating the data. The method includes the step of extracting features from the data. The method further includes the step of splitting the data into a training dataset and a testing dataset. The method further includes the step of detecting unusual patterns or relationships between the features in the data. Further the method includes the step of assessing the unusual patterns or relationships detected for predicting the worst affected cells or the base stations in the network.
[0008] In an embodiment, the features comprising one of timestamps, call parameters, geographic coordinates, and network load metrics.
[0009] In an embodiment, the method includes pre-processing the data for data definition, data normalization, or data cleaning.
[0010] In an embodiment, the method includes annotating the data with a target variable for identifying the worst affected cells or base stations in the network.
[0011] In one embodiment, the method includes assessing performance of the one or more machine learning models using the testing dataset.
[0012] In an embodiment, the method includes identifying areas or cells at risk of the call releases due to one of a poor signal strength, congestion, and resource unavailability in the network.
[0013] In an embodiment, the one or more processor (205) trains one or more machine learning models using the training dataset.
[0014] In another aspect of the present invention, the system of predicting the worst affected cells or the base stations in the network is disclosed. The system includes a probing unit configured to collect data from the base stations and aggregate the data. The system further includes a pre-processing unit configured to extract features from the data and split the data into a training dataset and a testing dataset. The system further includes an anomaly detection unit configured to detect unusual patterns or relationships between the features in the data. The system further includes a prediction unit configured to assess the unusual patterns or relationships detected for predicting the worst affected cells or the base stations in the network.
[0015] In another aspect of the invention, a non-transitory computer-readable medium having stored thereon computer-readable instructions is disclosed. The computer- readable instructions are executed by a processor. The processor is configured to collect data from the base stations and aggregate the data. Further, the processor is configured to extract features from the data and split the data into a training dataset and a testing dataset. Further, the processor is configured to detect unusual patterns or relationships between the features in the data. Further, the processor is configured to assess the unusual patterns or relationships detected for predicting worst affected cells or base stations in the network.
[0016] Other features and aspects of this invention will be apparent from the following description and the accompanying drawings. The features and advantages described in this summary and in the following detailed description are not all-inclusive, and particularly, many additional features and advantages will be apparent to one of ordinary skill in the relevant art, in view of the drawings, specification, and claims hereof. Moreover, it should be noted that the language used in the specification has been principally selected for readability and instructional purposes and may not have been selected to delineate or circumscribe the inventive subject matter, resort to the claims being necessary to determine such inventive subject matter.
BRIEF DESCRIPTION OF THE DRAWINGS
[0017] The accompanying drawings, which are incorporated herein, and constitute a part of this disclosure, illustrate exemplary embodiments of the disclosed methods and systems in which like reference numerals refer to the same parts throughout the different drawings. Components in the drawings are not necessarily to scale, emphasis instead being placed upon clearly illustrating the principles of the present disclosure. Some drawings may indicate the components using block diagrams and may not represent the internal circuitry of each component. It will be appreciated by those skilled in the art that disclosure of such drawings includes disclosure of electrical components, electronic components or circuitry commonly used to implement such components.
[0018] FIG. 1 is an exemplary block diagram of an environment of predicting worst affected cells or base stations in a network, according to one or more embodiments of the present invention;
[0019] FIG. 2 is an exemplary block diagram of a system for predicting the worst affected cells or the base stations in the network, according to one or more embodiments of the present invention;
[0020] FIG. 3 is an exemplary block diagram of an architecture implemented in the system of the FIG. 2, according to one or more embodiments of the present invention;
[0021] FIG. 4 is a flowchart diagram of predicting the worst affected cells or the base stations in the network, according to one or more embodiments of the present invention; and
[0022] FIG. 5 is a schematic representation of a method of predicting the worst affected cells or the base stations in the network, according to one or more embodiments of the present invention.
[0023] The foregoing shall be more apparent from the following detailed description of the invention.
DETAILED DESCRIPTION OF THE INVENTION
[0024] Some embodiments of the present disclosure, illustrating all its features, will now be discussed in detail. It must also be noted that as used herein and in the appended claims, the singular forms "a", "an" and "the" include plural references unless the context clearly dictates otherwise. [0025] Various modifications to the embodiment will be readily apparent to those skilled in the art and the generic principles herein may be applied to other embodiments. However, one of ordinary skill in the art will readily recognize that the present disclosure including the definitions listed here below are not intended to be limited to the embodiments illustrated but is to be accorded the widest scope consistent with the principles and features described herein.
[0026] A person of ordinary skill in the art will readily ascertain that the illustrated steps detailed in the figures and here below are set out to explain the exemplary embodiments shown, and it should be anticipated that ongoing technological development will change the manner in which particular functions are performed. These examples are presented herein for purposes of illustration, and not limitation. Further, the boundaries of the functional building blocks have been arbitrarily defined herein for the convenience of the description. Alternative boundaries can be defined so long as the specified functions and relationships thereof are appropriately performed. Alternatives (including equivalents, extensions, variations, deviations, etc., of those described herein) will be apparent to persons skilled in the relevant art(s) based on the teachings contained herein. Such alternatives fall within the scope and spirit of the disclosed embodiments.
[0027] FIG. 1 illustrates an exemplary block diagram of an environment 100 of predicting worst affected cells or base stations in a network 105, according to one or more embodiments of the present disclosure. In this regard, the environment 100 includes a User Equipment (UE) 110, a server 115, the network 105 and a system 120 communicably coupled to each of monitoring the network 105. The cell refers to a geographic area served by a base station. Each cell allows the UE 110 to connect to the network 105 and access services. The network 105 is divided into multiple cells to ensure coverage over a wide area. The cells are crucial for managing user traffic efficiently, optimizing coverage, and ensuring seamless handovers as users move from one cell to another. The base stations are responsible for managing communication between the UE 110 and the core network. The base stations are part of the Radio Access Network (RAN) in the network 105 and plays the key role in handling wireless communication includes, but not limited to, data transmission, signal processing, and control functions.
[0028] The worst affected cells refer to the cells or the base stations experiencing performance degradation or ability to serve users effectively. The cells facing challenges like network congestion, interference, hardware failure, and high user density are termed as the worst affected cells.
[0029] As per the illustrated embodiment and for the purpose of description and illustration, the UE 110 includes, but not limited to, a first UE 110a, a second UE 110b, and a third UE 110c, and should nowhere be construed as limiting the scope of the present disclosure. In alternate embodiments, the UE 110 may include a plurality of UEs as per the requirement. For ease of reference, each of the first UE 110a, the second UE 110b, and the third UE 110c, will hereinafter be collectively and individually referred to as the “User Equipment (UE) 110”.
[0030] In an embodiment, the UE 110 is one of, but not limited to, any electrical, electronic, electro-mechanical or an equipment and a combination of one or more of the above devices such as a smartphone, virtual reality (VR) devices, augmented reality (AR) devices, laptop, a general-purpose computer, desktop, personal digital assistant, tablet computer, mainframe computer, or any other computing device.
[0031] The environment 100 includes the server 115 accessible via the network 105. The server 115 may include, by way of example but not limitation, one or more of a standalone server, a server blade, a server rack, a bank of servers, a server farm, hardware supporting a part of a cloud service or system, a home server, hardware running a virtualized server, one or more processors executing code to function as a server, one or more machines performing server-side functionality as described herein, at least a portion of any of the above, some combination thereof. In an embodiment, the entity may include, but is not limited to, a vendor, a network operator, a company, an organization, a university, a lab facility, a business enterprise side, a defense facility side, or any other facility that provides service.
[0032] The network 105 includes, by way of example but not limitation, one or more of a wireless network, a wired network, an internet, an intranet, a public network, a private network, a packet-switched network, a circuit-switched network, an ad hoc network, an infrastructure network, a Public-Switched Telephone Network (PSTN), a cable network, a cellular network, a satellite network, a fiber optic network, or some combination thereof. The network 105 may include, but is not limited to, a Third Generation (3G), a Fourth Generation (4G), a Fifth Generation (5G), a Sixth Generation (6G), a New Radio (NR), a Narrow Band Internet of Things (NB-IoT), an Open Radio Access Network (O-RAN), and the like.
[0033] The network 105 may also include, by way of example but not limitation, at least a portion of one or more networks 105 having one or more nodes that transmit, receive, forward, generate, buffer, store, route, switch, process, or a combination thereof, etc. one or more messages, packets, signals, waves, voltage or current levels, some combination thereof, or so forth. The network 105 may also include, by way of example but not limitation, one or more of a wireless network, a wired network, an internet, an intranet, a public network, a private network, a packet-switched network, a circuit- switched network, an ad hoc network, an infrastructure network, a Public-Switched Telephone Network (PSTN), a cable network, a cellular network, a satellite network, a fiber optic network, a VOIP or some combination thereof.
[0034] The environment 100 further includes the system 120 communicably coupled to the server 115 and the UE 110 via the network 105. The system 120 is configured to manage CNFs in the network 105. As per one or more embodiments, the system 120 is adapted to be embedded within the server 115 or embedded as an individual entity.
[0035] Operational and construction features of the system 120 will be explained in detail with respect to the following figures. [0036] FIG. 2 is an exemplary block diagram of the system 120 of predicting the worst affected cells or the base stations in the network 105, according to one or more embodiments of the present invention.
[0037] As per the illustrated embodiment, the system 120 includes one or more processors 205, a memory 210, a user interface 215, and a database 220. For the purpose of description and explanation, the description will be explained with respect to one processor 205 and should nowhere be construed as limiting the scope of the present disclosure. In alternate embodiments, the system 120 may include more than one processor 205 as per the requirement of the network 105. The one or more processors 205, hereinafter referred to as the processor 205 may be implemented as one or more microprocessors, microcomputers, microcontrollers, digital signal processors, central processing units, state machines, logic circuitries, single board computers, and/or any devices that manipulate signals based on operational instructions.
[0038] As per the illustrated embodiment, the processor 205 is configured to fetch and execute computer-readable instructions stored in the memory 210. The memory 210 may be configured to store one or more computer-readable instructions or routines in a non- transitory computer-readable storage medium, which may be fetched and executed to create or share data packets over a network service. The memory 210 may include any non-transitory storage device including, for example, volatile memory such as RAM, or non-volatile memory such as disk memory, EPROMs, FLASH memory, unalterable memory, and the like.
[0039] In an embodiment, the user interface 215 includes a variety of interfaces, for example, interfaces for a graphical user interface, a web user interface, a Command Line Interface (CLI), and the like. The user interface 215 facilitates communication of the system 120. In one embodiment, the user interface 215 provides a communication pathway for one or more components of the system 120. Examples of such components include, but are not limited to, the UE 110 and the database 220. [0040] The database 220 is one of, but not limited to, a centralized database, a cloudbased database, a commercial database, an open-source database, a distributed database, an end-user database, a graphical database, a No-Structured Query Language (NoSQL) database, an object-oriented database, a personal database, an in-memory database, a document-based database, a time series database, a wide column database, a key value database, a search database, a cache databases, and so forth. The foregoing examples of database 220 types are non-limiting and may not be mutually exclusive e.g., a database can be both commercial and cloud-based, or both relational and open-source, etc.
[0041] In order for the system 120 to predict the worst affected cells in the network 105, the processor 205 includes one or more modules. In one embodiment, the one or more modules/units includes, but not limited to, a probing unit 225, a pre-processing unit 230, a model training unit 235, an anomaly detection unit 240, and a prediction unit 245 communicably coupled to each other of monitoring the network 105.
[0042] In one embodiment, the one or more modules may be used in combination or interchangeably for predicting the worst affected cells or the base stations in the network 105.
[0043] The probing unit 225, the pre-processing unit 230, the model training unit 235, the anomaly detection unit 240, and the prediction unit 245, in an embodiment, are implemented as a combination of hardware and programming (for example, programmable instructions) to implement one or more functionalities of the processor 205. In the examples described herein, such combinations of hardware and programming are implemented in several different ways. For example, the programming for the processor 205 may be processor-executable instructions stored on a non-transitory machine -readable storage medium and the hardware for the processor may comprise a processing resource (for example, one or more processors), to execute such instructions. In the present examples, the memory 210 may store instructions that, when executed by the processing resource, implement the processor. In such examples, the system 120 may comprise the memory 210 storing the instructions and the processing resource to execute the instructions, or the memory 210 may be separate but accessible to the system 120 and the processing resource. In other examples, the processor 205 may be implemented by electronic circuitry.
[0044] In an embodiment, the probing unit 225 is configured to collect data from the base stations and aggregate the data. Hereinafter, the base stations are referred as Next- Generation Node B (gNBs). Further the base stations or the gNBs can be used interchangeably without limiting the scope of the invention. In one embodiment, the base stations are configured to continuously transmit the data to the probing unit 225 on periodic basis. For example, the probing unit 225 receives the data from the one base stations in every one hour. In yet another embodiment, the probing unit 225 collects the data from the base stations via an interface. In one embodiment, the interface includes at least one of, but not limited to, one or more Application Programming Interfaces (APIs) which are used by the probing unit 225 for collecting the data. The one or more APIs are sets of rules and protocols that allow different entities to communicate with each other. The one or more APIs define the methods and data formats that entities can use to request and exchange information, enabling integration and functionality across various platforms.
[0045] The data includes, but is not limited to, signal strength metrics, traffic data, network load metrics, mobility patterns, and Quality of Service metrics (QoS). The signal strength metrics are measurements of received signal strength indicator and reference signal received power to assess network coverage and quality. The traffic data includes information on data usage patterns such as, but not limited to, download and upload speeds, the volume of data transmitted, and the types of applications being used, the network load metrics encompass data on the number of active users, resource utilization, and load balancing across the gNBs. The mobility patterns include information on user movement and handover events as devices transition between different the gNBs or cells. The QoS metrics consist of data on latency, jitter, packet loss, and throughput, which are used to evaluate the performance of different services. [0046] Upon collecting and aggregating the data from the gNBs, the pre-processing unit 230 is configured to extract features from the data. In one embodiment, the preprocessing unit 230 extract features from the data based on a predefined criteria such as the extracted features must include at least one of, but not limited to, the call parameters, the geographic coordinates, the network load metrics and timestamps. Herein, the predefined criteria is defined by at least one of, but not limited to, the user and the model based on the previous model training experience. The pre-processing unit is further configured to split the data into a training dataset and a testing dataset. The training dataset is a subset of the collected data used to train one or more machine learning model. The training dataset consists of input features and the corresponding labels, allowing the model to learn patterns and relationships within the data. The examples of the training dataset include, but not limited to, signal strength metrics and traffic patterns. The training dataset applied to the model to identify patterns and relationships in the data and adjusts the parameters of the one or more machine learning model to minimize prediction errors based on the features and labels in the training dataset. The parameters refer to the internal configurations within the one or more machine leaning models that are adjusted during the training process to improve the performance.
[0047] In an embodiment, the features include one of timestamps, call parameters, geographic coordinates, and network load metrics. The call parameters are specific metrics related to voice or video calls in the network 105. The call parameters include, but are not limited to, all setup time, call drop rate, and voice quality metrics. The geographic coordinates are specific latitude and longitude values that pinpoint the physical location of the user or the network node, the geographic coordinates include, but not limited to, location-based services and network optimization. The network load metrics are measurements that provide insight into the utilization of the network resources. The network load metrics may include, but not limited to, active users, resources utilization, data throughput, and latency.
[0048] The testing dataset is a separate subset of the collected data used to evaluate the performance of the trained model. The testing dataset helps assess how well the model generalizes to new, unseen data. The examples of testing dataset include, but not limited to, new user movement data and new QoS measurements. The model predicts outcomes based on the features in the testing dataset, and the results are compared to the actual labels to measure accuracy, precision, recall, and other performance metrics.
[0049] Subsequent to extracting the features, the pre-processing unit 230 is configured to pre-process the data collected from the base stations. The pre-processing includes at least one of, but not limited to, data definition, data normalization, or data cleaning. The data definition involves specifying the structure, format, and meaning of the data collected from the network 105. The data definition includes identifying the types of data being collected such as, but not limited to, signal strength, user location, call parameters. The data normalization is the process of transforming data into a consistent format, ensuring that the data is comparable and may be analyzed effectively. The data normalization may involve adjusting data values to a common scale without distorting differences in the ranges of values. The data cleaning refers to the process of identifying and correcting errors, inconsistencies, and inaccuracies in the data. The data normalization is critical in the network 105, where the data from various sources may be noisy or incomplete. The data normalization includes, but not limited to, removing duplicate entries, handling missing values, and correcting errors.
[0050] In one embodiment, the pre-processing unit 230 is configured to preprocess the data to ensure the data consistency and quality within the system 120. The main goal of, the pre-processing unit 230 is to achieve a standardized data format of the data with no errors. The pre-processing unit 230 eliminates duplicate data and inconsistencies which reduces manual efforts. The pre-processing unit 230 ensures that the pre-processed data is stored appropriately in at least one of, the database 220 and the pre-processed data is ready for subsequent retrieval and analysis.
[0051] Furthermore, the pre-processing unit 230 annotates the data with a target variable for identifying the worst affected cells or the gNBs in the network 105. In one embodiment, the target variables are one or more parameters related to the cells or the gNBs which includes, at least one of, but not limited to, a signal strength, a throughput, and a dropped call rate. The data annotation involves adding labels or tags to data points, which helps in identifying specific features or outcomes in the dataset. In this regard, the pre-processing unit 230 annotates the data with the target variable to highlight which cells or the gNBs are the worst affected in the network 105. For example, the target variable such as the signal strength indicates the strength of the signal in the cell. If the strength of the signal is less as compared to the predefined threshold of the signal strength, then the cell or the gNBs is identified as the worst affected cells or the gNBs. By annotating the data, the one or more machine learning models may be trained to predict network 105 performance issues, allowing them to recognize patterns associated with worst affected the gNBs based on the historical data and enabling proactive measures to mitigate issues before escalating, ultimately supporting operational efficiency by allowing network operators to prioritize resources and interventions for the cells or the gNBs that require the most attention.
[0052] Thereafter, the model training unit 235 configured to train the training dataset using the one or more machine learning models. The model training unit 235 is configured to enable the one or more machine learning models to learn patterns and relationships that can predict outcomes related to network performance. The model training unit 235 is responsible for training the one or more machine learning algorithms using the dataset. The model training unit 235 prepares the training dataset by selecting relevant features and preprocesses the data through normalization to ensure consistent input for training Features and labels are fed into the model, which learns by making predictions, calculating loss, and adjusting parameters through backpropagation and optimization algorithms like gradient descent. The process of training the dataset is repeated over multiple epochs, sometimes using batch training for efficiency, while validation sets monitor performance to prevent overfitting. After evaluating the model's accuracy and precision on the testing dataset, the trained model is deployed in the network management system to enable real-time predictions and proactive management of the network 105. [0053] For example, let us consider that the training data is provided to the one or more models by the model training unit 235. Further, the one or more models generates outputs based on the provided training data. Furthermore, based on the generated outputs, the anomaly detection unit 240 assesses performance of the one or more machine learning models. Herein, the assessing refers to the process of evaluating, measuring, or estimating the quality, value, or performance of the one or more machine learning models. Based on the assessing, the model with the superior performance as compared to the other models is selected by the anomaly detection unit 240 to detect the anomaly.
[0054] Upon training the one or more machine learning models using the dataset from the model training unit 235, the anomaly detection unit 240 is configured to detect unusual patterns or relationships between the historical data or real-time data. The anomaly detection unit 240 identifies unusual patterns in the network data by collecting metrics from sources like the gNBs and user devices, including signal strength and latency. The networks data generates the information and metrics within the network environment. The anomaly detection unit 240 extracts the feature and establishes a standard for normal behavior using historical data. The anomaly detection unit 240 then applies algorithms such as, but not limited to, statistical methods, isolation forests, and clustering techniques to detect deviations in real-time. Upon detecting an anomaly, the anomaly detection unit 240 generates alerts for network operators, detailing the issue and affected the gNBs. The detection process involves collecting metrics and data from various sources, extracting relevant features, and establishing the standard for normal behavior using the historical data. Algorithms such as, but not limited to, statistical methods, isolation forests, and clustering techniques are then applied to analyze real-time data against the standard for normal behaviour, identifying significant deviations that may indicate anomalies. Furthermore, a feedback mechanism allows operators to validate the anomalies, refining the detection algorithms and improving future accuracy to maintain network performance and reliability.
[0055] For example, let us consider that the features in the data includes at least one of, but not limited to, call parameters, and network load metrics. Further, the call parameters may include at least one of, but not limited to, a call drop and the network load metric include at least one of, but not limited to, a latency. Herein, the relationships between two or more features, shows how changes in one metric are linked to changes in another. More particularly, the relationships due to the changes in the latency are linked to the call drop. In order to detect, the unusual patterns or relationships between the features in the data, the anomaly detection unit 240 compares the features with the standard for normal pattern or behavior of the features. Based on comparison, the anomaly detection unit 240 detects the unusual patterns or relationships between the features in the data such as even though the latency is less, the number of the call drops are more in the particular cell.
[0056] In an embodiment, the anomaly detection unit 240 assesses performance of the one or more machine learning models using the testing dataset. The anomaly detection unit 240 assesses the performance of the one or more machine learning models using the testing dataset, which consists of data that the models have not encountered during training. By applying the trained models to the testing dataset, the anomaly detection unit 240 makes predictions about whether data points are normal or anomalous. The anomaly detection unit 240 compares the predictions to the actual labels, which indicate the true classifications. Various performance metrics are calculated to quantify the models' effectiveness including accuracy, precision, recall, providing insight into how well the models generalize to new data. The process of evaluating the performance metrics of the anomaly detection unit 240 which helps to identify strengths and weaknesses, allowing for adjustments to improve accuracy such as, bit not limited to, refining feature selection or retraining with additional data. The example of the strength includes, but not limited to, real-time monitoring, predictive maintenance, enhanced user experience, and automated decision making. The examples of the weaknesses include, but not limited to, false positives, dependency on historical data, and complexity of implementation. Based on the evaluation results, the anomaly detection unit 240 select the best-performing model for deployment or tune hyperparameters to enhance performance, ensuring that the models reliably detect unusual patterns in the network 105 and contribute to effective network management and operational efficiency.
[0057] Upon detecting the unusual patterns or relationships between the features in the data, the prediction unit 245 is configured to assess the unusual patterns or relationships detected for predicting the worst affected cells or the gNBs in the network 105 that are likely to experience call releases.
[0058] The call releases occur when the active call is terminated prematurely due to various reasons, including poor network conditions, inadequate signal strength, or resource unavailability in the network 105. Frequent call releases may lead to the negative user experience and dissatisfaction among customers. By identifying specific cells or gNBs at the higher risk of causing call releases, the prediction unit 245 enables proactive measures to be taken by network operators.
[0059] The prediction unit 245 assesses unusual patterns or relationships in network data to predict which cells or the gNBs are likely to experience call releases. The prediction unit 245 analyzes various metrics such as, but not limited to, signal strength, user mobility patterns, and traffic load metrics to identify correlations with past call drops. By employing the one or more machine learning algorithms, the prediction unit 245 recognizes patterns that indicate potential connectivity issues, such as low signal strength combined with high traffic. Continuous real-time monitoring allows the prediction unit 245 to apply predictive models trained on historical data to forecast at-risk the gNBs. The prediction unit 245 generating alerts for network operators when unusual conditions arise, like sudden drops in signal quality or congestion during peak hours. For example, if the gNBs in the densely populated area is nearing capacity during the major event, the prediction unit 245 can predict call releases, enabling operators to take proactive measures like reallocating resources or optimizing load balancing. The capability of the prediction unit 245 helps maintain service quality and enhances user experience within the network 105. [0060] In yet another example, let us consider that there are 1 lakh users connected with the base station the network 105 and the threshold value for handling the capacity of the users by the base station is 2 lakhs users. Herein, the threshold is set based on assessing the unusual patterns or relationships between the features in the data. In one scenario at a peak time the number of the number of the users connected with the base station are increasing. So, the prediction unit 245 identifies the pattern of the increasing number of the users with respect to time using the one or more models. Based on the identified pattern, the prediction unit 245 predicts that in some time interval the signal strength of the base station will be degraded due to the high traffic of the user which will lead to various call drops. So, prediction unit 245 infers the base station as the worst affected cell or the base station.
[0061] Subsequently, the prediction unit 245 identifies areas or cells at risk of the call releases due to one of a poor signal strength, congestion, and resource unavailability in the network 105. The prediction unit 245 monitors signal strength metrics collected from user devices and the gNBs. If the signal strength falls below the predefined threshold, then the prediction unit 245 identifies areas or cells at risk of the call releases due to at least one of the poor signal strengths and indicates that users may struggle to maintain the stable connection, leading to potential call drops. The predefined threshold represents the minimum signal strength required for reliable communication in the network 105 when the predefined threshold is exceeded, users may expect the high- quality connection. For example, if the cell shows a consistent drop in signal quality during peak usage times, the prediction unit 245 may be flagged as at risk. The prediction unit 245 assesses traffic load metrics, which include the number of active users and the volume of data being transmitted. The congestion refers to the situation where the demand for network resources exceeds the available capacity, leading to a degradation of service quality. The congestion results in delays, dropped calls, or failures in data transmission, as the struggles to accommodate the number of active users and the data requirements. The example of congestion includes, but not limited to, mobile network congestion, wifi network congestion, data center congestion, and transport network congestion. Further,
Y1 the prediction unit 245 evaluates the availability of network resources, such as bandwidth and processing power. When the availability of network resources is insufficient to handle the current load, calls may be released because the network cannot maintain the required quality of service. Overloading of the gNBs or temporary unavailability of certain resources due to the maintenance or other issues can lead to such situations.
[0062] In one embodiment, upon prediction, the one or more models is retrained based on updated data received from the base stations due to which the one or more models are getting advances and leads to increase in the accuracy of the prediction. In particular, the updated data is collected from the base stations on periodic basis. The updated data is fed to the one or more models by the model training unit 235, so that based on the updated data, the model is trained again.
[0063] FIG. 3 is an exemplary block diagram of an architecture implemented in the system of the FIG. 2, according to one or more embodiments of the present invention;
[0064] The architecture 300 includes a probing unit 305, a data consumer 310, a processing hub 315, a data integration 320, a data pre-processing 325, a model training 330, a prediction module 335, a processing hub interface 340, and the database 220.
[0065] The process begins with the probing unit 305, which acts as the data collection component, gathering information from gNodeBs. The probing unit 305 data includes key features such as, but not limited to, timestamps, call parameters, geographic coordinates, and network load metrics. After the data collection, the data is passed to the data integration 320, where the collected data is aggregated and organized for further analysis. The data aggregation and organization process performed by the data integration 320 and consolidates the data collected from the gNBs preparing for further analysis like feature extraction and model training. Extraction of features is essential for understanding patterns in the network 105, such as, but not limited to, identifying potential issues with signal strength or network congestion. [0066] Once the data is received from the probing unit 305, the data integration 320 undergoes aggregation, where data from the gNBs is combined and aligned correctly. The aggregation process may include, but not limited to, merging data, performing initial consistency checks, and preparing for the pre-processing. The data integration 320 ensures that the data is consistent and ready for the one or more machine learning operations, where the data collected from the gNBs is aggregated for subsequent analysis.
[0067] Upon data integration, the data moves to the data pre-processing 325, where the data pre-processing 325 performs essential tasks such as, but not limited to, data definition, data normalization, and data cleaning are performed. The data definition ensures that the data filed are clearly defined and in the proper format. The data normalization involves scaling the data to ensure that all features are on a consistent scale, which improves the performance of the one or more machine learning models. The data cleaning and performed involves removing or handling missing data and outliers to ensure the quality of the database 220. The present step preparing the database 220 for model training by ensuring that is clean, normalized, and ready for use.
[0068] Thereafter, the data is split into the training dataset and the testing dataset. The data ensures that the one or more machine learning models are trained on one portion of the data while another portion is reserved for evaluating the performance of the models. The training dataset is used to build the predictive model, and the testing dataset is used to evaluate the accuracy of the predictions.
[0069] Upon splitting the data, the model training 330 uses the one or more machine learning models to train on the training dataset. Specifically, the model training 330 involves analyzing the data to extract patterns, trends, and relationships from the features, such as, but not limited to, call parameters, geographic coordinates, and network load metrics. During the model training 330 phase, the one or more machine learning model learns from the data to identify underlying patterns that may later be used to make predictions or detect anomalies. The choice of the one or more machine learning algorithms applied at the stage depends on the complexity and volume of the data. [0070] After the model is trained, the prediction module 335 is responsible for detecting unusual patterns or relationships between the features in the data. The prediction module 335 identifies potential anomalies or deviations from normal behavior such as, but not limited to, increased call drops, poor signal strength, congestion, or resource unavailability. Based on the detections, the gNodeBs or cells at the highest risk of experiencing issues such as, call releases, are predicted.
[0071] The prediction module 335 is then tested using the testing dataset to assess its accuracy. The prediction module 335 evaluates how well the model performs at identifying affected cells and the gNBs by comparing predictions with actual data. The various metrics such as, but not limited to, precision, recall, accuracy, and the root mean square error to assess the model’s effectiveness.
[0072] Both the training results and predictions generated by the one or more machine learning models are stored in the database 220. The database 220 holds historical data and can be used to re-train models over time or for further analysis. The database 220 ensures that there is the persistent record of the predictions, which can be accessed for future comparisons or investigations.
[0073] Thereafter, the anomaly detected results are sending and predictions to the data consumers 310. The data consumers 310 could be network operators or automation systems that monitor network performance and take proactive actions based on the predictions. For example, the data consumers may use this information to adjust network resources, troubleshoot poor signal areas, or mitigate congestion before leads to significant service degradation.
[0074] The processing hub interface 340 serves as the central communication link between various components within the architecture 300. The processing hub interface 340 role is to manage the flow of the data between modules such as the data integration 320, the data pre-processing 325, the model training 330, and the prediction module 335. The processing hub interface 340 ensures that data is transferred efficiently between the processing hub 315, enabling seamless interaction.
[0075] FIG. 4 is an exemplary flowchart diagram of predicting the worst affected cells or the gNBs in the network 105, according to one or more embodiments of the present invention.
[0076] At step 405, the process begins with the collection and aggregation of the raw data from the gNBs in the network 105. The data includes, but is not limited to, call release reasons, timestamps, call parameters, network load metrics, and other relevant information. The purpose of this step is to gather essential information that will later be used for training and prediction.
[0077] At step 410, upon collecting and aggregation of the data, the raw data is processed to extract key features such as, but not limited to, timestamps, geographic coordinates, call parameters, and network load metrics. The data is then divided into the training dataset and the testing dataset. This step ensures the creation of meaningful input features for the one or more machine learning models and prepares the data for training and testing.
[0078] At step 415, once the data is processed to extract the features, the training dataset is used to train the one or more machine learning models. The prediction module 335 identifies the patterns, relationships, and connections between variables in the data which includes, but not limited to, factors contributing to call releases or network congestion. The objective is to develop the prediction module 335 capable of recognizing patterns linked to potential network issues, thereby enabling early detection and response to problems.
[0079] At step 420, upon training the model, the model training 330 is evaluated against the testing dataset to assess the accuracy and performance. Various metrics, such as precision, recall, and root mean square error, are used to determine how effectively the model can predict the worst-affected cells or gNodeBs. The goal is to ensure the model's capability to accurately identify potential network issues and forecast which cells or the gNBs are most at risk.
[0080] At step 425, upon evaluating against the testing dataset to assess the accuracy and performance, the trained model produces insights by identifying cells or the gNBs that are at the highest risk of facing issues, such as call releases. These insights are based on the unusual patterns or anomalies detected during the training and testing phases. The objective is to predict the worst-affected areas in the network 105 and highlight potential problems before they result in disruptions.
[0081] At step 430, upon predicting the wort affected cell or the gNbs in the network 105, the one or more machine learning model continuously adapts based on new data and changing network conditions to ensure the one or more machine learning model remains effective in predicting network issues as conditions evolve over time.
[0082] FIG. 5 is a schematic representation of a method 500 of predicting worst affected cells or the gNBs in the network 105, according to one or more embodiments of the present invention. For the purpose of description, the method 500 is described with the embodiments as illustrated in FIG. 2 and should nowhere be construed as limiting the scope of the present disclosure.
[0083] At step 505, the method 500 includes the step of collecting data from the gNBs and aggregating the data. The data includes, but not limited to, signal strength metrics, traffic data, network load metrics, mobility patterns, and QoS.
[0084] At step 510, the method 500 includes the step of extracting features from the data. The features include one of the timestamps, call parameters, geographic coordinates, and network load metrics.
[0085] The step 515, the method 500 includes the step of splitting the data into the training dataset and the testing dataset. The training dataset is the subset of the collected data used to train the one or more machine learning model. The training dataset applied to the model to identify patterns and relationships in the data and adjusts the one or more machine learning model parameters to minimize prediction errors based on the features and labels in the training dataset. The testing dataset is a separate subset of the collected data used to evaluate the performance of the trained model. The model predicts outcomes based on the features in the testing dataset, and the results are compared to the actual labels to measure accuracy, precision, recall, and other performance metrics.
[0086] At step 520, the method 500 includes the step of training the one or more machine learning models using dataset. The process of training the dataset is repeated over multiple epochs, sometimes using batch training for efficiency, while validation sets monitor performance to prevent overfitting. After evaluating the model's accuracy and precision on the testing dataset, the trained model is deployed in the network management system to enable real-time predictions and proactive management of the network 105.
[0087] At step 525, the method 500 includes the step of detecting unusual patterns or relationships between the features in the data. By collecting metrics from sources like the gNBs and user devices, including signal strength and latency. The anomaly detection unit 240 extracts feature and establishes the normal behavior baseline using historical data, then applies algorithms such as, but not limited to, statistical methods, isolation forests, and clustering techniques to detect deviations in real-time. Upon detecting an anomaly, the anomaly detection unit 240 unit generates alerts for network operators, detailing the issue and affected the gNBs. Further, the feedback loop allows operators to validate anomalies, refining detection algorithms and improving future accuracy to maintain network performance and reliability.
[0088] At step 530, the method 500 includes the step of assessing the unusual patterns or relationships detected for predicting worst affected cells or the gNBs in the network that are likely to experience call releases. The prediction unit 245 assesses unusual patterns or relationships in network data to predict which cells or the gNBs are likely to experience call releases. The prediction unit 245 analyzes various metrics such as, but not limited to, signal strength, user mobility patterns, and traffic load metrics to identify correlations with past call drops. By employing the one or more machine learning algorithms, the prediction unit 245 recognizes patterns that indicate potential connectivity issues, such as low signal strength combined with high traffic. Continuous real-time monitoring allows the prediction unit 245 to apply predictive models trained on historical data to forecast at-risk the gNBs. The prediction unit 245 generating alerts for network operators when unusual conditions arise, like sudden drops in signal quality or congestion during peak hours.
[0089] The present invention further discloses a non-transitory computer-readable medium having stored thereon computer-readable instructions. The computer-readable instructions are executed by the processor 205. The processor 205 is configured to collect data from the base stations and aggregate the data. Further, the processor 205 is configured to extract features from the data and split the data into the training dataset and the testing dataset. Further, the processor 205 is configured to detect unusual patterns or relationships between the features in the data. Further, the processor 205 is configured to assess the unusual patterns or relationships detected for predicting worst affected cells or the base stations in the network 105 that are likely to experience call releases.
[0090] A person of ordinary skill in the art will readily ascertain that the illustrated embodiments and steps in description and drawings (FIG.1-5) are set out to explain the exemplary embodiments shown, and it should be anticipated that ongoing technological development will change the manner in which particular functions are performed. These examples are presented herein for purposes of illustration, and not limitation. Further, the boundaries of the functional building blocks have been arbitrarily defined herein for the convenience of the description. Alternative boundaries can be defined so long as the specified functions and relationships thereof are appropriately performed. Alternatives (including equivalents, extensions, variations, deviations, etc., of those described herein) will be apparent to persons skilled in the relevant art(s) based on the teachings contained herein. Such alternatives fall within the scope and spirit of the disclosed embodiments. [0091] The present disclosure includes technical advancement by improving predictive accuracy with advanced the machine learning models and diverse data sources, such as call parameters and geographic coordinates. Automated anomaly detection and targeted risk identification address potential issues like poor signal strength and congestion. The invention robust the data pre-processing ensures high-quality inputs for reliable predictions, while systematic performance assessments facilitate continuous improvement. The scalable architecture optimizes resource allocation, reducing costs and boosting user satisfaction, and computer-readable instructions allow for easy integration across platforms, promoting widespread adoption and overall efficiency.
[0092] The present invention offers multiple advantages by enabling proactive issue resolution, allowing operators to identify and address potential problems like poor signal strength or congestion before they escalate. The results in improved network quality, fewer call drops, and higher customer satisfaction. By anticipating issues, the invention significantly reduces service disruptions, leading to more reliable services. It also optimizes resource utilization, ensuring efficient allocation of bandwidth and capacity, ultimately driving cost savings and fostering greater customer loyalty.
[0093] The present invention offers multiple advantages over the prior art and the above listed are a few examples to emphasize on some of the advantageous features. The listed advantages are to be read in a non-limiting manner.
REFERENCE NUMERALS
[0094] Environment- 100
[0095] User Equipment (UE)- 110
[0096] Server- 115
[0097] Network- 105
[0098] System -120
[0099] Processor- 205
[00100] Memory- 210
[00101] User interface- 215
[00102] Database - 220
[00103] Probing unit - 225
[00104] Pre-processing unit - 230
[00105] Model training unit - 235
[00106] Anomaly detection unit - 240
[00107] Prediction unit - 245
[00108] Probing unit - 305
[00109] Data consumers - 310
[00110] Processing hub - 315
[00111] Data integration - 320
[00112] Data pre-processing - 325
[00113] Model training - 330
[00114] Prediction module - 335
[00115] Processing hub interface - 340

Claims

CLAIMS We Claim:
1. A method (500) of predicting worst affected cells or base stations in a network (105), the method (500) comprising the steps of: collecting, by a one or more processor (205), data from the base stations and aggregating the data; extracting, by the one or more processor (205), features from the data; splitting, by the one or more processor (205), the data into a training dataset and a testing dataset; detecting, by the one or more processor (205), unusual patterns or relationships between the features in the data; and assessing, by the one or more processor (205), the unusual patterns or relationships detected for predicting the worst affected cells or the base stations in the network (105).
2. The method (500) as claimed in claim 1, the features comprising one of timestamps, call parameters, geographic coordinates, and network load metrics.
3. The method (500) as claimed in claim 1 , comprising pre-processing, by the one or more processor (205), the data for data definition, data normalization, or data cleaning.
4. The method (500) as claimed in claim 1, comprising annotating, by the one or more processor (205), the data with a target variable for identifying the worst affected cells or the base stations in the network (105).
5. The method (500) as claimed in claim 1, comprising assessing, by the one or more processor (205), performance of the one or more machine learning models using the testing dataset.
6. The method (500) as claimed in claim 1, comprising, identifying, by the one or more processor (205), areas or cells at risk of the call releases due to one of a poor signal strength, congestion, and resource unavailability in the network (105).
7. The method (500) as claimed in claim 1, wherein the one or more processor (205) trains one or more machine learning models using the training dataset.
8. A system (120) for predicting worst affected cells or base stations in a network (105), the system (120) comprising: a probing unit (225) configured to collect data from base stations and aggregate the data; a pre-processing unit (230) configured to extract features from the data, and split the data into a training dataset and a testing dataset; an anomaly detection unit (240) configured to detect unusual patterns or relationships between the features in the data; and a prediction unit (245) configured to assess the unusual patterns or relationships detected for predicting worst affected cells or the base stations in the network (105).
9. The system (120) as claimed in claim 8, wherein the features comprise one of timestamps, call parameters, geographic coordinates, and network load metrics.
10. The system (120) as claimed in claim 8, wherein the pre-processing unit (230) pre-processes the data for data definition, data normalization, or data cleaning.
11. The system (120) as claimed in claim 8, wherein the pre-processing unit (230) annotates the data with a target variable for identifying the worst affected cells or the base stations in the network (105).
12. The system (120) as claimed in claim 8, wherein the anomaly detection unit (240) assesses performance of the one or more machine learning models using a testing dataset.
13. The system (120) as claimed in claim 8, wherein the prediction unit (245) identifies areas or cells at risk of the call releases due to one of a poor signal strength, congestion, and resource unavailability in the network (105).
14. The system (120) as claimed in claim 8, wherein a model training unit (235) is configured to train one or more machine learning models using the training dataset.
15. A non-transitory computer-readable medium having stored thereon computer- readable instructions that, when executed by a processor (205), cause the processor (205) to: collect data from base stations and aggregate the data; extract features from the data, and split the data into a training dataset and a testing dataset; detect unusual patterns or relationships between the features in the data; and assess the unusual patterns or relationships detected for predicting worst affected cells or the base stations in the network (105).
PCT/IN2024/051995 2023-10-07 2024-10-07 Method and system of predicting worst affected cells or base stations in a network Pending WO2025074431A1 (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
IN202321067386 2023-10-07
IN202321067386 2023-10-07

Publications (1)

Publication Number Publication Date
WO2025074431A1 true WO2025074431A1 (en) 2025-04-10

Family

ID=95284296

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/IN2024/051995 Pending WO2025074431A1 (en) 2023-10-07 2024-10-07 Method and system of predicting worst affected cells or base stations in a network

Country Status (1)

Country Link
WO (1) WO2025074431A1 (en)

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20200169895A1 (en) * 2018-11-26 2020-05-28 Samsung Electronics Co., Ltd. Methods and apparatus for coverage prediction and network optimization in 5g new radio networks
CN114154765A (en) * 2022-01-07 2022-03-08 中国联合网络通信集团有限公司 Cell prediction method, cell prediction device, electronic device and storage medium

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20200169895A1 (en) * 2018-11-26 2020-05-28 Samsung Electronics Co., Ltd. Methods and apparatus for coverage prediction and network optimization in 5g new radio networks
CN114154765A (en) * 2022-01-07 2022-03-08 中国联合网络通信集团有限公司 Cell prediction method, cell prediction device, electronic device and storage medium

Similar Documents

Publication Publication Date Title
CN111212038B (en) Open data API gateway system based on big data artificial intelligence
JP7145764B2 (en) Network advisor based on artificial intelligence
US10581667B2 (en) Method and network node for localizing a fault causing performance degradation of a service
WO2020135510A1 (en) Burst load prediction method and device, storage medium and electronic device
US20240333615A1 (en) Network analysis using dataset shift detection
TW202016805A (en) System and method of learning-based prediction for anomalies within a base station
CN119892603A (en) Method, system, equipment and medium for preventing and repairing cross-regional network faults
WO2025223075A1 (en) Data processing method and computer program product
CN118708461A (en) An intelligent central control system for commercial tablet computers
Shi et al. Towards automatic troubleshooting for user-level performance degradation in cellular services
US20220014446A1 (en) Network device association with network management system
WO2025074431A1 (en) Method and system of predicting worst affected cells or base stations in a network
US20250330839A1 (en) Protocol-aware multi-domain network optimization using deep learning
WO2025074444A1 (en) Method and system of monitoring a network
WO2025074430A1 (en) Method and system of management of network functions (nfs) in a network
WO2025079108A1 (en) System and method for anomaly detection in a network
WO2025017715A1 (en) System and method for detecting anomalies in a communication network
WO2025099768A1 (en) System and method for monitoring service quality in the network slices
WO2025120657A1 (en) Method and system for predicting clear code failures count in the network function (nf)
WO2025074413A1 (en) Method and system for identifying anomalies in events subscription service management
WO2025074410A1 (en) Method and system for forecasting events in a network
WO2025079094A1 (en) System and method for forecasting one or more alerts
WO2025074417A1 (en) Method and system for generating the one or more predictions
WO2025074441A1 (en) Method and system for analysing wireless local area network (wlan) performance and subscription management
WO2025079103A1 (en) Method and system for anomaly detection in a network

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 24874255

Country of ref document: EP

Kind code of ref document: A1