Method and apparatus for intelligent clustering of cells and network entities using network intelligence-as-a-service

Abstract

Described are systems, apparatuses and methods for a machine learning k- means clustering in an Operations, Administration and Maintenance (OAM) module of a Radio Access Network to generate clusters of strongly-interfering cells together, while splitting apart weekly-interfering cells across different clusters.

Description

METHOD AND APPARATUS FOR INTELLIGENT CLUSTERING OF CELLS AND NETWORK ENTITIES USING NETWORK INTELLIGENCE-AS-A-SERVICE CROSS REFERENCE TO RELATED APPLICATIONS

[0001] This application is an International Application and claims foreignpriority to Indian Provisional Patent Application No.: 202441055175, filed on July 19, 2024, the entirety of which is incorporated herein by reference. DESCRIPTION OF THE RELATED TECHNOLOGY a. Field of the Disclosure

[0002] The present disclosure relates to systems and methods for radioaccess networks. The present disclosure is related to the design, operation, administration and management of various network elements of 4G, 5G, and further generations of a radio access network system. SUMMARY OF THE DISCLOSURE

[0003] Described are implementations of a computer system, computersystem components, computer apparatus, a method, and computer program products configured to execute program instructions for the method for radio access network, and operation, administration and management of various network elements of 4G, 5G, and further generations of the radio access network system. The method is performed by a computer system that comprises one or more processors and a computer-readable storage medium encoded with instructions executable by at least one of the processors and operatively coupled to at least one of the processors.

[0004] In an implementation, an apparatus comprises: method for anArtificial Intelligence and Machine Learning (AI-ML) Operations, Administration and Maintenance (OAM) cell optimization in a RAN comprising:modeling key performance indicators of a cell using the cell’s OAM fault, configuration, accounting, performance, and security(FCAPs) and the OAM FCAPS of other strongly interfering cells; and clustering strongly-interfering cells together, while splitting apart weekly- interfering cells across different clusters, wherein the clustering is modeled and executed by a k-means clustering algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0005] FIG. 1 is a block diagram of a system architecture.

[0006] FIG. 2 shows an example of a User Plane Stack.

[0007] FIG. 3 shows an example of a Control Plane Stack.

[0008] FIG. 4A shows an example of high-level NG-RAN including a gNB CUand DU.

[0009] FIG. 4B shows an example of a Separation of CU-CP (CU-Control Plane)and CU-UP (CU-User Plane) in a 5G gNB.

[0010] FIG. 4C shows an example of a Separation of CU-CP (CU-Control Plane)and CU-UP (CU-User Plane) in a 4G ng-eNB.

[0011] FIG. 5 shows a DL (Downlink) Layer 2 Structure.

[0012] FIG. 6 shows an exemplary logical flow for implementing an RBallocation policy.

[0013] FIG. 7 shows an L2 Data Flow example.

[0014] FIG. 8A shows an example of an O-RAN architecture.

[0015] FIG. 8B shows an example of an O-RAN OAM architecture.

[0016] FIG. 8C shows an example of interfaces and network functions for anO-RAN OAM architecture.

[0017] FIG. 9A illustrates a PDU Session architecture comprising of multipleDRBs and multiple QoS Flows.

[0018] FIG. 9B illustrates a flow for PDU sessions, DRBs and GTP-U Tunnelsacross CU and DU.

[0019] FIG. 9C illustrates a CU and DU view on PDU session, DRBs and GTP-Utunnels for a 5G network architecture.

[0020] FIG. 10 shows a Resource Allocation MAC Scheduler, DL Data, andFlow Control Feedback for 5G Network.

[0021] FIG. 11A is an EN-DC architecture.

[0022] FIG. 11B is a NG-RAN architecture.

[0023] FIG. 12A illustrates a Non-Standalone NSA architecture and logicalflow.

[0024] FIG. 12B illustrates a Non-Standalone NSA architecture and logicalflow.

[0025] FIG. 13 illustrates a model of a clustering problem as a graph-theoretic problem.

[0026] FIG. 14 is a graph charting a performance evaluation of clusteringalgorithms. DETAILED DESCRIPTION OF THE DISCLOSURE

[0027] Reference is made to Third Generation Partnership Project (3GPP) andthe Internet Engineering Task Force (IETF) and related standards bodies in accordance with embodiments of the present disclosure. The present disclosureemploys abbreviations, terms and technology defined in accord with Third Generation Partnership Project (3GPP) and / or Internet Engineering Task Force (IETF) technology standards and papers, including the following standards and definitions. 3GPP and IETF technical specifications (TS), standards (including proposed standards), technical reports (TR) and other papers are incorporated by reference in their entirety hereby, define the related terms and architecture reference models that follow.

[0028] O-RAN.WG4.MP.0-R003-v13.00

[0029] 3GPP TS 23.203 V 17.2.02021-12-23

[0030] 3GPP TS 23.501 V 18.1.02023-04-05

[0031] 3GPP TS 28.552 V 18.6.02024-04-05

[0032] 3GPP TS 38.300 V 17.4.003-28-2023

[0033] 3GPP TS 36.321 V 17.4.02023-03-29

[0034] 3GPP TS 36.323 V 17.2.02023-01-13

[0035] 3GPP TS 36.331 V. 18.1.2024-04-01

[0036] 3GPP TS 38.321 V 17.4.02023-03-29

[0037] 3GPP TS 38.331 V 18.1.02024-04-01

[0038] 3GPP TS 38.401 V 17.4.02023-04-03

[0039] 3GPP TS 38.501 V 18.1.02023-04-05

[0040] 3GPP TS 38.42517.3.0, 2023-04-03

[0041] Acronyms3GPP: Third generation partnership project 5GC: 5G Core Network 5G NR: 5G New Radio 5QI: 5G QoS IdentifierACK: Acknowledgement ACLR: Gain and Adjacent Channel Leakage Ratio ADC: Analog-to-Digital Converter AM: Acknowledged Mode AMF: Access Mobility Function AMBR: Aggregate Maximum Bit Rate APN: Access Point Name ARP: Allocation and Retention Priority ASIC: Application Specific Integrated Circuit AWGN: Additive White Gaussian Noise BFW: Beamforming weight BS: Base Station CNF: Cloud-Native Network Function CP: Control Plane C-RAN: cloud radio access network CU: Centralized unit CU-CP: Centralized Unit – Control Plane CU-UP: Centralized Unit – User Plane CQI: Channel Quality Indicator DAC: Digital-to-Analog Converter DC: Dual Connectivity DCI: Downlink Control Information DDDS: DL Data Delivery Status DFE: Digital Front EndDL: Downlink DMRS: Demodulation Reference Signal DNN: Data Network Name DRB: Data Radio Bearer DU: Distributed unit eNB: evolved Node B eMBB: Enhanced Mobile Broadband EPC: Evolved Packet Core EN-DC EP: Endpoint Pod E-UTRAN: Evolved Universal Terrestrial Radio Access Network E-RAB: E-UTRAN Radio Access Bearer IoT: Internet of Things IP: Internet Protocol IWF: Interworking Function GBR: Guaranteed Bit Rate gNB: gNodeB (5G base station) GTP-U: General Packet Radio Service (GPRS) Tunnelling Protocol – User Plane GW: Gateway HA: High Availability L1: Layer 1 L2: Layer 2 L3: Layer 3 LC: Logical ChannelLTE: Long Term Evolution (4G) MAC: Medium Access Control MIMO: multiple-in multiple-out MME: Mobility Management Entity MR-DC: Multi-Radio Dual Connectivity M-plane: Management plane interface between SMO and O-RU NACK: Negative Acknowledgement NAS: Non-Access Stratum NB: Narrowband Near-RT RIC: Near-Real-Time RIC ng: Next Generation NIaaS: Network Intelligence as a Service NMS: Network Management System NR: New Radio NR-U: New Radio – User Plane NSA: Non-Standalone Architecture One-CU-UP: One O-RAN compliant Centralized Unit User Plane OFDM: orthogonal frequency-division multiplexing O-RAN: Open Radio Access Network PA: Power Amplifier PDB: Packet Delay Budget PDCP : Packet Data Convergence Protocol PDU: Protocol Data Unit PDCCH: Physical Downlink Control ChannelPDSCH: Physical Downlink Shared Channel PHY: Physical Layer PRG: Physical Resource block Group PUCCH: Physical Uplink Control Channel PUSCH: Physical Uplink Shared Channel QCI: QoS Class Identifier QFI: QoS Flow Id QFI: QoS Flow Identifier QoS : Quality of Service RAT: Radio Access Technology RB: Resource Block RDI: Reflective QoS Flow to DRB Indication RIC: RAN Intelligent Controller RLC: Radio Link Control RLC-AM: RLC Acknowledged Mode RLC-UM: RLC Unacknowledged Mode RMM: Radio resource management RQI: Reflective QoS Indication RRC: Radio Resource Control RSRP: Reference Signal Received Power RU: Radio Unit SA: Standalone Architecture SCTP: Stream Control Transmission Protocol SDAP: Service Data Adaptation ProtocolSDU: Service Data Unit S-GW: Serving Gateway SINR: Signal-to-Interference and Noise Ratio SMO: Service Management and Orchestration system SN: Secondary Node SR: Scheduling Request SRS: Sounding Reference Signal TCP: Transmission Control Protocol TEID: Tunnel Endpoint Identifier U-plane: User plane UPF: User Plane Function UE: user equipment UL: uplink UM: Unacknowledged Mode URLLC: Ultra Reliance Low Latency Communication

[0042] Described are implementations of technology for a cloud-based RadioAccess Networks (RAN), where a significant portion of the RAN layer processing is performed at a central unit (CU) and a distributed unit (DU). Both CUs and DUs are also known as the baseband units (BBUs). CUs are usually located in the cloud on commercial off the shelf servers, while DUs can be distributed. The RF and real-time critical functions can be processed in the remote radio unit (RU).

[0043] RAN Architectures

[0044] FIG. 1 is a block diagram of a system 100 for implementations asdescribed herein. System 100 includes a NR UE 101 and a NR gNB 106. The NR UE and NR gNB 106 are communicatively coupled via a Uu interface 120.

[0045] NR UE 101 includes electronic circuitry, namely circuitry 102, thatperforms operations on behalf of NR UE 101 to execute methods described herein. Circuity 102 can be implemented with any or all of (a) discrete electronic components, (b) firmware, and (c) a programmable circuit 102A.

[0046] NR gNB 106 includes electronic circuitry, namely circuitry 107, thatperforms operations on behalf of NR gNB 106 to execute methods described herein. Circuity 107 can be implemented with any or all of (a) discrete electronic components, (b) firmware, and (c) a programmable circuit 107A.

[0047] Programmable circuit 107A, which is an implementation of circuitry107, includes a processor 108 and a memory 109. Processor 108 is an electronic device configured of logic circuitry that responds to and executes instructions. Memory 109 is a tangible, non-transitory, computer-readable storage device encoded with a computer program. In this regard, memory 109 stores data and instructions, i.e., program code, that are readable and executable by processor 108 for controlling operations of processor 108. Memory 109 can be implemented in a random-access memory (RAM), a hard drive, a read only memory (ROM), or a combination thereof. One of the components of memory 109 is a program module, namely module 110. Module 110 has instructions for controlling processor 108 to execute operations described herein on behalf of NR gNB 106.

[0048] The term "module" is used herein to denote a functional operationthat can be embodied either as a stand-alone component or as an integrated configuration of a plurality of subordinate components. Thus, each of module 105 and 110 can be implemented as a single module or as a plurality of modules that operate in cooperation with one another.

[0049] While modules 110 are indicated as being already loaded intomemories 109, and module 110 can be configured on a storage device 130 for subsequent loading into their memories 109. Storage device 130 is a tangible, non- transitory, computer-readable storage device that stores module 110 thereon.Examples of storage device 130 include (a) a compact disk, (b) a magnetic tape, (c) a read only memory, (d) an optical storage medium, (e) a hard drive, (f) a memory unit comprising of multiple parallel hard drives, (g) a universal serial bus (USB) flash drive, (h) a random-access memory, and (i) an electronic storage device coupled to NR gNB 106 via a data communications network.

[0050] Uu Interface (120) is the radio link between the NR UE and NR gNB,which is compliant to the 5G NR specification.

[0051] UEs 101 can be dispersed throughout a wireless communicationnetwork, and each UE can be stationary or mobile. A UE includes: an access terminal, a terminal, a mobile station, a subscriber unit, a station, and the like. A UE can also include be a cellular phone (e.g., a smart phone), a personal digital assistant (PDA), a wireless modem, a wireless communication device, a handheld device, a laptop computer, a cordless phone, a wireless local loop (WLL) station, a tablet, a camera, a gaming device, a drone, a robot / robotic device, a netbook, a smartbook, an ultrabook, a medical device, medical equipment, a healthcare device, a biometric sensor / device, a wearable device such as a smart watch, smart clothing, smart glasses, a smart wristband, and / or smart jewelry (e.g., a smart ring, a smart bracelet, and the like), an entertainment device (e.g., a music device, a video device, a satellite radio, and the like), industrial manufacturing equipment, a global positioning system (GPS) device, or any other suitable device configured to communicate via a wireless or wired medium. UEs can include UEs considered as machine-type communication (MTC) UEs or enhanced / evolved MTC (eMTC) UEs. MTC / eMTC UEs that can be implemented as IoT UEs. IoT UEs include, for example, robots / robotic devices, drones, remote devices, sensors, meters, monitors, cameras, location tags, and the like, that can communicate with a BS, another device (e.g., remote device), or some other entity. A wireless node can provide, for example, connectivity for or to a network (e.g., a wide area network such as Internet or a cellular network) via a wired or wireless communication link.

[0052] One or more UEs 101 in the wireless communication network can be anarrowband bandwidth UE. As used herein, devices with limited communication resources, e.g. smaller bandwidth, are considered as narrowband UEs. Similarly, legacy devices, such as legacy and / or advanced UEs, can be considered as wideband UEs. Wideband UEs are generally understood as devices that use greater amounts of bandwidth than narrowband UEs.

[0053] The UEs 101 are configured to connect, for example, communicativelycouple, with an or RAN. In embodiments, the RAN can be an NG RAN or a 5G RAN, an E-UTRAN, an MF RAN, or a legacy RAN, such as a UTRAN or GERAN. The term “NG RAN” or the like refers to a RAN 110 that operates in an NR or 5G system, the term “E-UTRAN” or the like refers to a RAN that operates in an LTE or 4G system, and the term “MF RAN” or the like refers to a RAN that operates in an MF system 100. The UEs 101 utilize connections (or channels), respectively, each of which comprises a physical communications interface or layer. The connections can comprise several different physical DL channels and several different physical UL channels. As examples, the physical DL channels include the PDSCH, PMCH, PDCCH, EPDCCH, MPDCCH, R-PDCCH, SPDCCH, PBCH, PCFICH, PHICH, NPBCH, NPDCCH, NPDSCH, and / or any other physical DL channels mentioned herein. As examples, the physical UL channels include the PRACH, PUSCH, PUCCH, SPUCCH, NPRACH, NPUSCH, and / or any other physical UL channels mentioned herein.

[0054] The RAN can include one or more AN nodes or RAN nodes. Theseaccess nodes can be referred to as BS, gNBs, RAN nodes, eNBs, NodeBs, RSUs, MF- APs, TRxPs or TRPs, and so forth, and comprise ground stations (e.g., terrestrial access points) or satellite stations providing coverage in a geographic area (e.g., a cell). The term “NG RAN node” or the like refers to a RAN node that operates in an NR or 5G system (e.g., a gNB), and the term “E-UTRAN node” or the like refers to a RAN node that operates in an LTE or 4G system (e.g., an eNB). According to various embodiments, the RAN nodes can be implemented as one or more of a dedicated physical device such as a macrocell base station, and / or a low power base stationfor providing femtocells, picocells or other like cells having smaller coverage areas, smaller user capacity, or higher bandwidth compared to macrocells.

[0055] In some embodiments, all or parts of the RAN nodes can beimplemented as one or more software entities running on server computers as part of a virtual network, which can be referred to as a CRAN and / or a vBBU. In these embodiments, the CRAN or vBBU can implement a RAN function split, such as a PDCP split where RRC and PDCP layers are operated by the CRAN / vBBU and other L2 protocol entities are operated by individual RAN nodes; a MAC / PHY split where RRC, PDCP, RLC, and MAC layers are operated by the CRAN / vBBU and the PHY layer is operated by individual RAN nodes; or a “lower PHY” split where RRC, PDCP, RLC, MAC layers and upper portions of the PHY layer are operated by the CRAN / vBBU and lower portions of the PHY layer are operated by individual RAN nodes. This virtualized framework allows the freed-up processor cores of the RAN nodes to perform other virtualized applications. In some implementations, an individual RAN node can represent individual gNB-DUs that are connected to a gNB-CU 151via individual F1 interfaces. In these implementations, the gNB-DUs can include one or more remote radio heads (RRH), and the gNB-CU 151can be operated by a server that is located in the RAN or by a server pool in a similar manner as the CRAN / vBBU. One or more of the RAN nodes can be next generation eNBs (ng-eNBs), which are RAN nodes that provide E-UTRA user plane and control plane protocol terminations toward the UEs 101, and are connected to a 5GC via an NG interface. In MF implementations, the MF-APs are entities that provide MultiFire radio services, can be similar to eNBs in an 3GPP architecture.

[0056] In some implementations, access to a wireless interface can bescheduled, where a scheduling entity (e.g.: BS, gNB, and the like) allocates bandwidth resources for devices and equipment in its service area or cell. As scheduling entity can be configured to schedule, assign, reconfigure, and release resources for one or more subordinate entities. In some examples, a UE 101 (or other device) can function as master node scheduling entity, scheduling resourcesfor one or more secondary node subordinate entities (e.g., one or more other UEs 101). Thus, in a wireless communication network with a scheduled access to time—frequency resources and having a cellular configuration, a P2P configuration, and a mesh configuration, a scheduling entity and one or more subordinate entities can communicate utilizing the scheduled resources.

[0057] BS or gNB 106 can be equipped with T antennas and UE 101 can beequipped with R antennas, where in general T 1 and R 1. At BS, a transmitprocessor is configured to receive data from a data source for one or more UEs 101 and select one or more modulation and coding schemes (MCS) for each UE based on channel quality indicators (CQIs) received from the UE 101. The BS is configured to process (e.g., encode and modulate) the data for each UE 101 based on the MCS(s) selected for the UE 101, and provide data symbols for all UEs. A transmit processor is also configured to process system information (e.g., for static resource partitioning information (SRPI), and the like) and control information (e.g., CQI requests, grants, upper layer signaling, and the like) and can provide overhead symbols and control symbols. Processor 108 can also generate reference symbols for reference signals (e.g., the cell-specific reference signal (CRS)) and synchronization signals (e.g., the primary synchronization signal (PSS) and the secondary synchronization signal (SSS)). A transmit (TX) multiple-input multiple- output (MIMO) processor can be configured perform spatial processing (e.g., precoding) on the data symbols, the control symbols, the overhead symbols, and / or the reference symbols, if applicable, and can be configured to provide T output symbol streams to T modulators (MODs). Each modulator can be configured to process a respective output symbol stream (e.g., for OFDM, and the like) to obtain an output sample stream. Each modulator can further be configured to process (e.g., convert to analog, amplify, filter, and upconvert) the output sample stream to obtain a downlink signal. T downlink signals from modulators can be transmitted via T antennas.

[0058] An overview of 5G NR Stacks is as follows. 5G NR (New Radio) userand control plane functions with monolithic gNB 106 are shown in FIG.1 and FIG.2. For the user plane, PHY (physical), MAC (Medium Access Control), RLC (Radio Link Control), PDCP (Packet Data Convergence Protocol) and SDAP (Service Data Adaptation Protocol) sublayers are terminated in the gNB 106 on the network side. For the control plane, RRC (Radio Resource Control), PDCP, RLC, MAC and PHY sublayers are terminated in the gNB 106 on the network side and NAS (Non-Access Stratum) is terminated in the AMF (Access Mobility Function) on the network side. FIG.2 shows an example of a User Plane Stack as described in 3GPP TS 38.300. FIG. 3 shows an example of a Control Plane Stack as described in 3GPP TS 38.300.

[0059] An NG-RAN (NG-Radio Access Network) architecture from 3GPP TS38.401 is described below. F1 is the interface between gNB-CU 151 (gNB – Centralized Unit) and gNB-DU 152 (gNB – Distributed Unit), NG is the interface between gNB-CU 151 (or gNB) and 5GC (5G Core), E1 is the interface between CU- CP (CU-Control Plane) and CU-UP (CU-User Plane), and Xn is interface between gNBs.

[0060] A gNB 106 can comprise a gNB-CU-CP, multiple gNB-CU-UPs andmultiple gNB-DUs. The gNB-CU-CP is connected to the gNB-DU 152 through the F1- C interface and to the gNB-CU-UP through the E1 interface. The gNB-CU-UP is connected to the gNB-DU 152 through the F1-U interface and to the gNB-CU-CP through the E1 interface. One gNB-DU 152 is connected to one gNB-CU-CP and one gNB-CU-UP is connected to one gNB-CU-CP. FIG.4A shows an example of an NG-RAN Architecture as described in 3GPP TS 38.501. FIG.4B shows an example of a Separation of CU-CP (CU-Control Plane) and CU-UP (CU-User Plane) as described in 3GPP TS 38.401.

[0061] A Layer 2 (L2) of 5G NR is split into the following sublayers isdescribed in 3GPP TS 38.300):o Medium Access Control (MAC): The MAC sublayer offers LogicalChannels (LCs) to the RLC sublayer. This layer runs a MAC scheduler to schedule radio resources across different LCs (and their associated radio bearers). oRadio Link Control (RLC): The RLC sublayer offers RLC channels tothe PDCP sublayer. The RLC sublayer supports three transmission modes: RLC-Transparent Mode (RLC-TM), RLC-Unacknowledged Mode (RLC-UM) and RLC-Acknowledgement Mode (RLC-AM). RLC configuration is per logical channel. It hosts ARQ (Automatic Repeat Request) protocol for RLC-AM mode. oPacket Data Convergence Protocol (PDCP): The PDCP sublayer offersRadio Bearers (RBs) to the SDAP sublayer. There are two types of Radio Bearers: Data Radio Bearers (DRBs) for data and Signaling Radio Bearers (SRBs) for control plane. oService Data Adaptation Protocol (SDAP): The SDAP offers QoS Flowsto the 5GC (5G Core). This sublayer provides mapping between a QoS flow and a DRB. It marks QoS Flow Id in DL (downlink) as well as UL (uplink packets).

[0062] FIG. 4B shows an example of a Separation of 4G CU-CP (CU-ControlPlane) and CU-UP (CU-User Plane).The example shown in FIG.4B shows ng-eNB but legacy eNB also applies same architecture.

[0063] FIG. 5 shows a DL (Downlink) Layer 2 Structure as described in 3GPPTS 38.300. FIG.6 shows an UL (uplink) Layer 2 Structure in accord with 3GPP TS38.300. FIG.7 shows an L2 Data Flow example in accord with 3GPP TS 38.300 ([H] denotes headers or subheaders in FIG.7.)

[0064] O-RAN, which is based on disaggregated components and connectedthrough open and standardized interfaces, is based on 3GPP NG-RAN. An overview of O-RAN with disaggregated RAN (CU, DU, and RU), near-real-time RIC 160 and non-real-time RIC is shown in the figure below. Here, DU (Distributed Unit) and CU (Centralized Unit) are typically implemented using COTS (Commercial off-the-shelf) hardware.

[0065] FIGS. 8A-8B show an example of an O-RAN architecture. In FIG. 8A, the CUand the DU are connected using the F1 interface (with F1-C for control plane and F1- U for user plane traffic) over the midhaul (MH) path. One DU could host multiple cells (for example, one DU could host 24 cells) and each cell can support many users. For example, one cell can support 600 RRC connected users and out of these 600, there can be 200 Active users (i.e.; users which have data to send at a given point of time).

[0066] A cell site can comprise multiple sectors and each sector can supportmultiple cells. For example, one site could comprise three sectors and each sector could support 8 cells (with 8 cells in each sector on different frequency bands). One CU-CP could support multiple DUs and thus multiple cells. For example, a CU-CP could support 1000 cells and around 100,000 UEs. Each UE could support multiple DRBs and there could be multiple instances of CU-UP to serve these DRBs. For example, each UE could support 4 DRBs, and 400,000 DRBs (corresponding to 100,000 UEs) can be served by five CU-UP instances (and one CU-CP instance).

[0067] DU can be located in a private data center or it could be located at a cell-site too. CU can also be located in a private data center or even hosted on a public cloud system. DU and CU can be tens of kilometers away. CU can communicate with 5G core system which could also be hosted in the same public cloud system (or could be hosted by a different cloud provider). RU (Radio Unit) is located at cell-site and communicated with DU via a fronthaul (FH) interface.

[0068] The E2 nodes (CU and DU) are connected to the near-real-time RIC 160using the E2 interface. The E2 interface is used to send data (e.g., user, cell, slice KPMs) from the RAN, and deploy control actions and policies to the RAN at near- real-time RIC 160. The application or service at the near-real-time RIC 160 that deploys the control actions and policies to the RAN are called xApps. The near-real- time RIC 160 is connected to the non-real-time RIC 161 using the A1 interface.

[0069] SMO 159 manages multiple regional networks, and O-RAN NFs (O-CUs151, Near-RT RIC 160, O-DUs 152) can be deployed in a regional data center that is connected to multiple cell sites or in cell site which is close to localized O-RU 153 according to network requirements. Since SMO 159 Functions and O-RAN NFs are micro services and deployment-independent logical functions, SMO 159 Functions and O-RAN NFs can be composed of multiple deployment instances deployed in the same O-Cloud or in a different O-Cloud in regional data center, or in cell site according to network requirements (ex. capacity, latency, security, and so on) if the secure connection among SMO159 Functions and O-RAN NFs are available.

[0070] As shown in FIG. 8B, an O-RAN compliant SMO 159 defines TE&IV 163,RAN NF OAM 164, Non-RT RIC 161, and NFO165, FOCOM services 166. SMO 159 interacts with O-RAN NFs with O1 interface. SMO interacts with O-RU 153 with Open FH M-Plane interface and interacts O-Cloud via the O2 interface. O-RAN NF OAM 164 manages O-RAN NF CM, FM, PM and creates O-RAN NF inventory and topology in TE&IV 163. FOCOM / NFO 166 manages O-Cloud resources and creates O-Cloud resources inventory and topology in TE&IV 163. Analytics 172 / rApp 171 in Non-RT RIC 161 can subscribe O-RAN NFs PM / FM, O-Cloud PM / FM data based on O-RAN NF OAM and FOCOM 166 / NFO 165. Analytics / rApp in Non-RT RIC 161 can retrieve the O-RAN NF and O-Cloud resource inventory and topology.

[0071] PDU Sessions, DRBs, QoS Flows

[0072] In 5G networks, PDU connectivity service is a service that providesexchange of PDUs between a UE and a data network identified by a Data NetworkName (DNN). The PDU Connectivity service is supported via PDU sessions that are established upon request from the UE. This DNN defines the interface to a specific external data network. One or more QoS flows can be supported in a PDU session. All the packets belonging to a specific QoS flow have the same 5QI (5G QoS Identifier). FIG.9A illustrates a PDU Session architecture comprising of multiple DRBs. Each DRB can include multiple QoS flows (3GPP TS 23.501). FIG.9B illustrates a flow for PDU sessions, DRBs and GTP-U Tunnels across CU and DU. FIG. 9C illustrates a CU and DU view on PDU session, DRBs and GTP-U tunnels for a 5G network architecture.

[0073] As shown in FIGS. 9A-9C, a PDU session comprises the following:

[0074] A Data Radio Bearer (DRB) is between UE and CU in RAN and a NG-UGTP tunnel which is between CU and UPF (User Plane Function) in the core network. For the 3GPP’s 5G network architecture, the transport connection between the base station (i.e., CU-UP) and User Plane Function (UPF) uses a single GTP-U tunnel per PDU session. The PDU session is identified using GTP-U TEID (Tunnel Endpoint Identifier). The transport connection between DU and CU-UP uses a single GTP-U tunnel per DRB.

[0075] SDAP

[0076] The SDAP (Service Adaptation Protocol) Layer receives downlink datafrom the UPF across the NG-U interface. It maps one or more QoS Flow(s) onto a specific DRB. The SDAP header is present between the UE and the CU (when reflective QoS is enabled), and includes a field to identify the QoS flow in a specific PDU session. GTP-U protocol includes a field to identify the QoS flow and is present between CU and UPF (in the core network).

[0077] Procedures and functionality of the F1-U interface are defined in 3GPPTS 38.425. This F1-U interface supports NR User Plane (NR UP) protocol that provides support for flow control and reliability between CU-UP and DU for each DRB. FIG.10 shows a Resource Allocation (MAC Scheduler), DL Data, and FlowControl Feedback (DDDS) in 5G Networks. Downlink User Data (DUD) PDU are used to carry PDCP PDUs from CU-UP to DU for each DRB. Downlink Data Delivery Status (DDDS) PDU from DU to CU-UP. The DDDS message conveys Desired Buffer Size (DBS), Desired Data Rate (DDR) and some other parameters from DU to CU-UP for each DRB as part of flow control feedback.

[0078] An E-UTRAN architecture is illustrated in FIG. 11A. The E-UTRANcomprises eNBs, providing the E-UTRAN U-plane (PDCP / RLC / MAC / PHY) and control plane (RRC) protocol terminations towards the UE. The eNBs are interconnected with each other by the X2 interface. The eNBs are also connected by the S1 interface to the EPC (Evolved Packet Core), more specifically to the MME (Mobility Management Entity) by the S1-MME interface and to the Serving Gateway (S-GW) by the S1-U interface. The S1 interface supports a many-to-many relation between MMEs / Serving Gateways and eNBs.

[0079] E-UTRAN also supports MR-DC via E-UTRA-NR Dual Connectivity (EN-DC), in which a UE is connected to one eNB that acts as a MN and one en-gNB 106 that acts as a SN. An EN-DC architecture is illustrated in FIG.11A. The eNB is connected to the EPC 140 via the S1 interface and to the en-gNB 106 via the X2 interface. The en-gNB 106 might also be connected to the EPC 140 via the S1-U interface and other en-gNBs 106 via the X2-U interface. In EN-DC, an en-gNB 106 comprises gNB-CU 151and gNB-DU(s)152.

[0080] As shown in FIG. 11B, in the NG-RAN architecture, an NG-RAN node iseither: a gNB, providing NR user plane and control plane protocol terminations towards the UE; or an ng-eNB, providing E-UTRA user plane and control plane protocol terminations towards the UE. (3GPP TS 38.30017.3.0.)

[0081] As shown in FIG. 11B, the gNBs 106 and ng-eNBs are interconnectedwith each other by the Xn interface. The gNBs 106 and ng-eNBs are also connected by the NG interfaces to the 5GC, more specifically to the AMF (Access and Mobility Management Function) by the NG-C interface and to the UPF (User Plane Function) by the NG-U interface.

[0082] The gNB 106 and ng-eNB host functions such as functions for RadioResource Management: Radio Bearer Control, Radio Admission Control, Connection Mobility Control, Dynamic allocation of resources to UEs in both uplink and downlink (scheduling), connection setup and release; session Management; QoS Flow management and mapping to data radio bearers; and Dual Connectivity.

[0083] In an example, control information (e.g., scheduling information) canbe provided for broadcast and / or multicast operation. The UE can monitor different bundle sizes for the control channel depending on the maximum number of repetitions.

[0084] NSA (Non-Standalone) Architecture

[0085] In the 5G Standalone (SA) architectures, 5G gNB 106 communicateswith 5G Core (and not with 4G Evolved Packet Core 140). Similarly, in the 4G architecture, 4G eNB 116 communicates with 4G Evolved Packet Core (EPC) 140 and not with 5G Core.

[0086] E-UTRA-NR Dual Connectivity (EN-DC) is a dominant form of the Non-Standalone (NSA) architecture. As shown in FIG.12A, a 4G eNB 116 as well as 5G gNB 106 use 4G EPC 140 (Evolved Packet Core), and 5G core is not used in this network architecture. In the architecture below, DL (downlink) data goes from 4G EPC 140 to 5G CU-UP 151 where it could be split across two different transmission paths (or network legs): 1) 5G CU-UP 151to 4G DU 142 to NSA UE and 2) 5G CU-UP 151 to 5G DU 152 to NSA UE, and combined again at the NSA UE 101. Flow control between CU-UP 151 and DU as specified in the previous section is run 1) between5G DU 152 and 5G CU-UP 151, and 2) between 4G DU 142 and 5G CU-UP 151 in this architecture.

[0087] FIG. 12B shows another UL Split bearer in NSA Architecture. In thisvariant of the NSA Architecture, DL data from 4G EPC 140 is first sent to 4G CU-UP 141 where it is split across two transmission paths (or network legs): 1) 4G CU-UP 141 to 4G DU 142 to NSA UE 101 and 2) 4G CU-UP 141 to 5G DU 152 to UE 101, and then combined again at the NSA UE 101. For the UL split bearer in the NSA architecture, uplink data from the UE 101 towards the 5G DU 152 / 4G DU 142 is split at the UE 101. After splitting, some packets are sent on the 5G leg 154 and other on the 4G leg 144 of the network.

[0088] In Operations, Administration and Maintenance (OAM) cellmanagement for a RAN, the KPI for a given cell (or NF or UE) is impacted by the measurements and configuration data pertaining to that cell alone, and the KPI for the cell is also impacted by measurements and configuration data pertaining to significantly-interfering neighbor cells (or NFs or UEs). The OAM data includes Fault management data (FM), Configuration Management Data (CM), Performance Management Data (PM) and Trace Data for observability and configuration. To model the KPI of a given cell, the modeling as described herein does not just pertain to the cell-of-interest, but also pertains to the strongly-interfering neighbor cells, which are considered as input features.

[0089] Described herein are advantageous implementations of technology toform clusters of cells that are strongly interfering with each other for joint observability and joint optimization. The present disclosure provides a technique to intelligently generate clusters of strongly-interfering cells together, while splitting apart weakly-interfering cells across different clusters.

[0090] In the process, the disclosure describes a number of techniques forclustering: (i) based on unsupervised spectral clustering machine learning algorithms, and (ii) based on heuristics.

[0091] Network Intelligence as a Service (NIaaS) is a holistic AI frameworkfor telecommunication systems that offers a suite of AI-native micro-services that include data engineering framework services and data science framework services, AI-ML model training framework services, AI-ML life cycle management framework services, AI-ML inference framework services, AI test framework services, AI-ML visualization framework services, and so on.

[0092] The clustering technology described herein is offered as part of theNIaaS data engineering services. While the present disclosure focuses on intelligently clustering significantly-interfering neighbor cells, clustering as described herein can also be applied to other network entities and / or network functions.

[0093] In conventional RAN systems as of the present disclosure, clusteringwas done using (i) -means clustering and (ii) min- -cut. In AI-ML, -meansclustering is an unsupervised ML technique, where the different cells can be grouped into clusters based on a similarity score and measured by a distance metric with respect to the centroid of each cluster. However, in conventional systems as of the present disclosure, -means clustering has not been used to quantify the pair-wise inter-dependency between the cells for grouping cells together with stronger inter-dependencies as described herein.

[0094] Min-k-cut is an NP-Hard combinatorial optimization problem.Relevant approximation heuristics for the min- -cut can be used to partition thecells into groups, and towards also finding the optimal value of . The issue with such approximation heuristics is that the approximation guarantee for the optimal point may not have a reasonable margin of error from the optimal point, especially when the value of is an unknown and for large values of .

[0095] For example, let , , { , , … , } be the vector of RSRP bins(denoting increasing order of the quality) corresponding to the serving cell – neighbor cell pair, representing the impact of neighbor cell collectively on allthe UEs served by cell , as reported by the cell for any given time step . Here, is the total number of bins. This information is reported as (a) Performance Measurement (PM) Reports and (b) Trace.

[0096] Performance Measurement (PM) Reports are measurement reportsthat provide the distribution of RSRP per neighbor E-UTRA cell received by eNB from UEs served by cell , during a given measurement window , and is given by L1M.RSRPEutraNbr.Bin r (3GPP TS 28.552), where is the index of any bin, and each bin corresponds to a pre-defined measurement range. Assign L1M.RSRPEutraNbr.Bin r for the bin for the measurement window in, ,.(b) Trace is a measurement report (Uu: MeasurementReport RRC message [3GPP TS 38.331, TS 36.331]) reported by any UE served by the cell to the serving eNB, and this measurement report includes the RSRP and RSRQ of the UE with respect to any neighbor cell . If the reported RSRP is within the pre-defined range pertaining to any bin , then increment the bin in, ,by 1 during the duration of the reporting window . Upon expiry of the window , the counters are reset to 0 and the process repeats for the next time window.

[0097] FIG. 8C illustrations O-RAN interfaces and network functions. Asshown in FIG.8C, the OAM Management Service (MnS) producer of the O-RAN NFs reports the Performance Measurements and Trace data, consisting of serving cell – neighbor cell-specific RSRP bin vectors to the OAM MnS consumer in the SMO 159 via the O1 interface (O-RAN WG10 O1 interface), and the NIaaS framework 173 inside the SMO 159 / Non-RT RIC 161 platform consumes this information to perform clustering of cells. For the MnS producer of the O-RAN NFs to report this information, the MnS consumer in the SMO 159 initiates a subscription with job control parameters towards the MnS producer. Likewise, the RIC-related functions of the O-RAN NFs report the performance measurements and trace data, including serving cell – neighbor cell-specific RSRP bin vectors to the consumer sub-functions in the Near-RT RIC via E2SM-KPM (E2 Service Model – Key Performance Monitoring) and E2SM-RC (E2 Service Model – RAN Control) using E2 ApplicationProtocol (E2AP) over E2 interface (O-RAN WG3 E2SM-KPM and O-RAN WG3 E2SM- RC), and the NIaaS framework 173 inside the Near-RT RIC 161 platform consumes this information to perform clustering of cells. Towards this end, the Near-RT RIC 161 initiates subscription with job control parameters to the O-RAN NFs for generating the PMs and trace reports to the Near-RT RIC 161.

[0098] The KPI of a cell is impacted by interfering OAM PerformanceMeasurement, Configuration Management, Fault Management race fault for configuration, accounting, performance, and security (FCAPS) data of neighbor cells, as well. Accordingly, described herein are implementations to form clusters of cells that are strongly interfering with each other for joint observability and joint optimization. Clusters are advantageously formed as clusters of strongly-interfering cells together, while splitting apart weekly-interfering cells across different clusters. The OAM FCAPS data of strongly-interfering cells are used for modelling the KPI of any cell of interest, along with the cell’s own OAM FCAPS information, for further optimizing the config parameters of the cell jointly with other strongly-interfering cells towards improving the desired KPIs.

[0099] The efficiency of the clustering algorithm is determined by a metric,given by the ratio of the net inter-cluster interference among cells across clusters normalized to the net intra-cluster interference among cells within individual clusters.

[0100] Dataset preparation for graph theoretic modeling

[0101] In this section, the dataset preparation towards modeling theclustering problem as a graph theoretic problem is described. An illustration of modeling the clustering problem as a graph-theoretic problem is shown in FIG.13.

[0102] In the dataset preparation, let , , { , , … , } be the vector ofRSRP bins (denoting increasing order of the quality) corresponding to the serving cell – neighbor cell pair (that represents the impact of neighbor cellcollectively on all the UEs served by cell ) for any given time step . Here, is the total number of bins.

[0103] For a periodic window (for running / updating the clusteringalgorithms) that comprises of time-steps = { , , … }, let , ,{ } ; , , . . Let { , , … , } be the weight vector ofthe RSRP bins, such that 0 1, and { , ,…, } = 1. Moreover, << < . Let , 2 , …, . . Hence, . ( )=1 and = ( ). Thus, the weight of any bin is given by =( ).

[0104] Now, the weight of the edgethe vertex paircorresponding to the cells and is given by:

[0105] . + ; , , , , .. ,

[0107] Spectral Clustering

[0108] In this section, the clustering algorithm based on unsupervised MLspectral clustering is disclosed.

[0109] Let and be the minimum and maximum number of clusters.Let be the set of cells in the network.

[0110] Let be the upper-bound max threshold on the number of cells inany cluster.

[0111] 1>Construct an undirected graph = ( , ). Let be the set ofvertices, corresponding to cells, and be the set of edges between the vertices denoting the relationship between the cells. Let,be the edge between thevertices , in , denoting the relationship between the respective cells , .Let ( , ) be the weight of the relationship.

[0112] 1>Construct an adjacency matrix of dimension | | × | |corresponding to the graph .

[0113] 1>Repeat For each entry , in the matrix corresponding to anyrow and column 2>If == then3>Set,:=0 2>Else 3>Set , ( , )

[0114] 1>Construct a diagonal matrix of dimension | | × | |corresponding to the adjacency matrix of

[0115] 1>Repeat For each entry , in the matrix corresponding to anyrow and column 2>Set , , , and set , 0, , where 1 | |, and , where

[0116] matrix of the graph as : = , or thenormalized Laplacian matrix of the graph as given by . .

[0117] 1>Use Eigen Value Decomposition on or to derive | | EigenValues (given by = { , , … , | |} ) and the corresponding | | Eigen Vectors(given by = , , … , | |) or normalized Eigen Vectors (given by =, , … , )

[0118] 1>Repeat for2>Let be the solution including the clusters. Set { }2>Let ( ) be the set of Eigen values with the smallest non-zero Eigenvalues out of . Note that (k) .2>Let ( ) be the set of Eigen vectors corresponding to ( ). So, ( ) is amatrix with | | rows and columns.2>Form clusters ( , , , , … , , ) out of | | vertices, corresponding to| | Eigen vectors. Cluster | |( ) using -means clustering algorithmusing the Euclidean distance as the metric. The Euclidean distance between any twovectors (where = { , , , , … , , } ) and (where = { , , , , … , , })is ,, , , ,,2>Assign clusters to , i.e,. = , . Set = ( , ); ( ,) (, );

[0119] 1>End Repeat

[0120] 1>Determine the optimal number of clusters asarg min {, , ( , )} with being the optimal cluster set with clusters.

[0121] algorithm

[0122] In this section, the clustering algorithm based on heuristic min-K-cutis discussed.

[0123] Let and be the minimum and maximum number of clusters.Let be the set of cells in the network.

[0124] Let , denote the relationship between any cells , , where. Let ( , ) be the weight of the relationship.

[0125] Let be the upper-bound max threshold on the number of cells inany cluster.

[00126] 1>Repeat for2>Let be the cut solution including the clusters. Set { }2>Repeat for 13>Let,be the cluster corresponding to index for the cluster set with clusters. Set , : = { }3>Pick any random cell 3>Assign to cluster index . Set , : = , { }3>Compute the cut value,for cluster index in the cluster set with clusters, where , : = { , , } ( , ) , and where , is the set of edgescrossing the cut an,d,2>Compute the total cut value for the cut solution, thus far, as : =,.,. 2>Repeat for each cell 3>Determine the cluster index out of the clusters to which shall be assigned by: arg min;, + , , ; where , , is the addition to the cutvalue,index , as a result of adding to the cluster 3>Assign to cluster index . Set , : = , { }3>Update the cut value,for cluster index for the cluster set with clusters, where , , + , ,3>Update the total cut value for the cut solution, thus far, as : =,.3>Determine any other cell , where on cluster index that could be potentially re-assigned to a different cluster index , where , by:arg min;{ ( , )}, and, ,, ) ; here, , , is theaddition to the cut value,of adding to thecluster index 1 3> Set , , { } , reflecting a potential temporaryassignment of to cluster index 1. 3>Compute the temporary cut value,for cluster with index in the cluster set with clusters, where ,, + , , , , ; where , , is the reduction in the cutthe cluster index3>Compute the temporary total cut value for the cut solution, as a result of this potential temporary assignment of to cluster index as ,, + , ,3>If < then4>Make the temporary assignment of to cluster index permanent by setting, ,4>Update the cut value,for cluster index for the cluster set with clusters, where, ,4>Update the total cut value for the cut solution by setting2>End Repeat 2>Set,. Note that is the total weight of the cut solution pertaining to clusters, corresponding to .1>End Repeat1>Determine the optimal number of clusters as arg min {, , ( , )} with being the optimal cluster set with clusters.

[0127] FIG. 14 is a graph charting a performance evaluation of clusteringalgorithms. In FIG.14, the performance of the spectral clustering algorithms are compared to traditional min-cut algorithms in grouping the cells into clusters. As discussed above, the optimal number of clusters advantageously minimizes the ratio of the net inter-cluster interference to the intra-cluster interference. For example, for a network topology comprising of 18 cells in an urban scenario with 43 dBm as the maximum transmission power of any cell deployed in B1 central-band frequency, the evaluation shows unsupervised ML-based spectral clustering yields a significantly lower net inter-cluster interference normalized to the intra-cluster interference and improves the objective metric (discussed above) by about 35 – 60%.

[0128] Implementation as described herein applies to grouping networkentities, of any type, based on inter-dependencies, again of any type, among them, such that the network entities with strong inter-dependencies are grouped together whereas the ones with weaker inter-dependencies are split apart. For example, CU- UPs that have high load balancing inter-dependency among each other are grouped together, whereas CU-UPs that have less load balancing inter-dependency among each other are grouped apart – based on appropriate modeling of the edge weights between the CU-UP network functions.

[0129] It will be understood that implementations and embodiments can beimplemented by computer program instructions. These program instructions can be provided to a processor to produce a machine, so that the instructions, which execute on the processor, create means for implementing the actions specified herein. The computer program instructions can be executed by a processor to cause a series of operational steps to be performed by the processor to produce acomputer-implemented process so that the instructions, which execute on the processor to provide steps for implementing the actions specified. Moreover, some of the steps can also be performed across more than one processor, such as might arise in a multi-processor computer system or even a group of multiple computer systems. In addition, one or more blocks or combinations of blocks in the flowchart illustration can also be performed concurrently with other blocks or combinations of blocks, or even in a different sequence than illustrated without departing from the scope or spirit of the disclosure.

Claims

CLAIMS 1. A method for Artificial Intelligence and Machine Learning (AI-ML) Operations, Administration and Maintenance (OAM) cell optimization comprising: modeling key performance indicators of a cell using the cell’s OAM fault, configuration, accounting, performance, and security(FCAPs) and the OAM FCAPS of other strongly interfering cells; and clustering strongly-interfering cells together, while splitting apart weekly- interfering cells across different clusters; wherein the clustering is modeled and executed by a k-means clustering algorithm.

2. The method of claim 1, wherein the k means clustering algorithm comprises an unsupervised machine learning spectral clustering algorithm.

3. The method of claim 1, further comprising: preparing a dataset where ,, { , , … , } is a vector of Reference Signal Received Power (RSRP)bins denoting and increasing order of quality corresponding to a serving cell c – a neighbor cell c^' pair that represents an impact of neighbor cell c^' collectively on all UEs served by the cell c for any given time step t, and R is the total number of bins; for the periodic window T for running or updating the k-means clusteringalgorithms, T comprises time-steps = { , , … } , , ,{ } ; , , . .,< < < . Let , 2 , …, . . Hence, . ( )=1 and=( ). such that a weight of any bin is given by =( ). giving the weight of the edge e E between the vertex pair corresponding to thecells c and c^' as .+ ; , , , , , where tmin{ } and max{ } and Set ,4. The method of claim 2, wherein the unsupervised machine learning spectral clustering algorithm method comprises: where and are a minimum and maximum number of clusters respectively and is a set of the cells in network; is an the upper-bound max threshold on the number of the cells in any cluster; constructing an undirected graph = ( , ) where a set of verticescorresponding to cells, and is a set of edges between the vertices denoting therelationship between the cells; and , is an edge between the vertices , in ,denoting the relationship between the respective cells , , and ( , ) be is aweight of the relationship; constructing an adjacency matrix of dimension | | × | | corresponding to thegraph ; repeating for each entry,in the matrix corresponding to any row and column if == thenset,:=0else set , ( , )end if;aof dimension | | × | | corresponding to theadjacency matrix of ; repeating for each entry,in the matrix corresponding to any row and column set , , , and set , 0, , where 1 | |, and , where(where , , , ).matrix of the graph as : = , or a normalizedLaplacian matrix of the graph as given by . . ;using an Eigen Value Decomposition on or to derive | | Eigen Values(given by = { , , … , | |} ) and the corresponding | | Eigen Vectors (given by= , , … , | | ) or normalized Eigen Vectors (given by =;where is the solution including the clusters. Set { } ,where ( ) is the set of Eigen values with the smallest non-zero Eigenvalues out of . Note that (k) ,where ( ) is the set of Eigen vectors corresponding to ( ) so that ( ) isa matrix with | | rows and columns;forming clusters ( , , , , … , , ) out of | | vertices, corresponding to the| | Eigen vectors and clustering | | rows from ( ) using the -means clusteringalgorithm using a Euclidean distance as the metric, wherein the Euclidean distancebetween any two vectors (where = { , , , , … , , } ) and (where ={ , , , , … , , }) isby,,assigning clusters to , i.e,. = , . Set = ( , ); ( ,);, (, );end Repeat; and determining an optimal number of clusters asarg min {,} with being the optimal cluster set with cluster ,( , )s.

5. The method of claim 1, wherein the k-means clustering algorithm comprises a heuristic Min-k-cut algorithm.

6. The method of claim 5, wherein heuristic Min-k-cut algorithm method comprises: and is the minimum and maximum number of clusters respectively, js a set of the cells in the network,,denotes a relationshipbetween any cells , , where , and ( , ) is a weight of the relationshipand is an upper-bound max threshold on the number of the cells in any cluster; repeating for ; is a cut solution including the clusters. Set { } ;repeating for 1,is the cluster corresponding to index for the cluster set with clusters. Set , : = { }picking any random cell assigning to cluster index and setting , : = , { }computing a cut value,for a cluster index in the cluster set withclusters, where , : = { , , } ( , ) , and where , is a set of edges crossingthe cut from cluster,and,;computing a total cut value for the cut solution, thus far, as : = , .setting,. repeating for each cell determining a cluster index out of the clusters to which shall be assigned by: arg min;, + , , ; where , , is the addition to the cut, value,index , as a result of adding to the clusterassigning to cluster index . Set , : = , { }updating the cut value,for cluster index for the cluster set withclusters, where , , + , ,value for the cut solution, thus far, as : =,. determining any other cell , where on the cluster indexthat can be re-assigned to a different cluster indexwhere , by: arg min;{ ( , )}, and, , ) ; here, , , is theaddition to the cut value,result of adding to the cluster index 1; setting , , { } , reflecting a potential temporaryassignment of to cluster index 1; computing a temporary cut value,for cluster with index in the cluster set with clusters, where ,, + , , , , ; where , , is a reduction in the cut value,cluster index , due to removing from the cluster index ;computing a temporary total cut value for the cut solution, as a result of the potential temporary assignment of to cluster index as ,, + , ,if < thenmaking the temporary assignment of to cluster index permanent by setting, ,, updating the cut value,for cluster index for the cluster set with clusters, where, ,, and cut value for the cut solution by setting; else cancelling and revoking the temporary assignment and setting ,{ }, , 0, 0end repeat; setting,. , where is a total weight of the cut solution pertaining toto . end repeat determining an optimal number of clusters asarg min {, ( , )} with being an optimal cluster set with clusters. , 7. A system for Artificial Intelligence and Machine Learning (AI-ML) Operations, Administration and Maintenance (OAM) cell optimization comprising: a Service Management and Orchestration (SMO) system comprising an OAM MnS consumer module and a Network Intelligence as a Service (NIaaS) module; the NIaaS being programmed to:model key performance indicators of a cell using the cell’s OAM fault, configuration, accounting, performance, and security(FCAPs) and the OAM FCAPS of other strongly interfering cells; and cluster strongly-interfering cells together, while splitting apart weekly- interfering cells across different clusters; wherein the clustering is modeled and executed by a k-means clustering algorithm.

8. The system of claim 7, wherein the k means clustering algorithm comprises an unsupervised machine learning spectral clustering algorithm.

9. The method of system of claim 7, further comprising: preparing a dataset where ,, { , , … , } is a vector of RSRP bins denoting and increasing order ofquality corresponding to a serving cell c – a neighbor cell c^' pair that represents an impact of neighbor cell c^' collectively on all UEs served by the cell c for any given time step t, and R is the total number of bins; for the periodic window T for running or updating the k-means clusteringalgorithms, T comprises time-steps = { , , … } , , ,{ } ; , , . .,where { , , … , } is the weight vector of the RSRP bins, such that 01, and { , ,…, } = 1, and., 2 , …, . . Hence, . ( )=1 and=( ). such that a weight of any bin is given by =( ).giving the weight of the edge e E between the vertex pair corresponding to thecells c and c^' as .+ ; , , , , , where t10. The system of claim 8, wherein the unsupervised machine learning spectral clustering algorithm method comprises: where and are a minimum and maximum number of clusters respectively and is a set of the cells in network; is an the upper-bound max threshold on the number of the cells in any cluster; constructing an undirected graph = ( , ) where a set of verticescorresponding to cells, and is a set of edges between the vertices denoting therelationship between the cells; and , is an edge between the vertices , in ,denoting the relationship between the respective cells , , and ( , ) be is aweight of the relationship; constructing an adjacency matrix of dimension | | × | | corresponding to thegraph ; repeating for each entry,in the matrix corresponding to any row and column if == thenset,:=0 else set , ( , )end if;constructing a diagonal matrix of dimension | | × | | corresponding to theadjacency matrix of ;repeating for each entry,in the matrix corresponding to any row and column set , , , and set , 0, , where 1 | |, and , where(whereconstructing a Laplacian matrix of the graph as : = , or a normalizedLaplacian matrix of the graph as given by . . ;using an Eigen Value Decomposition on or to derive | | Eigen Values(given by = { , , … , | |} ) and the corresponding | | Eigen Vectors (given by= , , … ,or normalized Eigen Vectors (given by =| | );repeating for where is the solution including the clusters. Set { } ,where ( ) is the set of Eigen values with the smallest non-zero Eigenvalues out of . Note that (k) ,where ( )isvectors corresponding to ( ) so that ( ) isa matrix with | | rows and columns;forming clusters ( , , , , … , , ) out of | | vertices, corresponding to the| | Eigen vectors and clustering | | rows from ( ) using the -means clusteringalgorithm using a Euclidean distance as the metric, wherein the Euclidean distancebetween any two vectors (where = { , , , , … , , } ) and (where ={ , , … , }) is give,, , n by , , , , , ,to , i.e,. = , . Set = ( , ); ( ,);, (, );end Repeat; anddetermining an optimal number of clusters asarg min {,} with being the optimal cluster set with clusters ,( , ).

11. The system of claim 7, wherein the k-means clustering algorithm comprises a heuristic Min-k-cut algorithm.

12. The system of claim 11, wherein heuristic Min-k-cut algorithm method comprises: and is the minimum and maximum number of clusters respectively, js a set of the cells in the network,,denotes a relationshipbetween any cells , , where , and () is a weight of the relationshipand is an upper-bound max threshold on the number of the cells in any cluster; repeating for ; is a cut solution including the clusters. Set { } ;repeating for 1,is the cluster corresponding to index for the cluster set with clusters. Set , : = { }picking any random cell assigning to cluster index and setting , : = , { }computing a cut value,for a cluster index in the cluster set withclusters, where , : = { , , } ( , ) , and where , is a set of edges crossingthe cut from cluster,and,; computing a total cut value for the cut solution, thus far, as : = , .setting,.determining a cluster index out of the clusters to which shall be assigned by:arg min;, + , , ; where , , is the addition to the cut, value,for cluster index , as a result of adding to the cluster assigning to cluster index . Set , : = , { }updating the cut value,for cluster index for the cluster set withclusters, where , , + , ,updating the total cut value for the cut solution, thus far, as : =,. determining any other cell , where on the cluster index that can be re-assigned to a different cluster indexwhere , by: arg min;{ ( , )}, and, , ) ; here, , , is theaddition to the cut value,result of adding to the cluster index 1; setting , , { } , reflecting a potential temporaryassignment of to cluster index 1; computing a temporary cut value,for cluster with index in the cluster set with clusters, where ,, + , , , , ; where , , is a reduction in the cut value,cluster index , due to removing from the cluster index ; computing a temporary total cut value for the cut solution, as a result of the potential temporary assignment of to cluster index as ,, + , ,if < thenmaking the temporary assignment of to cluster index permanent by setting, ,,updating the cut value,for cluster index for the cluster set with clusters, where, ,, andsetting ,{ }, , 0, 0setting,. , where is a total weight of the cut solution pertaining to clusters, corresponding to . end repeat determining an optimal number of clusters asarg min {,} with being an optimal cluster set with clusters. ,( , )

Images