AIML assisted optimization of downlink split-bearer operation

Abstract

A system and method that allows optimizing downlink split-bearer operations in Open Radio Access Network (O-RAN) systems to enhance packet delivery efficiency, reduce head-of-line blocking, and improve throughput under varying traffic conditions, while simultaneously dynamically integrating mid-haul latency, queue depth metrics, and per-DRB sequencing feedback utilizing Artificial Intelligence / Machine Learning (AI / ML)-driven adaptive splitting in disaggregated RAN architectures.

Description

AIML assisted Optimization of Downlink Split-Bearer OperationBACKGROUND1. Field of the Disclosure

[0001] The present disclosure pertains to wireless communication technologies. The system enhances downlink split-bearer operations in Open Radio Access Network (O-RAN) systems through the support of Artificial Intelligence and Machine Learning (AIML).2. Description of Related Art

[0002] An overview of Next Generation Radio Access Network (NG-RAN) architecture and 5G New Radio (NR) stacks is presented below. 5GNR (New Radio) user and control plane functions with monolithic gNodeB (gNB) are shown in FIGS. 1 A, IB and 2. For the user plane (shown in FIG. 1A), Physical (PHY), Medium Access Control (MAC), Radio Link Control (RLC), Packet Data Convergence Protocol (PDCP) and Service Data Adaptation Protocol (SDAP) are sublayers that originate in the User Equipment (UE) 101 and are terminated in the gNB 102 on the network side.

[0003] As shown in FIG. IB, which is a block diagram illustrating the user plane protocols stacks for a Protocol Data Unit (PDU) session, PDU layer 9010 corresponds to the PDU carried between the UE 101 and the data network (DN) 9011 over the PDU session. As shown in FIG. IB, UE 101 is connected to the 5G Access Network (AN) 902, which AN 902 is in turn connected via the N3 interface to the Intermediate User Plane Function (I-UPF) 903a portion of the UPF 903, which I-UPF 903a is in turn connected via the N9 interface to the PDU session anchor 903b portion of the UPF 903, and which PDU session anchor 903b is connected to the DN 9011. The PDU session can correspond to IPv4, IPv6, or both types of IP packets, when the PDU session is of type IPv4, IPv6 or IPv4v6, respectively. GPRS Tunneling Protocol User Plane (GTP-U) shown in FIG. IB supports tunnelling user plane data over N3 and N9 interfaces and provides encapsulation of end user PDUs for N3 and N9 interfaces.

[0004] For the control plane (shown in FIG. 2), Radio Resource Control (RRC), PDCP, RLC, MAC and PHY sublayers originate in the UE 101 and are terminated in the gNB 102 on the network side, and Non-Access Stratum (NAS) originate in the UE 101 and is terminated in the Access Mobility Function (AMF) 103 on the network side.

[0005] NG-Radio Access Network (NG-RAN) architecture is shown in FIGS. 3 and 4. As shown in FIG. 3, the NG-RAN 301 comprises a set of gNBs 302 connected to the 5GC 303 through the NG interface. Each gNB comprises gNB Central Unit (CU) 304 and one or more gNB Distributed Units (DU) 305 (FIG. 3). As shown in FIG. 4, which illustrates separation of CU-Control Plane (CU-CP) and CU-User Plane (CU-UP), El is the interface between gNB-CU-CP 304a and gNB-CU-UP 304b; Fl-C is the interface between gNB-CU-CP 304a and gNB-DU 305; and Fl-U is the interface between gNB-CU-UP 304b and gNB-DU 305. As shown in FIG. 4, gNB 302 may comprise a gNB-CU-CP 304a, multiple gNB-CU-UPs (or gNB-CU-UP instances) 304b and multiple gNB-DUs (or gNB-DU instances) 305. One gNB-DU 305 is connected to only one gNB-CU-CP 304a, and one gNB-CU-UP 304b is connected to only one gNB-CU-CP 304a.

[0006] Note that Fl -Application Protocol (Fl-AP) is running on Fl-C, and NR User Plane (NR-U) running on Fl-U.

[0007] An overview of Layer 2 (L2) of 5G NR is provided in connection with FIGS. 5-7. L2 of 5G NR is split into the following sublayers:

[0008] 1) MAC 501 in FIGS. 5-7: Logical Channels (LCs) are Service Access Points (SAPs) between the MAC and RLC layers. This layer runs a MAC scheduler to schedule radio resources across different LCs (and their associated radio bearers). For the downlink (DL) direction, the MAC layer processes and sends RLC PDUs received on LCs to the PHY as Transport Blocks (TBs). For the uplink (UL) direction, it receives TBs from the PHY, processes these and sends them to the RLC layer using the LCs.

[0009] 2) RLC 502 in FIGS. 5-7: The RLC sublayer presents RLC channels to the 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 Automatic Repeat Request (ARQ) protocol for RLC-AM mode.

[0010] 3) PDCP 503 in FIGS. 5-7: The PDCP sublayer presents Radio 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 the control plane.

[0011] 4) SDAP 504 in FIGS. 5-7: The SDAP maps Quality of Service (QoS) flows within a PDU session to a specific DRB.

[0012] Referring now to FIGS. 5 - 7, FIG. 5 is a block diagram illustrating DL L2 structure, FIG. 6 is a block diagram illustrating UL L2 structure, and FIG. 7 is a block diagram illustrating L2 data flow example where “H” denotes headers or sub-headers.

[0013] Open Radio Access Network (0-RAN) is based on disaggregated components, which are connected through open and standardized interfaces based on 3 GPP NG-RAN. An overview of O-RAN with disaggregated CU, DU, and Radio Unit (RU), near-real-time Radio Intelligent Controller (RIC) and non-real-time RIC is illustrated in FIG. 8.

[0014] As shown in FIG. 8, the CU (shown split as O-CU-CP 801a and O-CU-UP 801b) and the DU (shown as O-DU 802) are connected using the Fl interface (with Fl-C for control plane and Fl-U for user plane traffic) over a mid-haul (MH) path. One DU can host multiple cells (e.g., one DU could host 24 cells) and each cell may support many users. In one example, one cell supports 800 Radio Resource Control (RRC)-connected users and out of these 800, there can be subsets of 250 Active users (i.e., users that have data to send at a given point of time).

[0015] A cell site can comprise multiple sectors, and each sector can support multiple cells. As an example, one site comprises three sectors and each sector could support eight cells (with each cell being on a different frequency band in each sector). One CU-CP could support multiple DUs and thus multiple cells. For example, a CU-CP could support 500 cells and around 100,000 different User Equipment (UE). 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 UE) may be served by five CU-UP instances (and one CU-CP instance).

[0016] The DU could be in a private data center, or it could be located at a cell site. The CU could also be in a private data center or even hosted on a public cloud system. The DU and CU, which are typically located at different physical locations, could be located many kilometers from each other. The CU communicates with a 5G core system, which could also be hosted in the same public cloud system (or could be hosted by a different cloud provider). A RU (shown as O-RU 803 in FIG. 8) is located at a cell-site and communicates with the DU via a front-haul (FH) interface.

[0017] The E2 nodes (CU and DU) are connected to the near-real-time RIC 132 using the E2 interface. The E2 interface is used to send data (e.g., user and / or cell KPMs) from the RAN, and deploy control actions and policies to the RAN at near-real-time RIC 132. The applications or services at the near-real-time RIC 132 that deploys the control actions and policies to the RAN are called xApps. During the E2 setup procedures, the E2 node advertises the metrics it can expose, and an xApp in the near-RT RIC can send a subscription message specifying key performance metrics which are of interest. The near-real-time RIC 132 is connected to the non-real-time RIC 133, which is shown as part of Service Management and Orchestration (SMO) Framework 805 in FIG. 8 using the Al interface. The applications that are hosted at non-RT-RIC are called rApps. Also shown in FIG. 8 are Open evolved Node B (O-eNB) 806, which is shown as being connected to the near-real-time RIC 132 and the SMO Framework 805 and O-Cloud 804, which is shown as being connected to the SMO Framework 805.

[0018] As in FIG. 8B, E2 node (which is DU or CU) and Near-RT -RIC establish E2 session using E2 SETUP REQUEST and E2 SETUP RESPONSE. Near-RT -RIC can subscribe to certain parameters from the E2 node on behalf the xApp running at Near-RT -RIC using the RIC SUBSCRIPTION REQUEST. The E2 node acknowledges this message by sending RIC SUBSCRIPTION RESPONSE to the Near-RT -RIC. As part of this, xApp running at the Near-RT -RIC also provides the event triggers to the E2 node (e.g., it could ask the E2 node to REPORT subscribed parameters periodically to the xApp or to REPORT these subscribed parameters based on certain events to the xApp). The E2 node communicates subscribed parameters to the Near-RT-RIC and the xApp using RIC INDICATION as shown in FIG. 8B. After analysing received parameters from the E2 nodes and based on network operator policies, the Near-RT-RIC can send RIC CONTROL REQUEST to take an action at the E2 node (e.g.influence mobility decision). The E2 node acknowledges this message by sending RIC CONTROL ACKNOWLEDGE to Near-RT-RIC while E2 node functions as instructed by the Near-RT-RIC.

[0019] PDU sessions, DRBs, and QoS flows will now be discussed. In 5G networks, PDU connectivity service is a service that provides exchange of PDUs between a UE and a Data Network (DN) identified by a Data Network Name (DNN). The PDU Connecitivity service is supported via PDU sessions that are established upon request from the UE. The DNN defines the interface to a specific external data network. One or more QoS flows can be supported in a PDU session. LI?? the packets belonging to a specific QoS flow have the same 5G QoS Identifier (5QI). A PDU session comprises the following: DRBs that are between UE and CU in RAN; and an NG-U GTP tunnel that is between CU and User Plane Function (UPF) in the core network. FIG. 9 illustrates an example PDU session comprising multiple DRBs, where each DRB may comprise multiple QoS flows. In FIG. 9, three components are shown for the PDU session 901 : UE 101; AN 902; and UPF 903, which includes Packet Detection Rules (PDRs) 9031.

[0020] 3GPP 5G network architecture is illustrated in FIGS. 10 and 11. FIG. 10 is provided in the context of multiple PDU sessions involving multiple DRBs and QoS Flow Identifiers (QFIs), which PDU sessions are implemented involving UE 101, gNB 102, UPF 903, and DNNs 9011a and 901 lb. FIG. 11 is provided in the context of Radio Resource Management (RRM) for connecting UE 101 to the network via RU 306 with a MAC Scheduler 1001. The following should be noted:

[0021] 1) The transport connection between the base station (i.e., CU-UP 304b of FIG. 11) and the UPF 903 uses a single GTP-U tunnel per PDU session, as shown in FIGS. 10 and 11. The PDU session is identified using GTP-U Tunnel Endpoint Identifier (TEID).

[0022] 2) The transport connection between the DU 305 and the CU-UP 304b of FIG. 11 uses a single GTP-U tunnel per DRB (see also FIG. 10 and FIG. 11). The DU is provided with an UL GTP-U TEID and the CU is provided with the corresponding DL GTP-U TEID to allow for data communication for that DRB between DU and CU-UP.

[0023] 3) SDAP:a) The SDAP (Service Adaptation Protocol) 504 Layer receives downlink data from the UPF 903 across the NG-U interface (see FIG. 11).b) The SDAP 504 maps one or more QoS Flow(s) onto a specific DRB.c) The SDAP header is present between the UE 101 and the CU (when reflective QoS is enabled), and includes a field to identify the QoS flow within a specific PDU session.

[0024] 4) GTP-U protocol includes a field to identify the QoS flow and is present between CU and UPF 903 (in the core network).

[0025] 5) One (logical) DU (or RLC) queue exists per DRB (or per logical channel) for RLC PDUs that are to be transmitted for the first time, as shown in FIG. 11. Separate logical queues may exist in DU for packets that are to be retransmitted to UE.

[0026] In this section, standardized 5QI to QoS characteristics mapping is discussed. One-to-one mapping of standardized 5QI values to 5G QoS characteristics is specified in Table 1. The first column represents the 5QI value. The second column lists the different resource types, i.e., as one of Non-GBR, GBR, Delay-critical GBR. The third column (“Default Priority Level”) represents the priority level Priority5QI, for which lower the value the higher the priority of the corresponding QoS flow. The fourth column represents the Packet Delay Budget (PDB), which defines an upper bound for the time that a packet can be delayed between the UE and the N6 termination point at the UPF. The fifth column represents the Packet Error Rate (PER). The sixth column represents the maximimum data burst volume for delay -critical GBR types. The seventh column represents averaging window for GBR, delay critical GBR types. Note that only a subset of 5QI values are shown in Table 1.5QI Value Resource Default Packet Packet Default Default Example Type Priority Delay Error Rate Maximum Averaging Services Level Budget Data Burst Window(Note 3) Volume(Note 2)1 GBR Note 1 20 lOO mS 10‘2N / A 2000 mS Conversational (Note 11, Voice Note 13)2 GBR Note 1 40 150 mS 10‘3N / A 2000 mS Conversational (Note 11, Video (live Note 13) streaming)3 GBR Note 1 30 50 mS (Note 10'3N / A 2000 mS Real Time 11, Note 13) Gaming (V2X messages (see TS 23287

[0121] ) Electricity distribution - medium voltage. Process automation monitoring 4 GBR Note 1 50 300 mS 10'6N / A 2000 mS Non- (Note 11, Conversational Note 13) Video (Buffered Streaming) 5 (Note 9, GBR Note 1 7 75 mS (Note 10'2N / A 2000 mS Mission Note 12) 7, Note 8) Critical user plane Push To Talk voice (e.g , MCPTT) (Note 12) GBR Note 1 20 100 mS IO'2N / A 2000 mS Non-Mission- (Note 10, Critical user Note 13) plane Push- To-Talk voice (Note 12) GBR Note 1 15 lOO mS 10'3N / A 2000 mS Mission (Note 10, Critical Video Note 13) user plane (Note 14) GBR Note 171 GBR Note 1 56 150 mS 10 ° N / A 2000 mS “Live” Uplink (Note 11, Streaming Note 13, (e.g , TS Note 15) 26.238

[0076] ) 72 GBR Note 1 56 300 mS IO'4N / A 2000 mS “Live” Uplink (Note 11, Streaming Note 13, (e g , TS Note 15) 26.238

[0076] ) 73 GBR Note 1 56 300 mS 10'8N / A 2000 mS “Live” Uplink (Note 11 , Streaming Note 13, (e.g , TS Note 15) 26.238

[0076] ) 74 GBR Note 1 56 500 mS 10'8N / A 2000 mS “Live” Uplink (Note 11, Streaming Note 13, (e.g , TS Note 15) 26.238

[0076] ) 76 GBR Note 1 56 500 mS IO'4N / A 2000 mS “Live” Uplink(Note 11, StreamingNote 13, (e.g , TS Note 15) 26.238

[0076] ) 5 Non-GBR 10 lOO mS 10 ° N / A N / A IMS Note 1 (Note 10, Signalling Note 13)6 Non-GBR 60 300 mS 10 ° N / A N / A Video Note 1 (Note 10, (Buffered Note 13) Streaming) TCP-based (e.g , www, e- mail, chat, ftp, p2p file sharing, progressive video, and the like) 7 Non-GBR 70 lOO mS 10'3N / A N / A Voice, Video Note 1 (Note 10, (Live Note 13) Streaming) Interactive GamingTable 1

[0027] For example, as shown in Table 1, 5QI value 1 is of resource type GBR with the default priority value of 20, PDB of 100ms, PER of 0.01, and averaging widnow of 2000 ms.Conversational voice falls under this category. Similarly, as shown in Table 1, 5QI value 7 is of resource type Non-GBR with the default priority value of 70, PDB of 100ms and PER of 0.001. Voice, video (live streaming), and interactive gaming fall under this category.

[0028] A block diagram for an example RRM (Radio Resource Management) with a MAC Scheduler is shown in FIG. 11.

[0029] FIG. 12 shows CU-UP sending DL User Data (DUD) to DU. FIG. 13 shows flow control feedback from CU-UP to DU. This is indicated as DL Data Delivery Status (DDDS) in FIG. 13.

[0030] DU can send Assistance Information Data (AID) to CU-UP as shown in FIG. 14 and this provides radio information to CU-UP over the Fl-U interface.

[0031] E-UTRA-NR Dual Connectivity (EN-DC) architecture is shown in FIG. 15 and FIG. 16. In this architecture, the UE has a single RRC (Radio Resource Control) state, based on MN (Master Node) RRC and a single C-plane (Control plane) connection towards the MME(Mobility Management Entity) of 4G EPC (Evolved Packet Core) as shown in FIG. 15. LTE eNB is the MN (denoted as Master eNB or MeNB) and 5G gNodeB is the SN (Secondary Node), denoted as SgNB (or Secondary gNB), in this EN-DC architecture. NR RRC messages can be routed from SgNB to MeNB via the X2-C interface and to the UE via the LTE-Uu air-interface.

[0032] FIG. 16 shows the data (user) plane for this EN-DC architecture. MeNB is connected to S-GW (Serving Gateway) of 4G EPC via the Sl-U interface. Similarly, the SgNB is connected to the S-GW via the Sl-U interface. For SN terminated split bearer, 4G EPC communicates data with SgNB via the Sl-U interface and for MN terminated split bearer, 4G EPC communicated data with MeNB via the Sl-U interface. X2-U is used for user plane traffic between MeNB and SgNB.

[0033] FIG. 17 shows MCG (Master Cell Group), SCG (Secondary Cell Group) and split bearers from a UE perspective. In the EN-DC architecture, MCG consists of group of 4G LTE cells and SCG comprises group of 5G NR cells. As in FIG. 17, NR-PDCP is used for at higher layers, and LTE (E-UTRA) RLC / MAC / PHY and NR RLC / MAC / PHY are used at lower layers for split bearer.

[0034] FIG. 18 shows MCG, SCG and split bearers from base station perspective. For SN terminated split bearer, NR-PDCP is used for at higher layers, and LTE (E-UTRA) RLC / MAC / PHY and NR RLC / MAC / PHY are used at lower layers for split bearer.

[0035] EN-DC architecture with DL split bearer operation is described in FIG. 19. In this FIG.19, DL data (i.e., IP packets or NR PDCP SDUs) is communicated from 4G EPC to SgNB-CU-UP across the Sl-U interface and the DL traffic splitting operation for split bearer is carried out at the SgNB-CU-UP after doing NR PDCP processing at the SgNB-CU-UP. Note that incoming DL NR PDCP SDUs (or IP packets) at the SgNB-CU-UP are transformed to NR PDCP PDUs after NR PDCP processing at the SgNB-CU-UP.

[0036] In FIG. 19, VI is the interface between MeNB-DU and MeNB-CU. SI -Control plane (Sl-C) interface exists between MeNB-CU and 4G EPC for SN terminated split bearers.

[0037] DDDS and AID messages are sent from SgNB-DU to SgNB-CU-UP across the Fl-U interface as shown in FIG. 19. Also, DDDS and AID messages are sent from MeNB-DU to SgNB-CU-UP via the X2-U interface.

[0038] Note that UE can start with a MCG bearer (i.e., along the UE - MeNB - 4G EPC) path and MeNB can decide to switch this bearer to SN terminated split bearer. Alternatively, MeNB can directly decide to establish a SN terminated split bearer for a UE.

[0039] As part of DL traffic splitting operation at the SgNB-CU-UP, some of these NR PDCP PDUs are communicated from SgNB-CU-UP to SgNB-DU (across the Fl-U interface as in FIG.19) and eventually to the UE via the 5GNR-Uu air-interface. These are also processed via NR RLC and NR MAC at SN (i.e., at SgNB-DU) as shown in FIG. 18.

[0040] Some other NR PDCP PDUs are communicated from SgNB-CU-UP to MeNB-DU (across the X2-U interface as in FIG. 19) and eventually to the UE via the 4G LTE-Uu airinterface. Note that these NR PDCP PDUs are also processed via E-UTRA RLC and E-UTRA MAC at MN (i.e.. at MeNB-DU) as shown in FIG. 18 for split bearers.

[0041] Some of the steps for secondary node (i.e.. SgNB in the EN-DC architecture) addition and DL traffic splitting procedures are shown in FIG. 20 and FIG. 21. As shown in FIG. 20, MCG bearer is already established for this UE and MeNB decides to convert this to SgNB terminated split bearer.

[0042] As part of Step 1) in FIG. 20, MeNB, specifically MeNB.CU-CP of MeNB, sends ‘SgNB Addition Request’ to SgNB.CU-CP. It contains information about the bearer for which SN (i.e., SgNB) split bearer needs to be established. It also informs DL SgNB UE-AMBR and UL SgNB UE-AMBRto SgNB.CU-CP.

[0043] In step 2), SgNB.CU-CP sends ‘Bearer Context Setup Request’ message to SgNB.CU-UP using the E1AP protocol. As part of step 3), SgNB.CU-UP sends ‘Bearer Context Setup Response’ message to SgNB.CU-CP. Along with other parameters, QoS information about bearer is shared with SgNB.CU-UP as part of steps 2) and 3).

[0044] In step 4), ‘UE Context Setup Request’ message is sent from SgNB.CU-CP to SgNB.DU across the Fl-C interface using the F1AP protocol. In step 5), SgNB.DU responds with ‘ E Context Setup Response’ message to SgNB.CU-CP. Along with other parameters, QoS information about the bearer is shared with SgNB.DU as part of steps 5) and 6) in FIG. 20.

[0045] In step 6), ‘Bearer Context Setup Request’ is sent from SgNB.CU-CP to SgNB.CU-UP. In step 7), ‘Bearer Context Setup Response’ message is sent from SgNB.CU-UP to SgNB.CU-CP as in FIG. 20. In step 8), SgNB.CU-CP sends ‘SgNB Addition Request Acknowledge’ message to MeNB (and specifically to MeNB.CU-CP of MeNB).

[0046] GTP-U tunnel, which is provided to carry traffic between SgNB.CU-UP and SgNB.DU for 5G leg of SgNB terminated split bearer, is also established as part of the steps shown in FIG.20.

[0047] FIG. 21 shows some additional steps where GTP-U tunnel is provided to carry traffic on a 4G leg is established between SgNB.CU-UP and MeNB (or specifically between SgNB.CU-UP and MeNB. DU for disaggregated base station architecture). It also shows where UE carries out random access procedures with SgNB to directly start receiving data via SgNB (for the 5G leg).

[0048] In step 9) of FIG. 21, MeNB, specifically MeNB.CU-CP of MeNB, sends RRC Connection Reconfiguration to (EN-DC) UE. In step 10), this UE responds with ‘RRC Reconfiguration Complete’ message. In step 11), MeNB, specifically MeNB.CU-CP of MeNB, sends ‘SgNB Reconfiguration Complete’ to SgNB.CU-CP. As part of step 12), UE carries out a random access procedure with SgNB. In step 13), MeNB, specifically MeNB.CU-CP of MeNB, sends ‘SN Status Transfer’ to SgNB-CU-CP’ (if the bearer is using RLC AM).

[0049] Data forwarding from MeNB (specifically from MeNB.CU-UP of MeNB) to SgNB.CU-UP happens as part of steps 14) and 15) in FIG. 21. Next, SgNB.CU-UP starts splitting traffic across 5G and 4G legs of the network. DL Data towards 4G leg is sent as part of steps 16) and 17). DL data towards 5G leg is sent as part of steps 18) and 19). As part of step 20), MeNB interacts with 4G EPC to carry out path switch to make it SN terminated split bearer. 4G EPC starts sending DL data towards SgNB.CU-UP as part of step 21) after the path switch operation. SgNB.CU-UP continues to split traffic across 4G and 5G legs of the network.

[0050] The inter-DU interface can be used for DU to DU communication. One such interface is denoted as the D2 interface. Protocol used for inter-DU communication is called D2-AP (D2-Application Layer) here. This is used for control plane information exchange between MeNB.DU and SgNB.DU in the EN-DC architecture.

[0051] In this section, a general overview of reinforcement learning (RL) is provided.Reinforcement Learning is a feedback-based machine learning technique in which an agent learns to behave in an environment by performing actions and seeing the results of the actions. For each good action, the agent gets a positive feedback or reward and for each bad action, the agent gets a negative feedback or penalty. The goal of the agent is to use RL algorithms to learn the best policy as it interacts with the environment so that, given any state, it will always take the most optimal action to produce the least cost (or the maximum reward) in the long run.

[0052] Some of the terms used in connection with reinforcement learning technique are listed below:AgentQ: An entity that interacts with the environment and acts upon it.Environment(): A situation in which an agent is present or surrounded by. In RL, a stochastic environment is assumed, which means it is random in nature.Action(): Actions are the moves taken by an agent within the environment.State(): State is a situation returned by the environment after each action taken by the agent.Cost(): A feedback returned to the agent from the environment to evaluate the action of the agent.Policy(): Policy is a strategy applied by the agent for the next action based on the current state.Value(): It is an expected long-term cost with a discount factor.Q-value(): It is similar to Value() but it takes one additional parameter as the current action ‘a’.

[0053] In this section, a Markov Decision Process (MDP) is discussed. A MDP is used to formalize the RL problems. If the environment is completely observable, then its dynamics can be modelled as a Markov Process. In MDP, which is illustrated in FIG. 22, the agent constantly interacts with the environment and performs actions. At each action, the environment generates a new state and responds with a reward (or penalty). MDP contains four elements: States, Actions, Costs (or Rewards), and Transition Probabilities. A finite MDP is when there are finite number of states, finite costs, and finite number of actions. Provided below is a simple set notation to represent each of the elements in an MDP:S: A set of finite States S.A: A set of finite Actions A.C(s, a): Immediate cost (or expected immediate cost) incurred after transitioning from state s to state s', due to action ‘a’.P: represents the Transition probability matrix corresponds to state space S and action space A.P(s’|s, a): Transition Probability of landing in state s' when action ‘a’ is taken at state s.

[0054] In this section, some of the approaches used in RL, such as, Value-based approach (Value iteration methods), Q-learning, Deep Q Neural Network (DQN), and Policy-based approach (Policy iteration methods) are discussed.

[0055] The value-based approach is about finding an optimal value function, which is the optimal value at a state under any policy it. The below system of equations for the state space are called Bellman equations or optimality equations. These characterize the values and the optimal policies in infinite-horizon models:V(s’) = min{C(s, a) + y / P(sf| s, a)V(s')}a / *sWhere, V(s) is value at state s, C(s,a) is immediate cost at state s for action a, y is a discount factor, P(s’|s,a) is a transition probability of landing in state s’ when action a is taken at state s, and V(s’) is value at state s’.

[0056] Q-learning involves learning the value function Q (s, a), which quantifies whether it is advantageous to take an action "a" at a particular state "s”. The main objective of Q-learning is to learn the policy that can inform the agent what actions should be taken for minimizing the overall cost. The goal of the agent in Q-learning is to optimize the value of Q where the value of Q-learning can be derived from the Bellman equation. Instead of using a value at each state, a Q-value, Q(s,a), is used for a pair of state and action. Q-value specifies which action is more beneficial than the other actions and according to the best Q-value, the agent takes its next move.

[0057] After performing an action “a”, the agent will incur a cost C(s, a), and the agent ends up at a certain state. The Q-value equation is as follows:Q(s, a) = C(s, d) -I- y ^(P(s'|s, a)mina[Q(s' , a)])s'

[0058] The flowchart shown in FIG. 23 illustrates the workings of Q-learning. In block 2301, the Q-table is initialized. In block 2302, an action to perform is selected. In block 2303, the selected action is performed. In block 2304, the associated cost for the action is found. In block 2305, the Q-table is updated.

[0059] Deep Q Neural Network (DQN) is Q-learning using Neural networks. For a big state space environments, it is a challenging and complex task to define and update a Q-table. To solve such an issue, a DQN algorithm can be used. In this approach, instead of defining a Q-table, neural network approximates the Q-values for each action and state.

[0060] A policy-based approach is used to find an optimal policy for the minimum future cost without using the value function. This approach involves two types of policies: 1) Deterministic policy where the same action is produced by the policy for any given state, and 2) Stochastic policy where for each state there is an inverse (probability) distribution over set of actions possible at that state.

[0061] For DL split bearer in NSA architecture, the data split at the DRB level is typically performed at the SgNB.CU-UP into a 4G leg and a 5G leg to corresponding DUs as illustrated in Fig. 19. For simplicity, one active DRB for UE may also be assumed (but the methods specified here are applicable for more generic scenarios also).

[0062] For each UE, a PDCP reordering queue is maintained at the PDCP layer for each DRB. The packets received in a specific window will be reordered and sent to upper layers in the UE. When the data is sent from 4G DU and 5G DU to the UE for a particular DRB , the received packets at UE may be out of order. There are multiple reasons for this including, for example, mid-haul delay between the 4G DU and the SgNB.CU-UP can be different from the mid-haul delay between the 5G DU and the SgNB.CU-UP. Additionally, RLC queuing delays at corresponding DUs can be different, and the channel conditions of the active cells under the 4G and 5G DUs can also be different for the UE. Packet reordering can lead to serious throughput degradation especially for applications running over TCP.

[0063] Measuring packet reordering involves analyzing the arrival sequence of packets at the UE in downlink and quantifying the extent of out-of-order delivery. Several techniques and metrics are commonly used to measure packet reordering, for example, packet sequence number analysis, inter packet arrival time analysis, and the like. By examining the sequence numbers of received packets, one can determine the degree of reordering. Metrics such as the gap size (the difference between expected and received sequence numbers) and the number of out-of-order packets can provide insights into the level of packet reordering. As part of this, UE need to have the capability of measuring the packet reordering measurements.

[0064] Accordingly, there is a need for a system and a method that overcomes, alleviates, and / or mitigates one or more of the aforementioned and other deleterious effects of prior art relating to degradation due to packet reordering in a split bearer configuration.SUMMARY

[0065] The problem domain pertains to the optimization of downlink split-bearer operations in Open Radio Access Network (0-RAN) systems, particularly in scenarios involving dual connectivity across 4G and 5G Distributed Units (DUs). Conventional approaches to split-bearer operations rely on static allocation policies or simple heuristics based on coarse metrics such as link quality indicators or cell load. These methods fail to effectively account for dynamic variations in mid-haul delays, queue depths, and packet sequencing at the Data Radio Bearer (DRB) level. As a result, packet reordering issues arise at the User Equipment (UE), leading tohead-of-line blocking, throughput degradation, and compromised Quality of Service (QoS), especially for delay-sensitive applications. Furthermore, existing Artificial Intelligence and Machine Learning (AI / ML) solutions in O-RAN systems primarily focus on cell-wide optimizations or aggregated Performance Indicators (Pls) without addressing granular DRB-level sequencing feedback or per-UE reordering metrics, leaving significant inefficiencies unaddressed.

[0066] The inventive concept introduces a dynamic, closed-loop optimization framework leveraging AI / ML techniques to address these limitations. By incorporating packet sequencing feedback from the UE, mid-haul latency variance, queue depth metrics, and per-DRB delivery status into the decision-making process, the system dynamically adjusts the distribution of downlink packets across multiple radio legs. This is achieved through reinforcement learningbased policies implemented as near-real-time xApps on the Radio Intelligent Controller (RIC) or directly hosted at CU-UP. The solution enables fine-grained control over split-bearer operations, preemptively mitigating packet reordering issues and improving instantaneous throughput under bursty or delay-sensitive traffic conditions. Additionally, the system enhances spectral efficiency and QoS by correlating real-time network metrics with adaptive split-bearer policies, ensuring optimal packet sequencing and delivery across heterogeneous RAN architectures.

[0067] According, what is desired is a system and a method for accounting for out of order packets in a split bearer O-RAN wireless system that minimizes throughput degredation.

[0068] It is further desired to provide a system and a method that utilizes Artificial Intelligence and Machine Learning (AIML) to optimize spit bearer operation in O-RAN networks.

[0069] It is still further desired to provide a system and a method that uses AIML assisted optimization of DL split bearer operation in an O-RAN network to minimize throughput degredation due to packet data misorder received at UE.

[0070] It is yet further desired to provide a system and a method that minimizes throughput degredation of received data packets from a 4G DU and a 5G DU in an O-RAN wireless system due to mismatched delays in the split system leading to misordered data packets.

[0071] These and other objectives are achieved by considering packet reordering information, delay at the DRB level (at DU and across midhaul for each UE), CQI and rank at the UE level, and cell load measurements at the SgNB.CU-UP in deciding a policy. Packet reordering information at the DRB level, latency at the DRB level for multiple UEs along with CQI, rank, cell load and other parameters are communicated to the CU-UP to obtain a split of data across different (4G and 5G) legs. In particular, however, Reinforcement Learning (RL)-based DL split bearer policy is used to improve split-bearer performance.

[0072] One method communicates packet reordering information from a UE to the corresponding MeNB.CU-CP (or LTE CU-CP) by enhancing the RRC protocol. Packet reordering measurements for a DRB from each UE are communicated to its corresponding MeNB. CU-CP with RRC protocol using an enhanced UEInformationRequest message and an enhanced UEInformationResponse message. As in FIG. 24 and FIG. 25, MeNB.CU-CP sends enhanced UEInformationRequest message to UE asking for PDCP reordering related measurements and the UE provides this information to MeNB.CU-CP with enhanced UEInformationResponse mssage. MeNB.CU-CP could ask for this information periodically from UE. Alternatively, MeNB.CU-CP could ask for this information aperiodically from the UE. For example, the packet reordering measurements received as part of UEInformationResponse message (at the MeNB.CU-CP) that exceed a threshold or where misordering ratio is above a threshold can cause the UEInformationRequest message to be sent more frequently from MeNB.CU-CP to UE than a defined periodicity.

[0073] Another method considers communication of packet reordering information as part of a Minimization of Drive Test (MDT) report from the UE to the MeNB.CU-CP over RRC. In this method, the packet reordering information at the DRB level is sent as part of MDT measurements.

[0074] As part of Immediate MDT procedure (when UE is in RRC Connected state), the packet reordering measurements at DRB level are communicated to MeNB.CU-CP over RRC using enhanced UEInformationRequest and UEInformationResponse messages as in FIG. 26 and FIG.27

[0075] Still another method considers communication of packet reordering information from UE to SgNB.DU (i.e. 5G NR DU) using a new MAC CE. The reserved LC ID values (e.g. LC codepoint / index from 37 - 42 from FIG. 28) can be used for this packet reordering MAC CE (as part of MAC CE header).

[0076] An example is shown in FIG. 29 for one of the MAC CE format where for a split-bearer DRB, the DRB ID is sent using the six bit field in the first octet and the next three octets are used to encode the packet reordering information.

[0077] Another MAC CE format is defined such that different DRB IDs are used in MAC subheader to identify different PDCP reordering measurements sent over uplink for different DRBs for that UE (as in FIG. 30).

[0078] For a UE with up to 8 split-bearer DRBs, it is shown in FIG. 30. Here, the first octet indicates a configuration status of the split bearer feature for DRB i, where i is the ascending order of the DRB ID among the DRBs configured with the split bearer feature. For the DRBs with the split bearer feature configured the corresponding packet reordering information will be sent in the next three octets in the ascending order of configured DRB ids.

[0079] Yet another method considers an RL-based approach to optimize DL split bearer operation. Using the measurements from a number of UEs, the splitting policy is derived at the CU-UP (e.g. at SgNB.CU-UP for SN terminated split bearer) for DL data split across LTE and 5G NR legs. Essentially, the DL split bearer splitting problem is mapped to a reinforcement learning (RL) based problem. This is accomplished by forming a Markov Decision Process (MDP) which includes, states, actions, transition probabilities, costs / rewards. All these quantities are be defined for this splitting problem.

[0080] For this application the following terms and definitions shall apply:

[0081] The term “network” as used herein includes both networks and internetworks of all kinds, including the Internet, and is not limited to any particular type of network or inter-network.

[0082] The terms “first” and “second” are used to distinguish one element, set, data, object or thing from another, and are not used to designate relative position or arrangement in time.

[0083] The terms “coupled”, “coupled to”, “coupled with”, “connected”, “connected to”, and “connected with” as used herein each mean a relationship between or among two or more devices, apparatus, files, programs, applications, media, components, networks, systems, subsystems, and / or means, constituting any one or more of (a) a connection, whether direct or through one or more other devices, apparatus, files, programs, applications, media, components, networks, systems, subsystems, or means, (b) a communications relationship, whether direct or through one or more other devices, apparatus, files, programs, applications, media, components, networks, systems, subsystems, or means, and / or (c) a functional relationship in which the operation of any one or more devices, apparatus, files, programs, applications, media, components, networks, systems, subsystems, or means depends, in whole or in part, on the operation of any one or more others thereof.

[0084] As used herein, the phrases "at least one", "one or more" ,"or" and "and / or" are open-ended expressions that are both conjunctive and disjunctive in operation. For example, each of the expressions "at least one of A, B and C" ,"at least one of A, B, or C" ,"one or more of A, B, and C", "one or more of A, B, or C", "A, B, and / or C", and "A, B, or C" means A alone, B alone, C alone, A and B together, A and C together, B and C together, or A, B and C together.

[0085] In one configuration, a method for minimizing throughput degredation of misordered data packets received by a User Equipment (UE) from a 4G Distributed Unit (DU) and a 5G DU in a split bearer Open Radio Access Network (O-RAN) wireless system due to mismatched delays in the 4G and 5G systems is provided, the method comprising the steps of: transmitting a UEInformationRequest message from a 4G Central Unit-Control Plane (CU-CP) server to the UE via enhanced Radio Resource Control (RRC) protocol, and transmitting packet reordering information from the UE to the 4G CU-CP server via RRC protocol using a modified UEInformationResponse message. The method is provided such that triggers for the UEInformationRequest message are periodic or event based. The method is further provided such that when the trigger is periodic, the UEInformationRequest message is sent at the end of a measurement collection period and Packet Data Convergence Protocol (PDCP) reordering measurements are transmitted in the modified UEInformationResponse message. The method is still further provided such that when the trigger is event based, reordering measurements are received for the number of reordered packets at Data Radio Bearer (DRB) level, wherein if thenumber of reordered packets exceeds a threshold or exceeds a misording ratio, the modified UEInformationResponse message is transmitted based on the event of exceeding the threshold or the misording ratio.

[0086] In another configuration, a method for optimizing downlink split-bearer delivery in a disaggregated radio access network using packet sequencing feedback from a user equipment and network delivery metrics is provided comprising a packet sequencing feedback module receiving sequencing feedback from the user equipment and a network delivery metrics module obtaining network delivery metrics for downlink packet transmissions. The method further comprises an artificial intelligence and machine learning policy module selecting a distribution of downlink packets across plural radio legs based on the sequencing feedback and the delivery metrics and the artificial intelligence and machine learning policy module dynamically adjusting the distribution of downlink packets across the plural radio legs so as to reduce out-of-order delivery at the user equipment. The method is provided such that the packet sequencing feedback module, the delivery metrics module, and the artificial intelligence and machine learning policy module cooperate to optimize downlink split-bearer delivery in the disaggregated radio access network.

[0087] In a further configuration a method is provided for optimizing downlink split-bearer delivery in a radio access network comprising receiving feedback from a user equipment, obtaining network metrics for downlink transmissions, selecting a distribution of downlink packets across multiple radio legs based on the feedback and the network metrics, and dynamically adjusting the distribution of downlink packets across the multiple radio legs to reduce out-of-order delivery at the user equipment.

[0088] In still another configuration a system is provided for optimizing downlink split-bearer delivery in a radio access network, comprising a feedback receiver configured to receive feedback from a user equipment and a metrics module configured to obtain network metrics for downlink transmissions. The system further comprises a policy module configured to select a distribution of downlink packets across multiple radio legs based on the feedback and the network metrics and the policy module further configured to dynamically adjust the distributionof downlink packets across the multiple radio legs to reduce out-of-order delivery at the user equipment.

[0089] The above-described and other features and advantages of the present disclosure will be appreciated and understood by those skilled in the art from the following detailed description, drawings, and appended claims.DESCRIPTION OF THE DRAWINGS

[0090] FIGS. 1A and IB illustrate a 5G NR user and control plane function with monolithic gNB according to the prior art.

[0091] FIG. 2 illustrates the control plane, RRC, PDCP, RLC, MAC and PHY sublayers that terminate in the gNB, and NAS that terminates in the AMF according to the prior art.

[0092] FIGS. 3 and 4 illustrate NG-RAN architecture according to the prior art.

[0093] FIGS. 5 - 7 illustrate an overview of L2 of 5G NR according to the prior art.

[0094] FIG. 8A illustrates and overview of 0-RAN structure according to the prior art.

[0095] FIG. 8B is a flow map for communication via the E2 node of a CU or DU and a Near-RT-RIC for control request according to the prior art.

[0096] FIG. 9 illustrates an example PDU session comprising multiple DRBs, where each DRB may comprise multiple QoS flows according to the prior art.

[0097] FIGS. 10 and 11 illustrate 3GPP 5G network architecture according to the prior art.

[0098] FIG. 12 illustrates a CU-UP sending DL User Data (DUD) to a DU according to the prior art.

[0099] FIG. 13 illustrates flow control feedback from a CU-UP to a DU according to the prior art.

[0100] FIG. 14 illustrates a DU sending Assistance Information Data (AID) to a CU-UP according to the prior art.

[0101] FIG. 15 illustrates E-UTRA-NR Dual Connectivity (EN-DC) architecture according to the prior art.

[0102] FIG. 16 illustrates data (user) plane for the EN-DC architecture according to the prior art.

[0103] FIG. 17 illustrates a Master Cell Group, Secondary Cell Group and split bearers from a UE perspective according to the prior art.

[0104] FIG. 18 illustrates MCG, SCG and split bearers from base station perspective according to the prior art.

[0105] FIG. 19 illustrates E-UTRA-NR Dual Connectivity architecture with downlink split bearer operation according to the prior art.

[0106] FIG. 20 illustrates a MCG bearer established for a UE and the MeNB converts to SgNB terminated split bearer according to the prior art.

[0107] FIG. 21 illustrates a GTP-U tunnel carrying traffic on a 4G leg established between SgNB.CU-UP and MeNB according to the prior art.

[0108] FIG. 22 illustrates a Markov Decision Process in which an agent constantly interacts with the environment and performs actions according to the prior art.

[0109] FIG. 23 is a flowchart illustrating the workings of Q-learning according to the prior art.

[0110] FIGS. 24 and 25 illustrate packet reordering information communicated to LTE CU-CP over RRC using enhanced UEInformationRequest and UEInformationResponse messages according to one configuration.

[0111] FIGS. 26 and 27 illustrate packet reordering information communicated to LTE CU-CP over RRC using enhanced UEInformationRequest and UEInformationResponse messages according to one configuration.

[0112] FIG. 28 illustrates different LC IDs used in MAC sub header to identify different MAC CEs sent over uplink.

[0113] FIGS. 29 and 30 illustrate two different formats to send MAC CE from the UE according to FIG. 28.DETAILED DESCRIPTION

[0114] The present detailed description provides an in-depth explanation of the subject matter, which pertains to the optimization of downlink split-bearer operations in Open Radio Access Network (0-RAN) systems, leveraging Artificial Intelligence and Machine Learning (A1ML) techniques. The subject matter is situated in the broader field of wireless communication technologies, particularly focusing on enhancing data delivery efficiency and minimizing throughput degradation in disaggregated RAN architectures. While specific embodiments and examples are provided herein to illustrate the principles and applications of the described concepts, these are not intended to limit the scope of the subject matter. Instead, they serve as representative implementations to aid in understanding the described ideas.

[0115] Certain well-understood details, such as standard protocols, interfaces, and configurations within 0-RAN and 5G systems, may be omitted to enhance brevity and clarity. Such omissions are intended to emphasize the novel aspects of the described subject matter without compromising the overall disclosure. Furthermore, the described subject matter includes various modifications, rearrangements, and adaptations that may be apparent to those skilled in the art, provided they align with the scope of the appended claims.

[0116] While the discussion for simplicity is directed to a one-to-one map between QoS flow to a DRB where one QoS flow per DRB and one active DRB for UE are assumed, the suggested methods are applicable when there are multiple QoS flows mapped to a DRB and multiple bearers for a given UE.

[0117] Method 1A: Communication of packet reordering information from a UE to the MeNB.CU-CP (i.e. LTE CU-CP) overRRC.

[0118] The packet reordering measurements at the DRB level from each UE is communicated to MeNB.CU-CP over RRC using UEIiiformationRequest andUEInformationResponse messages. These messages will be enhanced for this purpose as shown in FIG. 24 and FIG. 25.

[0119] The triggers for UEInformationRequest can be be periodic, or event based. In the case of periodic trigger, the UEInformationRequest will be sent periodically and the PDCP reordering measurements will be sent to the MeNB.CU-CP as part of UEInformationResponse message. In the case the event based trigger, as part of UEInformationResponse message, the reordering measurements are received for the number of reordered packets for a DRB (based on certain events). The system determines if the absolute number of out of order PDCP packets exceeds a certain threshold or if misordering ratio (i.e., the ratio of number of out of order packets versus the total number of packets received) is above a threshold. If so, the UEInformationRequest message will be sent based on that event rather than at the defined periodicity.

[0120] Method IB: Communication of packet reordering information as part of MDT (Minimization of Drive Test) report from UE to MeNB.CU-CP over RRC.

[0121] In this method, the packet reordering information at the DRBlevel is sent as part of MDT measurements. There are two modes for the MDT measurements: Logged MDT and Immediate MDT. In the case of Immediate MDT, the measurements are performed by the UE in RRC-CONNECTED state. In Immediate MDT, the RAN measurements and UE measurements can be configured. The configuration for UE measurements is based on the existing RRC measurement procedures for configuration and reporting.

[0122] For the case of Immediate MDT, as part of MDT measurement procedure UE should be able to measure packet reordering measurements in this enhanced MDT method. That is, each (enhanced) MDT-capable UE should be able to measure the packet reordering information at the DRB level. In this case, the measurement trigger can be periodic or event based. In the case of periodic trigger, the MDT measurements will be sent to MeNB.CU-CP periodically and at the the end of measurement collection period. In the case of the event based trigger, when the packet reordering measurements received as part of UEInformationResponse message exceeds a certain threshold or the misordering ratio (i.e., the ratio of number of out oforder packets to the total number of packets received) exceeds a threshold, theI JEInformationRequest message will be sent more frequently than the defined periodicity.

[0123] As part of Immediate MDT procedure, the packet reordering measurements at DRB level from each UE will be communicated to MeNB.CU-CP over RRC using UEInformationReqi est and UEInformationResponse messages. These messages will be enhanced for this purpose as shown in FIG. 26 and FIG. 27.

[0124] Method IC Communication of packet reordering information from UE to 5GNR DU using an enhanced MAC Control Element (MAC CE):

[0125] In this method, the packet reordering measurements at the DRB level for each UE are sent to a DU using an enhanced MAC CE. A new MAC CE is defined for this purpose. FIG.28 shows different LC IDs used in MAC sub header to identify different MAC CEs sent over UL (uplink) from UE to DU. The reserved LC ID values, as shown in the FIG. 28, will be used for this Packet reordering MAC CE. Two different formats are defined to send this MAC CE from the UE are shown in FIGS. 29 and 30.

[0126] In the first format, for the DRB with split bearer feature enabled as shown in FIG.29, the DRB ID will be sent using the six-bit field in the first octet where the next three octets are used to encode the packet reordering information.

[0127] In the second format, the first octet shown in FIG. 30 indicates the configuration status of the split bearer feature for DRB i, where i is the ascending order of the DRB ID among the DRBs configured with the split bearer feature. For the DRBs with the split bearer feature configured, the corresponding packet reordering information will be sent in the next three octets in the ascending order of configured DRB ids.

[0128] Similarly, the packet reordering measurements will be sent to the LTE DU, instead of the NR DU over a MAC CE. A similar new LTE MAC CE is defined for this purpose.

[0129] Communication of packet reordering information to SgNB.CU-UP (i.e. 5GNR CU-UP) in Method IA, Method IB and Method IC:

[0130] In Method IA and Method IB, the packet reordering measurements from the MeNB.CU-CP will be sent to the SgNB.CU-CP over X2-C interface. The SgNB.CU-CP will send these measurements to the SgNB.CU-UP over the El interface. X2-C and El interfaces are therefore enhanced for this purpose.

[0131] In Method IC, the packet reordering measurements at DRB level from the NR DU (or the LTE DU) will be sent to the SgNB.CU-UP over the Fl-U (or X2-U) interface using AID PDU. The Fl-U (or X2-U) interface is therefore enhanced for this purpose.

[0132] Communication of other relevant parameters to the SgNB.CU-UP in Method IA, Method IB and Method IC:

[0133] Here the DRB level (for example, DL latency), UE level (for example, CQI and rank) and Cell level (for example, cell load and cell bandwidth) parameters are communicated from the eNB.DU to the SgNB.CU-UP. The X2-U interface is enhanced for this purpose.

[0134] Similarly, the DRB level (for example, DL latency, packet reordering measurements received from the UE ), UE level (for example, CQI and rank ), Cell level (for example, cell load and ceee bandwidth ) parameters are communicated from the SgNB.DU to the SgNB.CU-UP. The Fl-U interface is therefore enhanced for this purpose.

[0135] In another embodiment, the measurements cell load, CQI, rank, latency from the MeNB.DU to the MeNB.CU-CP will be communicated over a VI interface. The MeNB.CU-CP communicates these measurements to the SgNB.CU-CP over the X2-C interface. The SgNB.CU-CP communicates these parameters to the SgNB.CU-UP over the El interface. The measurements from the SgNB.DU will be communicated to the SgNB.CU-CP over the Fl-C interface. The SgNB.CU-CP communicates these measurements to the SgNB.CU-UP over the El interface. The X2-C, VI, Fl-C and El interfaces are therefore enhanced for this purpose.

[0136] Method ID: Reinforcement Learning (RL) based approach to optimize DL split bearer operation.

[0137] Using the measurements from the number of UEs, a splitting policy is derived at the SgNB.CU-UP for DL data split across the LTE and the 5G NR legs.

[0138] The DL split bearer splitting problem is mapped to a reinforcement learning (RL) based problem. In order to do this, a Markov Decision Process (MDP) is used. A MDP comprises states, actions, transition probabilities, and costs / rewards. These quantities are all defined here for the current splitting problem.

[0139] States (S): For a particular UE (e.g., UEi), the state Si is represented as following:- Packet reordering measurements for each split DRB for each UE, for example the number of out of order packets in an interval of time or the misordering ratio (the ratio of number of out of order packets to the total number of packets received for a given time interval).CQI as reported by the UE and available at the 4G DU and the 5G DU.Rank as reported by the UE and available at the 4G DU and the 5G DU.- Delay experienced by the packets of a given split-bearer DRB at the 4G DU and the 5G DU.Cell load in terms of percentage of PRBs utilized for 4G and 5G cells (and available at the 4G DU and the 5G DU respectively).Achieved throughput for that DRB .

[0140] The range of values taken by each state variable is quantified (or classified) to n levels, where n is a finite value (e.g., n=4 or 16 or higher number, depending on the parameter being quantized). State S is the array (or vector) of states of all the UEs in the system. For example, it is represented as S= (SI, S2..., SK) forK UEs

[0141] Action (A): The action space is considered in the range of 0 to 100 for each UE, which is a finite number. UEi has action space as an integer number from the range of values in 0 to 100, i.e. ai(t)c {0,1,2,...,100}. Here, ai(t)=z (e.g. z = 20) implies the data split at the SgNB.CU-UP is z% on LTE leg and (100 - z)% on 5G NR leg, i.e., the SgNB.CU-UP splits data as (z%, (100-z)%) on the LTE and NR legs, respectively. The action space can be further reduced by quantizing it to finite number of action levels, for example, range of values for ai(t) as a multiple of 10, i.e., ai(t)c {0,10,20,...,100}. The common action across all UEs in the cell is represented as an array (or vector) of actions for each individual UE i.e., (al, a2... , aK) for K UEs.

[0142] Transition probabilities (P): For the K UEs in the system (i.e. in a cell), the realized state at time t is represented as s= (si, s2.. ,,sK) and the corresponding realized action at time t is a= (al,a2.. ,,aK). Transition probability is the probability of reaching to state u= (ul,u2.. ,,uK) at time t+1. It is mathematically represented as P(S(t+l) =(ul,u2.. ,,uK)|S(t)= (si, s2... ,sK), A(t)= (al,a2... ,aK)). Note that the state space is finite, its cardinality | S |=b and the action space is also finite. So, the transition probability matrix is of finite size too. For example, if there six state variables in the state space and they are quantized into nl, n2, n3, n4, n5 and n6 levels respectively, the state space cardinality is |S|=b= (nl*n2*n3*n4*n5*n6)*K for K UEs. If the number of possible actions at each UE state are H, it implies H*K possibilities. So, the transition probability matrix is of size (nl*n2*n3*n4*n5*n6*H)*K.

[0143] Cost (C(s,a)): At time t, the immediate cost for UE ‘i’ when this UEi is in state ‘s’ with action ‘a’ is represented as C(si(t)=s, ai(t)=a)= C((s(t), a(t)). The cost of all the UEs in the system C(s, a)= C((sl, s2.. ,,sK), (al, a2..., aK)) is simply a sum over all the individual costs,i.e., C(s, a) = Xjli C'(s(t), a(t)). For each UE, the cost can be written as a function of costs associated with the state variables as shown below. For GBR (Guaranteed Bit Rate) and non- GBR bearers, the cost function is defined separately. Goal is to minimize the overall cost function across all active DRBs (or UEs). It is expected that usually splitting to be applied for non-GBR bearers though application of this for GBR bearres in some scenarios is not ruled out.

[0144] The cost C1(s(t), a(t)) for UEi is defined as follows:C1(s(t), a(t))= Wj * CTHP+ W2* CpR’ (s(t), a(t)) + w3* CDELLTE '(s(t),a(t)) +w4 * CDELNR'CSCI), a(t))where w1(w2, w3, w4are the weights associated with the cost funtions defined below.

[0145] For GBR bearer the throughput cost function will be CTHP‘(s(t), a(t)) = CTHPGBR'CSCO / a(t)) and it is defined as follows:

[0146] CTHpGBR’(s(t), a(t)): The immediate cost associated with the throughput for GBR DRB is a function of the guaranteed target throughput and the achieved throughput. After taking action ‘a’, if the achieved throughput meets (or exceeds) the target throughput, then theassociated immediate cost is zero. After taking action ‘a’, if the achieved (or observed) throughput goes below the target throughput, then the squared difference of these observed and target throughput values is the immediate cost. In another variant, the absolute difference of target and achieved throughput is defined as the immediate cost if the achieved throughput goes below the target throughput, and zero if the achieved throughput is more than the target throughput.

[0147] For non-GBR bearer the throughput cost function will be CTHP‘(s(t), a(t)) = ^THPNGBR1(^Ct)>a(t)) and it is defined as follows:

[0148] CTHPNGBR'(S(1-)’a(t)): The immediate cost associated with the throughput for non-GBR DRB is a function of the achieved throughput. After taking action ‘a’, if the achieved throughput increases compared to the previously achieved throughput, then the immediate cost is zero. After taking action ‘a’, if the achieved throughput decreases, then the squared difference between the two instances of achieved throughput i.e., before and after the action ‘a’ is the immediate cost.

[0149] CPR'(s(t), a(t)): The immediate cost associated with the packet reordering measurements for a DRB is a function of the packet reordering measurements at t and t+1. For example, if the number of reordering packets at time t is ‘X’, and after taking action ‘a’ the number of reordering packets at time t+1 is ‘ Y’, then the immediate cost at time t is a function of X and Y, denoted as f(X, Y). If the number of packets reordered decreases after taking an action ‘a’, then the immediate cost is zero. If the number of reordered packets increases after taking action a, then the squared difference is taken as the immediate cost. In another configuration, the immediate cost is defined as the absolute difference as C(S(t), a(t))= max(0, Y-X).

[0150] DU delay budget for a DRB at CU-UP is defined as the packet delay budget for that DRB (i.e. PDB as per standardized QoS tables) minus “sum of midhaul delay, backhaul delay and processing delay at CU-UP and in the core network”.

[0151] CDELLTE'(s(t)’a(t)): For a particular DRB , after taking action ‘a’, if more than the threshold percentage (e.g., X as threshold) of packets are served at LTE DU within their LTEDU delay budget, then the delay-related immediate cost is zero. After taking action ‘a’, if Y (where, Y<X) percentage of packets are served within their LTE DU delay budget, then the square of the average DU delay for sending the remaining X-Y packets is the immediate cost.

[0152] CDELNR’ a(t)): For a particular DRB, after taking action ‘a’, if more than the threshold percentage (e.g., X as threshold) of packets are served in their 5G NR DU delay budget, then the delay-related immediate cost is zero. After taking action ‘a’, if Y (where, Y<X) percentage of packets are received in their NR DU delay budget, then the square of the average 5G NR DU delay for sending the remaining X-Y packets is the immediate cost.

[0153] For other services (with less stringent delay requirements), the delay-related cost functions are based on average delay as defined below:

[0154] CDELLTE‘(s(t), a(t)): The immediate cost associated with the average delay on an LTE leg for a given DRB after taking action ‘a’ is defined as follows. If the average delay decreases, then the immediate cost is zero. If the average delay increases, then the squared increase in average delay will be the immediate cost.

[0155] CDELNR'(s(t), a(t)): The immediate cost associated with the average delay on a NR leg for a given DRB after taking action ‘a’ is defined as follows: If the average delay decreases, then the immediate cost is zero. If the average delay increases, then the squared increase in average delay will be the immediate cost.

[0156] Splitting policies are derived at the CU-UP, which minimizes the overall cost function. The overall cost function is obtained by summing the individual cost functions across all the active UEs. Individual cost functions for each UE can be calculated using a value iteration method.

[0157] From a value function, it can be observed that for each state there is an associated immediate cost, and additional discounted costs associated with the subsequent state action pairs. The discount factor essentially determines how much the RL agents weigh costs in the distant future relative to those in the immediate future. Normally, discount factor ranges from 0 to 1, i.e., Y e [0,1).

[0158] For value iteration for an unknown transition probability matrix, initialize the matrix with zero and update it as follows: At state ‘s’, after taking action ‘a’, the system lands in state ‘u’, then update P(u|s,a) = 1. Later at the same state ‘s’, taking the same action ‘a’, the system lands at ‘v’, then update P(u|s, a) = P(v|s, a) = 0.5. Update the transition probabilities based on the different states it has landed on, and based on a number of times it has landed on each state. For example, for state ‘s', taking action ‘a’ after a long run (e.g. 100 times), it landed on u, v, w in 20, 30, 50 time,s respectively. The updated transition probabilities are given as follows: P(u|s, a) = 0.2, P(v|s, a) = 0.3, P(w|s, a) = 0.5. This can be referred to as a learning phase of the MDP.Step 1: Initialize Voc V (e.g., initialize with zero) for each state. Specify s>0 (a small positive value) and set n=0 (iteration).Step 2: For each state scS, compute Vn+i(s, a) by,Vn+i(s,a) = min{C(s, a) AyY P(u|s, a) Vn(s')}aeA / ‘s'Step 3: If ||Vn+i(s, a) - Vn(s, a)|| < £*(1- y) / 2y, go to Step 4. Otherwise increment n by 1 and return to Step 2.Step 4: For each srS, choose an optimal action as argmin{ C(s, a)+ y P(u|s, a) Vn(u)} aeAand stop.

[0159] In conventional disaggregated radio access networks, optimizing downlink splitbearer delivery is challenged by the lack of real-time, granular feedback regarding packet sequencing and network delivery metrics. Static or heuristic-based split-bearer policies often fail to adapt to dynamic changes in latency, throughput, cell load, and mid-haul delays, resulting in out-of-order packet delivery at the user equipment (UE). This leads to head-of-line blocking, increased retransmissions, and degraded throughput and quality of service. Existing solutions do not incorporate per-DRB packet reordering measurements or leverage advanced artificial intelligence and machine learning (AI / ML) techniques for dynamic policy adjustment, limiting their effectiveness in heterogeneous and rapidly changing network environments.

[0160] The inventive solution introduces a system comprising a packet sequencing feedback module, a delivery metrics module, and an AI / ML policy module. Sequencingfeedback from the UE, including packet reordering measurements at the DRB level, and network delivery metrics such as latency, throughput, and cell load, are collected and analyzed. The AI / ML policy module, employing reinforcement learning algorithms, selects and dynamically adjusts the distribution of downlink packets across plural radio legs in response to real-time feedback and network metric changes. Sequencing feedback is received periodically or in response to event triggers, enabling closed-loop optimization. The policy module is implemented as a near-real-time application on a radio intelligent controller, ensuring adaptive and efficient downlink split-bearer delivery that reduces out-of-order packet delivery and improves overall network performance.

[0161] In one embodiment, the packet sequencing feedback module is implemented as a software application integrated into the user equipment, capable of periodically transmitting packet reordering measurements to the network. Alternatively, the module can operate in an event-driven mode, where feedback is triggered by specific conditions such as exceeding a threshold of out-of-order packets. In another embodiment, the delivery metrics module (at CU-UP) is configured to collect latency, throughput, and cell load data from both 4G and 5G Distributed Units (DUs) using enhanced interfaces such as Fl-U and X2-U. The artificial intelligence and machine learning policy module can employ reinforcement learning algorithms to dynamically adjust the distribution of downlink packets, with the policy being derived based on real-time data from multiple user equipment and network nodes. In yet another embodiment, the distribution of downlink packets is adjusted in response to mid-haul delay variations or queue depth metrics, ensuring optimal packet sequencing and minimizing throughput degradation. The modules can be implemented as near-real-time applications hosted on a Radio Intelligent Controller (RIC), with the capability to interact with xApps for deploying control actions and policies or at CU-UP (such as at SgNB.CU-UP for SN terminated split bearer). Additionally, the system may support various configurations, such as hosting the modules in a private data center or a public cloud, enabling scalability and adaptability to different network architectures.

[0162] By incorporating a packet sequencing feedback module (at CU-UP or at RIC) that receives sequencing feedback from the user equipment, the system enables real-time monitoring of received packet sequence numbers for each Data Radio Bearer (DRB). This arrangementallows the detection and quantification of out-of-order packet delivery, which is necessary for addressing head-of-line blocking issues in the reordering queue of the user equipment.

[0163] The delivery metrics module (at CU-UP or at RIC), which obtains network delivery metrics such as latency, throughput, and cell load, provides granular insights into the performance of downlink transmissions. This module ensures that the system can dynamically adapt to varying network conditions, such as mid-haul delays or queue depth fluctuations, thereby improving the efficiency of packet delivery.

[0164] Artificial intelligence and machine learning policy module leverages advanced algorithms, including reinforcement learning, to select and adjust the distribution of downlink packets across multiple radio legs. This approach enables the system to optimize packet routing decisions based on real-time feedback and metrics, reducing packet misorder and enhancing throughput.

[0165] Dynamic adjustment of the distribution of downlink packets across plural radio legs ensures that the system can respond to rapid changes in network conditions, such as midhaul jitter or cell load variations. This adaptability minimizes packet misordering and improves the quality of service experienced by the user equipment.

[0166] Implementing the artificial intelligence and machine learning policy module as a near-real-time application on a Radio Intelligent Controller (RIC or at CU-UP allows for low-latency decision-making and policy enforcement. This configuration ensures that the optimization of downlink split-bearer delivery is performed with minimal delay, maintaining high spectral efficiency and reducing retransmissions.

[0167] The cooperation between the packet sequencing feedback module, delivery metrics module, and artificial intelligence and machine learning policy module creates a closed-loop system that continuously optimizes downlink split-bearer delivery. This integration addresses the limitations of static allocation ratios and heuristic-based methods, providing a more responsive and adaptive solution to packet sequencing challenges in disaggregated radio access networks.

[0168] While the present disclosure has been described with reference to one or more exemplary embodiments, it will be understood by those skilled in the art that various changes can be made and equivalents can be substituted for elements thereof without departing from the scope of the present disclosure. In addition, many modifications can be made to adapt a particular situation or material to the teachings of the disclosure without departing from the scope thereof. Therefore, it is intended that the present disclosure not be limited to the particular embodiment(s) disclosed as the best mode contemplated, but that the disclosure will include all embodiments falling within the scope of the appended claims.

[0169] Glossary of Terms

[0170] Artificial Intelligence and Machine Learning (AIML) - Computational techniques that enable systems to learn from data, make decisions, and optimize processes without explicit programming. AIML is used to enhance dynamic control and policy optimization in 0-RAN networks.

[0171] Open Radio Access Network (O-RAN) - A disaggregated and open architecture for radio access networks that allows interoperability between different vendors' hardware and software components. It includes elements such as Central Units (CUs), Distributed Units (DUs), and Radio Units (RUs).

[0172] Central Unit-Control Plane (CU-CP) - A logical entity within the O-RAN architecture responsible for managing control plane functions, such as signaling and resource allocation.

[0173] Central Unit-User Plane (CU-UP) - A logical entity within the O-RAN architecture responsible for handling user data traffic and implementing data plane functions.

[0174] Distributed Unit (DU) - A component in the O-RAN architecture that hosts lower-layer protocol functions, including Physical (PHY), Medium Access Control (MAC), and Radio Link Control (RLC) sublayers.

[0175] Radio Unit (RU) - A hardware component in the O-RAN architecture responsible for radio frequency processing and interfacing with antennas.

[0176] Data Radio Bearer (DRB) - A logical channel used to carry user data between the User Equipment (UE) and the network. DRBs are associated with specific Quality of Service (QoS) flows.

[0177] Packet Data Convergence Protocol (PDCP) - A sublayer in the 5G protocol stack responsible for header compression, encryption, and reordering of data packets.

[0178] Radio Resource Control (RRC) - A control plane protocol that manages signaling between the User Equipment (UE) and the network, including connection setup, maintenance, and release.

[0179] Quality of Service (QoS) - A set of performance metrics that define the service quality for data flows, including latency, throughput, and error rates.

[0180] Minimization of Drive Test (MDT) - A feature in wireless networks that allows collection of network performance data from User Equipment (UE) to optimize network operations.

[0181] Logical Channel Identifier (LC ID) - A unique identifier used to distinguish different logical channels in the Medium Access Control (MAC) layer.

[0182] Medium Access Control Control Element (MAC CE) - A signaling mechanism within the MAC layer used to exchange control information, such as packet reordering metrics.

[0183] GPRS Tunneling Protocol User Plane (GTP-U) - A protocol used to encapsulate and transport user data packets between network nodes in 4G and 5G systems.

[0184] Markov Decision Process (MDP) - A mathematical framework used in reinforcement learning to model decision-making processes, consisting of states, actions, transition probabilities, and rewards.

[0185] Reinforcement Learning (RL) - A machine learning technique where an agent learns optimal policies by interacting with an environment and receiving feedback in the form of rewards or penalties.

[0186] Head-of-Line Blocking - A phenomenon where out-of-order packet delivery causes delays in processing subsequent packets in a queue.

[0187] Downlink Data Delivery Status (DDDS) - A feedback mechanism used to report the status of downlink data delivery, including buffer depths and transmission rates.

[0188] Assistance Information Data (AID) - Information exchanged between network nodes to provide radio conditions and other metrics for optimizing data delivery.

[0189] Physical Resource Blocks (PRBs) - The smallest unit of resource allocation in the radio spectrum, used for scheduling data transmissions in wireless networks.

[0190] Channel Quality Indicator (CQI) - A metric reported by the User Equipment (UE) to indicate the quality of the radio channel, used for adaptive modulation and coding.

[0191] Split Bearer - A configuration in dual connectivity scenarios where data flows are split across multiple radio legs, such as 4G and 5G, to enhance throughput and reliability.

[0192] Misordering Ratio - A metric that quantifies the extent of packet reordering, calculated as the ratio of out-of-order packets to the total number of packets received.

[0193] Near-Real -Time Radio Intelligent Controller (Near-RT RIC) - A component in the 0-RAN architecture that hosts xApps for dynamic control and optimization of network operations.

[0194] Non-Real-Time Radio Intelligent Controller (Non-RT RIC) - A component in the 0-RAN architecture that hosts rApps for analytics and long-term policy management.

[0195] Service Management and Orchestration (SMO) Framework - A framework in O-RAN systems that manages network services and orchestrates resources across different components.

[0196] E2 Interface - A standardized interface in O-RAN systems used for communication between the Near-RT RIC and network nodes, enabling data exchange and policy deployment.

[0197] X2 Interface - A standardized interface used for communication between LTE and 5G nodes in dual connectivity scenarios.

[0198] Fl Interface - A standardized interface in O-RAN systems that connects Central Units (CUs) and Distributed Units (DUs) for control and user plane traffic.

[0199] VI Interface - An interface used for communication between MeNB (Master eNodeB) and MeNB.CU in (4G standalone as well as) dual connectivity architectures.

[0200] Packet Delay Budget (PDB) - The maximum allowable delay for a packet to be transmitted between the User Equipment (UE) and the network.

[0201] Packet Error Rate (PER) - The percentage of packets that fail to be delivered correctly within a specified time frame.

[0202] Guaranteed Bit Rate (GBR) - A QoS parameter that ensures a minimum data rate for certain types of traffic, such as voice or video.

[0203] Non-GBR - A QoS parameter for traffic that does not require a guaranteed minimum data rate, such as best-effort data.

[0204] Deep Q Neural Network (DQN) - A reinforcement learning algorithm that uses neural networks to approximate Q-values for large state spaces.

[0205] Evolved Packet Core (EPC) - The core network architecture for LTE systems, responsible for managing data and control plane functions.

[0206] 5G Core (5GC) - The core network architecture for 5G systems, designed to support advanced features such as network slicing and ultra-reliable low-latency communication.

[0207] Random Access Procedure - A process used by the User Equipment (UE) to establish initial communication with the network.

[0208] Event-Based Trigger - A mechanism that initiates a specific action or process based on the occurrence of predefined events, such as exceeding a threshold.

[0209] Periodic Trigger - A mechanism that initiates a specific action or process at regular intervals.

Claims

CLAIMSWhat is claimed is:

1. A method for optimizing downlink split-bearer delivery in a disaggregated radio access network using packet sequencing feedback from a user equipment and network delivery metrics, comprising:a packet sequencing feedback module receiving sequencing feedback from the user equipment;a delivery metrics module obtaining network delivery metrics for downlink packet transmissions;an artificial intelligence and machine learning policy module selecting a distribution of downlink packets across plural radio legs based on the sequencing feedback and the delivery metrics;wherein the artificial intelligence and machine learning policy module dynamically adjusts the distribution of downlink packets across the plural radio legs so as to minimize out-of-order delivery at the user equipment; andwherein the packet sequencing feedback module, the delivery metrics module, and the artificial intelligence and machine learning policy module cooperate to optimize downlink splitbearer delivery in the disaggregated radio access network.

2. The method of claim 1, wherein the sequencing feedback from the user equipment comprises packet reordering measurements at a data radio bearer level, or the delivery metrics comprise at least one of: latency, throughput, or cell load associated with downlink packet transmissions.

3. The method of claim 1, wherein the artificial intelligence and machine learning policy module:employs a reinforcement learning algorithm to select and adjust the distribution of downlink packets, oris implemented as a near-real-time application on a radio intelligent controller.

4. The method of claim 1, wherein the distribution of downlink packets is dynamically adjusted in response to changes in mid-haul delay or queue depth metrics, or the packet sequencing feedback module receives sequencing feedback periodically or in response to an event trigger.

5. A method for optimizing downlink split-bearer delivery in a radio access network comprising:receiving feedback from a user equipment;obtaining network metrics for downlink transmissions;selecting a distribution of downlink packets across multiple radio legs based on the feedback and the network metrics; anddynamically adjusting the distribution of downlink packets across the multiple radio legs to reduce out-of-order delivery at the user equipment.

6. The method of claim 5, whereinthe feedback from the user equipment comprises packet reordering information, or the feedback is received periodically or in response to an event, orthe network metrics comprise at least one of latency, throughput, or load associated with downlink transmissions, orthe dynamically adjusting the distribution of downlink packets is performed in response to changes in the feedback or the network metrics.

7. The method of claim 5, wherein selecting the distribution of downlink packets comprises utilizing an artificial intelligence or machine learning techniques.

8. The method of claim 5, wherein the method is implemented as a near-real-time application on a radio intelligent controller.

9. A system for optimizing downlink split-bearer delivery in a radio access network, comprising:a feedback receiver configured to receive feedback from a user equipment; a metrics module configured to obtain network metrics for downlink transmissions; and a policy module configured to select a distribution of downlink packets across multiple radio legs based on the feedback and the network metrics;wherein the policy module is further configured to dynamically adjust the distribution of downlink packets across the multiple radio legs to reduce out-of-order delivery at the user equipment.

10. The system of claim 9, whereinthe feedback from the user equipment comprises packet reordering information, or the feedback is received periodically or in response to an event, orthe network metrics comprise at least one of latency, throughput, or load associated with downlink transmissions, orthe dynamically adjusting the distribution of downlink packets by the policy module is performed in response to changes in the feedback or the network metrics.

11. The system of claim 9, wherein the policy module is configured to utilize an artificial intelligence or machine learning technique to select the distribution of downlink packets.

12. The system of claim 9, wherein the policy module is configured to dynamically adjust the distribution of downlink packets in response to changes in the feedback or the network metrics.

13. The system of claim 9, wherein the policy module is implemented as a near-real-time application on a radio intelligent controller.

14. A method for communicating packet reordering information from a user equipment (UE) to a master evolved Node B central unit-control plane (MeNB.CU-CP) over a Radio Resource Control (RRC) protocol, comprising:transmitting, from the MeNB.CU-CP to the UE over the RRC protocol, an enhanced UEInformationRequest message requesting Packet Data Convergence Protocol (PDCP) reordering measurements for a data radio bearer;measuring, at the UE, packet reordering for the data radio bearer; andtransmitting, from the UE to the MeNB.CU-CP over the RRC protocol, an enhanced UEInformationResponse message comprising packet reordering information including the PDCP reordering measurements for the data radio bearer.

15. The method of claim 14, further comprising:transmitting, from the MeNB.CU-CP to a secondary gNodeB central unit-control plane (SgNB.CU-CP), the packet reordering information over an enhanced X2-C interface.

16. The method of claim 15, further comprising:transmitting, from the SgNB.CU-CP to a secondary gNodeB central unit-user plane (SgNB.CU-UP), the packet reordering information over an enhanced El interface.

17. A method for communicating packet reordering information as part of a Minimization of Drive Test (MDT) report sent from a User Equipment (UE) to a master evolved Node B central unit-control plane (MeNB.CU-CP) over a Radio Resource Control (RRC) protocol, comprising:transmitting, from the MeNB.CU-CP to the UE over the RRC protocol, an enhanced UEInformationRequest message requesting packet reordering information for a data radio bearer as part of Minimization of Drive Test (MDT) measurements;measuring, at the UE, packet reordering for the data radio bearer as part of the MDT measurements; andtransmitting, from the UE to the MeNB.CU-CP over the RRC protocol, an enhanced UEInformationResponse message comprising the packet reordering information for the data radio bearer as part of the MDT report.

18. The method of claim 17, further comprising:transmitting, from the MeNB.CU-CP to a secondary gNodeB central unit-control plane (SgNB.CU-CP), the packet reordering information included in the Minimization of Drive Test report over an enhanced X2-C interface.

19. The method of claim 18, further comprising:transmitting, from the SgNB.CU-CP to a secondary gNodeB central unit-user plane (SgNB.CU-UP), the packet reordering information over an enhanced El interface.

20. A method for communicating packet reordering information from a user equipment (UE) to a secondary gNodeB Distributed Unit (SgNB.DU), comprising:measuring, at the UE, packet reordering measurements at a data radio bearer level for the UE;generating, at the UE, a Medium Access Control - control element (MAC CE) comprising packet reordering information corresponding to the packet reordering measurements, wherein the MAC CE is enhanced for communicating the packet reordering information; and transmitting, from the UE to the SgNB.DU over physical uplink shared channel, the MAC CE carrying the packet reordering information.

21. The method of claim 20, further comprising:transmitting, from the SgNB.DU to a secondary gNodeB central unit (SgNB.CU), the packet reordering information over an Fl-U interface using an enhanced Assistance Information Data protocol data unit (AID PDU).

22. A method for communicating packet reordering information from a user equipment (UE) to a master evolved Node B Distributed Unit (MeNB.DU), comprising:measuring, at the UE, packet reordering measurements at a data radio bearer level for the UE;generating, at the UE, a Medium Access Control - control element (MAC CE) comprising packet reordering information corresponding to the packet reordering measurements, wherein the MAC CE is enhanced for communicating the packet reordering information; and transmitting, from the UE to the MeNB.DU over physical uplink shared channel, the MAC CE carrying the packet reordering information.

23. The method of claim 22, further comprising:transmitting, from the MeNB.DU to a secondary gNodeB central unit-user plane (SgNB.CU), the packet reordering information over an enhanced X2-U interface.

Images