US-12621703-B2 - Systems and methods for radio scheduling with blind traffic characterization

US12621703B2US 12621703 B2US12621703 B2US 12621703B2US-12621703-B2

Abstract

Systems and methods described herein provide intelligent scheduling in a radio access network (RAN). A RAN device receives traffic characterization model parameters and also receives data for a communication session with a User Equipment (UE) device. The RAN device identifies traffic characteristics for the communication session based on the traffic characterization model parameters. The traffic characteristics include a predicted level of periodicity for a future time interval. The RAN device selects a scheduling discipline for the communication session based on the projected level of periodicity for the future time interval, and implements the selected scheduling discipline for the communication session.

Inventors

Vishwanath Ramamurthi
Arda Aksu
Wei David Huang

Assignees

VERIZON PATENT AND LICENSING INC.

Dates

Publication Date: 20260505
Application Date: 20230221

Claims (20)

1 . A method comprising: receiving, by a network device in a radio access network (RAN), traffic characterization model parameters, wherein the traffic characterization model parameters are based on training data; receiving, by the network device and after receiving the traffic characterization model parameters, data packets for a communication session with a User Equipment (UE) device; performing, by the network device, blind traffic characterization for the communication session based on the traffic characterization model parameters, wherein traffic characteristics identified by the blind traffic characterization include a predicted level of periodicity for a future time interval; selecting, by the network device, a scheduling discipline for the communication session based on the predicted level of periodicity for the future time interval; and implementing, by the network device, the selected scheduling discipline for the communication session.
2 . The method of claim 1 , wherein selecting the scheduling discipline includes: selecting between one of dynamic scheduling (DS) and semi-persistent scheduling (SPS).
3 . The method of claim 1 , wherein receiving the data packets for the communication session includes: receiving the data packets for an active communication session between the UE device and a data network in real-time.
4 . The method of claim 1 , wherein the traffic characteristics identified by the blind traffic characterization further include a validity duration for the predicted level of periodicity.
5 . The method of claim 1 , wherein selecting the scheduling discipline includes: performing quality of service (QoS) flow to data radio bearer (DRB) mapping based on analysis of QoS requirements and traffic characteristics.
6 . The method of claim 1 , wherein performing the blind traffic characterization for the communication session includes: identifying, by the network device, the traffic characteristics for the communication session without assistance of additional traffic information from an application server or a user device.
7 . The method of claim 1 , wherein the data packets for the communication session includes one or more of: packet arrival time stamps, packet sizes, and UE device subscription information.
8 . The method of claim 1 , wherein selecting the scheduling discipline includes: assigning a value corresponding to a packet importance ranking.
9 . The method of claim 1 , wherein selecting the scheduling discipline includes: selecting the scheduling discipline for the communication session is further based on packet sizes and jitter in the communication session for the future time interval.
10 . The method of claim 1 , further comprising: selecting, by the network device, a resource pool for implementing the selected scheduling discipline based on a packet importance ranking of packets in the communication session.
11 . A radio access network (RAN) device comprising: one or more processors configured to: receive traffic characterization model parameters, wherein the traffic characterization model parameters are based on training data; receive, after receiving the traffic characterization model parameters, data packets for a communication session with a User Equipment (UE) device; perform blind traffic characterization for the communication session based on the traffic characterization model parameters, wherein traffic characteristics identified by the blind traffic characterization include a predicted level of periodicity for a future time interval; select a scheduling discipline for the communication session based on the predicted level of periodicity for the future time interval; and implement the selected scheduling discipline for the communication session.
12 . The RAN device of claim 11 , wherein, when selecting the scheduling discipline, the one or more processors are further configured to: select between one of dynamic scheduling (DS) and semi-persistent scheduling (SPS).
13 . The RAN device of claim 11 , wherein, when receiving the data packets for the communication session, the one or more processors are further configured to: receive the data packets for an active communication session between the UE device and a data network in real-time.
14 . The RAN device of claim 11 , wherein the traffic characteristics identified by the blind traffic characterization further include a validity duration for the predicted level of periodicity.
15 . The RAN device of claim 11 , wherein the RAN device includes a gNodeB.
16 . The RAN device of claim 11 , wherein, when performing the blind traffic characterization for the communication session, the one or more processors are further configured to: identify the traffic characteristics for the communication session without assistance of additional traffic information from an application server or a user device.
17 . The RAN device of claim 11 , wherein the RAN device includes a central unit (CU) device.
18 . A non-transitory, computer-readable storage medium storing instructions, executable by a processor of a network device, for: receiving, by the network device in a radio access network (RAN), traffic characterization model parameters, wherein the traffic characterization model parameters are based on training data; receiving, by the network device and after receiving the traffic characterization model parameters, data packets for a communication session with a User Equipment (UE) device; performing, by the network device, blind traffic characterization for the communication session based on the traffic characterization model parameters, wherein traffic characteristics identified by the blind traffic characterization include a predicted level of periodicity for a future time interval; selecting, by the network device, a scheduling discipline for the communication session based on the predicted level of periodicity for the future time interval; and implementing, by the network device, the selected scheduling discipline for the communication session.
19 . The non-transitory, computer-readable storage medium of claim 18 , wherein the instructions for selecting the scheduling discipline are further for: selecting between one of dynamic scheduling (DS) and semi-persistent scheduling (SPS).
20 . The non-transitory, computer-readable storage medium of claim 18 , wherein the instructions are further for: selecting a resource pool for implementing the selected scheduling discipline based on a packet importance ranking of packets for the communication session.

Description

CROSS-REFERENCE TO RELATED APPLICATION This application is a continuation-in-part of U.S. patent application Ser. No. 17/931,949, filed on Sep. 14, 2022, and titled “Systems and Methods for XR-Aware Radio Scheduling,” the contents of which are incorporated herein by reference. BACKGROUND New cellular networks (e.g., Fifth Generation (5G) networks) can provide various services and applications, to user devices, with optimized latency and quality of service. In the context of wireless networks, scheduling may generally be referred to as allocating resources for data transmission, such as transmissions over a radio interface between a base station and an end device. Optimal resource allocation for certain application traffic depends on the network knowing the application traffic characteristics, its quality of service (QoS) requirements, and/or network conditions. Application feedback (such as high-level data or metadata) would be useful for selecting a particular scheduling discipline, as well as for configuring scheduling parameters for a selected scheduling algorithm. However, an application traffic type or application metadata is not always available to the network to enable the correct choice of an appropriate scheduling scheme. BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a diagram illustrating an exemplary environment in which embodiments may be implemented; FIG. 2 is a diagram illustrating a generic architecture and communications for blind traffic characterization based on artificial intelligence (AI)/machine learning (ML) (TCAM); FIG. 3A is a diagram illustrating an example TCAM system architecture according to an implementation; FIG. 3B is a diagram illustrating an example TCAM system architecture according to another implementation; FIG. 4 is a diagram illustrating quality of service (QoS) mapping to data radio bearers (DRBs), according to an implementation; FIG. 5 is a chart illustrating an example of mapping DRBs to semi-persistent scheduling (SPS) bins; FIG. 6 is a diagram illustrating exemplary components of a device that may correspond to one or more of the devices illustrated and described herein; and FIGS. 7 and 8 are flow diagrams illustrating an exemplary process for dynamically implementing a scheduling discipline for a communication session, according to an implementation. DETAILED DESCRIPTION OF EXAMPLE EMBODIMENTS The following detailed description refers to the accompanying drawings. The same reference numbers in different drawings may identify the same or similar elements. Also, the following detailed description does not limit the invention. Dynamic and semi-persistent scheduling schemes are two different scheduling mechanisms available in mobile networks and suited for different kinds of traffic. Semi-persistent scheduling (SPS) is typically used for services like voice (e.g., voice-over-IP), which have periodic traffic (e.g., fixed size packets arriving at regular periodic intervals) in both uplink (UL) and downlink (DL). With SPS-based resource allocation, a serving base station allocates at least a part of resources and transport formats to the end device semi-statically over a certain time interval. Using SPS for periodic traffic ensures regular availability of network resources for a service (e.g., an extended reality (XR) service, a gaming service, an industrial Internet of Things (IoT) service, etc.), thus ensuring a good user experience. Also, underutilized radio resources with SPS are minimal. Furthermore, there is little to no overhead associated with requesting grants that are typically needed for aperiodic traffic. Dynamic scheduling (DS) uses control information to adjust the amount of UL/DL data that can be transferred over a radio interface. Dynamic scheduling is more suited for applications that involve a dynamic traffic pattern with no periodicity or having periodicity but with large jitter and variable packet sizes. For example, dynamic scheduling is more suited for a web browsing service because the traffic characteristics are not periodic and do not typically follow a specific pattern. In contrast with SPS, dynamic scheduling allows more flexibility in resource allocation for dynamic traffic. Using SPS for dynamic traffic would result in highly inefficient use of limited radio resources. Thus, mobile networks can benefit from a correct selection of SPS or DS for certain application traffic. The optimal selection depends, for example, on the traffic characteristics, QoS requirements, and network conditions. If such information is not available, the network may not be able to select an optimal scheduling scheme. Systems and methods described herein provide an intelligence function to optimize exchange of application traffic over a RAN. The systems and methods may implement blind traffic characterization (e.g., without the assistance of additional traffic information or metadata from an application server or a user device) based on artificial intelligence (AI)/machi