5G QoS Framework

Introduction to 5G QoS Framework

Quality of Service (QoS) is a fundamental networking mechanism that is essential for both 5G and LTE because it allows mobile network operators to manage and prioritize different types of data traffic. Without QoS, all data—from a critical video call to a background email sync—would be treated with the same level of importance. This "best-effort" approach would lead to a chaotic and unpredictable user experience, especially when the network is congested.

QoS mechanisms like QoS Flows in 5G and EPS Bearers in LTE help ensure a smooth and reliable user experience by intelligently allocating network resources based on the specific needs of each application.

Here’s how QoS directly improves the user experience in several common scenarios:

1. Clear and Uninterrupted Voice and Video Calls

QoS identifies the voice packets and assigns them to a high-priority flow (e.g., a dedicated bearer in LTE with QCI 1, or a QoS Flow in 5G with 5QI 1). This ensures the packets are scheduled immediately and given a guaranteed bit rate, preserving the low-latency, consistent data stream needed for a clear, real-time conversation. The user experiences a crisp, uninterrupted call.

2. Smooth, High-Resolution Video Streaming

QoS maps the video stream to a flow that guarantees a specific bit rate. This ensures the streaming service receives the necessary data throughput to fill its buffer continuously, even if other, less critical applications are also using the network. The user enjoys a smooth, high-definition viewing experience without frustrating interruptions.

3. Responsive and Competitive Online Gaming

Gaming traffic is assigned to a QoS flow designed for ultra-low latency. These packets are given the highest scheduling priority to minimize their travel time through the network. This ensures that the user's actions are reflected in the game almost instantaneously, providing a fluid and competitive gaming experience.

4. Efficiently Multitasking on Your Device

QoS differentiates between these tasks.

• The video conference gets a high-priority, low-latency QoS Flow to ensure perfect quality.

• The OS update and photo sync are assigned to a lower-priority, non-guaranteed "best-effort" flow. They will use whatever network capacity is left over.

• Email syncing might get a slightly higher priority than the large download but will still be treated as less important than the video call.

  
  

The 5G Quality of Service (QoS) framework represents a significant evolution from the bearer-based model used in 4G LTE. Designed to be more granular, flexible, and aligned with the diverse requirements of 5G applications, the new framework is centered around the concept of QoS Flows.

The QoS concept will be flow-based. Packets are classified and marked with QoS Flow Identifier (QFI). There will be two types of flows: One with standardized QoS profiles and the other with operator-specific QoS profiles.

For the first one, only the QFI value is used in the network. For the latter one, QoS attributes are also signaled between the network elements. The 5G QoS flows are mapped in the Access Network to Data Radio Bearers (DRBs), unlike 4G where the mapping is 1:1 between EPC and radio bearers.

Source: 5G QoS, Ericsson

QoS (Quality of Service) parameters and characteristics

QoS (Quality of Service) parameters and characteristics are used to define the specific packet forwarding treatment that a QoS Flow receives from the UE to the User Plane Function (UPF). These parameters are collectively bundled into a QoS Profile, which is referenced by the 5G QoS Identifier (5QI).

The 5G QoS Identifier (5QI) is a scalar value that serves as a reference to a set of standardized or pre-configured QoS characteristics. These characteristics are the fundamental performance targets for a QoS Flow and are often directly mapped from a standardized 5QI value. Each QoS Flow is assigned a 5QI, which dictates how the flow should be treated by the network.

image source: https://juejin.cn/post/7087204368310599694

1. Resource Type

This parameter defines the fundamental nature of the resource allocation for the QoS Flow. There are three types :

• Non-GBR (Non-Guaranteed Bit Rate): This is for "best-effort" traffic that does not require a reserved data rate. The network does not guarantee any specific throughput. This is suitable for applications like web browsing, email, and file downloads.

• GBR (Guaranteed Bit Rate): This is for traffic that requires a dedicated, reserved data rate to function properly. The network commits to providing a minimum throughput for this flow. It's essential for services like video streaming and real-time voice calls (VoNR).

• Delay-Critical GBR: This is a stricter form of GBR for services that are not only bit-rate sensitive but also extremely latency-sensitive. This is designed for URLLC (Ultra-Reliable Low-Latency Communication) use cases, such as industrial automation, remote surgery, and vehicle-to-everything (V2X) communication.

2. Priority Level

This is a scalar value that determines the scheduling priority for a QoS Flow relative to other flows. A lower numerical value indicates a higher priority. For example, a QoS Flow for IMS signaling (critical for call setup) will have a higher priority (lower number) than a flow for standard internet traffic. This ensures that in times of network congestion, critical packets are processed first.

3. Packet Delay Budget (PDB)

PDB defines the maximum acceptable delay for a packet from the moment it leaves the UE to when it reaches the UPF (or vice versa). This is a critical parameter for latency-sensitive applications. For instance:

• Conversational Voice (5QI 1) has a PDB of 100ms.

• Real-time Gaming (5QI 3) has a tighter PDB of 50ms.

• URLLC services like discrete automation (5QI 82) may have a PDB as low as 10ms.

4. Packet Error Rate (PER)

PER specifies the maximum acceptable rate of packets that can be lost or delivered with errors. It essentially defines the required reliability of the data flow. A lower PER means higher reliability.

• A service like watching a standard video stream can tolerate a higher PER.

• A service like IMS signaling (5QI 5) requires an extremely low PER (e.g., 10⁻⁶) to ensure call control messages are never lost.

Flow and Session Level Parameters

These parameters control bit rates and resource allocation policies.

5. Allocation and Retention Priority (ARP)

The ARP is a crucial policy parameter that governs how resources are allocated and retained, especially during network congestion. It consists of three components:

• Priority Level: Similar to the QoS Flow Priority Level, this value (1-15) determines the priority of the resource request.

• Pre-emption Capability: A boolean value indicating whether this QoS Flow is allowed to pre-empt, or "kick off," another existing flow with a lower retention priority.

• Pre-emption Vulnerability: A boolean value indicating whether this QoS Flow can be pre-empted by another flow with a higher allocation priority.

ARP is vital for ensuring that high-priority services, like an emergency call, can always get network resources, even if it means terminating a less critical service like a file download.

6. Guaranteed Flow Bit Rate (GFBR) and Maximum Flow Bit Rate (MFBR)

These parameters are only applicable to GBR QoS Flows.

• GFBR: The minimum bit rate that the network guarantees to provide for the flow.

• MFBR: The maximum bit rate that the flow can reach if network resources are available. The flow is allowed to burst up to this rate but is not guaranteed anything beyond the GFBR.

7. Aggregate Maximum Bit Rate (AMBR)

AMBR is applicable only to Non-GBR QoS Flows and defines a "speed cap" for all combined Non-GBR traffic for a user. There are two types:

• Session-AMBR: The maximum aggregate bit rate for all Non-GBR flows within a single PDU Session.

• UE-AMBR: The maximum aggregate bit rate for all Non-GBR flows across all PDU Sessions for a single UE.

Dynamic QoS Parameter

8. Reflective QoS Attribute (RQA)

The RQA is a parameter that enables a more dynamic and efficient way of managing QoS for uplink traffic. When RQA is enabled for a QoS flow:

1. The network sends a downlink packet with a specific QoS Flow Identifier (QFI).

2. The UE's radio stack inspects this packet and creates a derived uplink QoS rule that maps the corresponding uplink traffic to the same QFI.This allows the UE to apply the correct QoS treatment to its uplink data without needing explicit signaling from the network for every new flow, reducing overhead and latency.

The Core Concept: QoS Flows

In 5G, the QoS framework is built upon QoS Flows, which are the most granular level of traffic handling. A QoS Flow is a pipeline that provides a specific QoS treatment to a defined set of data packets. Unlike LTE's rigid bearer system, a single PDU (Protocol Data Unit) session in 5G can contain multiple QoS Flows, each with distinct QoS characteristics tailored to the application generating the data.

This new model provides several key advantages:

• Flexibility: Applications with multiple traffic types (e.g., video, voice, and data in a conferencing app) can have separate QoS Flows for each, ensuring that latency-sensitive data is prioritized without impacting other data streams.

• Granularity: QoS is applied at the flow level rather than the bearer level, allowing for more precise control over traffic differentiation.

• Efficiency: It eliminates the need for establishing and maintaining multiple bearers for a single user session, simplifying network management.

Mapping Data to QoS Flows

The process of directing traffic into the correct QoS Flow is managed by the User Plane Function (UPF) in the 5G Core (5GC). This mapping is based on packet filters defined in the SDF (Service Data Flow) templates.

Here’s how it works:

1. SDF Definition: An SDF represents a single IP flow or a group of IP flows. It is defined by a set of packet filters, such as source/destination IP address, port number, and protocol ID.

2. SDF to QoS Flow Mapping: The Session Management Function (SMF) provides mapping rules to both the UE and the UPF. These rules link each SDF to a specific QoS Flow.

3. Packet Forwarding: In the uplink, the UE classifies IP packets based on these rules and directs them to the appropriate Data Radio Bearer (DRB) associated with a QoS Flow. In the downlink, the UPF performs a similar function, identifying incoming packets based on the SDF templates and forwarding them to the corresponding QoS Flow.

1. Data Network (DN)

• Packets from services like Netflix, WhatsApp Video, or IMS (voice/video calls) originate from servers in the DN.

• Each service flow has specific QoS requirements (e.g., latency for voice, throughput for streaming).

2. UPF (User Plane Function)

• Traffic Classification: Incoming packets from DN to UPF arrive on the N6 interface.

• SDF/TFT Templates: The UPF examines packet headers using Service Data Flow (SDF) and Traffic Flow Template (TFT) rules to classify packets according to their service—identifying each as, for example, streaming video, IMS media, or best-effort data.

• QoS Flow Mapping: Based on policies received from the SMF and PCF, the UPF assigns packets to the correct QoS Flow and inserts the corresponding QFI (QoS Flow Identifier).

• QoS Enforcement: UPF enforces prescribed bit rates, priorities, and packet marking through PFCP-controlled policies, prioritizing critical flows (e.g., voice) and throttling or dropping less important traffic if necessary.

• Packet Forwarding: Enriched with QFI, packets are sent on the N3 interface to the gNB.

3. gNB (5G Radio Access Network Node)

• SDAP Layer: The gNB’s SDAP reads the QFI, mapping each flow to the appropriate Data Radio Bearer (DRB), as configured for 5G QoS.

• Radio Scheduler: Each DRB is prioritized according to its 5QI (QoS characteristics like priority, delay, and reliability), ensuring premium services like voice/video get fast and reliable radio resources.

• Lower RAN Layers (PDCP, RLC, MAC, PHY): These layers compress headers, segment data, sequence packets, multiplex different flows, and transmit over the air to the UE—all tailored to the QoS demands identified earlier.

4. UE (User Equipment)

• SDAP Layer: The QFI is inspected so that packets are delivered to the correct application with proper QoS handling (e.g., low-latency for voice, guaranteed bitrate for streaming).

• TFT Rules: Ensure the received IP flows are segregated and routed to the respective app/service instance, preserving end-to-end QoS behavior.

Flow Summary Table

Stage	QoS Actions (Downlink)	Examples
Data Network	Originates service flows, labels priorities	Netflix, WhatsApp Video, IMS Media
UPF	Classifies, assigns QFI, enforces policy, forwards	Maps Netflix/Voice to QFI, shapes/schedules traffic
gNB	Maps QFI to DRB, schedules resources, handles radio	Allocates DRB/quota for Netflix, Voice, WhatsApp
UE SDAP/TFT	Recognizes QFI, delivers to app per QoS needs	Ensures streaming quality, low-latency voice calls

Roles of Core/RAN network functions

In the 5G Quality of Service (QoS) framework, several network functions and protocol layers collaborate to ensure that data traffic receives the appropriate treatment. The Session Management Function (SMF), User Plane Function (UPF), and Access and Mobility Management Function (AMF) in the 5G Core, along with the SDAP and PDCP layers in the radio protocol stack, each have distinct responsibilities in managing QoS Flows.

Core Network Functions

• Session Management Function (SMF) The SMF is the central control point for session management and QoS. Its primary roles include :

  • QoS Flow Management It handles the creation, modification, and deletion of QoS Flows for a PDU session.

  • Policy Enforcement The SMF receives QoS policy rules from the Policy Control Function (PCF) and translates them into specific instructions for the UPF and the UE. This includes defining the QoS profile for each flow, such as its 5G QoS Identifier (5QI), bit rate, and priority.

  • UPF Configuration It instructs the UPF on how to handle data packets by providing Packet Detection Rules (PDRs) for mapping traffic to QoS Flows, enforcing QoS, and marking packets.

• User Plane Function (UPF) As the primary component in the user plane, the UPF is responsible for the actual handling and forwarding of data packets. Its key QoS-related functions are :

  • Packet Filtering and Routing The UPF inspects incoming data packets and uses the PDRs provided by the SMF to classify and map them to the correct QoS Flow.

  • QoS Enforcement It applies the QoS policies for each flow, which includes rate limiting (for GBR flows), packet marking, and prioritizing traffic based on its 5QI.

  • Usage Reporting The UPF monitors the traffic volume for each QoS Flow and reports this data to the SMF for charging and policy control purposes.

• Access and Mobility Management Function (AMF) The AMF is mainly responsible for connection and mobility management, but it plays a crucial supporting role in the QoS framework.

  • Relaying QoS Information The AMF acts as a bridge between the SMF and the RAN (gNB). It forwards the necessary QoS requirements for each QoS Flow from the SMF to the gNB during the PDU session setup or modification process. This enables the gNB to configure the appropriate radio resources.

Radio Protocol Stack Layers

• Service Data Adaptation Protocol (SDAP) The SDAP layer is a new addition in the 5G NR user plane protocol stack, specifically designed for QoS Flow management.

  • Mapping QoS Flows to Data Radio Bearers (DRBs) The primary function of the SDAP layer is to map one or more QoS Flows to a single Data Radio Bearer (DRB). This many-to-one mapping is a key difference from LTE, where there is a one-to-one correspondence between an EPS bearer and a DRB.

  • QoS Flow ID (QFI) Tagging In the downlink, the SDAP layer at the gNB adds a QoS Flow Identifier (QFI) to the header of each packet. In the uplink, the UE's SDAP layer performs the same function. This QFI allows the receiving entity (UE in the downlink, UPF in the uplink) to identify which QoS Flow the packet belongs to.

• Packet Data Convergence Protocol (PDCP) The PDCP layer sits just below the SDAP layer and is responsible for several functions that contribute to the overall QoS.

  • Header Compression It compresses the IP headers of packets to reduce overhead and improve radio efficiency, which is especially important for services like VoIP that use small packets.

  • Ciphering and Integrity Protection PDCP encrypts user data and provides integrity protection for control plane data to ensure security.

  • In-sequence Delivery and Reordering It numbers the packets to ensure they are delivered in the correct order to the upper layers and reorders any packets that arrive out of sequence, which is critical for maintaining the quality of real-time services.

1. UE to AMF: PDU Session Establishment Request The process begins when the User Equipment (UE) sends a NAS (Non-Access Stratum) message to the AMF to request the establishment of a new PDU session. This request contains information such as the PDU session type (e.g., IPv4, IPv6) and may include a specific PDU Session ID.

2. AMF to SMF: Create SM Context Request The AMF receives the request and, after authenticating the UE, selects an appropriate SMF to manage the session. It then forwards the PDU session establishment request to the chosen SMF, asking it to create a new Session Management (SM) context.

3. SMF to PCF: SM Policy Control Create The SMF needs policy information to correctly establish the QoS Flows. It contacts the PCF, requesting the creation of a new policy association for the session. This request provides the PCF with details about the user and the requested service.

4. PCF to SMF: SM Policy Control Response The PCF evaluates the request based on operator policies, the user's subscription data, and real-time network conditions. It then generates the applicable Policy and Charging Control (PCC) rules and sends them back to the SMF. These PCC rules define the specific QoS parameters (e.g., 5QI, bit rates) for the service.

5. SMF to AMF: N1/N2 Message Transfer With the policy rules from the PCF, the SMF now has all the information needed to configure QoS. It does the following :

   • Constructs a NAS Message for the UE containing the QoS Rules that map uplink data packets to the correct QoS Flows.

   • Creates QoS Profiles for the Radio Access Network (gNB), detailing the QoS parameters for each flow.

   • It sends both the NAS message and the QoS profiles to the AMF within an N1N2MessageTransfer request.

6. AMF to gNB and UE: NGAP and NAS Signaling The AMF forwards the information to the RAN.

   • The QoS profiles are sent to the gNB via an NGAP message, allowing the gNB to configure the necessary Data Radio Bearers (DRBs).

   • The NAS message containing the uplink QoS rules is relayed to the UE.

7. SMF to UPF: PFCP Session Establishment/Modification Request Simultaneously, the SMF configures the user plane by sending a PFCP (Packet Forwarding Control Protocol) message to the UPF. This message instructs the UPF to install the necessary Packet Detection Rules (PDRs) and Forwarding Action Rules (FARs) that map downlink data packets to the corresponding QoS Flows and enforce the defined QoS policies (e.g., rate limiting).

8. UPF Confirmation and Data Forwarding The UPF acknowledges the request from the SMF and applies the rules. It is now ready to receive downlink data, classify it according to the PDRs, enforce the QoS policies, and forward the traffic toward the gNB.

Comparison with the LTE QoS Framework

The LTE QoS model is based on the concept of an EPS (Evolved Packet System) Bearer. An EPS Bearer is a pipe that provides a specific QoS treatment between the UE and the Packet Data Network Gateway (P-GW).

Source image: https://devopedia.org/5g-quality-of-service

Here’s a comparison of the key differences:

Feature	4G LTE QoS	5G NR QoS
Basic Unit	EPS Bearer	QoS Flow
Granularity	Bearer-level (one-to-one mapping between an EPS Bearer and a DRB)	Flow-level (multiple QoS Flows can be multiplexed onto a single DRB)
Identifier	QCI (QoS Class Identifier)	5QI (5G QoS Identifier)
Flexibility	Rigid. A UE typically has one default bearer and may have one or more dedicated bearers for specific services like VoLTE.	Highly flexible. A single PDU session can support multiple QoS Flows, each with its own characteristics.
Signaling	Establishing a dedicated bearer requires significant RRC and NAS signaling, which can be slow.	More efficient signaling. QoS Flows can be added, modified, or removed with less overhead.

Real-World Example: From LTE Bearer to 5G QoS Flow

Let's consider a user on a video call while browsing the web.

• In 4G LTE:

  • Web browsing traffic would use the default EPS bearer, typically assigned QCI 9 (non-GBR).

  • To ensure the quality of the video call, the network would establish a dedicated EPS bearer with a GBR-type QCI (e.g., QCI 2 for conversational video).

  • This results in two separate bearers, each with its own DRB, requiring separate management and signaling.

• In 5G NR:

  • A single PDU session is established for the user.

  • The web browsing traffic is mapped to a QoS Flow with 5QI 9 (non-GBR).

  • The video call traffic is mapped to a separate QoS Flow with 5QI 2 (GBR, for live video).

  • Both of these QoS Flows can be multiplexed onto the same DRB, simplifying resource management while still ensuring that the video packets receive priority scheduling based on their 5QI. The network can differentiate the packets within the same DRB and apply the appropriate QoS treatment.

This example highlights how the 5G QoS framework achieves a finer level of control and greater efficiency, making it better suited for the diverse and demanding applications of the 5G era.

References

• 3GPP 38.300 - NR and NG-RAN Overall description; Stage-2 => Chapter 12

• 3GPP 23.501 - 5G;System architecture for the 5G System (5GS) => Section 5.7

• 3GPP 24.501 - 5G;System architecture for the 5G System (5GS) => Section 9.11.4.12 ~ 9.11.4.14

• 3GPP 38.413 - 5G; NG-RAN;NG Application Protocol (NGAP) => Section 8.2