PDU Set Marking

Description

PDU Set marking is a protocol mechanism introduced in 3GPP Release 19 that involves marking Protocol Data Units (PDUs) carrying a payload with PDU Set Information. This concept is part of the broader work on PDU Set handling, which was developed to support advanced applications such as Extended Reality (XR), cloud gaming, and immersive media streaming. In these applications, data is often organized into application-level units (e.g., video frames, audio frames, or control messages) that consist of multiple PDUs. The PDU Set represents a group of PDUs that carry a single application-level data unit, and the marking mechanism allows the network to identify which PDUs belong to the same set.

The need for PDU Set marking arises from the limitations of traditional per-PDU QoS handling. In 5G networks, QoS is typically applied on a per-PDU basis, with each PDU being treated independently according to its QoS flow. However, for many advanced applications, the loss or delay of a single PDU within an application-level unit can render the entire unit useless. For example, in XR video streaming, a video frame is encoded into multiple PDUs; if one PDU is lost, the entire frame may be undecodable, wasting network resources and degrading user experience. PDU Set marking allows the network to treat the set of PDUs as a cohesive unit, enabling more intelligent handling such as selective dropping, prioritized delivery, or coordinated scheduling.

The marking process involves adding PDU Set Information to the PDUs as they traverse the protocol stack. This information typically includes a PDU Set Sequence Number, which identifies the set to which the PDU belongs, and may include other parameters such as the total number of PDUs in the set, the importance of the PDU within the set, or the remaining time budget for the set. The marking can be performed by the application, the UE, or the network, depending on the deployment scenario and the capabilities of the entities involved. In the uplink, the UE may mark PDUs based on information provided by the application; in the downlink, the UPF or another core network entity may perform the marking based on information received from the application server.

From an architectural perspective, PDU Set marking interacts with several layers of the protocol stack. At the application layer, the application provides information about the PDU Set structure to the lower layers. At the transport layer (e.g., SDAP, PDCP), the marking information is carried in protocol headers or as part of the PDU payload. At the network layer, the marking is used by the core network (e.g., SMF, UPF) to apply PDU Set-aware QoS policies. The RAN (gNB) uses the marking to make scheduling and resource allocation decisions, such as prioritizing PDUs from a set that is close to its deadline or dropping PDUs from a set that is already partially lost.

PDU Set marking is specified in 3GPP TS 26.522 (PDU Set handling for media services) and TR 26.822 (Study on PDU Set handling). These specifications define the format of the PDU Set Information, the procedures for marking and unmarking, and the interactions with QoS frameworks. The mechanism is designed to be flexible and extensible, allowing future applications to define their own PDU Set semantics while reusing the core marking framework.

In the context of 5G-Advanced and 6G, PDU Set marking is a key enabler for latency-critical and bandwidth-intensive applications. It allows the network to make more informed decisions about resource allocation, packet dropping, and scheduling, ultimately improving the quality of experience for end users. As applications become more sophisticated and demand more from the network, mechanisms like PDU Set marking will become increasingly important for delivering efficient and reliable services.

Purpose & Motivation

PDU Set marking exists to address the limitations of per-PDU QoS handling in 5G networks, particularly for advanced applications like XR, cloud gaming, and immersive media streaming. In these applications, data is organized into application-level units (e.g., video frames) that consist of multiple PDUs. Traditional per-PDU QoS treats each PDU independently, which can lead to inefficient resource usage and degraded user experience when a single PDU loss renders an entire application-level unit useless.

The motivation for PDU Set marking comes from the need to treat groups of PDUs as cohesive units, enabling the network to make more intelligent decisions about scheduling, dropping, and resource allocation. For example, if a PDU Set is partially lost, the network may decide to drop the remaining PDUs of that set to free up resources for other sets, rather than wasting resources on an incomplete set. Similarly, the network may prioritize PDUs from a set that is close to its deadline, ensuring timely delivery of critical data.

Previous approaches to QoS handling did not provide a standardized way to identify and manage PDU Sets. Application-level solutions were proprietary and not integrated with the network's QoS framework. PDU Set marking fills this gap by providing a standardized, network-integrated mechanism for marking PDUs with PDU Set Information. This allows the network to apply PDU Set-aware QoS policies, improving efficiency and user experience for advanced applications. The mechanism is specified in 3GPP TS 26.522 and TR 26.822, and is part of the broader effort to enhance 5G-Advanced and 6G networks for emerging applications.