Service Data Unit Transfer Delay

Description

SDU transfer delay is a Quality of Service (QoS) performance parameter that quantifies the time taken to successfully transfer a Service Data Unit (SDU) from a source point to a destination point in a telecommunications network. An SDU is the unit of data that a protocol layer receives from the layer above it and is required to transmit; for example, an IP packet handed from the IP layer to a PDCP entity is an SDU from PDCP's perspective. The transfer delay measures the total elapsed time between the moment the transfer of that SDU begins and the moment its successful transfer is confirmed. The definition is adopted from ITU-T Recommendation X.140, which provides a general framework for defining QoS parameters in data communication networks.

In the 3GPP architecture, SDU transfer delay is one of several performance metrics associated with bearer services and QoS flows. It differs subtly from packet delay budget (PDB) and one-way packet delay, in that it is defined around the successful transfer of a complete SDU rather than individual link-level packet traversals. This makes it a service-level measurement rather than a purely network-level one. It can be measured end-to-end across the full user-plane path — from a User Equipment (UE) through the Radio Access Network (RAN), the Core Network (CN), and out to an external Packet Data Network (PDN) or Data Network (DN) — or it can be scoped to a specific segment or bearer.

Measurement of SDU transfer delay in 3GPP systems is typically performed at reference points where SDUs can be observed and time-stamped. For instance, an SDU can be time-stamped when it enters a PDCP entity on the transmitter side and again when it is successfully received and acknowledged on the receiver side. GTP-U extension headers support the transfer of timestamp information over the N3 and N9 interfaces to support such measurements in 5G. The 3GPP defines related QoS monitoring mechanisms in TS 23.501 and TS 38.415 for 5G systems, allowing the network to measure uplink, downlink, and round-trip packet delays, which feed into the same conceptual masonry from which SDU transfer delay originates.

SDU transfer delay is a key input to QoS class design. In UMTS, the four traffic classes (conversational, streaming, interactive, background) were characterised by delay and error tolerance. In LTE, QCIs and their associated Packet Delay Budgets and Packet Error Loss Rates served a similar function. In 5G, 5QIs and QoS Flows define delay budgets as a core attribute. SDU transfer delay provides a measurable complement to these static budgets, allowing operators to detect violations, perform SLA assurance, and trigger corrective network actions such as admission control or load balancing.

Because the term's definition has remained essentially stable since its introduction in Rel-4, it serves as a consistent QoS vocabulary anchor across all generations of 3GPP systems, from UMTS through LTE and 5G NR to beyond. Engineers and students should understand it as a service-level delay metric distinct from link-level latency, and as one of the foundational performance parameters in the ITU-T X.140 tradition that 3GPP adopted for its own QoS specifications.

Purpose & Motivation

SDU transfer delay exists to give network operators, service providers, and standards bodies a precise, measurable definition of how long the network takes to deliver a complete unit of user data from ingress to egress. Without such a definition, QoS discussions around 'latency' would remain vague and incomparable between implementations. The metric's adoption from ITU-T X.140 grounded 3GPP QoS work in an internationally recognised framework, ensuring interoperability of measurement and comparison across vendors and operators.

The term addresses the need for a service-level, SDU-based delay measure that is distinct from lower-layer packet latency. Lower-layer metrics measure time spent traversing a specific link or node, while SDU transfer delay captures whether the user's data actually arrived successfully within an observed interval. This distinction is critical for SLA enforcement and for verifying that a bearer or QoS flow is meeting its contracted delay characteristics.

Historically, before Rel-4, 3GPP QoS terminology was less formalised across all service data types; Rel-4's alignment with ITU-T X.140 consolidated this vocabulary and provided a stable reference point that has persisted through every subsequent release, enabling long-term comparability of delay metrics across generations from UMTS to 5G and beyond.