Average Window

Description

The Average Window (AW) is a fundamental parameter in 5G's Policy and Charging Control (PCC) framework, defined in 3GPP TS 29.513. It operates as a configurable time interval, typically specified in seconds, over which specific traffic measurements—such as uplink and downlink data rates—are averaged. This averaging is not a simple arithmetic mean but a continuous, sliding-window calculation performed by the Policy Control Function (PCF) or the User Plane Function (UPF) based on real-time usage reports. The core mechanism involves monitoring packet flows associated with a PDU Session or specific Service Data Flow (SDF), accumulating byte counts within the window duration, and computing the average rate to compare against predefined policy thresholds, such as those in Usage Monitoring Control or QoS policy rules.

Architecturally, the AW is a key component within the PCC rule set that the PCF provisions to the Session Management Function (SMF). The SMF then enforces these rules by configuring the UPF accordingly. When the UPF performs traffic detection and measurement, it applies the AW to smooth the instantaneous data rate. For example, if an AW of 60 seconds is configured for a guaranteed bit rate (GBR) flow, the UPF will calculate the average data rate over the most recent 60-second window to determine if the flow is complying with its GBR commitment or if congestion actions are needed. This prevents the network from overreacting to very short-term bursts, which are normal in packet-switched traffic.

The role of the AW extends across multiple PCC functions, including usage monitoring for fair usage policies, dynamic policy enforcement for application-aware shaping, and QoS verification. It is integral to the 5G network's ability to provide consistent quality of service, implement tiered service plans, and manage network congestion intelligently. By decoupling policy decisions from momentary fluctuations, the AW enhances network stability, improves user experience predictability, and allows for more sophisticated, time-aware charging models. Its configuration is highly flexible, allowing network operators to tailor the responsiveness of their control systems to different services, from low-latency applications requiring shorter windows to background services where longer averaging is appropriate.

Purpose & Motivation

The Average Window was introduced to address the limitations of instantaneous or very short-term measurement in policy enforcement, which could lead to erratic network behavior and poor user experience. In pre-5G systems, policy triggers based on momentary data rates could cause unnecessary throttling, re-authorization requests, or charging events during normal traffic bursts, such as when a user opens a web page or starts a video stream. This 'jitter' in policy application was inefficient and could degrade perceived service quality. The AW provides a smoothing mechanism, ensuring that policy decisions reflect sustained usage patterns rather than transient spikes.

Historically, as networks evolved towards more granular and dynamic QoS control with the advent of 5G and network slicing, the need for stable, predictable policy enforcement became critical. The AW allows the 5G core network to implement sophisticated service-level agreements (SLAs) and charging models that are based on average performance over time, which aligns better with user experience and commercial offerings. It solves the problem of network overreaction, enabling more reliable congestion management, fair resource sharing among users, and accurate usage monitoring for billing purposes. By incorporating the AW into the PCC architecture, 3GPP provided operators with a essential tool for implementing intelligent, time-aware network policies that balance responsiveness with stability.