Per Hop Behaviour

Description

Per Hop Behaviour (PHB) is a core construct within the Differentiated Services (DiffServ) architecture, standardized by the IETF and adopted by 3GPP for IP-based bearer management. It defines the packet forwarding behavior that a network node (a "hop") must apply to a specific class of traffic. A PHB is not a service but a rule set implemented in a router's or node's queuing and scheduling mechanisms. It is invoked based on the DiffServ Code Point (DSCP) value in the IP header's Type of Service (ToS) or Traffic Class field. When a packet arrives, the node examines its DSCP, maps it to a specific PHB, and then applies the corresponding treatment through its traffic conditioning functions.

The implementation of a PHB involves several key components within a network node: classifiers, meters, markers, droppers, and queues. The classifier sorts packets based on the DSCP. The PHB then dictates actions such as the scheduling priority for transmission from an output port, the queue into which the packet is placed, and the drop precedence used during congestion. For example, the Expedited Forwarding (EF) PHB provides a low-loss, low-latency, and assured bandwidth service by ensuring packets are serviced from a high-priority queue with minimal queuing delay. The Assured Forwarding (AF) PHB group provides several classes, each with three levels of drop precedence, allowing for more granular treatment during congestion.

In the 3GPP architecture, PHBs are crucial for implementing the QoS characteristics of an EPS Bearer or 5G QoS Flow across the IP-based transport network (e.g., the S1-U, N3, or N9 interfaces). The Packet Data Network Gateway (PGW) or User Plane Function (UPF) marks the DSCP of downlink IP packets based on the QoS rules associated with the bearer/flow. Each router in the core and backhaul transport network is configured to recognize these DSCP values and apply the corresponding PHB. This creates a cohesive, hop-by-hop treatment that collectively realizes the end-to-end performance objectives like guaranteed bit rate, packet delay budget, and packet error loss rate. The PHB model is stateless and scalable, as each node makes independent decisions based only on the packet header, avoiding the need for per-flow signaling across the core network.

Purpose & Motivation

PHB exists to enable scalable Quality of Service (QoS) in large IP networks, such as the Internet and 3GPP core transport. Prior to DiffServ, the Integrated Services (IntServ) model used per-flow signaling (RSVP), which did not scale for service provider networks with millions of simultaneous flows. The PHB-based DiffServ model was developed to overcome this by aggregating traffic into a limited number of behavior aggregates (classes) and applying simple, fast processing at each router.

3GPP adopted this model to manage the diverse QoS requirements of different applications (voice, video, web browsing, IoT) over a shared IP infrastructure. It solves the problem of providing predictable packet delivery for real-time services like VoLTE or mission-critical IoT alongside best-effort data. By standardizing the mapping between 3GPP QoS parameters (QCI/5QI) and DSCP/PHB, it ensures consistent treatment from the core network gateway through the transport network to the radio access network's edge.

The creation of PHB was motivated by the need for a simple, deployable mechanism for service differentiation. It allows network operators to define service level agreements (SLAs) and engineer their networks to meet them without maintaining state for every single user session. This is fundamental to the 3GPP's all-IP evolution, allowing efficient support of converged services with varying sensitivity to delay, jitter, and loss over a cost-effective, packet-switched backbone.

Key Features

• Defines forwarding treatment (scheduling, queuing, dropping) at a single network node.
• Activated by the DiffServ Code Point (DSCP) in the IP packet header.
• Key component of the scalable IETF DiffServ architecture.
• Includes standardised behaviours like Expedited Forwarding (EF) and Assured Forwarding (AF).
• Enables aggregation of traffic flows into a few classes for scalable QoS management.
• Implemented in router queues and schedulers without per-flow state.

Evolution Across Releases

Defining Specifications