Header Compression

Description

Header Compression (HC) is a performance-enhancing protocol layer that operates between the Packet Data Convergence Protocol (PDCP) layer and the core network protocols in 3GPP radio access networks (UTRAN, E-UTRAN, NG-RAN). Its primary function is to suppress the redundancy found in consecutive packet headers belonging to the same packet flow before they are sent over the air interface. A typical VoIP packet, for example, may have a 40-byte IPv4/UDP/RTP header for only 15-30 bytes of voice payload, representing a huge overhead. HC algorithms identify static fields (like source/destination IP addresses) and predictable fields (like sequence numbers) in the headers. For the first packet in a flow, a full header is sent, and the compressor and decompressor establish a shared context—a stored copy of that header. For subsequent packets, only the changing fields (deltas) and a context identifier are transmitted, drastically reducing the header to a few bytes.

The 3GPP specifications adopt and profile robust header compression algorithms standardized by the IETF, primarily RObust Header Compression (ROHC), defined in RFC 5795. The PDCP layer in 3GPP systems is responsible for executing HC. The PDCP entity at the transmitting end (e.g., in the UE for uplink or gNB for downlink) acts as the compressor. It classifies packets into flows, establishes and maintains compression contexts, and sends compressed packets. The corresponding PDCP entity at the receiving end acts as the decompressor, using the context to reconstruct the original full headers before passing the packets up the stack. The ROHC framework is highly robust to packet loss and error-prone links due to its stateful operation and feedback mechanisms, allowing the decompressor to request context updates if synchronization is lost.

HC is applied on a per-radio bearer basis and is crucial for the efficiency of all real-time services. For Voice over LTE (VoLTE) and Voice over NR (VoNR), it is mandatory to use ROHC profiles for RTP/UDP/IP to make the service spectrally efficient. It also significantly benefits other low-latency, small-packet applications like gaming and IoT messaging. By reducing the header overhead, HC decreases the required radio resource blocks per packet, allowing the network to serve more users, reduce packet transmission time (lowering latency), and improve overall system capacity and user experience, especially at cell edges where radio conditions are poor.

Purpose & Motivation

Header Compression was introduced to address the severe inefficiency of transmitting small-payload, real-time traffic over bandwidth-constrained and expensive radio links. In early 3G networks, as packet-switched services emerged, it became apparent that protocols like IP, UDP, and RTP—designed for wired networks with abundant bandwidth—carried significant header overhead. For voice services migrating from circuit-switched to packet-switched (VoIP), this overhead could constitute 60-80% of the total packet size, wasting scarce radio resources and increasing latency due to longer transmission times.

The motivation for standardizing HC within 3GPP, starting in Release 4, was to enable efficient VoIP and other real-time multimedia services over UMTS. The limitations of previous approaches (like simple Van Jacobson compression for TCP/IP) were their lack of robustness on lossy wireless links and support for the full suite of Internet protocols. 3GPP's adoption of the IETF's ROHC framework provided a robust, standardized solution capable of compressing headers for IP, UDP, RTP, and ESP (for IPSec) flows, even in the presence of high packet error rates. This was a critical enabler for the all-IP core network vision and the eventual deployment of VoLTE, making packet-switched voice a viable and efficient alternative to traditional circuit-switched voice.

Key Features

• Dramatically reduces IP/UDP/RTP header overhead from ~40-60 bytes to 1-4 bytes per packet
• Implements IETF RObust Header Compression (ROHC) framework and profiles
• Operates within the PDCP layer in 3GPP radio protocol stacks (LTE, NR)
• Stateful compression based on shared context between compressor and decompressor
• Includes robust feedback mechanisms for error recovery and context synchronization
• Essential for spectral efficiency of VoLTE, VoNR, and other real-time low-latency services

Evolution Across Releases

Defining Specifications