HeaderKompression

Header compression is a networking technique that reduces the size of protocol headers in packets transmitted over a link, saving bandwidth and reducing latency on low-capacity or high-latency connections. It targets the typically repetitive and slowly changing header fields, such as IP addresses, ports, and protocol flags, to minimize overhead without altering the payload data.

In practice, a compressor and a decompressor establish a shared context. For each packet, the compressor sends a compact representation containing only the fields that have changed since the previous packet, along with small updates to the context. The decompressor uses the maintained context to reconstruct the full, uncompressed header. There are stateless variants that do not maintain long-term state and rely on per-packet structures, and stateful variants that preserve context across many packets to achieve higher compression.

A widely known approach is Robust Header Compression (ROHC), an IETF-standardized framework that supports headers for IPv4, IPv6, UDP, RTP, and some other protocols. ROHC is designed to handle packet loss and reordering more gracefully than earlier schemes and to operate across a range of link technologies, including cellular networks. Other forms of header compression are used in specific environments, such as IPv6 over low-power networks, where headers can be compressed to suit constrained devices and links.

Limitations include the potential for decompression errors after loss of synchronization, the added processing complexity, and diminished effectiveness when headers change frequently or when high packet loss disrupts context maintenance. Security and transparency considerations also influence deployment.