Internet checksum

Internet checksum
Name	Internet checksum
Type	checksum
Introduced	1970s
Author	David D. Clark, Jon Postel
Related	Cyclic redundancy check, Fletcher's checksum, Adler-32

Internet checksum The Internet checksum is a simple error-detection method used in early ARPANET and Internet Protocol era protocols, designed to detect data corruption in transit. It operates on fixed-size words and produces a short fixed-length summary for inclusion in packet headers used by systems in TCP/IP stacks, BSD Unix, Linux kernel, Microsoft Windows, Cisco Systems, and other networking implementations. The checksum influenced later schemes in telecommunications and data storage but remains notable for simplicity and historical significance in IETF standards authored by figures like Jon Postel and contributors from organizations such as DARPA and BBN Technologies.

Background and Purpose

The method was conceived during the development of RFC specifications in the 1970s within the milieu of ARPANET research at institutions like RAND Corporation, MIT, Stanford University, and University of California, Los Angeles. Designers sought a lightweight mechanism compatible with constrained hosts including early VAX machines and DEC PDP-11 systems. The checksum's role in negotiation and error reporting tied it to protocols standardized by the Internet Engineering Task Force and working groups chaired by figures from USC/ISI and NASA research centers. It aimed to provide protection against accidental corruption on links such as those used by Ethernet pioneers at Xerox PARC and early satellite experiments coordinated with NASA Deep Space Network operations.

Algorithm and Implementation

The algorithm sums fixed-size 16-bit words using one’s-complement arithmetic and stores the one’s-complement of that sum in a header field. Implementations were embedded in stacks for Berkeley Software Distribution and GNU projects, and optimized for targets like Intel 80386 and ARM architectures in products from Sun Microsystems and Hewlett-Packard. Kernel developers at organizations including Google, Red Hat, Canonical (company), and NetBSD implemented word-aligned loops and unrolled additions to exploit cache and instruction pipelines. On platforms influenced by RISC-V research and SPARC designs, assembly-level routines leveraged carry handling peculiarities of the architecture. Implementations must treat odd-length payloads and pseudo-headers used by TCP and UDP carefully; the pseudo-header concept appears in foundational RFC 791 and RFC 793 documents authored and discussed in arenas like IETF meetings.

Variants and Related Checksums

Related algorithms and successors include Fletcher's checksum, Adler-32, and polynomial-based Cyclic redundancy check families standardized by bodies such as ISO and IEC. Academic evaluations by researchers at Bell Labs, MITRE Corporation, and Carnegie Mellon University compared the Internet checksum to alternatives like MD5 in integrity contexts and to SHA-1 and SHA-2 families for cryptographic use. Specialized variants appear in implementations for PPP links, ATM segmentation, and file systems developed at Oracle Corporation and IBM. Networking vendors such as Juniper Networks and Arista Networks sometimes provide hardware offload that implements checksums alongside CRC32 options in silicon from vendors like Broadcom and Intel Corporation.

Uses in Networking Protocols

The checksum is used in protocols in the TCP/IP suite including IP header, UDP, and ICMP headers; its pseudo-header application for TCP and UDP couples protocol and addressing information. Router and switch firmware from Cisco Systems, Huawei, and Extreme Networks implement checksum verification and generation, while virtualization platforms from VMware and Xen Project expose offload features to guest operating systems. Protocol analyses by institutions like IETF and IAB document interactions with tunneling protocols such as GRE and IPsec and with encapsulations used in MPLS deployments common in enterprise networks managed by operators including AT&T and Verizon Communications.

Limitations and Error Detection Properties

Analyses at University of Cambridge and ETH Zurich show the checksum detects many common transmission errors but is weak against certain patterns, particularly when bit errors cancel in one’s-complement arithmetic. It cannot provide cryptographic integrity or authenticity and is vulnerable to deliberate modification contrasted with schemes endorsed by NIST and used in TLS and IPsec. Studies comparing error-detection strength to CRC polynomials and to Fletcher algorithms were conducted by teams at Columbia University and TU Delft; these show higher detection rates for burst errors with carefully chosen CRC polynomials used in standards from IEEE (e.g., Ethernet frames) and ITU-T recommendations.

Performance and Optimization

Because it uses simple additions and end-around carry, the checksum is computationally cheap and amenable to software and hardware acceleration. Network interface controllers from Intel Corporation and Broadcom provide checksum offload, reducing CPU load in high-throughput environments operated by companies like Amazon Web Services and Microsoft Azure. Optimizations include loop unrolling, vectorized SIMD implementations for AVX and NEON instruction sets, and zero-copy techniques used in high-performance computing clusters at Argonne National Laboratory and Oak Ridge National Laboratory. Despite its modest detection capability, the balance of low cost and ubiquity keeps it in many legacy and constrained systems where more complex mechanisms from IETF or ISO are impractical.

Category:Checksum algorithms