Checksum

A checksum is a small datum derived from a larger data set, used to detect errors in transmission or storage. It is computed by applying a checksum algorithm to the data and is often stored or transmitted with the data. When the data are later read or received, the checksum is recomputed and compared to the original value to verify integrity. Checksums are intended to catch accidental changes such as bit flips, dropped or reordered bytes, and small corruption, but they are not designed to resist intentional tampering.

Common types of checksums include additive checksums, parity checks, and more fault-tolerant methods such as cyclic redundancy checks (CRC). CRCs use polynomial division over a finite field and are widely adopted in networks, disk drives, and other storage systems because they detect common error patterns efficiently. The Internet checksum, specified in RFC 1071, is another widely used method in TCP/IP protocols. In contrast to cryptographic hash functions, traditional checksums are designed for error detection and are not meant to provide security guarantees against intentional modification.

Applications of checksums include verifying downloaded files, ensuring data integrity during backups and replications, and validating data blocks in communication protocols. They are frequently stored in accompanying files or headers so that a recipient can confirm that the data remained unchanged in transit or storage. Limitations include the possibility of collisions (two different data sets yielding the same checksum) and the fact that many checksums do not protect against deliberate tampering.

See also: CRC, parity check, hash function, error-detecting code.