prvsum

Prvsum is a term used in privacy-preserving data analysis to denote a protocol or function that computes the sum of a set of private numeric inputs without disclosing the individual values. The name is often written as prvsum or private sum and is used in discussions of secure aggregation.

The concept arose within the fields of secure computation and federated analytics, where organizations want to derive aggregate statistics from distributed data without exposing individual records. It is associated with privacy-preserving techniques that balance data utility with confidentiality, and it is sometimes described as a specialized form of secure aggregation.

Several approaches are used to realize prvsum. Additive secret sharing splits each input into random shares distributed among computation servers, which sum the shares and reconstruct the final total. Homomorphic encryption allows numbers to be encrypted and added in encrypted form, with decryption yielding the aggregate sum. Secure multi-party computation protocols enable a coalition of parties to compute the sum without revealing their inputs to others. In all cases, raw values remain with their owners or are protected by cryptographic means during computation.

Common applications include federated analytics, smart metering, healthcare data aggregation, and sensor networks, where the goal is to produce overall statistics without exposing individual measurements. Security considerations focus on threat models (for example, semi-honest versus malicious actors) and possible leakage through side information or collusion among parties. In practice, prvsum is often used alongside other privacy techniques, such as differential privacy, to provide stronger guarantees.

See also: differential privacy, secure multi-party computation, homomorphic encryption, secret sharing, privacy-preserving data analysis.