Concepts
Homomorphic encryption
Computing on data while it stays encrypted
How it works
Homomorphic encryption lets someone perform calculations on ciphertext so that decrypting the result gives the same answer as calculating on the plaintext. Partially homomorphic schemes support one operation (unpadded RSA and ElGamal support multiplication; Paillier supports addition). Fully homomorphic encryption (FHE), first shown possible in 2009, supports arbitrary computation but is much slower.
Strengths
- Data can be processed by an untrusted party, such as a cloud provider, without being decrypted
- Supports privacy-preserving analytics
Weaknesses and known attacks
- Fully homomorphic schemes carry large performance and ciphertext-size overheads.
- Malleability, the property that makes it work, is a weakness in ordinary encryption.
- Still a specialised technology with limited standardisation.
Use it for
- Privacy-preserving computation on sensitive data in the cloud
Don't use it for
- General-purpose encryption where AES-GCM is enough
Exam tip
Homomorphic encryption protects data in use, not just at rest or in transit. It is a CCSP and CASP+ favourite for cloud privacy questions.