Homomorphic Operations and Noise Management
What You'll Learn
You'll discover how to perform addition and multiplication directly on encrypted data, and why these operations can't continue forever.
How Operations Work on Ciphertexts
In homomorphic encryption, you can perform mathematical operations on ciphertexts without decrypting them first. The two fundamental operations are:
Addition: When you add two encrypted numbers, the result is an encryption of the sum of the original numbers. Enc(5) + Enc(3) = Enc(8)
Multiplication: When you multiply two encrypted numbers, the result is an encryption of the product. Enc(5) × Enc(3) = Enc(15)
This seems magical—you're computing on data you can't see!
The Noise Problem
Here's the catch: every homomorphic operation adds noise to the ciphertext. Think of it like making photocopies of photocopies—each generation gets a bit blurrier.
- Each addition adds a small amount of noise
- Each multiplication adds significantly more noise
- When noise grows too large, decryption produces garbage instead of the correct answer
This noise accumulation creates a computation depth limit—you can only perform a certain number of operations before the ciphertext becomes unusable.
The Practical Impact
Imagine calculating (2 + 3) × 4 × 5 × 6 on encrypted data. The initial addition works fine. The first multiplication adds noise. The second multiplication adds even more. By the third multiplication, the noise might be so large that decryption fails completely.
This is why partially homomorphic schemes (supporting unlimited additions or unlimited multiplications) differ from fully homomorphic schemes, which use sophisticated techniques like bootstrapping to periodically "refresh" ciphertexts and reduce noise.
Key Takeaway: Homomorphic operations enable computation on encrypted data, but noise accumulation limits how many operations you can chain together before decryption becomes impossible.