The Threat of Timing Attacks

Constant-Time Cryptography

If a password verification algorithm checks a password character by character and returns 'false' immediately upon the first wrong letter, an attacker can precisely measure how many microseconds the server took to respond.

By timing the response, the attacker maps out exactly how many characters were correct. Modern cryptographic algorithms must be written in 'constant-time' to ensure every calculation takes the exact same duration regardless of the input data.

Constant-Time Implementations

Preventing timing attacks requires re-writing string comparison frameworks to unconditionally evaluate every single index of an array before returning the result, enforcing a completely rigid, unvarying execution timeline.

Everyday Example

Suppose a bouncer checks names on a VIP list. If a guest says 'Adam', and the first letter on the paper is 'B', the bouncer instantly says NO (taking 1 second). If a guest says 'Brad', the bouncer checks B, then R, then realizes the list says 'Brian', taking 3 seconds to say NO. An attacker systematically measures these seconds to map out the exact letters written on the VIP list!

The Deep Mathematics

Vulnerable string evaluation functions (`strcmp`) deploy an early-exit optimization loop `if (a[i] != b[i]) return false;`. Evaluating strings sequentially produces a measurable timing variance O(k) relative to the length of identical prefix subsets. Mitigation requires strict bitwise accumulation arrays evaluating length vectors continuously.