sidechannel

A side channel is a source of information leakage that arises from the physical implementation of a system rather than from its intended interfaces. Side-channel information can be inferred from measurements of a device’s operation, such as how long a computation takes, how much power it consumes, the electromagnetic emissions it radiates, acoustic sounds it makes, or its thermal profile. These measurements may reveal secret data processed by the device, including cryptographic keys, passwords, or user inputs.

Common side channels include timing (data-dependent operation durations), power consumption (and related power or energy analyses), electromagnetic emanations (EM analysis), acoustic emissions (sound from hardware or cooling systems), and thermal emissions. Microarchitectural channels, such as cache timing and branch predictor behavior, can also leak information even when external interfaces appear secure. Side-channel attacks combine careful observation with statistical or cryptanalytic techniques to extract secrets from measured data. Notable areas include differential power analysis, timing attacks, and more recently cache and speculative-execution side channels.

Defenses focus on reducing or masking the information that can be learned from measurements. Approaches include constant-time and constant-power implementations, masking or blinding of sensitive data, avoiding data-dependent control flow or memory access patterns, noise injection, physical shielding, and tamper-resistant hardware. Verification and auditing of implementations, along with architectural changes to minimize leakage, are also important. Side-channel research remains a dynamic field as new hardware and software optimizations create potential leakage paths.