Applied cryptography and key storage
All AmbiSecure engineering content tagged Cryptography, newest first. Newer entries reflect current thinking; earlier entries come from the engineering archive.
Cryptography is the mathematical foundation under every credential AmbiSecure ships: symmetric ciphers (AES) for secure messaging, asymmetric algorithms (RSA, ECC) for signatures and key agreement, hashes and MACs for integrity, and key-derivation functions for session keys. In embedded identity the interesting problems are rarely the primitives themselves — they are key storage, entropy, side-channel resistance, and protocol composition.
This category gathers practical engineering notes on applied cryptography for smart cards and secure elements: how session keys are derived, why hardware entropy matters, how secure channels such as SCP03 are constructed, and where naive implementations leak. The guiding principle is that a correct algorithm on an insecure substrate is still insecure — the key must live in tamper-resistant silicon.
Articles tagged Cryptography
Frequently asked questions
Why does key storage matter more than the algorithm?
Standard algorithms like AES-128 and ECDSA are not the weak point — extractable keys are. A key held in a secure element that never exposes it to software resists both malware and physical attack, whereas the same algorithm with a key in application memory can be dumped.
What is a key-derivation function used for?
A KDF (for example the SP 800-108 counter-mode KDF used in SCP03) turns a long-term static key plus context into fresh per-session keys, so a compromise of one session's keys does not reveal the master key or other sessions.
Is hardware entropy really necessary?
Yes. Deterministic or low-entropy random number generation has broken real deployments (predictable keys, repeated nonces). A hardware TRNG in the secure element provides the unpredictability that key generation and challenge freshness depend on.
What is a hardware true random number generator?
A source of randomness derived from a physical process in silicon rather than computed by an algorithm. It seeds the deterministic generators that produce keys, which is why its quality bounds the security of everything above it.
Why do standards move away from algorithms that still look unbroken?
Because margin erodes. Key sizes and primitives are retired while attacks are still theoretical, so deployed devices with long service lives are not left holding the weakest link when practice catches up.