{
  "id": "post-quantum-cryptography",
  "title": "Post-quantum Cryptography",
  "lang": "en",
  "source_lang": "en",
  "summary": "Post-quantum cryptography develops classical cryptographic systems intended to remain secure against attacks by both classical and quantum computers.",
  "category": "computer-science",
  "subcategory": "networks-security",
  "type": "concept",
  "aliases": [
    "Quantum-resistant cryptography",
    "Quantum-safe cryptography",
    "PQC"
  ],
  "generated_at": "2026-11-02",
  "model": "pro/gpt-6.1-sol",
  "url": "https://aiwiki.page/en/post-quantum-cryptography/",
  "translations": {
    "zh-Hans": {
      "url": "https://aiwiki.page/zh/post-quantum-cryptography/",
      "json": "https://aiwiki.page/zh/post-quantum-cryptography.json"
    }
  },
  "sections": [
    {
      "heading": "",
      "text": "Post-quantum cryptography is the branch of cryptography concerned with systems designed to withstand attacks by both conventional computers and sufficiently powerful quantum computers. Its principal focus is the replacement of widely deployed public-key methods whose mathematical foundations would be vulnerable to quantum algorithms. Post-quantum algorithms normally run on ordinary computers and communicate over ordinary networks; their use does not require quantum hardware.\n\nThe field matters because public-key cryptography supports confidential communication, software authentication, financial transactions, and digital identity throughout the Internet. A cryptographically relevant quantum computer could undermine many existing mechanisms for establishing secret keys and verifying signatures. Moreover, information intercepted before such a machine exists could remain valuable when decryption becomes possible. Transitioning cryptographic infrastructure therefore involves both a mathematical problem—identifying durable security assumptions—and an engineering problem: replacing algorithms across protocols, devices, certificates, and long-lived records.\n\nPost-quantum research encompasses lattice-based, code-based, hash-based, and other constructions, together with security proofs, cryptanalysis, implementation techniques, and migration strategies. The publication of the first three United States National Institute of Standards and Technology post-quantum standards in August 2024 marked an important stage in the field’s movement from research into deployment. Standardization does not establish that an algorithm is unbreakable; it identifies concrete constructions and parameters suitable for implementation and continued public analysis."
    },
    {
      "heading": "Quantum algorithms and the public-key threat",
      "text": "The central quantum threat arises from Shor’s algorithm, introduced by Peter Shor in 1994. It solves integer factorization and discrete logarithm problems in polynomial time on an appropriate quantum computer. These problems underpin RSA, finite-field Diffie–Hellman key exchange, and elliptic-curve cryptography. For RSA, factoring the public modulus allows an attacker to reconstruct the private key. For Diffie–Hellman and elliptic-curve systems, computing discrete logarithms can reveal private values from public keys. The resulting attacks affect both confidentiality and digital authentication.\n\nPolynomial-time complexity does not imply that present-day quantum hardware can perform these attacks at cryptographically significant scales. A practical attack requires extensive fault-tolerant computation, including enough logical qubits and sufficiently reliable quantum operations. The physical resources needed depend on the cryptosystem, circuit design, hardware error rates, and the chosen quantum error-correction scheme. Estimates consequently vary substantially. Counting physical qubits alone does not establish whether a device can break a deployed cryptographic key.\n\nQuantum computing has a different effect on symmetric cryptography. Grover’s algorithm, published in 1996, provides a quadratic speedup for unstructured search. In an idealized model, exhaustive search over a \\(k\\)-bit key space can require approximately \\(2^{k/2}\\) quantum oracle evaluations rather than \\(2^k\\) classical trials. This is not the exponential improvement that Shor’s algorithm provides for factoring. Increasing symmetric key sizes can therefore compensate for important classes of generic quantum attack, although actual resource estimates must include circuit depth, implementation cost, and limits on parallelization. A 256-bit symmetric key is commonly used when seeking a substantial margin against such attacks.\n\nThe timing of the threat differs across applications. Recorded traffic protected by vulnerable public-key key establishment can be collected now and attacked later, a scenario often called “harvest now, decrypt later.” Classical forward secrecy does not by itself prevent this outcome: a quantum attacker could compute the ephemeral Diffie–Hellman secrets from recorded public exchanges. By contrast, a future ability to forge signatures does not automatically reveal the contents of old encrypted messages, but it can threaten software updates, identity credentials, and the future verification of signed records. These distinctions make the lifetime of confidentiality, the lifetime of authentication, and the expected replacement time of a system separate parts of its threat model."
    },
    {
      "heading": "Cryptographic functions and mathematical foundations",
      "text": "Post-quantum cryptography is not a single replacement algorithm. It comprises several cryptographic functions, each with its own security definitions and operational requirements. Public-key encryption lets anyone encrypt to a recipient’s public key, while only the holder of the corresponding private key can decrypt. A key-encapsulation mechanism, or KEM, instead establishes a shared secret: an encapsulation algorithm produces a ciphertext and a secret using a public key, and a decapsulation algorithm recovers that secret using the private key. The secret is then processed into keys for symmetric encryption and integrity protection.\n\nA digital signature serves a different purpose. A private signing key produces a signature on a message, and a public verification key allows others to check it. Signatures support authentication, software distribution, certificate systems, and document validation. A KEM does not automatically authenticate the parties exchanging secrets, and a signature does not conceal the signed message. Network protocols must combine these mechanisms with transcript binding, key derivation, and authenticated symmetric encryption.\n\nThe main mathematical families differ in the problems they present to an attacker. Lattice systems exploit the difficulty of finding unusually short vectors or solving noisy algebraic equations in high-dimensional spaces. Code-based systems use the difficulty of decoding general linear error-correcting codes. Hash-based signatures build on carefully specified security properties of cryptographic hash functions. Multivariate systems use polynomial equations over finite fields, while isogeny systems use maps between elliptic curves. None is considered secure merely because its underlying problem lacks a known quantum algorithm: the precise construction, parameters, and available structural attacks all matter.\n\nSecurity rests on computational assumptions rather than proof that all efficient attacks are impossible. A reduction may show that an attacker breaking a particular scheme could also solve a stated mathematical problem. Such a result establishes a conditional relationship, not the unconditional hardness of that problem. Some reductions concern worst-case problems, while practical attackers encounter samples drawn from particular distributions; connecting those settings is a central research achievement and also a source of technical qualifications.\n\nPost-quantum cryptography should also be distinguished from quantum key distribution. The latter uses quantum states and specialized physical channels to establish keys under particular assumptions. Post-quantum algorithms operate through conventional computation and communication. They can therefore be incorporated into existing software libraries, processors, and network protocols, although their larger messages and different operational requirements can still demand substantial changes."
    },
    {
      "heading": "Lattice-based constructions",
      "text": "A mathematical lattice is a discrete set of points generated by integer combinations of basis vectors. In sufficiently high dimensions, problems involving short lattice vectors can be computationally difficult even when the basis is known. Modern lattice-based cryptography frequently expresses this difficulty through noisy linear equations rather than directly exposing a geometric shortest-vector problem. Ajtai’s work in 1996 established influential connections between average-case cryptographic problems and worst-case lattice problems, while Oded Regev introduced the Learning With Errors framework in 2005.\n\nA simplified Learning With Errors sample has the form\n\n\\[\\mathbf{b}=A\\mathbf{s}+\\mathbf{e}\\pmod q,\\]\n\nwhere \\(A\\) is public, \\(\\mathbf{s}\\) is secret, and \\(\\mathbf{e}\\) is a small random error vector. Without the error, ordinary linear algebra can often recover the secret. With appropriately chosen noise and parameters, recovering the secret or distinguishing these samples from random values is believed difficult. Ring and module variants organize related computations over polynomial rings, reducing key sizes and accelerating arithmetic. That additional structure is part of the security assumption and must be examined rather than treated as a free optimization.\n\nML-KEM, standardized in FIPS 203, is derived from CRYSTALS-Kyber and uses module-lattice techniques. Its three standardized parameter sets illustrate the communication costs:\n\nParameter set | Encapsulation key | Decapsulation key | Ciphertext | Shared secret\nML-KEM-512 | 800 bytes | 1,632 bytes | 768 bytes | 32 bytes\nML-KEM-768 | 1,184 bytes | 2,400 bytes | 1,088 bytes | 32 bytes\nML-KEM-1024 | 1,568 bytes | 3,168 bytes | 1,568 bytes | 32 bytes\n\nThe construction combines a structured public-key encryption mechanism with a transformation designed to provide security against chosen-ciphertext attacks. Polynomial multiplication is accelerated using a number-theoretic transform. Ciphertext compression reduces bandwidth but must be accounted for in correctness and security analysis.\n\nLattice-based signatures use different techniques. ML-DSA, standardized in FIPS 204 and derived from CRYSTALS-Dilithium, combines module-lattice assumptions with a Fiat–Shamir-style signing construction and rejection sampling. Its standardized signatures range from 2,420 to 4,627 bytes. Rejection sampling helps keep the distribution of released signatures from exposing the signing secret. Falcon, selected by NIST for standardization in 2022, instead uses NTRU lattices and Gaussian sampling to obtain compact signatures, with more demanding implementation requirements.\n\nLattice cryptanalysis includes lattice reduction, sieving, enumeration, and hybrid attacks. Security estimates depend on the dimension and shape of the relevant lattice, the secret and error distributions, and assumptions about memory and quantum speedups. There is no general theorem that converts a familiar parameter name into a fixed amount of real-world security independent of the attack model."
    },
    {
      "heading": "Code-based cryptography",
      "text": "Code-based cryptography derives its security from decoding problems associated with linear error-correcting codes. A code maps messages to longer codewords so that transmission errors can be detected or corrected. If a receiver knows a suitable decoding algorithm and the necessary code structure, the original message can be recovered despite some corruption. Cryptographic constructions arrange for legitimate users to possess an efficient decoding method while presenting attackers with a decoding problem believed to be difficult.\n\nRobert McEliece proposed his public-key cryptosystem in 1978. In its classic form, a binary Goppa code provides efficient private decoding, while transformations obscure the code’s useful structure in the public key. Encryption adds a controlled number of errors to an encoded message. An attacker who lacks the private structure must recover the underlying message or error pattern. The long history of the construction is an important source of confidence, though it does not eliminate the need to assess concrete parameters and specialized attacks.\n\nThe major practical disadvantage of Classic McEliece is public-key size. Its parameter sets have public keys ranging from hundreds of kilobytes to more than a megabyte, much larger than those of common elliptic-curve systems or standardized lattice KEMs. Ciphertexts are comparatively small. This asymmetry can be acceptable where a public key is distributed infrequently and reused extensively, but awkward where certificates, protocol messages, or device storage impose strict limits. Niederreiter-style formulations, which represent encryption through error syndromes, are important in modern code-based KEM designs.\n\nStructured code families can reduce key sizes substantially, but they introduce further questions about hidden algebraic or quasi-cyclic structure. HQC, short for Hamming Quasi-Cyclic, uses quasi-cyclic code-based techniques and a different approach from systems that depend on hiding a secret structured decoder. NIST selected HQC for standardization in March 2025 as an additional key-establishment mechanism, providing an alternative mathematical family to ML-KEM. Selection is a stage in a standards process, not the same event as publication of a final standard.\n\nThe principal generic attack family is information-set decoding. An attacker seeks a subset of coordinates with a favorable error pattern, solves a reduced linear problem, and checks the resulting candidate. Improvements in this approach, including quantum variants, influence recommended dimensions and error weights. Implementations must also prevent decoding success, failure, timing, or error messages from revealing information. Even where the mathematical problem remains difficult, an observable difference in handling malformed ciphertexts can create a practical attack that bypasses the intended decoding challenge."
    },
    {
      "heading": "Hash-based digital signatures",
      "text": "Hash-based signatures derive their security primarily from properties of cryptographic hash functions rather than number-theoretic or lattice problems. Their origins include Ralph Merkle’s tree constructions and Leslie Lamport’s one-time signature scheme from the late 1970s. In a basic Lamport scheme, the private key contains pairs of random secret values. The public key contains their hashes. Signing a message digest reveals one secret from each pair, chosen according to the corresponding digest bit. Verification hashes the revealed secrets and compares them with the appropriate public values.\n\nThis construction illustrates both the appeal and the main operational hazard. Its security depends on the difficulty of deriving unrevealed secret values from public hashes. However, signing different messages with the same one-time key can reveal enough material to enable forgery. A Merkle tree allows many one-time public keys to be authenticated under a single root value. Each signature includes an authentication path connecting the selected one-time key to that root, so verifiers need not store every public key separately.\n\nStateful schemes such as XMSS and LMS track which one-time keys have already been used. XMSS was specified in RFC 8391 in 2018, and LMS in RFC 8554 in 2019. Their state must survive crashes, backups, device replacement, and concurrent signing. Restoring an old snapshot can be dangerous if it causes an already-used signing index to be reused. Dividing signing work across machines likewise requires careful allocation of unused leaves. The state-management requirement is therefore part of the cryptographic security boundary, not simply an administrative detail.\n\nStateless constructions avoid requiring a persistent counter. SLH-DSA, standardized in FIPS 205 in August 2024, is based on SPHINCS+. It combines a hypertree of hash-based components with techniques that control the risk associated with selecting signing resources without a global usage record. Its parameter sets use SHA-2 or SHAKE and offer different trade-offs between speed and signature length. Public keys range from 32 to 64 bytes, while signatures range from 7,856 to 49,856 bytes.\n\nHash-based signatures are attractive because their assumptions are comparatively narrow and because they provide mathematical diversity alongside lattice signatures. Their disadvantages include larger signatures or substantial hashing work. Security analysis must still account for the exact hash properties used, multiple targets, the number of signatures, and quantum queries. The security of an entire construction cannot be inferred simply by taking the output length of its hash function and applying a single generic rule."
    },
    {
      "heading": "Other mathematical approaches and lessons from cryptanalysis",
      "text": "Multivariate public-key cryptography builds schemes from systems of polynomial equations over finite fields. Solving arbitrary multivariate quadratic systems is computationally difficult, but usable cryptosystems require special structure that allows the private-key holder to sign or decrypt efficiently. The design challenge is to conceal that structure without leaving exploitable algebraic relationships. Signature schemes based on Oil-and-Vinegar constructions are prominent examples. Their public keys can be large, while signatures may be relatively compact.\n\nThe history of multivariate cryptography demonstrates the gap between a difficult general problem and a secure particular construction. Rainbow, a layered Oil-and-Vinegar signature scheme, reached the finalist stage of NIST’s initial post-quantum standardization process. In 2022, Ward Beullens published attacks that sharply reduced its claimed security and enabled a practical break of an important proposed parameter set. This did not prove that all multivariate signatures are insecure; it showed that the specific structure introduced to make Rainbow efficient could be exploited.\n\nIsogeny-based cryptography uses relationships between elliptic curves rather than the elliptic-curve discrete logarithm problem used in conventional systems. An isogeny is a structure-preserving map between curves. Some proposed systems encode secrets through chains of such maps, with security depending on the difficulty of reconstructing the relevant relationships. These constructions attracted attention partly because certain candidates offered unusually small public keys and ciphertexts compared with other post-quantum families.\n\nSIKE, a KEM based on Supersingular Isogeny Diffie–Hellman, was a prominent example. In 2022, Wouter Castryck and Thomas Decru developed an efficient classical attack against its underlying key-exchange construction by exploiting additional mathematical information exposed by the protocol. Subsequent work improved the attack. The result invalidated SIKE as a secure deployment candidate, but it was not a general efficient solution to every isogeny problem. Other isogeny-based directions remain separate research subjects and require their own assumptions and analysis.\n\nThese episodes illustrate why public cryptanalysis is central to post-quantum design. A proposal may have small keys, an attractive performance profile, and a reduction to a specialized problem, yet still fail because the specialized problem is easier than expected. Conversely, the failure of one construction need not invalidate its entire mathematical family. Research therefore distinguishes broad problem classes, particular hardness assumptions, protocol transformations, and implementation choices. Diversity can reduce dependence on one assumption, but deploying a larger collection of immature algorithms also increases implementation and maintenance complexity."
    },
    {
      "heading": "Security definitions, reductions, and parameter selection",
      "text": "The word “secure” has meaning only relative to a specified adversary, objective, and resource limit. For a KEM, an important goal is indistinguishability under adaptive chosen-ciphertext attack, commonly abbreviated IND-CCA. Informally, an attacker should not distinguish a real encapsulated secret from a random value even when allowed to request decapsulation of other ciphertexts. For signatures, a standard goal is existential unforgeability under chosen-message attack: obtaining valid signatures on selected messages should not enable an attacker to produce a valid signature on a new message.\n\nQuantum security introduces distinctions that do not occur in purely classical models. An adversary may perform quantum computation while communicating with a real protocol through ordinary classical messages. In some theoretical settings, it may also query an idealized hash function in a superposition of inputs. The latter setting is formalized by the quantum random-oracle model. A security argument established only for classical queries does not automatically remain valid under quantum queries. Nor should theoretical superposition access be assumed for every deployed network interface.\n\nReductions connect the success of a cryptographic attacker to the success of an algorithm solving an underlying problem. Their quantitative quality matters. A loose reduction may require conservative parameters because a modest advantage against the scheme translates into a much smaller advantage against the assumed hard problem. Conversely, a tight proof does not rescue an assumption that later becomes easy. Security evaluation combines proofs, the best known attacks, concrete estimates, implementation constraints, and the amount of independent scrutiny a design has received.\n\nNIST’s post-quantum security categories use reference tasks rather than a universal count of “security bits.” Categories 1, 3, and 5 are benchmarked against key search for AES-128, AES-192, and AES-256, respectively; categories 2 and 4 use collision-search benchmarks associated with SHA-256 and SHA-384. Comparisons depend on how computation, memory, circuit depth, and parallel resources are measured. A category is therefore not an assertion that every classical and quantum attack requires precisely the same number of elementary operations.\n\nCorrectness also enters the security analysis. Some lattice and code-based systems have a small probability that an honestly generated ciphertext does not decapsulate as intended. Parameters must make such events suitably unlikely, and implementations must not turn exceptional behavior into information about the secret key. More broadly, mathematical resistance to quantum attacks does not guarantee protection against weak randomness, protocol misuse, compromised endpoints, or classical cryptanalysis. “Post-quantum” identifies an intended adversary capability, not a blanket certification of security."
    },
    {
      "heading": "Research history and standardization",
      "text": "Post-quantum cryptography draws on ideas older than the practical quantum-computing threat. McEliece’s code-based proposal appeared in 1978, and early hash-based signature ideas emerged in the same period. Shor’s 1994 result transformed the strategic significance of such alternatives by showing that widely used factoring and discrete-logarithm systems were not secure against sufficiently capable quantum computation. Subsequent advances in lattice cryptography supplied additional constructions with useful security reductions and increasingly practical performance.\n\nResearch activity became more organized during the 2000s. The first International Workshop on Post-Quantum Cryptography took place in 2006, and a dedicated research community developed around algorithms, proofs, cryptanalysis, and implementation. Different families matured at different rates: some had long histories but inconvenient key sizes, while others offered better performance with newer assumptions. Benchmarking and open implementations became important because abstract operation counts did not capture memory use, message sizes, or protection against side channels.\n\nNIST announced its call for post-quantum algorithm submissions in 2016. In December 2017, it accepted 69 candidates into the first round of evaluation. The process combined public specifications, reference implementations, performance testing, security analyses, and successive rounds of selection. In July 2022, NIST selected CRYSTALS-Kyber for key establishment and CRYSTALS-Dilithium, Falcon, and SPHINCS+ for signatures. Further evaluation continued for additional KEMs, and a separate process sought additional signature algorithms.\n\nOn 13 August 2024, NIST published three final standards: FIPS 203 for ML-KEM, FIPS 204 for ML-DSA, and FIPS 205 for SLH-DSA. These standardized algorithms are derived from the selected research submissions, but their specifications are the authoritative definitions. Implementations cannot assume that every earlier Kyber, Dilithium, or SPHINCS+ version is byte-for-byte compatible with the corresponding standard. Exact encodings, parameter choices, checks, and domain-separation rules affect interoperability and security.\n\nAlgorithm standards are only one layer of deployment. Protocol specifications must define how algorithms are negotiated, how public keys and signatures are encoded, and how results enter key schedules. The Internet Engineering Task Force develops such integrations for Internet protocols. Validation programs, test vectors, cryptographic libraries, and hardware support address further layers. A standardized primitive can be correctly implemented yet used insecurely by a protocol, or securely specified yet deployed incorrectly through incompatible formats and configuration errors."
    },
    {
      "heading": "Implementation costs and protocol integration",
      "text": "Post-quantum systems change the balance between processor time, memory, and communication. Conventional elliptic-curve public keys can occupy only a few dozen bytes, whereas post-quantum public keys, ciphertexts, and signatures may occupy kilobytes or substantially more. These increases affect certificate chains, handshake fragmentation, packet loss, embedded storage, and network amplification defenses. A fast arithmetic operation does not necessarily produce a fast connection if the resulting messages require additional packets or exceed an intermediary’s limits.\n\nTransport Layer Security illustrates the distinction between key establishment and authentication. Replacing its classical ephemeral key agreement with a post-quantum KEM can protect newly established traffic against future quantum recovery, while leaving certificate signatures classical. That arrangement does not provide post-quantum authentication, but it addresses a different and immediate concern: the long-term confidentiality of recorded sessions. Migrating signatures additionally changes certificate formats, chain sizes, verification workloads, and trust infrastructure. Similar issues arise in virtual private networks, secure messaging, and remote administration.\n\nImplementations must resist side-channel attacks that reveal secrets through timing, power consumption, electromagnetic emissions, or processor behavior. Constant-time programming seeks to avoid secret-dependent branches and memory access patterns. Lattice schemes require care around sampling, rejection loops, polynomial arithmetic, and decapsulation checks. Code-based schemes require careful handling of decoding. Signature implementations must prevent secret information from leaking through sampled values or exceptional execution paths. Fault attacks add another concern by deliberately disturbing computations and analyzing the results.\n\nRandomness and input validation remain essential. Key generation and signing may depend on high-quality random seeds, while deterministic variants still require sound initial entropy and exact adherence to their specifications. Malformed public keys, ciphertexts, or signatures must be processed according to the algorithm and protocol rules. KEMs using implicit rejection derive an alternative secret when ciphertext validation fails rather than exposing a simple validity distinction to an attacker. Such mechanisms must be implemented without inadvertently leaking the very branch they are intended to conceal.\n\nConstrained devices and hardware security modules present additional challenges. A module may have fixed command interfaces, limited memory, or hardware acceleration designed around RSA and elliptic curves. Adding a post-quantum algorithm can therefore require changes beyond its cryptographic arithmetic. Testing uses known-answer vectors, independent implementations, malformed-input tests, and differential comparisons. Formal verification can establish selected properties of code or protocol models, but only within explicitly stated assumptions about compilers, hardware, randomness, and the surrounding system."
    },
    {
      "heading": "Hybrid deployment, migration, and long-lived trust",
      "text": "Migration begins with understanding where vulnerable cryptography is embedded. Public-key operations occur not only in visible network connections but also in boot firmware, package managers, backup systems, identity services, smart cards, industrial controllers, and signed archives. Cryptographic inventories therefore distinguish algorithms, key sizes, certificate dependencies, implementation libraries, device lifetimes, and the value of protected information over time. A system whose confidential records remain sensitive for decades has a different exposure from one protecting short-lived data, even if both use the same protocol today.\n\nA hybrid key-establishment construction combines a conventional mechanism with a post-quantum one. Its intended benefit is to preserve confidentiality if at least one component remains secure under the construction’s assumptions. For example, an elliptic-curve shared secret and a KEM shared secret can enter a carefully specified key-derivation process. The security claim belongs to the complete combiner and protocol, not simply to the presence of two algorithms. Public values, negotiated choices, identities, and the protocol transcript must be bound appropriately to prevent substitution or downgrade attacks.\n\nHybrid signatures raise a separate set of questions. A format may require both a conventional and a post-quantum signature, or it may accept either according to a policy. These choices have fundamentally different security consequences. Requiring both can preserve authenticity when one scheme fails, subject to the details of composition and verification. Accepting either can allow an attacker to exploit whichever scheme is weaker unless additional policy constraints apply. Certificate validation, trust-anchor handling, and compatibility with older software make signature migration more than a matter of attaching a second byte string.\n\nCrypto-agility denotes the ability to replace cryptographic algorithms and parameters without redesigning an entire system. Useful agility includes explicit algorithm identifiers, updateable implementations, manageable key lifecycles, and protocol mechanisms that prevent negotiation from selecting an unauthorized weaker option. It does not mean enabling every available algorithm. Long-lived devices can be especially difficult to transition when firmware verification depends on an immutable classical public key or when update mechanisms cannot accommodate larger signatures.\n\nArchived signatures and long-term evidence require treatment distinct from live communication. If a signing algorithm later becomes forgeable, a verifier may need trustworthy evidence that a document and its signature existed before the compromise. Timestamping, evidence records, and periodic renewal of archival protections address that problem under their own trust assumptions. Similarly, replacing a vulnerable key-exchange mechanism today does not restore confidentiality to traffic already collected under the old mechanism. Migration therefore has different effects on future sessions, stored ciphertext, software trust, and historical records, each governed by the cryptographic operations that originally protected it."
    }
  ],
  "links": [
    "cryptography",
    "quantum-computer",
    "internet",
    "shors-algorithm",
    "rsa-cryptosystem",
    "logical-qubit",
    "quantum-error-correction",
    "grovers-algorithm",
    "encryption",
    "digital-signature",
    "authentication",
    "cryptographic-hash-function",
    "quantum-key-distribution",
    "lattice-based-cryptography",
    "error-correcting-code",
    "internet-engineering-task-force",
    "transport-layer-security",
    "side-channel-attack",
    "formal-verification"
  ],
  "sources": [],
  "license": "https://creativecommons.org/licenses/by/4.0/"
}