{
  "id": "quantum-key-distribution",
  "title": "Quantum Key Distribution",
  "lang": "en",
  "source_lang": "en",
  "summary": "Quantum key distribution uses quantum signals and authenticated classical communication to establish shared secret keys with security quantified under explicit physical and operational assumptions.",
  "category": "computer-science",
  "subcategory": "networks-security",
  "type": "concept",
  "aliases": [
    "QKD",
    "quantum key establishment"
  ],
  "generated_at": "2026-11-04",
  "model": "pro/gpt-6.1-sol",
  "url": "https://aiwiki.page/en/quantum-key-distribution/",
  "translations": {
    "zh-Hans": {
      "url": "https://aiwiki.page/zh/quantum-key-distribution/",
      "json": "https://aiwiki.page/zh/quantum-key-distribution.json"
    }
  },
  "sections": [
    {
      "heading": "",
      "text": "Quantum key distribution (QKD) is a family of cryptographic protocols that enables two parties to establish a shared secret key using quantum signals and authenticated classical communication. Its security derives from properties of quantum mechanics, rather than solely from the presumed difficulty of a mathematical problem. Under specified assumptions, a protocol either produces matching keys with a quantifiable bound on an adversary’s knowledge or aborts. QKD distributes keys; it does not itself encrypt messages, authenticate previously unknown parties, or protect compromised endpoints."
    },
    {
      "heading": "Physical principles",
      "text": "Classical information can generally be copied without disturbing it. Quantum information behaves differently: an arbitrary unknown quantum state cannot be perfectly copied, and measurements that extract information about certain possible states necessarily disturb them. These properties, including the no-cloning theorem, constrain an eavesdropper’s ability to learn encoded information without affecting observable statistics.\n\nMany QKD systems encode information in photons, using polarization, phase, or arrival time. In discrete-variable protocols, the signals represent qubits or other finite-dimensional quantum systems. Continuous-variable protocols instead encode information in continuous properties of an optical field.\n\nThe security principle is more precise than the claim that “any observation is detected.” Ordinary noise and malicious interference can produce similar effects, and some attacks exploit losses or imperfections rather than simply introducing errors. A security proof relates measured statistics to a conservative bound on an adversary’s information. The parties then shorten their partially secret data sufficiently to obtain a secure final key.\n\nQKD normally requires two communication channels:\n\n- A quantum channel, such as an optical fiber or free-space optical link, carrying the quantum signals.\n- An authenticated classical channel, carrying protocol messages that may be publicly readable but must not be alterable without detection.\n\nThe classical channel need not be confidential. Its authentication is essential: without it, an attacker could impersonate each participant to the other and establish separate keys with both."
    },
    {
      "heading": "A typical protocol: BB84",
      "text": "The BB84 protocol, proposed by Charles H. Bennett and Gilles Brassard in 1984, is the best-known prepare-and-measure QKD protocol. In its idealized form, a sender conventionally called Alice prepares one of four quantum states, belonging to two incompatible measurement bases. A receiver, Bob, independently chooses a basis for each measurement.\n\nA typical implementation proceeds through several stages.\n\n- Preparation and measurement. Alice chooses random bit values and bases, prepares the corresponding signals, and sends them to Bob. Bob measures the received signals using independently selected bases.\n- Sifting. Over the authenticated classical channel, the parties disclose their basis choices, but not the associated secret bit values. They retain suitable outcomes, usually those obtained with matching bases.\n- Parameter estimation. They disclose and discard selected data to estimate errors and other relevant channel statistics. If these statistics do not support secure key generation, they abort.\n- Information reconciliation. Classical processing corrects discrepancies between the retained strings. This often uses techniques related to error-correcting codes. Information revealed during reconciliation must be included in the security analysis.\n- Error verification. The parties check that their reconciled strings agree, allowing for a quantified probability of an undetected mismatch.\n- Privacy amplification. They apply a suitable compression procedure to reduce the adversary’s possible information, producing a shorter final key.\n\nPrivacy amplification commonly uses randomly selected universal hash functions. Its purpose differs from error correction: reconciliation makes the parties’ strings agree, while privacy amplification makes the resulting key sufficiently independent of the adversary.\n\nIn a simple intercept-and-resend attack on ideal BB84, an eavesdropper measures each signal in a randomly chosen legitimate basis and sends a replacement. Among sifted bits, this attack introduces an expected error rate of 25%. That figure illustrates a particular attack; it is not a universal detection threshold or a complete security argument."
    },
    {
      "heading": "Security definitions and assumptions",
      "text": "Modern QKD security is usually expressed through a composable definition: the generated key should remain secure when used as a component of another cryptographic task. Two central requirements are:\n\n- Correctness: the probability that the parties accept different keys is very small.\n- Secrecy: the accepted key is close to an ideal uniformly random key independent of the adversary’s information, including quantum information retained for later measurement.\n\nSecurity proofs may allow an adversary to control the quantum channel, coordinate attacks across many signals, and postpone measurements using quantum memory. Such proofs do not require the adversary’s computational resources to be bounded. Nevertheless, their guarantees depend on the stated model.\n\nImportant assumptions concern the characterization of sources and detectors, the isolation of laboratories, the quality of random choices, classical authentication, and the correctness of processing software. “Unconditional security,” a traditional description of QKD, therefore does not mean security without assumptions.\n\nKey rates and finite data\n\nThe final key is generally shorter than the sifted data because errors must be corrected and possible information leakage removed. For an idealized, asymptotic single-photon BB84 setting with appropriate symmetry assumptions and ideal one-way reconciliation, a familiar achievable secret fraction is\n\n\\[r \\geq 1-2h_2(Q),\\]\n\nwhere \\(Q\\) is the quantum bit error rate and\n\n\\[h_2(Q)=-Q\\log_2 Q-(1-Q)\\log_2(1-Q)\\]\n\nis the binary entropy function.\n\nThis expression becomes positive below an error rate of approximately 11%. That value is not a universal QKD operating threshold: different protocols, device models, reconciliation methods, and security arguments yield different bounds.\n\nReal systems process finite blocks of data. Statistical uncertainty, verification costs, authentication costs, and chosen security parameters reduce the usable key length. A reported rate is therefore meaningful only when it is clear whether it refers to transmitted signals, detected events, sifted bits, or final secret bits.\n\nAuthentication and initial trust\n\nQKD cannot establish authenticated identities from nothing. With information-theoretically secure authentication, the participants typically begin with a small pre-shared secret used in a message authentication code. Successful QKD can generate more secret material than authentication consumes, allowing part of the output to authenticate later sessions.\n\nAlternatively, digital signatures can authenticate the classical exchange. In that case, authentication depends on computational assumptions during the exchange. Later compromise of the signature mechanism does not by itself reveal a previously established QKD key, provided the exchange was correctly authenticated when it occurred and the other security assumptions held."
    },
    {
      "heading": "Main protocol families",
      "text": "Prepare-and-measure protocols\n\nIn prepare-and-measure schemes, one participant prepares states and another measures them. BB84 and related protocols belong to this family.\n\nPractical transmitters often use attenuated laser pulses rather than ideal single-photon sources. Some pulses contain multiple photons, creating opportunities for photon-number-splitting attacks in which an adversary exploits the extra photons. Decoy-state QKD varies pulse intensities so that the participants can estimate security-relevant contributions from different photon-number components. Decoy methods are central to many practical discrete-variable systems.\n\nEntanglement-based protocols\n\nEntanglement-based QKD uses correlations between measurements of entangled quantum systems. Artur Ekert’s 1991 proposal connected key distribution with Bell inequalities. Other entanglement-based approaches reproduce correlations closely related to BB84.\n\nAn entangled source can, in suitable protocols, be located between the participants and need not itself be trusted. However, using entanglement does not automatically make every component untrusted or every implementation device-independent.\n\nContinuous-variable protocols\n\nContinuous-variable QKD typically encodes data in optical-field quadratures and uses homodyne or heterodyne detection. These techniques can employ components related to those used in coherent optical communications.\n\nTheir security analysis must account for factors such as optical loss, excess noise, detector behavior, phase-reference handling, and finite sample sizes. Performance comparisons with discrete-variable systems depend on the operating conditions and the security assumptions used.\n\nMeasurement-device-independent and device-independent protocols\n\nMeasurement-device-independent QKD allows participants to send signals to an untrusted measurement station. Properly implemented, it removes detector-side trust assumptions covered by its security model. It does not eliminate requirements concerning the users’ state preparation, randomness, or laboratory security.\n\nDevice-independent QKD seeks to certify secrecy from observed nonlocal correlations without relying on detailed models of the internal operation of the quantum devices. It still requires assumptions such as secure laboratories, appropriate independence of choices, and authenticated classical communication. Demanding detection efficiencies and low noise make it technologically more restrictive than conventional QKD."
    },
    {
      "heading": "Implementation vulnerabilities",
      "text": "A secure abstract protocol can have an insecure physical implementation. Sources may emit unintended information, detectors may respond outside their modeled operating regime, and optical components may admit injected light. Such weaknesses are forms of side-channel attack when they expose information through behavior not represented in the intended protocol model.\n\nExamples include detector-blinding attacks, timing-dependent detector responses, and Trojan-horse attacks in which injected light probes internal settings. Countermeasures include monitoring, filtering, improved component characterization, security proofs that incorporate imperfections, and protocol architectures that reduce specific trust requirements.\n\nQKD also does not protect stored keys from malware, prevent an authorized endpoint from disclosing plaintext, or guarantee availability. An attacker can block signals or introduce enough disturbance to force an abort, producing a denial of service without learning the key."
    },
    {
      "heading": "Distance, networks, and deployment",
      "text": "Optical attenuation is a central limitation. As fiber length increases, fewer transmitted photons reach the receiver, reducing key rates and making background noise more significant. Ordinary optical amplifiers cannot transparently amplify arbitrary unknown quantum states in the way they amplify classical signals while preserving everything required for QKD security.\n\nFor an ideal pure-loss bosonic channel with transmissivity \\(\\eta\\), the repeaterless secret-key capacity is\n\n\\[K=-\\log_2(1-\\eta)\\]\n\nbits per channel use. At small \\(\\eta\\), it is approximately \\(1.44\\eta\\). This bound helps characterize the fundamental rate–loss trade-off for the specified channel model; it is not a direct prediction of a practical system’s throughput.\n\nSeveral network architectures address distance or coverage:\n\n- Trusted-node networks connect QKD links through intermediate sites. These sites must be trusted for the key material they handle.\n- Satellite links use free-space propagation to connect distant ground stations, subject to visibility, weather, pointing, and architectural constraints.\n- Twin-field protocols use interference at an intermediate station to obtain favorable rate–distance scaling under their assumptions, without requiring that station to be trusted as a key holder.\n- Quantum repeaters aim to extend quantum links through entanglement distribution and related operations. They are distinct from ordinary trusted relays and require additional quantum capabilities.\n\nQKD links and networks have been demonstrated and deployed, including fiber and satellite systems. Their integration with conventional communications requires key-management interfaces, authenticated control systems, compatible cryptographic equipment, and explicit accounting for trusted nodes."
    },
    {
      "heading": "Relationship to encryption and post-quantum cryptography",
      "text": "QKD is a key-establishment mechanism, not a replacement for all of cybersecurity. Once keys are established, applications normally protect data using symmetric cryptography.\n\nA one-time pad can provide perfect secrecy when its key is uniformly random, at least as long as the message, used only once, and kept secret. QKD can supply such key material, but the required key rate grows with the message rate, and message integrity still needs a separate mechanism. Many systems instead use QKD-generated keys with the Advanced Encryption Standard, retaining the computational security characteristics of that cipher.\n\nQKD differs fundamentally from post-quantum cryptography. Post-quantum algorithms operate on conventional computers and communication networks, using mathematical problems believed to resist attacks by both classical and quantum computers. QKD requires specialized physical links and equipment but can provide information-theoretic key-secrecy guarantees within its model.\n\nA sufficiently capable quantum computer running Shor’s algorithm would threaten widely used public-key systems based on factoring or discrete logarithms. QKD addresses key establishment rather than all the functions those systems perform: it does not directly replace digital signatures or solve identity management. Its suitability depends on achievable secret-key rates, distance, equipment and operating costs, physical infrastructure, and the trust assumptions of the complete system."
    }
  ],
  "links": [
    "cryptography",
    "quantum-mechanics",
    "no-cloning-theorem",
    "photon",
    "qubit",
    "authentication",
    "bb84",
    "error-correcting-code",
    "message-authentication-code",
    "digital-signature",
    "side-channel-attack",
    "denial-of-service-attack",
    "cybersecurity",
    "symmetric-key-cryptography",
    "one-time-pad",
    "perfect-secrecy",
    "advanced-encryption-standard",
    "post-quantum-cryptography",
    "quantum-computer",
    "shors-algorithm"
  ],
  "sources": [],
  "license": "https://creativecommons.org/licenses/by/4.0/"
}