Fact-checked by Grok 2 weeks ago

International Data Encryption Algorithm

The International Data Encryption Algorithm (IDEA) is a symmetric-key that processes 64-bit blocks into 64-bit blocks using a 128-bit , consisting of 8 full rounds followed by a half-round output transformation. Developed by Xuejia Lai and James L. Massey at the Swiss Federal Institute of Technology in , it was first proposed in 1990 as the Proposed Encryption Standard (PES) and improved and finalized as IDEA in 1991, with the goal of replacing the aging (DES) by providing stronger security through a balanced mix of operations from incompatible algebraic groups. IDEA's core strength lies in its use of three pairwise incompatible group operations on 16-bit sub-blocks: bitwise exclusive-OR (XOR), addition modulo $2^{16}, and multiplication modulo $2^{16} + 1 (with special handling for zero in multiplication). These are applied in a generalized Feistel network structure, where each round splits the 64-bit block into four 16-bit sub-blocks and applies a novel "MA-box" transformation for diffusion and confusion, derived from 52 unique 16-bit sub-keys generated by rotating and partitioning the 128-bit user key. Decryption mirrors encryption but uses inverse sub-keys, ensuring the same algorithm works in reverse with minimal additional computation. The design deliberately avoids reliance on any single mathematical structure to resist cryptanalytic attacks like differential and linear cryptanalysis, which were emerging threats at the time. Commercialized by Ascom Systec Ltd., which held patents on IDEA (Swiss patent PCT/CH91/00117 and international equivalents) until their expiration in 2012, the algorithm saw adoption in secure communications software, notably as the default symmetric cipher in (PGP) version 2.0 and later releases up to PGP 5.0, replacing the insecure BassOmatic from PGP 1.0. It was also integrated into other tools like early versions of GnuPG under royalty-free licensing terms granted by Ascom for non-commercial use, contributing to its role in promoting and file encryption during the . From a security perspective, the full 8.5-round IDEA remains unbroken by any practical attack better than exhaustive key search across its $2^{128} key space, with only theoretical weaknesses identified for reduced-round variants and rare "weak keys" occurring with negligible probability ($2^{-77}). However, its 64-bit block size imposes fundamental limitations in modern contexts, capping secure data volumes at approximately $2^{32} blocks (about 32 GB) in modes like due to birthday paradox collisions, making it unsuitable for high-volume without modes providing or larger effective blocks. As a result, IDEA has been largely supplanted by and other 128-bit block ciphers in standards like OpenPGP (RFC 4880), though it retains and occasional use in legacy systems.

History and Development

Origins and Designers

The International Data Encryption Algorithm (IDEA) was designed by Xuejia Lai, a cryptographer, and James L. Massey, an American cryptographer and professor at the Swiss Federal Institute of Technology in Zurich (). Their collaboration stemmed from Lai's doctoral work under Massey's supervision at , where Lai's thesis focused on the design principles and security analysis of block ciphers, including the foundational Lai-Massey scheme that underpins IDEA's structure. Development of the algorithm began in the late 1980s as part of broader research at into secure symmetric-key block ciphers capable of replacing aging standards like the . Initially proposed in 1990 as the Proposed Encryption Standard (PES), the cipher evolved through iterative improvements informed by emerging cryptanalytic techniques. By 1991, an enhanced version named the Improved Proposed Encryption Standard (IPES) was introduced, which was later renamed IDEA to reflect its international applicability and finalized design. The primary motivation for IDEA's creation was to overcome DES's vulnerabilities, including its 56-bit effective key length, which was increasingly susceptible to brute-force attacks, and to provide robust protection against newly discovered threats such as differential cryptanalysis, introduced by Eli Biham and in 1990 and subsequently analyzed by Lai, Massey, and collaborator Sean Murphy in their refinement of PES. To ensure long-term security, the designers specified a 128-bit , significantly extending resistance to exhaustive search compared to DES, while maintaining a 64-bit block size for compatibility with existing systems. This focus on algebraic diversity and resistance to known attacks positioned IDEA as a forward-looking alternative in the evolving landscape of symmetric .

Evolution from PES

The Proposed Encryption Standard (PES), introduced by Xuejia Lai and James L. Massey in 1990, served as the direct predecessor to the International Data Encryption Algorithm (IDEA). PES was a symmetric-key operating on 64-bit blocks using a 128-bit key, incorporating three key algebraic operations—modular addition, bitwise XOR, and modular multiplication—to achieve through an iterated structure based on the Lai-Massey scheme. This design emphasized mixing incompatible group operations to resist known cryptanalytic attacks at the time, marking an early application of the Lai-Massey paradigm in construction. Following the discovery of differential cryptanalysis by Eli Biham and in 1990, which demonstrated vulnerabilities in PES after just a few rounds, Lai and colleagues refined the algorithm to enhance its resistance. The resulting improvements, developed between 1990 and , introduced targeted modifications to the round function and , increasing the number of rounds from PES's 8 full rounds to IDEA's 8.5 rounds (eight full rounds plus a final half-round output transformation) to promote better effects and properties. These changes ensured that IDEA maintained the core operations of modular addition (modulo $2^{16}), XOR, and multiplication (modulo $2^{16} + 1) while fortifying the cipher against differential attacks requiring significantly more chosen plaintexts. The first full specification of IDEA, originally termed the Improved Proposed Encryption Standard (IPES), was published by Lai, Massey, and Sean Murphy in at EUROCRYPT '91. IDEA's development also involved patent protection to support its . An initial was filed in on May 18, 1990 (CH A 1690/90), followed by filings in the United States on May 16, 1991 (leading to US Patent 5,214,703, issued May 25, 1993) and on the same date (EP 0482154 B1, issued June 30, 1993). These patents covered the specific round structure and key derivation process, enabling licensed implementations in standards like ISO/IEC 9979. Later evolutions, such as IDEA NXT introduced in 2005 as a patented extension with a wider and additional nonlinear components, built on this foundation but addressed emerging threats like advanced .

Design Principles

Cipher Specifications

The International Data Encryption Algorithm (IDEA) is a symmetric-key that employs a modified Lai-Massey scheme to process data securely. This structure combines bitwise operations, modular addition, and modular multiplication to achieve , distinguishing it from Feistel-based designs like . IDEA operates on fixed-size blocks of 64 bits for both input and output, dividing each into four 16-bit sub-blocks or words for internal processing. The cipher uses a 128-bit , which is expanded into 52 16-bit subkeys to drive the transformations across its rounds. It consists of eight full rounds followed by a final output transformation, often described as 8.5 rounds in total, providing a balance between computational efficiency and resistance to cryptanalytic attacks. The design of IDEA prioritizes a harmonious integration of security, processing speed, and implementation simplicity, making it suitable for both software and hardware environments prevalent in the early 1990s. Unlike DES, which shares the same 64-bit block size but relies on a shorter 56-bit effective key length, IDEA's 128-bit key was engineered to withstand brute-force attacks feasible with the era's computational resources.

Core Operations

The International Data Encryption Algorithm (IDEA) employs three fundamental algebraic operations on 16-bit sub-blocks to achieve its transformations: bitwise XOR (denoted ⊕), modular addition (denoted ⊞), and modular multiplication (denoted ⊠). These operations are applied to mix sub-blocks with key sub-blocks, ensuring both in the cipher's design. Developed by Xuejia Lai and James L. Massey, these primitives were selected for their computational efficiency on 16-bit processors common in the early , while providing incompatible algebraic structures that resist straightforward algebraic attacks. Bitwise XOR (⊕) performs bit-by-bit addition modulo 2 without carry, combining two 16-bit words such that each output bit is 1 if the corresponding input bits differ. Formally, for 16-bit words a = (a_{15}, \dots, a_0) and b = (b_{15}, \dots, b_0) in binary, the result is a \oplus b = (a_{15} \oplus b_{15}, \dots, a_0 \oplus b_0), where \oplus denotes modulo-2 addition. This operation draws from the vector space over GF(2), enabling linear mixing of bits essential for initial diffusion. Modular addition (⊞) adds two 16-bit words $2^{16} = 65{,}536, wrapping around to prevent beyond 16 bits. It is defined as: a \boxplus b = \begin{cases} a + b & \text{if } a + b < 2^{16}, \\ a + b - 2^{16} & \text{otherwise}. \end{cases} This (from \mathbb{Z}/2^{16}\mathbb{Z}) introduces carry effects that propagate changes across bits, contributing to avalanche-like . Modular multiplication (⊠) multiplies two 16-bit words $2^{16} + 1 = 65{,}537, a Fermat prime ensuring the is a field. To handle zero, which lacks an inverse, 0 is mapped to $2^{16} before multiplication, and if the result equals $2^{16}, it is remapped to 0. Formally: a \boxtimes b = (a \cdot b) \mod (2^{16} + 1), with the mapping $0 \mapsto 2^{16} and $2^{16} \mapsto 0. For decryption, the modular inverse of a non-zero sub-key is computed using the extended Euclidean algorithm, as the modulus's primality guarantees invertibility. This , from the of \mathbb{Z}/(2^{16}+1)\mathbb{Z}, provides non-linear mixing incompatible with XOR and . The choice of these operations ensures non-commutativity and incompatibility between algebraic groups—XOR in a over GF(2), addition in a of $2^{16}, and in a of 65{,}537—promoting thorough intermixing of data and keys to thwart cryptanalytic attacks relying on . Their implementation is straightforward in hardware and software, with multiplications optimized via table lookups or shifts on 16-bit architectures.

Encryption Process

Round Structure

The International Data Encryption Algorithm (IDEA) processes a 64-bit plaintext block through 8 full rounds followed by a half-round output transformation, utilizing a Lai-Massey scheme to ensure balanced diffusion and confusion via three incompatible group operations: bitwise XOR (⊕), addition modulo $2^{16} (+), and multiplication modulo $2^{16} + 1 (×). The plaintext is split into four 16-bit words P_1, P_2, P_3, P_4. Each full round employs six 16-bit subkeys K_1 to K_6, transforming the input words into output words for the next iteration, while the half-round uses four additional subkeys to finalize the 64-bit ciphertext C_1 C_2 C_3 C_4. This structure promotes security by mixing operations that do not commute or associate, preventing algebraic simplifications in cryptanalysis. For multiplication modulo $2^{16} + 1 = 65537, if an operand is 0 it is replaced by $2^{16}; if the result is $2^{16} it is replaced by 0, ensuring the operation is a bijection on 16-bit words. In a full round, the transformation begins with key mixing on the input words:
  • Compute Y_1 = P_1 \times K_1
  • Compute Y_2 = P_2 + K_2
  • Compute Y_3 = P_3 + K_3
  • Compute Y_4 = P_4 \times K_4
Next, XOR the results from opposite halves to form inputs to the round's core:
  • Y_5 = Y_1 \oplus Y_3
  • Y_6 = Y_2 \oplus Y_4
The central component is the key-dependent MA-box (multiplication-addition box), a 2×2 structure that applies the remaining subkeys to Y_5 and Y_6 for nonlinear :
  • Y_7 = Y_5 \times K_5
  • Y_8 = Y_6 + Y_7
  • Y_9 = Y_8 \times K_6
  • Y_{10} = Y_7 + Y_9
The MA-box outputs Y_9 and Y_{10}, which are then XORed with the initial transformed words to produce the round outputs, with the middle two crossed:
  • P'_1 = Y_1 \oplus Y_9
  • P'_2 = Y_3 \oplus Y_9
  • P'_3 = Y_2 \oplus Y_{10}
  • P'_4 = Y_4 \oplus Y_{10}
These P'_i become the inputs to the subsequent round. The MA-box's crossed operations ensure that a single-bit change in input affects all output bits with high probability, achieving full in minimal steps. The logical flow of the MA-box can be diagrammed as: compute Y_7 = Y_5 \times K_5, add Y_6 to Y_7 to get Y_8, multiply Y_8 by K_6 to get Y_9, add Y_9 to Y_7 to get Y_{10}; the outputs Y_9 and Y_{10} then cross-XOR with the pre-MA transformed words to balance the halves. After 8 full rounds, the half-round (output transformation) applies a truncated version without the central XORs or MA-box, using subkeys K_{49} to K_{52}, with crossed additions:
  • C_1 = P_1 \times K_{49}
  • C_2 = P_3 + K_{50}
  • C_3 = P_2 + K_{51}
  • C_4 = P_4 \times K_{52}
where the P_i are the outputs from the 8th round. This simplifies the final step while maintaining consistency with the round operations. The following pseudocode illustrates one full round (subkeys indexed per round as K{{grok:render&&&type=render_inline_citation&&&citation_id=1&&&citation_type=wikipedia}} to K{{grok:render&&&type=render_inline_citation&&&citation_id=6&&&citation_type=wikipedia}}):
function full_round(P1, P2, P3, P4, K):
    Y1 = multiply_mod(P1, K[1])  // × mod 2^16 + 1, with 0 → 2^16 handling
    Y2 = add_mod(P2, K[2])       // + mod 2^16
    Y3 = add_mod(P3, K[3])       // + mod 2^16
    Y4 = multiply_mod(P4, K[4])  // × mod 2^16 + 1, with 0 → 2^16 handling
    Y5 = Y1 XOR Y3
    Y6 = Y2 XOR Y4
    Y7 = multiply_mod(Y5, K[5])
    Y8 = add_mod(Y6, Y7)
    t1 = multiply_mod(Y8, K[6])
    t2 = add_mod(Y7, t1)
    P1_new = Y1 XOR t1
    P2_new = Y3 XOR t1
    P3_new = Y2 XOR t2
    P4_new = Y4 XOR t2
    return P1_new, P2_new, P3_new, P4_new
The overall encryption iterates this 8 times, followed by the half-round.

Key Schedule

The key schedule of the International Data Encryption Algorithm (IDEA) expands the 128-bit master key into 52 distinct 16-bit subkeys used across eight full rounds and a final half-round . The process begins by loading the 128-bit key into a and partitioning it into eight consecutive 16-bit words, conventionally denoted as K_1 through K_8. These initial words directly provide the first eight subkeys, extracted sequentially as 16-bit blocks from the . To generate subsequent subkeys, the entire 128-bit register undergoes a cyclic left rotation by 25 bits, after which the next eight 16-bit words are extracted in the same manner. This rotation-and-extraction step is repeated eight more times, yielding a total of nine sets of eight subkeys, though only 52 are utilized: six per full round for the eight rounds (48 subkeys) and four for the output transformation. The choice of a 25-bit shift ensures the subkeys are unique and avoids repetition, as the of 25 and 128 is , providing a full through the key material without periodic overlaps. Subkeys are indexed as Z_i^{(r)} for the i-th subkey in round r, where the first four subkeys in each round (Z_1^{(r)} to Z_4^{(r)}) are used for the initial key mixing (two multiplications and two additions) and the remaining two (Z_5^{(r)}, Z_6^{(r)}) for multiplications in the MA-box. When these multiplication subkeys are zero, they are specially interpreted as $2^{16} (equivalent to -1 \mod 2^{16}+1) during the operations to maintain invertibility, though no adjustment occurs in the schedule itself.

Decryption Process

The decryption process in the International Data Encryption Algorithm (IDEA) mirrors the encryption structure, employing the same sequence of 8 full rounds followed by a half-round output , but with the 52 subkeys applied in reverse order and each subkey inverted according to the operation it governs. This ensures that applying the decryption steps to recovers the original uniquely. The inversion leverages the algebraic properties of the underlying operations: and XOR over \mathbb{Z}/2^{16}\mathbb{Z}, and over \mathbb{Z}/(2^{16}+1)\mathbb{Z}, where $2^{16}+1 = 65537 is prime, guaranteeing invertibility for nonzero elements. To prepare the decryption subkeys from those generated for encryption, the sequence is reversed such that the subkeys for the first decryption derive from the last encryption 's subkeys (and vice versa), with specific applied based on usage. For subkeys involved in (Z_1, Z_4, Z_5, Z_6 per ), the is computed modulo $2^{16}+1 using the to find Z_i^{-1} such that Z_i^{-1} \cdot Z_i \equiv 1 \pmod{2^{16}+1}. For subkeys in (Z_2, Z_3), the is taken as -Z_i \mod 2^{16}, effectively replacing with in the decryption . Subkeys for XOR remain unchanged, as XOR is its own . The final half- uses the inverted subkeys from the encryption's output transformation (last four subkeys), and the initial decryption transformation uses the inverses of the first four encryption subkeys, adjusted to reverse the input mixing precisely. These inversions are precomputed once per , yielding the full set of 52 decryption subkeys. The decryption begins by loading the 64-bit ciphertext into four 16-bit words P_1, P_2, P_3, P_4, which serve as input to the first inverse round using the inverted subkeys from the eighth encryption round. Each full round applies the inverted Lai-Massey transformation: multiplication by the inverse subkey for P_1 and P_4, addition (now subtraction) for P_2 and P_3, followed by the modified addition box (MA-box) with XOR inverses. After 8 such rounds, the half-round inverts the initial encryption transformation by applying the remaining four inverted subkeys in reverse operational order, yielding the plaintext words. The inverse of addition is formally a \ominus b = (a - b) \mod 2^{16}, while inverse multiplication solves for x in x \cdot b \equiv a \pmod{2^{16}+1}. Computations for multiplicative inverses rely on the for efficiency, as the modulus is prime. The zero-handling in multiplication is similarly inverted. The invertibility of the entire process is guaranteed by the Lai-Massey scheme's design, which composes bijective operations from distinct algebraic groups, ensuring that each round and the overall permutation is perfectly reversible without loss. This structure avoids the need for separate decryption algorithms, as the same core routine suffices with adjusted subkeys, promoting efficient implementations.

Security Analysis

Resistance to

The International Data Encryption Algorithm (IDEA) was designed with inherent resistance to differential cryptanalysis, a technique introduced in 1991 that targets probabilistic relationships between plaintext pairs and their corresponding ciphertexts. By incorporating non-linear operations such as modular multiplication modulo $2^{16} + 1 (the prime ), IDEA disrupts the propagation of differences across rounds, making high-probability differentials difficult to exploit. An early analysis by the algorithm's designers demonstrated that breaking the full 8.5-round IDEA via differential cryptanalysis would require approximately $2^{47} chosen s, a complexity deemed impractical even with computational resources. IDEA also exhibits strong resistance to linear cryptanalysis, which seeks linear approximations between plaintext, ciphertext, and bits to recover the key with fewer known plaintexts than . The use of modular multiplications in its core operations introduces non- that breaks potential linear approximations, as these operations do not preserve in the same way as XOR or alone. Studies have confirmed that the probability of useful linear trails in IDEA is sufficiently low to prevent effective key recovery attacks, with the best approximations yielding biases too small for practical exploitation across the full . The most advanced known on the full IDEA remains the biclique , introduced in 2012, which extends meet-in-the-middle techniques using biclique structures to partition the key space more efficiently. This achieves a of $2^{126.1} operations for recovering a 128-bit key, requiring $2^{126.1} data and negligible memory beyond that, representing only a minor improvement over exhaustive search ($2^{128}). As of 2025, this remains infeasible with current or foreseeable hardware, affirming IDEA's security margin against generic key-recovery methods. Other cryptanalytic efforts have targeted reduced-round variants of IDEA but have not threatened the full design. For instance, a meet-in-the-middle attack on 6-round IDEA achieves key recovery with a time complexity of approximately $2^{112} encryptions and 16 known plaintexts, while attacks on 7.5-round IDEA require $2^{114} time and $2^{63} chosen plaintexts. No full-round practical breaks have emerged, with all known attacks limited to theoretical demonstrations on fewer rounds. A theoretical limitation arises from IDEA's 64-bit block size, which exposes it to attacks in modes like after approximately $2^{32} blocks of data (about 32 GB), potentially allowing collisions that leak information. However, this bound pertains to mode-of-operation security rather than the cipher's strength and can be mitigated by rotation or using authenticated modes, leaving the core algorithm secure. Recent academic assessments as of continue to view the original IDEA as unbroken against standard , though proposals for enhancements—such as integrating chaotic maps like the Hanon map into or round functions—aim to further bolster resistance for specific applications like audio . These modifications expand the effective and diffusion properties without altering the base structure. Exceptions like weak keys exist but are addressed separately and do not undermine the general resistance profile.

Weak Keys and Vulnerabilities

In 1993, researchers identified classes of weak keys in the International Data Encryption Algorithm (IDEA) due to its simple , where certain 128-bit keys cause subkeys in specific rounds to become all zeros or all ones, resulting in linear relations that simplify . These weak keys arise from patterns such as repeated 16-bit words or large numbers of zero bits in the key, with one prominent class containing approximately $2^{51} keys that exhibit differential weaknesses. Subsequent analysis expanded this to additional classes totaling up to $2^{64} weak keys overall. The impact of these weak keys enables distinguishing attacks on the full or key recovery on reduced-round variants with low computational effort; for instance, membership in the primary $2^{51}-key class can be tested using just two encryptions and solving a of 16 nonlinear equations involving 12 variables, achievable in roughly $2^{32} operations. Such vulnerabilities stem directly from the key schedule's lack of , allowing attackers to exploit predictable subkey structures without requiring extensive data. To mitigate these issues, the original analysis proposed minor modifications to the , such as introducing additional bit rotations or permutations to ensure full diffusion across subkeys and eliminate the linear relations in all rounds. Alternative redesigns, including enhanced round constants or nonlinear transformations in the schedule, have also been suggested to prevent formation while preserving IDEA's efficiency. Beyond weak keys, IDEA's 64-bit block size introduces risks in certain modes of operation for high-volume data; encrypting more than approximately $2^{32} blocks (about 32 GB) under the same key can lead to ciphertext collisions via the birthday paradox, potentially enabling attacks that recover plaintext information, as demonstrated in analyses of similar 64-bit ciphers. IDEA itself has no inherent major side-channel vulnerabilities in its design, though implementations must address timing or power analysis risks common to block ciphers. Given the $2^{128} key space, the probability of randomly selecting a weak key is about $2^{-64}, rendering them negligible for most practical uses with strong random key generation. These mitigations and the low incidence make modified versions of IDEA suitable for legacy applications where compatibility is prioritized over adopting newer algorithms.

Implementations and Usage

Patent Status and Availability

The International Data Encryption Algorithm (IDEA) was protected by key patents filed in the early 1990s. The primary U.S. patent, No. 5,214,703, was filed on January 7, 1992, by inventors Xuejia Lai and James L. Massey, and granted on May 25, 1993, to assignee Ascom Tech AG of . A corresponding patent, No. EP 0482154 B1, was filed on May 16, 1991, and granted on June 22, 1994. Ownership of the patents was initially held by Ascom Tech AG, with licensing responsibilities later transferred to MediaCrypt AG, a related entity focused on cryptographic solutions. Prior to expiration, commercial implementations required licensing fees from MediaCrypt AG, while non-commercial use was permitted without charge; for instance, (PGP) version 2.0 incorporated IDEA under a non-commercial agreement with Ascom. The patents have since expired, rendering IDEA fully available in the worldwide. The patent expired on May 16, 2011, and the U.S. patent on January 7, 2012, eliminating all royalty obligations for both commercial and non-commercial applications. Open-source implementations of IDEA are now freely available in established cryptographic libraries. The library provides support for IDEA through its legacy provider, enabling symmetric encryption via the EVP_CIPHER-IDEA interface. Similarly, the Crypto++ C++ library includes a dedicated IDEA class for block cipher operations. Reference implementations derive from the original algorithm description, including outlined in the 1991 proposal paper by Lai and Massey. No standardized specifications exist for IDEA, though software optimizations are common in modern libraries.

Modern Applications and Legacy Use

The International Data Encryption Algorithm (IDEA) saw early adoption as the default symmetric cipher in (PGP) version 2.0, released in 1992, where it replaced the insecure BassOmatic algorithm from PGP 1.0 to enable secure and encryption. This integration positioned IDEA as a key component in pioneering open-source privacy tools during the 1990s, supporting encrypted communications in early secure systems and protection applications. In the OpenPGP standard defined by RFC 4880, IDEA is listed as an optional (ID 1) for with legacy PGP implementations, such as those from version 2.6 and earlier, but implementations are not required to generate new content using it and should prefer stronger alternatives like or CAST5. As of 2025, IDEA is rarely deployed in new systems due to its 64-bit block size, which exposes it to practical attacks like SWEET32 when encrypting large volumes of , making exhaustive search over 2^64 possibilities increasingly feasible with modern computational resources. It has been superseded by the () with 128-bit or 256-bit keys, which offers superior security and was selected by NIST in 2001 as the federal block cipher standard, while IDEA received no such approval for U.S. government use. Modern protocols like prioritize larger block sizes to mitigate collision risks, contributing to IDEA's decline, as 64-bit ciphers are deprecated in TLS configurations to prevent vulnerabilities in extended streams. Despite its obsolescence, IDEA persists in legacy contexts, including optional support in GnuPG versions 1.4.13 and later (via 1.6) for decrypting archived PGP data, as well as in some embedded systems and academic cryptographic tools where compatibility with historical implementations is needed. Originally designed for efficiency on resource-constrained 16-bit processors—where its operations like modular multiplication modulo 65,537 enable strong with relatively low overhead compared to contemporaries like —IDEA performs slower than on 64-bit hardware today, lacking dedicated instruction set support like AES-NI that accelerates modern bulk .

References

  1. [1]
    A Proposal for a New Block Encryption Standard - ResearchGate
    Aug 7, 2025 · Conference PaperPDF Available. A Proposal for a New Block Encryption Standard. March 1999; Lecture Notes in Computer Science 90. DOI:10.1007/3 ...
  2. [2]
    [PDF] pdf
    Modern block ciphers covered in chronological order are DES (§7.4),. FEAL (§7.5), and IDEA (§7.6), followed by SAFER, RC5, and other ciphers in §7.7, col-.
  3. [3]
    RFC 4880: OpenPGP Message Format
    PGP 2.X always used IDEA with Simple string-to-key conversion when ... Note that PGP V2.6 uses 13 bits of compression. If an implementation uses ...
  4. [4]
    [PDF] Pre-Draft Comments on NIST SP 800-197A
    Security Bounds: Current modes often have “birthday bound” limitations (ex., around 2^64 blocks for a 128-bit block cipher like AES), beyond which security may ...
  5. [5]
    A Proposal for a New Block Encryption Standard - SpringerLink
    Download book PDF · Advances in Cryptology ... About this paper. Cite this paper. Lai, X., Massey, J.L. (1991). A Proposal for a New Block Encryption Standard.
  6. [6]
    On the design and security of block ciphers - Research Collection
    Theses · Doctoral Thesis. On the design and security of block ciphers. On the design and security of block ciphers. OPEN ACCESS. Downloads.Missing: PhD | Show results with:PhD
  7. [7]
    [PDF] Lai-Massey Scheme Revisited - Cryptology ePrint Archive
    Abstract. Lai-Massey scheme is a well-known block cipher structure which has been used in the design of the ciphers PES, IDEA, WIDEA,. FOX and MESH.Missing: original | Show results with:original
  8. [8]
  9. [9]
    [PDF] United States Patent - Googleapis.com
    May 16, 1991 · 11 Patent Number: (45) Date of Patent: 5,214,703. May 25, 1993. 4,850,019 7/1989 Shimizu et al. ....................... 380/29. 5,020,106 5 ...
  10. [10]
    [PDF] Lai-Massey Scheme and Quasi-Feistel Networks (Extended Abstract)
    The Lai-Massey scheme was originally used in the IDEA cipher [2,3]. But in this paper, by Lai-Massey scheme, we refer to the version given by Vaudenay in [20].
  11. [11]
    [PDF] New Data-Efficient Attacks on Reduced-Round IDEA
    Abstract. IDEA is a 64-bit block cipher with 128-bit keys which is widely used due to its inclusion in several cryptographic packages such as PGP.
  12. [12]
    New Attacks on IDEA with at Least 6 Rounds | Journal of Cryptology
    Nov 16, 2013 · 2 Description of IDEA and Notations. IDEA [22] is a 64-bit, 8.5-round block cipher with 128-bit keys. It uses a composition of XOR ...<|control11|><|separator|>
  13. [13]
    [PDF] Improved DST Cryptanalysis of IDEA - Bilkent University Computer ...
    The first is the transformation part: T : (P1,P2,P3,P4) → (P1 K1,P2 Ш K2,P3 Ш K3,P4 K4). In the second part, two inputs of the MA-box are calculated as p = (P1.<|control11|><|separator|>
  14. [14]
    [PDF] A New Meet-in-the-Middle Attack on the IDEA Block Cipher - Bilkent ...
    Abstract. In this paper we introduce a novel meet-in-the-middle attack on the IDEA block cipher. The attack consists of a precomputation and.
  15. [15]
    [PDF] A note on Weak Keys of PES, IDEA and some Extended Variants
    Oct 7, 2002 · The block ciphers IDEA and PES use three group operations: addition mod- ulo 216, represented by Ш, bitwise exclusive-or, denoted ⊕, and ...
  16. [16]
    [PDF] Key-Schedule Cryptanalysis of IDEA, G-DES, GOST, SAFER, and ...
    A key schedule expands a short master key to a large expanded key for encryption/decryption. This paper presents new attacks on these key schedules.<|control11|><|separator|>
  17. [17]
    None
    ### Core Operations in IDEA: Definitions and Rationale
  18. [18]
    Narrow-Bicliques: Cryptanalysis of Full IDEA - SpringerLink
    We apply and extend the recently introduced biclique framework to IDEA and for the first time describe an approach to noticeably speed-up key-recovery for ...
  19. [19]
    Sweet32: Birthday attacks on 64-bit block ciphers in TLS and ...
    However, for a block cipher with 64-bit blocks, the birthday bound corresponds to only 32 GB, which is easily reached in practice. When the amount of data ...Missing: IDEA | Show results with:IDEA
  20. [20]
  21. [21]
    [PDF] Improvement the International Data Encryption Algorithm(IDEA ...
    Jun 30, 2024 · The objective of this research is to improve and secure IDEA by integrating the ideas of Chaos Theory. Using Chaos Theory to produce strong, ...Missing: cryptanalysis | Show results with:cryptanalysis
  22. [22]
    Weak Keys for IDEA | SpringerLink
    Large classes of weak keys have been found for the block cipher algorithm IDEA, previously known as IPES [2]. IDEA has a 128-bit key and encrypts blocks of ...
  23. [23]
    [PDF] On the Practical (In-)Security of 64-bit Block Ciphers - Sweet32
    Mar 8, 2016 · 64-bit ciphers are vulnerable to collision attacks, like 3DES in HTTPS and Blowfish in OpenVPN, which can recover session cookies and ...
  24. [24]
    Cryptography - UBC Theoretical Physics
    Ascom Systec is the holder of the patents but licensing is now handled by MediaCrypt AG . PGP has a license to use it for non-commercial use only. Return to ...
  25. [25]
    PGP User's Guide, Volume II: Special Topics
    IDEA(tm) is a trademark of Ascom-Tech AG. There is no license fee required for noncommercial use of IDEA. Commercial users of IDEA may obtain licensing ...
  26. [26]
    2008-September.txt - GnuPG and GNUTLS Mailing List Archives
    A: IDEA is patented in the USA (US 5,214,703), Europe (EP-B-0482154)and Japan (JP 3225440) by Ascom Systec AG. This patent expires 25 May 2010 (USA) or 16 May ...
  27. [27]
    EVP_CIPHER-IDEA - OpenSSL Documentation
    DESCRIPTION¶. Support for IDEA symmetric encryption using the EVP_CIPHER API. Algorithm Names¶. The following algorithms are available in the legacy provider:.
  28. [28]
    draft-irtf-cfrg-cipher-catalog-01 - IETF Datatracker
    International Data Encryption Algorithm (IDEA) IDEA was first published in 1992. ... IDEA was used in earlier PGP versions). Intellectual Property Rights ...
  29. [29]
    A Detailed Overview of IDEA-128 Encryption : Python - MojoAuth
    VPN Services: Some Virtual Private Network (VPN) providers may use IDEA-128 for securing user data during transmission.
  30. [30]
    SWEET32: Birthday attacks against TLS ciphers with 64bit block size ...
    Dec 30, 2019 · Legacy block ciphers having block size of 64 bits are vulnerable to a practical collision attack when used in CBC mode.
  31. [31]
    Why not IDEA - GnuPG
    Dec 20, 2016 · Note that the IDEA patent has expired and IDEA is now directly supported since GnuPG 1.4.13 respective 2.x with Libgcrypt 1.6.
  32. [32]
    What is difference between AES-128 vs IDEA-128 - Compile7
    A common misconception is that IDEA will always be faster than AES. However, this often isn't the case on modern x86 CPUs, which feature AES-NI instructions.