1. Introduction
If practical quantum computers are ever built, the current public-key cryptography, which relies heavily on the hardness assumptions of factoring integers and solving discrete logarithms, will be vulnerable to quantum attacks. Given the escalating risks posed by quantum computing in recent years, the crypto community has shifted its research focus towards post-quantum cryptography (PQC). Constructing cryptographic schemes based on lattices is one of the promising PQC routines. It has driven several nations to launch professional organizations to start the standardizations of PQC schemes.
In 2016, the US National Institute of Standards and Technology (NIST) initiated a standardization competition for post-quantum cryptography primitives, including public-key encryption (PKE), key encapsulation mechanisms (KEMs), and digital signatures. Notably, lattice-based schemes occupied a significant portion of the submissions, accounting for 26 out of 64 in the initial round [
1], 12 out of 26 in the second round [
2], and ultimately, 7 out of 15 in the third round [
3]. In 2022, NIST finally selected lattice-based schemes named Kyber [
4] (official name is ML-KEM [
5]) and Dilithium [
6] (official name is ML-DSA [
7]) as the standardized candidates [
8].
The Chinese Association for Cryptologic Research (CACR) also initiated a PQC competition to standardize PQC schemes between 2018 and 2019. In the second round of the Chinese National cryptographic algorithms design contest, lattice-based schemes accounted for 11 out of 14 among public-key schemes [
9].
Most of these lattice-based schemes are “small lattice systems”, which are based on algebraically structured lattices, such as ideal lattices and module lattices, with polynomial rings as their underlying algebraic structures. The most common one is the cyclotomic ring , where is defined as the m-th cyclotomic polynomial.
For the lattice-based schemes, the learning with error (LWE) problem [
10] is one of the most common hardness assumptions to construct public-key encryption or key encapsulation mechanisms. But for those “small lattice systems”, they are based on variants of LWE, which are over cyclotomic rings
, where
is an
m-th root of unity, e.g., a ring learning with error (RLWE) problem [
11] or module learning with error (MLWE) problem [
12]. The most popular cyclotomic polynomial used in lattice-based crypto is the power-of-two cyclotomic polynomial:
, where
and
are power-of-two integers, and
is the Euler function. At this time, its corresponding cyclotomic ring is
. In fact, the analysis in [
11,
12] is mainly in the case of
. Through the NIST round 3, Kyber [
4], Saber [
13], and Dilithium [
6] use
as their underlying polynomial ring. There are some advantages of choosing power-of-two cyclotomic rings. (1) They are simple but useful:
, where
n is a power of two, is one of the simplest cyclotomic rings. And
is one of the best understood and the most widely studied cyclotomic rings in algebraic number theory, and there are no improved attacks that have been proposed against the schemes based on {R,M}LWE over
. (2) Most {R,M}LWE-based schemes use suitable parameters such that number theoretic transform (NTT) can be utilized to compute the polynomial multiplication in
. As we know, NTT-based schemes are very efficient due to the remarkable memory efficiency and speed of NTT, outperforming any other algorithm for multiplication in polynomial rings.
However, some disadvantages cannot be ignored in their practical application. The main focus should be on the inflexibility of selecting parameters. Take RLWE-based schemes as an example. The security level is directly influenced by the ring dimension
n of RLWE-based schemes. Since
n is a power of two, to achieve a higher security level, it is inconvenient to find a polynomial of some particular degree up to the next power of two. To reach 128-bit security, the ring dimension
n should be somewhere around 700 [
14]. There are two power-of-two integers: 512 and 1024 which are close to 700, but the former integer leads to insufficient security and the latter one leads to redundant security.
A natural question to ask in this point is as follows.
Motivating question 1: Are there ever flexible ways to use other cyclotomic rings rather than power-of-two cyclotomic rings?
Considering 128-bit security in the post-quantum era, it is interesting but meaningful to be able to construct lattice-based schemes over other cyclotomic rings as alternatives. For motivating question 1, the answer to the question is affirmative. The work in [
15] shows that for any cyclotomic polynomial
, RLWE can work entirely in the ring
. There also have been some schemes using trinomial cyclotomic rings. For example, Falcon Round 1 used
, where
[
16]. NewHope-Compact, an RLWE-based scheme [
17], and NTTRU, an NTRU-like scheme [
18], use
with a prime
q. Scabbard applies
with a power-of-two
q due to its hardness of ring learning with rounding (RLWR) [
19]. Later, the work in [
14] instantiated NTRU over some trinomial cyclotomic rings of the form
with various
n in order to select flexible parameters. The fact is that
is the
-th cyclotomic ring if
n is of the form
,
.
There is a gap for schemes based on module lattices, especially MLWE-based schemes. One exception is that the work in [
20] provided a variant scheme of Kyber; however, over power-of-three cyclotomic rings. Actually, no one has applied trinomial cyclotomics to MLWE-based schemes. Undoubtedly, MLWE-based schemes take into account the security of LWE-based schemes and the efficiency of RLWE-based schemes. Therefore, there will be a balance between security and efficiency by adjusting the parameters. Changing the sampling number
k is a major way to achieve different security levels for MLWE-based schemes. But, the increase in
k will lead to a more complex implementation. In addition,
is still widely used in MLWE-based schemes. For example, Kyber, an outstanding representative of MLWE-based schemes, and the only NIST-standardized KEM candidate, is based on the power-of-two cyclotomic ring
. Kyber’s supporting documentation has mentioned that “One could consider using Kyber with a ring that is not
”, as
may be exploited in the inflexibility of selecting parameters. Such a sentence is also applicable to other MLWE-based schemes. Hence, it leads to the following question.
Motivating question 2: Could we extend the known power-of-two MLWE-based schemes (e.g., Kyber) to the cases over trinomial cyclotomic rings, with appropriate selection of parameters so as to achieve a practical security level and matching error probabilities?
We answer motivating question 2 in the affirmative by proposing a variant scheme of Kyber, named Tyber, which is constructed over trinomial cyclotomic rings , where n is a positive integer of the form , with in this paper. The modulus q is chosen as a prime number, in order to be suitable for NTT. The security level of our Tyber is aimed at NIST security levels I, III, and V, while it can also achieve negligible error probabilities.
1.1. Related Works
There is a line of recent works that use trinomial cyclotomic rings of the form
.
Table 1 shows their detailed descriptions.
The early version of Falcon, i.e., Falcon Round 1 [
16], used
for its parameter set of
. Later, Espitau et al. [
21] proposed Mitaka, which is a simpler, parallelizable and maskable variant of Falcon, and its underlying polynomial rings include trinomial cyclotomic rings. Then, the Gaussian sampling and smoothing parameters of Mitaka were studied and optimized in subsequent work [
22]. Lyubashevsky and Seiler [
18] proposed a variant of NTRU, named NTTRU, by offering a new ring structure
. There have even been further improvements since then. Duman et al. [
14] extended the rings
with various
n in order to select flexible parameter sets. Additionally, Liang et al. [
23] proposed compact and efficient NTRU-based KEMs over trinomial cyclotomic rings with the aid of lattice-based error correction codes. Recently, Bai et al. [
24] designed compact PKEs based on the module-NTRU hardness assumption over trinomial cyclotomic rings. As for RLWE-based schemes, Alkim et al. [
17] improved NewHope and presented NewHope-Compact by offering a parameter set for NIST security level III, over the trinomial cyclotomic ring
. Similarly, Liang et al. [
25] proposed NewHope-Unified, which used
as its underlying ring for
. This can be extended to the case of RLWR-based schemes. For example, Bermudo Mera et al. [
19] introduced a suite of post-quantum KEMs, named Scabbard, and it contained an RLWR-based KEM applying
.
1.2. Our Contributions
We propose Tyber, a variant scheme of Kyber over trinomial cyclotomic rings of the form , where , . Our Tyber includes an IND-CPA secure public key encryption and an IND-CCA secure key encapsulation mechanism. The parameter sets of Tyber are provided, featuring quantum security of 128, 192, and 256 bits (actually achieving 129, 197, and 276 bits) with matching and negligible error probabilities. When compared to Kyber, our Tyber exhibits stronger quantum security, by 22, 31, and 44 bits, than Kyber for three security levels. All analysis and conclusions in this paper can be extended to any other power-of-two MLWE-based schemes.
2. Preliminaries
2.1. Notation and
Definitions
Let represent the ring of rational integers, with n and q being positive integers. We define as the quotient ring and it comprises the set . Furthermore, we denote as the group of invertible elements within . For a given real number x, we use the notation to represent the integer closest to x. Additionally, we introduce the notation for the ring and for the quotient ring . Elements in or are polynomials, denoted by regular font letters, such as . All the vectors in this paper are column vectors by default. Bold lowercase letters represent polynomial vectors over or while bold uppercase letters are polynomial matrices. For example, and , whose transposes are denoted by and , respectively. A polynomial f in (or ) has two equivalent representations: a power series form and a column vector form , where (or ) for . A function is said to be negligible if it satisfies for any positive c and sufficiently large . Such a function is denoted by negl.
Cyclotomics. Additional information regarding cyclotomic polynomials is available in [
26]. Given a positive integer
m, the
m-th root of unity is denoted by
. The
m-th cyclotomic polynomial, labeled
, is expressed as
. This type of polynomial is monic, irreducible, and has a degree of
over the polynomial ring
, where
represents the Euler function. The
m-th cyclotomic field is
, with its associated ring of integers being
. Some important types of cyclotomic polynomials are mentioned in this paper: (1) Power-of-two cyclotomic polynomials
with
m of the form
and
; (2) Trinomial cyclotomic polynomials with
m of the form
and
.
Modular reductions. Let be a positive integer. We define the modulo operation with signed remainder as follows. For even , represents the unique element in the range satisfying . For odd , represents the unique element in the range satisfying . For any , represents the unique element in the range satisfying . It is simply written as if the exact representation is not important.
Sizes of elements. For any element w in the ring , represents . We define the norm and the norm for any vector as follows: the norm is given by , while the norm is computed as . Furthermore, for a vector , we introduce the norm as and the norm as .
Sets and distributions. For a given set D, we utilize the notation to indicate that x is sampled uniformly from D. Furthermore, when referring to a probability distribution , the notation signifies that x is selected in accordance with the distribution . The centered binomial distribution , parameterized by a positive integer , is defined as follows: Sample uniformly from and output the sum . The distribution is defined as . Sampling a polynomial or a polynomial vector means sampling each coefficient according to individually.
Compression function. The compression function is formulated as , while the decompression function is defined as . When they deal with a polynomial (vector), the procedure is applied to each coefficient individually. For any , is an element close to x, i.e., .
Module learning with error (MLWE). Let
n be a power of two. The underlying hardness assumption of Kyber [
4,
27] is module learning with error (MLWE) [
12] over the ring
. The hard problem module learning with errors (MLWE) over
is to distinguish uniform samples
from the samples
, where
and
with
and
for all
i. The MLWE problem over
is hard if the advantage
of any probabilistic polynomial time adversary
is negligible, where
2.2. Cryptographic Primitives
A public-key encryption scheme contains PKE = (KeyGen, Enc, Dec), with a message space
. The key generation algorithm KeyGen returns a pair of a public key and a secret key
. The encryption algorithm Enc takes a public key
and a message
to produce a ciphertext
c. The deterministic decryption algorithm Dec takes a secret key
and a ciphertext
c, and outputs either a message
or a special symbol ⊥ to indicate a rejection. The decryption error probability of PKE, which is denoted as
, is defined as E[
Pr[Dec(
,Enc(
))]
]
. The advantage of an adversary
against indistinguishability under chosen-plaintext attacks (IND-CPA) for public-key encryption is defined as
A key encapsulation mechanism consists of three algorithms, which are defined as KEM = (KeyGen, Encaps, Decaps) with a key space
. The key generation algorithm KeyGen returns a pair of a public key and a secret key
. The encapsulation algorithm Encaps takes a public key
to produce a ciphertext
c and a key
. The deterministic decapsulation algorithm Decaps inputs a secret key
and a ciphertext
c, and outputs either a key
or a special symbol ⊥ to indicate a rejection. The correctness error
of KEM is defined as Pr[Decaps
Encaps(
)]
. The advantage of an adversary
against indistinguishability under chosen-ciphertext attacks (IND-CCA) for the key encapsulation mechanism is defined as
2.3. Kyber
In 2017, Bos et al. [
27] proposed a lattice-based cryptography suite called Cryptographic Suite for Algebraic Lattices (CRYSTALS for short). The algorithms of CRYSTALS are designed based on the MLWE problem over a module lattice, meaning that the algorithms take into account the security of LWE-based schemes and the efficiency of RLWE-based schemes. Among them, Kyber is an IND-CCA secure key encapsulation mechanism (KEM). Kyber follows a common construction framework. Specifically, it has two steps: the first step is to construct an IND-CPA secure public key encryption (Kyber.CPAPKE); The second step is to transform the IND-CPA secure PKE into an IND-CCA secure KEM (Kyber.CCAKEM) by using a variant of Fujisaki–Okamoto transform [
28,
29]. More precisely, Kyber is based on the MLWE problem over power-of-two cyclotomic ring
, where
n is a power of two. In the first round of the NIST PQC competition, Kyber’s modulus was chosen to be 7681, but it was changed after the first round, and adjusted from 7681 to 3329 [
4]. Additionally, Kyber’s secret distribution has been different from the ciphertext noise distribution for Kyber512 since the third round. In 2022, NIST finally selected MLWE-based Kyber (official name is ML-KEM) as the only standardized KEM candidate [
8].
5. Comparisons
As illustrated in
Table 3, we provide concise comparisons between our scheme and the NIST-standardized candidate Kyber [
4].
n is the polynomial dimension.
q is the modulus.
means the underlying cyclotomic polynomial used in the schemes. The magnitudes of the public key (
), ciphertext (
), and bandwidth (B.W., i.e.,
) are quantified in bytes. “Sec.C” denotes classical security and “Sec.Q” denotes quantum security, both of which are expressed in bits.
means the error probability.
Upon comparison, our scheme utilizes trinomial cyclotomic rings, so there is more flexibility when selecting parameters. The dimension n in our scheme can take values of the form . However, Kyber suffers from the inflexibility of selecting parameters due to its underlying power-of-two cyclotomic rings, since n can only be .
Although Kyber has a more compact public key and ciphertext for the three security levels, Kyber actually achieves quantum security of 107, 166, and 232 bits, respectively, which is far less than 128, 192, and 256 bits, respectively. Note that Kyber768 has a quantum security of 166 bits, which has a very large margin for quantum security of 128 bits, resulting in larger security redundancy. Another important point is that the error probability of Kyber1024 is only , which actually does not match its security requirement as 232-bit quantum security.
According to
Table 3, our scheme stands out with the practical and reliable security guarantees, since our scheme achieves the target quantum security of 128, 192, and 256 bits (actually achieving 129, 197, and 276 bits). The error probabilities of our scheme are precisely calibrated to satisfy the targeted security level for each parameter set, making them negligible in comparison to the specified security level, as they are substantively lower than
,
, and
, respectively. When compared to Kyber, Tyber648, Tyber972, and Tyber1296 exhibit stronger quantum security, by 22, 31, and 44 bits, than Kyber512, Kyber768, and Kyber1024, respectively. In addition, Tyber972 and Tyber1296 demonstrate significantly lower error probabilities when compared to Kyber768 and Kyber1024, respectively.
Note that Tyber uses different moduli,
and
, in order to achieve a balanced integrated performance for the three security levels. However, to adapt to different moduli we need two suites of NTT algorithms with different primitive roots of unity, resulting in more complicated implementation and more memory usage. In addition, according to the studies in
Section 4.1, the trinomial cyclotomic rings used in Tyber lead to lower error probabilities due to their more complicated structures, but the error probabilities can be controlled in a negligible range by choosing parameter sets carefully.