DOCSLIB.ORG

On Low-Complexity Lattice Reduction Algorithms for Large-Scale MIMO

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
On Low-Complexity Lattice Reduction Algorithms for Large-Scale MIMO 1 On Low-complexity Lattice Reduction Algorithms for Large-scale MIMO Detection: the Blessing of Sequential Reduction Shanxiang Lyu, Jinming Wen, Jian Weng and Cong Ling Abstract—Lattice reduction is a popular preprocessing strategy low-complexity reduction algorithm for large-scale systems. in multiple-input multiple-output (MIMO) detection. In a quest Moreover, an efficient reduction algorithm for large-scale for developing a low-complexity reduction algorithm for large- problems may also find its applications to cryptanalysis [9] scale problems, this paper investigates a new framework called sequential reduction (SR), which aims to reduce the lengths of and image processing [10]. all basis vectors. The performance upper bounds of the strongest The principle of designing a reduction algorithm varies reduction in SR are given when the lattice dimension is no larger depending on the desired basis properties: to make all the than 4. The proposed new framework enables the implementation basis vectors short, or to make the condition number of the of a hash-based low-complexity lattice reduction algorithm, which reduced basis small. There are several popular types of lattice becomes especially tempting when applied to large-scale MIMO detection. Simulation results show that, compared to other reduc- reduction strategies, such as Minkowski reduction, Korkine- tion algorithms, the hash-based SR algorithm exhibits the lowest Zolotareff reduction (KZ) [11], Gauss reduction [12], Lenstra– complexity while maintaining comparable error performance. Lenstra–Lovász (LLL) reduction [13], Seysen reduction [14], Index Terms—lattice reduction, MIMO, large-scale, hash- etc. They yield reduced bases with shorter or more orthogonal based. basis vectors, and provide a trade-off between the quality of the reduced basis and the computational effort required for finding it. In essence, a reduction algorithm aims to I. INTRODUCTION find a unimodular matrix to transform an input basis into The number of antennas has been scaled up to tens or another one with better property. The process involves a series hundreds in multiple-input multiple-output (MIMO) systems of elementary operations noted as reflection, swapping, and to fulfill the performance requirements needed by the next translation. These operations vary for distinct algorithms. generation communication systems [1]. A critical challenge Much work has been done to advance conventional reduc- that comes with very large arrays is to design reliable and tion algorithms. Regarding KZ, refs. [15]–[17] give some prac- computationally efficient detectors. Though the well-known tical implementations and improve the performance bounds. maximum likelihood detector (MLD) provides optimal er- As for blockwise KZ, its faster implementations and the ror performance, it suffers from exponential complexity that expected basis properties are given in [18]. Researchers have grows with the number of transmit antennas [2]. In the also been constructing and analyzing the variants of LLL with past two decades, lattice-reduction-aided suboptimal detec- great effort. For instance, the size reduction step is optimized tion techniques have been well investigated [3]–[5], whose in [16], [19], [20], the implementation order of swaps is instantaneous complexity does not depend on constellation simplified in [21]–[23], and the fixed complexity versions of arXiv:1912.06278v1 [cs.IT] 13 Dec 2019 size and noise realizations, but collect the same diversity as LLL are given in [19], [22], [24]. In contrast, the direction on the MLD for MIMO systems [6]–[8]. Although conventional Seysen reduction has few follow-up studies [25], [26], partly lattice reduction algorithms suffice for small-scale MIMO because of the fact that Seysen reduction has unsatisfactory systems, there is still an avenue to pursue a more practical performance in high dimensions. While LLL and blockwise KZ are still the default choices in This work was supported in part by the National Natural Science Foun- cryptography to reduce a basis in hundreds of dimensions, the dation of China under Grants 61902149, 61932010 and 11871248, in part by element-based lattice reduction (ELR) proposed in [27] has the Major Program of Guangdong Basic and Applied Research under Grant 2019B030302008, in part by the Natural Science Foundation of Guangdong become more attracting in large MIMO, which preprocesses Province under Grants 2019B010136003 and 2019B010137005, in part by a large basis with even lower complexity than LLL. Later the Fundamental Research Funds for the Central Universities under Grant ref. [27] has been generalized to ELR+ [28] for small-scale 21618329, and in part by the National Key R&D Plan of China under Grants 2017YFB0802203 and 2018YFB1003701. This paper was presented in part problems, but the theoretical characterization of ELR and at the 8th International Conference on Wireless Communications and Signal ELR+ has not been given a rigorous treatment, even for small Processing (WCSP), Yangzhou, China, Oct. 2016. dimensions. It is noteworthy that ELR and ELR+ have totally S. Lyu, J. Wen and J. Weng are with the College of Cyber Security, Jinan University, Guangzhou 510632, China (e-mail: [email protected], different structures with LLL variants, and one might be lured [email protected], [email protected]). S. Lyu is also with the into the belief that ELR and ELR+ can be tuned to arrive State Key Laboratory of Cryptology, P.O. Box 5159, Beijing, 100878, China. at more sophisticated methods. Nevertheless, no analytical C. Ling is with the Department of Electrical and Electronic Engineering, Imperial College London, London SW7 2AZ, United Kingdom (e-mail: skills can be inherited from LLL/KZ literature [12], [13], [email protected]). which makes the performance analysis of the new algorithms 2 complicated. vector space spanned by vectors in B . B x and x Γ π Γ ( ) πB⊥Γ ( ) In this work, we investigate a general form of ELR and denote the projection of x onto span(BΓ) and the orthogonal + ELR which we refer to as sequential reduction (SR). We complement of span(BΓ), respectively. x denotes rounding derive the objective function from a MIMO detection task, x to the nearest integer, x denotes getting⌊ ⌉ the absolute value and present the general form of an SR algorithm which can of x, and x denote the| Euclidean| norm of vector x. N and Z solve/approximate the smallest basis problem. Unlike KZ or respectivelyk k denotes the set of natural numbers and integers. Minkowski reduction, SR reduces the basis vectors by using The set of n n integer matrices with determinants 1 is × ± sub-lattices so as to avoid a basis expansion process. The denoted by GLn(Z). strongest algorithm in SR tries to minimize the length of basis vectors with the aid of a closest vector problem (CVP) oracle. II. PRELIMINARIES We show bounds on the basis lengths and orthogonal defects for small dimensions. After that, the feasibility of applying A. Lattices SR to reduce a large dimensional basis is analyzed, and we An n-dimensional lattice Λ is a discrete additive subgroup actually construct a hash-based algorithm for this task. Our in the real field Rn. Similarly to the fact that any finite- simulation results then show the plausibility of using SR in dimensional vector space has a basis, a lattice has a basis. large-scale MIMO systems. To consider a square matrix for simplicity, a lattice generated n n Preliminary results of this work have been partly presented by basis B = [b1, ..., bn] R × is defined as in a conference paper [29]. Compared with [29], this work ∈ contains the following new contributions: Λ(B)= v v = c b ; c Z . The performance bounds on small dimensional bases | i i i ∈ • i [n] are rigorously analyzed (Theorems 2 and 3). Unlike the X∈ results in [29] that rely on an assumption about covering The dual lattice of Λ is defined as Λ† = n radius, these bounds hold for all input bases. u R u, v Z, v Λ . One basis of Λ† is given by Comparisons with other types of strong&weak reduction { ∈ | h i ∈ ∀ ∈ } • B−⊤. are made (Section III-B, Section IV-D), including η- The Gram-Schmidt (GS) basis of B, referred to as B∗, is Greedy reduction, KZ and its variants, Minkowski reduc- i 1 found by using bi∗ = π b∗,...,b∗ (bi)= bi j−=1 µi,j bj∗, 1 i−1 − tion, LLL and its variants, and Seysen reduction. { 2 } where µi,j = bi, b∗ / b∗ . A Hash-based SR algorithm is constructed (Section IV). h j i || j || P • The ith successive minimum of an n dimensional lattice More specifically, the nearest neighbor search problem is Λ(B) is the smallest real positive number r such that Λ approximately solved with the aid of hashing, and not contains i linearly independent vectors of length at most r: through a brute-force search. The theoretical studies are supported with more simula- λ (B) = inf r dim(span((Λ (0, r))) i , • i tion results (Section V), these include the comparisons { | ∩B ≥ } in which (t, r) denotes a ball centered at t with radius r. with major lattice-reduction-aided MIMO detection al- B gorithms, and the BER performance tested for various The orthogonality defect (OD) can alternatively quantify the channels . goodness of a basis The types of bases feasible for using SR-Hash is dis- n • i=1 bi cussed (Appendix A). We numerically show that the dual η(B)= || || . (1) det(BT B) of large-scale Gaussian random bases have dense pairwise Q| | angles. From Hadamard’s inequality,p we know that η(B) 1. As ≥ It is worth mentioning that SR is emerging as a new building the determinant of a given basis is fixed, the parameter is block in lattice-reduction-aided MIMO detection. Thus, the proportional to the product of the lengths of the basis vectors. proposed SR variants may also benefit list sphere decoding A necessary condition for reaching the smallest orthogonality [30] and Klein’s sampling algorithm [31].
Load more

Recommended publications
  • Arxiv:Cs/0601091V2
    Communication Over MIMO Broadcast Channels Using Lattice-Basis Reduction1 Mahmoud Taherzadeh, Amin Mobasher, and Amir K. Khandani Coding & Signal Transmission Laboratory Department of Electrical & Computer Engineering University of Waterloo Waterloo, Ontario, Canada, N2L 3G1 Abstract A simple scheme for communication over MIMO broadcast channels is introduced which adopts the lattice reduction technique to improve the naive channel inversion method. Lattice basis reduction helps us to reduce the average transmitted energy by modifying the region which includes the constellation points. Simulation results show that the proposed scheme performs well, and as compared to the more complex methods (such as the perturbation method [1]) has a negligible loss. Moreover, the proposed method is extended to the case of different rates for different users. The asymptotic behavior (SNR ) of the symbol error rate of the proposed method and the −→ ∞ perturbation technique, and also the outage probability for the case of fixed-rate users is analyzed. It is shown that the proposed method, based on LLL lattice reduction, achieves the optimum asymptotic slope of symbol-error-rate (called the precoding diversity). Also, the outage probability for the case of fixed sum-rate is analyzed. I. INTRODUCTION In the recent years, communications over multiple-antenna fading channels has attracted the attention of many researchers. Initially, the main interest has been on the point-to-point Multiple-Input Multiple-Output (MIMO) communications [2]–[6]. In [2] and [3], the authors have shown that the capacity of a MIMO point-to-point channel increases linearly with the arXiv:cs/0601091v2 [cs.IT] 18 Jun 2007 minimum number of the transmit and the receive antennas.
    [Show full text]
  • Solving Hard Lattice Problems and the Security of Lattice-Based Cryptosystems
    Solving Hard Lattice Problems and the Security of Lattice-Based Cryptosystems † Thijs Laarhoven∗ Joop van de Pol Benne de Weger∗ September 10, 2012 Abstract This paper is a tutorial introduction to the present state-of-the-art in the field of security of lattice- based cryptosystems. After a short introduction to lattices, we describe the main hard problems in lattice theory that cryptosystems base their security on, and we present the main methods of attacking these hard problems, based on lattice basis reduction. We show how to find shortest vectors in lattices, which can be used to improve basis reduction algorithms. Finally we give a framework for assessing the security of cryptosystems based on these hard problems. 1 Introduction Lattice-based cryptography is a quickly expanding field. The last two decades have seen exciting devel- opments, both in using lattices in cryptanalysis and in building public key cryptosystems on hard lattice problems. In the latter category we could mention the systems of Ajtai and Dwork [AD], Goldreich, Goldwasser and Halevi [GGH2], the NTRU cryptosystem [HPS], and systems built on Small Integer Solu- tions [MR1] problems and Learning With Errors [R1] problems. In the last few years these developments culminated in the advent of the first fully homomorphic encryption scheme by Gentry [Ge]. A major issue in this field is to base key length recommendations on a firm understanding of the hardness of the underlying lattice problems. The general feeling among the experts is that the present understanding is not yet on the same level as the understanding of, e.g., factoring or discrete logarithms.
    [Show full text]
  • Lattice Reduction for Modular Knapsack⋆
    Lattice Reduction for Modular Knapsack? Thomas Plantard, Willy Susilo, and Zhenfei Zhang Centre for Computer and Information Security Research School of Computer Science & Software Engineering (SCSSE) University Of Wollongong, Australia fthomaspl, wsusilo, [email protected] Abstract. In this paper, we present a new methodology to adapt any kind of lattice reduction algorithms to deal with the modular knapsack problem. In general, the modular knapsack problem can be solved using a lattice reduction algorithm, when its density is low. The complexity of lattice reduction algorithms to solve those problems is upper-bounded in the function of the lattice dimension and the maximum norm of the input basis. In the case of a low density modular knapsack-type basis, the weight of maximum norm is mainly from its first column. Therefore, by distributing the weight into multiple columns, we are able to reduce the maximum norm of the input basis. Consequently, the upper bound of the time complexity is reduced. To show the advantage of our methodology, we apply our idea over the floating-point LLL (L2) algorithm. We bring the complexity from O(d3+"β2 + d4+"β) to O(d2+"β2 + d4+"β) for " < 1 for the low den- sity knapsack problem, assuming a uniform distribution, where d is the dimension of the lattice, β is the bit length of the maximum norm of knapsack-type basis. We also provide some techniques when dealing with a principal ideal lattice basis, which can be seen as a special case of a low density modular knapsack-type basis. Keywords: Lattice Theory, Lattice Reduction, Knapsack Problem, LLL, Recursive Reduction, Ideal Lattice.
    [Show full text]
  • Lattice Reduction in Two Dimensions: Analyses Under Realistic Probabilistic Models
    Lattice reduction in two dimensions : analyses under realistic probabilistic models Brigitte Vallée, Antonio Vera To cite this version: Brigitte Vallée, Antonio Vera. Lattice reduction in two dimensions : analyses under realistic proba- bilistic models. 2008. hal-00211495 HAL Id: hal-00211495 https://hal.archives-ouvertes.fr/hal-00211495 Preprint submitted on 21 Jan 2008 HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. Lattice reduction in two dimensions: analyses under realistic probabilistic models Brigitte Vallee´ and Antonio Vera CNRS UMR 6072, GREYC, Universite´ de Caen, F-14032 Caen, France The Gaussian algorithm for lattice reduction in dimension 2 is precisely analysed under a class of realistic probabilistic models, which are of interest when applying the Gauss algorithm “inside” the LLL algorithm. The proofs deal with the underlying dynamical systems and transfer operators. All the main parameters are studied: execution parameters which describe the behaviour of the algorithm itself as well as output parameters, which describe the geometry of reduced bases. Keywords: Lattice Reduction, Gauss’ algorithm, LLL algorithm, Euclid’s algorithm, probabilistic analysis of algo- rithms, Dynamical Systems, Dynamical analysis of Algorithms. 1 Introduction The lattice reduction problem consists in finding a short basis of a lattice of Euclidean space given an initially skew basis.
    [Show full text]
  • Algorithms for the Approximate Common Divisor Problem
    Algorithms for the Approximate Common Divisor Problem Steven D. Galbraith1, Shishay W. Gebregiyorgis1, and Sean Murphy2 1 Mathematics Department, University of Auckland, New Zealand. [email protected],[email protected] 2 Royal Holloway, University of London, UK. [email protected] Abstract. The security of several homomorphic encryption schemes depends on the hardness of the Approximate Common Divisor (ACD) problem. In this paper we review and compare existing algorithms to solve the ACD prob- lem using lattices. In particular we consider the simultaneous Diophantine approximation method, the orthogonal lattice method, and a method based on multivariate polynomials and Coppersmith’s algorithm that was studied in detail by Cohn and Heninger. One of our main goals is to compare the multivariate polynomial approach with other methods. We find that the multivariate polynomial approach is not better than the orthogonal lattice algorithm for practical cryptanalysis. Another contribution is to consider a sample-amplification technique for ACD samples, and to consider a pre- processing algorithm similar to the Blum-Kalai-Wasserman (BKW) algorithm for learning parity with noise. We explain why, unlike in other settings, the BKW algorithm does not give an improvement over the lattice algorithms. This is the full version of a paper published at ANTS-XII in 2016. Keywords: Approximate common divisors, lattice attacks, orthogonal lattice, Coppersmith’s method. 1 Introduction The approximate common divisor problem (ACD) was first studied by Howgrave-Graham [HG01]. Further inter- est in this problem was provided by the homomorphic encryption scheme of van Dijk, Gentry, Halevi and Vaikun- tanathan [DGHV10] and its variants [CMNT11,CNT12,CS15].
    [Show full text]
  • A Hybrid Lattice-Reduction and Meet-In-The-Middle Attack Against NTRU
    A hybrid lattice-reduction and meet-in-the-middle attack against NTRU Nick Howgrave-Graham [email protected] NTRU Cryptosystems, Inc. Abstract. To date the NTRUEncrypt security parameters have been based on the existence of two types of attack: a meet-in-the-middle at- tack due to Odlyzko, and a conservative extrapolation of the running times of the best (known) lattice reduction schemes to recover the private key. We show that there is in fact a continuum of more efficient attacks between these two attacks. We show that by combining lattice reduction and a meet-in-the-middle strategy one can reduce the number of loops in attacking the NTRUEncrypt private key from 284.2 to 260.3, for the k = 80 parameter set. In practice the attack is still expensive (depen- dent on ones choice of cost-metric), although there are certain space/time tradeoffs that can be applied. Asymptotically our attack remains expo- nential in the security parameter k, but it dictates that NTRUEncrypt parameters must be chosen so that the meet-in-the-middle attack has complexity 2k even after an initial lattice basis reduction of complexity 2k. 1 Introduction It is well known that the closest vector problem (CVP) can be solved efficiently in the case that the given point in space is very close to a lattice vector [7, 18]. If this CVP algorithm takes time t and a set S has the property that it includes at least one point v0 S which is very close to a lattice vector, then clearly v0 can be found in time O∈( S t) by exhaustively enumerating the set S.
    [Show full text]
  • A Lattice Basis Reduction Algorithm ∗
    A Lattice Basis Reduction Algorithm ∗ Franklin T. Luk Department of Mathematics, Hong Kong Baptist University Kowloon Tong, Hong Kong Sanzheng Qiao Department of Computing and Software, McMaster University Hamilton, Ontario L8S4K1 Canada Wen Zhang Institute of Mathematics, School of Mathematical Science Fudan University, Shanghai, 200433 P.R. China Abstract In this paper, we give a definition of an optimally reduced basis for a lattice in the sense that an optimally reduced basis is a shortest basis for the lattice. Then we present an algorithm for computing an approximation of an optimally reduced basis for a lattice using a novel unimodular transformation. To compare lattice bases, we propose a quantitative measure of the degree of the linear independence of lattice basis vectors. Keywords Lattice, lattice basis reduction, unimodular transformation, linear independence. 1 Introduction A lattice is a set of discrete points representing integer linear combinations of linearly indepen- dent vectors. The set of linearly independent vectors generating a lattice is called a basis for the lattice. A set of lattice points does not uniquely determine a basis. This leads to the problem of finding a “nearly” orthogonal basis. Intuitively, shorter basis vectors are “more” orthogonal. A basis reduction algorithm finds a reduced basis, that is, a basis whose vectors are reduced in length. The lattice reduction problem arises from fields such as integer programming [2], cryptology [6], number theory [4], and information theory [1]. In this paper, after a short intro- duction to lattices and bases in Section 2, various definitions of reduced basis are described in Section 3. They include the definitions of Minkowski-reduced basis, Hermite size-reduced basis, HKZ-reduced basis, and LLL-reduced basis.
    [Show full text]
  • Lattice Reduction in Two Dimensions: Analyses Under Realistic Probabilistic Models
    LATTICE REDUCTION IN TWO DIMENSIONS: ANALYSES UNDER REALISTIC PROBABILISTIC MODELS BRIGITTE VALLEE´ AND ANTONIO VERA Abstract. The Gaussian algorithm for lattice reduction in dimension 2 is precisely analysed under a class of realistic probabilistic models, which are of interest when applying the Gauss algorithm “inside” the LLL algorithm. The proofs deal with the underlying dynamical systems and transfer operators. All the main parameters are studied: execution parameters which describe the behaviour of the algorithm itself as well as output parameters, which describe the geometry of reduced bases. 1. Introduction The lattice reduction problem consists in finding a short basis of a lattice of Euclidean space given an initially skew basis. This reduction problem plays a pri- mary rˆolein many areas of computational mathematics and computer science: for instance, modern cryptanalysis [14], computer algebra [18] integer linear program- ming [11] and number theory [5]. In the two-dimensional case, there exists an algorithm due to Lagrange and Gauss which computes in linear time a minimal basis of a lattice. This algorithm is in a sense optimal, from both points of view of the time-complexity and the quality of the output. It can be viewed as a generalization of the Euclidean Algorithm to the two dimensional-case. For n ≥ 3, the LLL algorithm [10] due to Lenstra, Lenstra and Lov`asz, computes a reduced basis of an n-dimensional lattice in polynomial time. However, the notion of reduction is weaker than in the case n = 2, and the exact complexity of the algorithm (even in the worst-case, and for small dimensions) is not precisely known.
    [Show full text]
  • Approximate Algorithms on Lattices with Small Determinant (Extended Abstract)
    Approximate Algorithms on Lattices with Small Determinant (Extended Abstract) Jung Hee Cheon and Changmin Lee Seoul National University, Korea fjhcheon,[email protected] Abstract. In this paper, we propose approximate lattice algorithms for solving the shortest vector problem (SVP) and the closest vector problem (CVP) on an n-dimensional Euclidean integral lattice L. Our algorithms run in polynomial time of the dimension and determinant of n2=4 lattices and improve on the LLL algorithm when the determinant of a lattice is lessp than 2 . More precisely, our approximate SVP algorithm gives a lattice vector of size ≤ 2 log det L and ourp approximate CVP algorithm gives a lattice vector, the distance of which to a target vector is 2 log det L times the distance from the target vector to the lattice. One interesting feature of our algorithms is that their output sizes are independent of dimensionp and become smaller as the determinant of L becomes smaller. For example, if det L = 2n n, a short vector outputted from n3=4 our approximatep SVP algorithm is of size 2 , which is asymptotically smaller than the size 2n=4+ n of the outputted short vectors of the LLL algorithm. It is similar to our approximate CVP algorithm. Keywords: LLL algorithm, SVP, CVP, Hermite normal form, Babai's nearest plane algorithm. 1 Introduction Lattices and the LLL algorithm. A lattice is a set of all the integral linear combinations of some linearly independent vectors in Rn. The Shortest Vector Problem (SVP) of a lattice is to find a shortest nonzero vector in the Euclidean norm given a basis of the lattice.
    [Show full text]
  • Fast Lattice Basis Reduction Suitable for Massive Parallelization and Its Application to the Shortest Vector Problem
    Fast Lattice Basis Reduction Suitable for Massive Parallelization and Its Application to the Shortest Vector Problem Tadanori Teruya1, Kenji Kashiwabara21, and Goichiro Hanaoka1 1 Information Technology Research Institute, National Institute of Advanced Industrial Science and Technology. AIST Tokyo Waterfront Bio-IT Research Building, 2-4-7 Aomi, Koto-ku, Tokyo, Japan. 2 Department of General Systems Studies, University of Tokyo. 3-8-1 Komaba, Meguro-ku, Tokyo, Japan. Abstract. The hardness of the shortest vector problem for lattices is a fundamen- tal assumption underpinning the security of many lattice-based cryptosystems, and therefore, it is important to evaluate its difficulty. Here, recent advances in studying the hardness of problems in large-scale lattice computing have pointed to need to study the design and methodology for exploiting the performance of massive parallel computing environments. In this paper, we propose a lattice ba- sis reduction algorithm suitable for massive parallelization. Our parallelization strategy is an extension of the Fukase–Kashiwabara algorithm (J. Information Processing, Vol. 23, No. 1, 2015). In our algorithm, given a lattice basis as input, variants of the lattice basis are generated, and then each process reduces its lattice basis; at this time, the processes cooperate and share auxiliary information with each other to accelerate lattice basis reduction. In addition, we propose a new strategy based on our evaluation function of a lattice basis in order to decrease the sum of squared lengths of orthogonal basis vectors. We applied our algorithm to problem instances from the SVP Challenge. We solved a 150-dimension problem instance in about 394 days by using large clusters, and we also solved problem instances of dimensions 134, 138, 140, 142, 144, 146, and 148.
    [Show full text]
  • Progress on LLL and Lattice Reduction
    Progress on LLL and Lattice Reduction Claus Peter Schnorr Abstract We surview variants and extensions of the LLL-algorithm of Lenstra, Lenstra Lovasz´ , extensions to quadratic indefinite forms and to faster and stronger reduction algorithms. The LLL-algorithm with Householder orthogonalisation in floating-point arithmetic is very efficient and highly accurate. We surview approx- imations of the shortest lattice vector by feasible lattice reduction, in particular by block reduction, primal-dual reduction and random sampling reduction. Segment re- duction performs LLL-reduction in high dimension mostly working with a few local coordinates. Key words: LLL-reduction, Householder orthogonalisation, floating-point arith- metic, block reduction, segment reduction, primal-dual reduction, sampling reduction, reduction of indefinite quadratic forms. 1 Introduction A lattice basis of dimension n consists of n linearly independent real vectors b1, ..., bn m ∈ R . The basis B = [b1, ..., bn] generates the lattice L consisting of all integer linear combinations of the basis vectors. Lattice reduction transforms a given basis of L into a basis with short and nearly orthogonal vectors. Unifying two traditions. The LLL-algorithm of Lenstra, Lenstra Lovasz´ [47] provides in polynomial time a reduced basis of proven quality. Its inventors focused on Gram-Schmidt orthogonalisation (GSO) in exact integer arithmetic which is only affordable for small dimension n. With floating-point arithmetic (fpa) available it is faster to orthogonalise in fpa. In numerical mathematics the orthogonalisation of Claus Peter Schnorr Fachbereich Informatik und Mathematik, Universit¨at Frankfurt, PSF 111932, D-60054 Frankfurt am Main, Germany. [email protected] – http: www.mi.informatik.uni-frankfurt.de 1 a lattice basis B is usually done by QR-factorization B = QR using Householder reflections [76].
    [Show full text]
  • Lattice Reduction: a Toolbox for the Cryptanalyst
    J. Cryptology (1998) 11: 161–185 © 1998 International Association for Cryptologic Research Lattice Reduction: A Toolbox for the Cryptanalyst Antoine Joux DGA=CELAR, Bruz, France Jacques Stern Laboratoire d’Informatique, Ecole Normale Sup´erieure, Paris, France Communicated by Andrew M. Odlyzko Received 19 May 1994 and revised 31 December 1997 Abstract. In recent years, methods based on lattice reduction have been used repeat- edly for the cryptanalytic attack of various systems. Even if they do not rest on highly sophisticated theories, these methods may look a bit intricate to practically oriented cryptographers, both from the mathematical and the algorithmic point of view. The aim of this paper is to explain what can be achieved by lattice reduction algorithms, even without understanding the actual mechanisms involved. Two examples are given. One is the attack devised by the second author against Knuth’s truncated linear congruential generator. This attack was announced a few years ago and appears here for the first time in complete detail. Key words. Lattices, Cryptanalysis, Knapsack cryptosystems. 1. Introduction 1.1. Historical Background A lattice is a discrete subgroup of Rn or equivalently the set L consisting of integral linear combinations ¸1b1 ¸pbp C¢¢¢C of a given set of independent n-dimensional vectors b1;:::;bp. The sequence .b1;:::;bp/ is said to be a basis of L and p is its dimension. From the mathematical point of view, the history of lattice reduction goes back to the theory of quadratic forms developed by Lagrange, Gauss, Hermite, Korkine and Zolotareff, and others (see [La], [G], [H], and [KZ]) and to Minkowski’s geometry of numbers [M].
    [Show full text]