The RSA cryptosystem, named after its inventors, Rivest, Shamir and Adleman, is the most widely known and widely used public-key cryptosystem in the world today. Compared to other public-key cryptosystems, such as elliptic curve cryptography, RSA requires longer keylengths and is computationally more expensive. In order to address these shortcomings, many variants of RSA have been proposed over the years. While the security of RSA has been well studied since it was proposed in 1977, many of these variants have not. In this thesis, we investigate the security of five of these variants of RSA. In particular, we provide detailed analyses of the best known algebraic attacks (including some new attacks) on instances of RSA with certain special private exponents, multiple instances of RSA sharing a common small private exponent, Multi-prime RSA, Common Prime RSA and Dual RSA.

The RSA cryptosystem, named after its inventors, Rivest, Shamir and Adleman, is the most widely known and widely used public-key cryptosystem in the world today. Compared to other public-key cryptosystems, such as elliptic curve cryptography, RSA requires longer keylengths and is computationally more expensive. In order to address these shortcomings, many variants of RSA have been proposed over the years. While the security of RSA has been well studied since it was proposed in 1977, many of these variants have not. In this thesis, we investigate the security of five of these variants of RSA. In particular, we provide detailed analyses of the best known algebraic attacks (including some new attacks) on instances of RSA with certain special private exponents, multiple instances of RSA sharing a common small private exponent, Multi-prime RSA, Common Prime RSA and Dual RSA.

NTRU is a public key cryptosystem, which was first introduced in 1996. It is a ring-based cryptosystem and its security relies on the complexity of a well-known lattice problem, i.e. shortest vector problem (SVP). There is no efficient algorithm known to solve SVP exactly in arbitrary high dimensional lattices. However, approximate solutions to SVP can be found by lattice reduction algorithms. LLL is the first polynomial time algorithm that finds reasonable short vectors of a lattice. The best known attacks on the NTRU cryptosystem are lattice attacks. In these attacks, the lattice constructed by the public key of the system is used to find the private key. The target vector, which includes private key of the system is one of the short vectors of the NTRU lattice. In this thesis, we study NTRU cryptosystem and lattice attacks on NTRU. Also, we applied an attack to a small dimensional NTRU lattice.

L’aléatoire est un ingrédient clé en cryptographie. Par exemple, les nombres aléatoires sont utilisés pour générer des clés, pour le chiffrement et pour produire des nonces. Ces nombres sont générés par des générateurs pseudo-aléatoires et des fonctions pseudo-aléatoires dont les constructions sont basées sur des problèmes qui sont supposés difficiles. Dans cette thèse, nous étudions certaines mesures de complexité des fonctions pseudo-aléatoires de Naor-Reingold et Dodis-Yampolskiy et étudions la sécurité de certains générateurs pseudo-aléatoires (le générateur linéaire congruentiel et le générateur puissance basés sur les courbes elliptiques) et de certaines signatures à base de couplage basées sur le paradigme d’inversion. Nous montrons que la fonction pseudo-aléatoire de Dodis-Yampolskiy est uniformément distribué et qu’un polynôme multivarié de petit dégré ou de petit poids ne peut pas interpoler les fonctions pseudo-aléatoires de Naor-Reingold et de Dodis-Yampolskiy définies sur un corps fini ou une courbe elliptique. Le contraire serait désastreux car un tel polynôme casserait la sécurité de ces fonctions et des problèmes sur lesquels elles sont basées. Nous montrons aussi que le générateur linéaire congruentiel et le générateur puissance basés sur les courbes elliptiques sont prédictibles si trop de bits sont sortis à chaque itération. Les implémentations pratiques de cryptosystèmes souffrent souvent de fuites critiques d’informations à travers des attaques par canaux cachés. Ceci peut être le cas lors du calcul de l’exponentiation afin de calculer la sortie de la fonction pseudo-aléatoire de Dodis-Yampolskiy et plus généralement le calcul des signatures dans certains schémas de signatures bien connus à base de couplage (signatures de Sakai-Kasahara, Boneh-Boyen et Gentry) basées sur le paradigme d’inversion. Nous présentons des algorithmes (heuristiques) en temps polynomial à base des réseaux qui retrouvent le secret de celui qui signe le message dans ces trois schémas de signatures lorsque plusieurs messages sont signés sous l’hypothèse que des blocs consécutifs de bits des exposants sont connus de l’adversaire. / Randomness is a key ingredient in cryptography. For instance, random numbers are used to generate keys, for encryption and to produce nonces. They are generated by pseudo-random generators and pseudorandom functions whose constructions are based on problems which are assumed to be difficult. In this thesis, we study some complexity measures of the Naor-Reingold and Dodis-Yampolskiy pseudorandom functions and study the security of some pseudo-random generators (the linear congruential generator and the power generator on elliptic curves) and some pairing-based signatures based on exponentinversion framework. We show that the Dodis-Yampolskiy pseudo-random functions is uniformly distributed and that a lowdegree or low-weight multivariate polynomial cannot interpolate the Naor-Reingold and Dodis-Yampolskiy pseudo-random functions over finite fields and over elliptic curves. The contrary would be disastrous since it would break the security of these functions and of problems on which they are based. We also show that the linear congruential generator and the power generator on elliptic curves are insecure if too many bits are output at each iteration. Practical implementations of cryptosystems often suffer from critical information leakage through sidechannels. This can be the case when computing the exponentiation in order to compute the output of the Dodis-Yampolskiy pseudo-random function and more generally in well-known pairing-based signatures (Sakai-Kasahara signatures, Boneh-Boyen signatures and Gentry signatures) based on the exponent-inversion framework. We present lattice based polynomial-time (heuristic) algorithms that recover the signer’s secret in the pairing-based signatures when used to sign several messages under the assumption that blocks of consecutive bits of the exponents are known by the attacker.