Algebraic attacks on certain stream ciphers

To encrypt data streams of arbitrary lengths, keystream generators are used in modern cryptography which transform a secret initial value, called the key, into a long sequence of seemingly random bits. Many designs are based on linear feedback shift registers (LFSRs), which can be constructed in such a way that the output stream has optimal statistical and periodical properties and which can be efficiently implemented in hardware. Particularly prominent is a certain class of LFSR-based keystream generators, called (ι,m)-combiners or simply combiners. The maybe most famous example is the E0 keystream generator deployed in the Bluetooth standard for encryption. To evaluate the combiner’s security, cryptographers adopted an adversary model where the design and some parts of the input and output are known. An attack is a method to derive the key using the given knowledge. In the last decades, several kinds of attacks against LFSR-based keystream generators have been developed. In 2002 a new kind of attacks came up, named ”algebraic attacks”. The basic idea is to model the knowledge by a system of equation whose solution is the secret key. For several existing combiners, algebraic attacks represent the fastest theoretical attacks publicly known so far. This thesis discusses algebraic attacks against combiners. After providing the required mathematical fundament and a background on combiners, we describe algebraic attacks and explore the two main steps (generating the system of equations and computing the solution) in detail. The efficiency of algebraic attacks is closely connected to the degree of the equations. Thus, we examine the existence of low-degree equations in several situations and discuss multiple design principles to thwart their existence. Furthermore, we investigate ”fast algebraic attacks”, an extension of algebraic attacks.To encrypt data streams of arbitrary lengths, keystream generators are used in modern cryptography which transform a secret initial value, called the key, into a long sequence of seemingly random bits. Many designs are based on linear feedback shift registers (LFSRs), which can be constructed in such a way that the output stream has optimal statistical and periodical properties and which can be efficiently implemented in hardware. Particularly prominent is a certain class of LFSR-based keystream generators, called (ι,m)-combiners or simply combiners. The maybe most famous example is the E0 keystream generator deployed in the Bluetooth standard for encryption. To evaluate the combiner’s security, cryptographers adopted an adversary model where the design and some parts of the input and output are known. An attack is a method to derive the key using the given knowledge. In the last decades, several kinds of attacks against LFSR-based keystream generators have been developed. In 2002 a new kind of attacks came up, named ”algebraic attacks”. The basic idea is to model the knowledge by a system of equation whose solution is the secret key. For several existing combiners, algebraic attacks represent the fastest theoretical attacks publicly known so far. This thesis discusses algebraic attacks against combiners. After providing the required mathematical fundament and a background on combiners, we describe algebraic attacks and explore the two main steps (generating the system of equations and computing the solution) in detail. The efficiency of algebraic attacks is closely connected to the degree of the equations. Thus, we examine the existence of low-degree equations in several situations and discuss multiple design principles to thwart their existence. Furthermore, we investigate ”fast algebraic attacks”, an extension of algebraic attacks