Correlation Of Binary Sequences Research Articles

On the correlation of binary sequences, II

This paper concerns the study of the correlation measures of finite binary sequences, more particularly the dependence of correlation measures of even order and correlation measures of odd order. These results generalize previous results due to Gyarmati (2005) [7] and to Anantharam (2008) [3] and provide a partial answer to a conjecture due to Mauduit (2003) [12]. The last part of the paper concerns the generalization of this study to the case of finite binary n -dimensional lattices.

On the correlation of binary sequences

In a series of papers Mauduit and Sárközy (partly with further coauthors) studied finite pseudorandom binary sequences. In particular, one of the most important applications of pseudorandomness is cryptography. If, e.g., we want to use a binary sequence E N ∈ { - 1 , + 1 } N (after transforming it into a bit sequence) as a key stream in the standard Vernam cipher [A. Menezes, P. van Oorschot, R. Vanstone, Handbook of Applied Cryptography, CRC Press, Boca Raton, 1997], then E N must possess certain pseudorandom properties. Does E N need to possess both small well-distribution measure and, for any fixed small k, small correlation measure of order k? In other words, if W ( E N ) is large, resp. C k ( E N ) is large for some fixed small k, then can the enemy utilize this fact to break the code? The most natural line of attack is the exhaustive search: the attacker may try all the binary sequences E N ∈ { - 1 , + 1 } N with large W ( E N ) , resp. large C k ( E N ) , as a potential key stream. Clearly, this attack is really threatening only if the number of sequences E N ∈ { - 1 , + 1 } N with (i) large W ( E N ) , resp. (ii) large C k ( E N ) is “much less” than the total number 2 N of sequences in { - 1 , + 1 } N , besides one needs a fast algorithm to generate the sequences of type (i), resp. (ii). The case (i) is easy, thus, for the sake of completeness, here we just present an estimate for the number of sequences E N with large W ( E N ) . The case (ii), i.e., the case of large correlation is much more interesting: this case will be studied in Section 2. In Section 3 we will sharpen the results of Section 2 in the special case when the order of the correlation is 2. Finally, in Section 4 we will study a lemma, which plays a crucial role in the estimation of the correlation in some of the most important constructions of pseudorandom binary sequences.

On the correlation of binary sequences

In a series of papers Mauduit and Sarkozy (partly with further coauthors) studied finite pseudorandom binary sequences. In particular, one of the most important applications of pseudorandomness is cryptography. If, e.g., we want to use a binary sequence E[email protected]?{-1,+1}^N (after transforming it into a bit sequence) as a key stream in the standard Vernam cipher [A. Menezes, P. van Oorschot, R. Vanstone, Handbook of Applied Cryptography, CRC Press, Boca Raton, 1997], then EN must possess certain pseudorandom properties. Does EN need to possess both small well-distribution measure and, for any fixed small k, small correlation measure of order k? In other words, if W(EN) is large, resp. Ck(EN) is large for some fixed small k, then can the enemy utilize this fact to break the code? The most natural line of attack is the exhaustive search: the attacker may try all the binary sequences E[email protected]?{-1,+1}^N with large W(EN), resp. large Ck(EN), as a potential key stream. Clearly, this attack is really threatening only if the number of sequences E[email protected]?{-1,+1}^N with(i)large W(EN), resp. (ii)large Ck(EN) is ''much less'' than the total number 2^N of sequences in {-1,+1}^N, besides one needs a fast algorithm to generate the sequences of type (i), resp. (ii). The case (i) is easy, thus, for the sake of completeness, here we just present an estimate for the number of sequences EN with large W(EN). The case (ii), i.e., the case of large correlation is much more interesting: this case will be studied in Section 2. In Section 3 we will sharpen the results of Section 2 in the special case when the order of the correlation is 2. Finally, in Section 4 we will study a lemma, which plays a crucial role in the estimation of the correlation in some of the most important constructions of pseudorandom binary sequences.

On the correlation of binary sequences

C. Mauduit conjectured that C2(E)CN3(EN) ?? Nc always holds with some constant 1/2< c = 1. This will be proved for c=2/3, more exactly if for a sequence EN =C {-1,+1}N we have C2 (EN) << N2/3 then C3 (EN) ?? N1/2. Indeed, a more general theorem is proved, involving correlation measures.