Cryptanalysis of Variants of RSA with Multiple Small Secret Exponents

In this paper, we analyze the security of two variants of the RSA public key cryptosystem where multiple encryption and decryption exponents are used with a common modulus. For the most well known variant, CRT-RSA, assume that n encryption and decryption exponents (el,dpl,dql) $$(e_l,d_{p_l},d_{q_l}...

Full description

Saved in:
Bibliographic Details
Published inProgress in Cryptology -- INDOCRYPT 2015 pp. 105 - 123
Main Authors Peng, Liqiang, Hu, Lei, Lu, Yao, Sarkar, Santanu, Xu, Jun, Huang, Zhangjie
Format Book Chapter
LanguageEnglish
Published Cham Springer International Publishing 2015
SeriesLecture Notes in Computer Science
Subjects
Online AccessGet full text
ISBN3319266160
9783319266169
ISSN0302-9743
1611-3349
DOI10.1007/978-3-319-26617-6_6

Cover

More Information
Summary:In this paper, we analyze the security of two variants of the RSA public key cryptosystem where multiple encryption and decryption exponents are used with a common modulus. For the most well known variant, CRT-RSA, assume that n encryption and decryption exponents (el,dpl,dql) $$(e_l,d_{p_l},d_{q_l})$$ , where l=1,⋯,n $$l=1,\cdots ,n$$ , are used with a common CRT-RSA modulus N. By utilizing a Minkowski sum based lattice construction and combining several modular equations which share a common variable, we prove that one can factor N when dpl,dql<N2n-38n+2 $$d_{p_l},d_{q_l}<N^{\frac{2n-3}{8n+2}}$$ for all l=1,⋯,n $$l=1,\cdots ,n$$ . We further improve this bound to dpl(ordql)<N9n-1424n+8 $$d_{p_l}(\mathrm {or}\,d_{q_l})<N^{\frac{9n-14}{24n+8}}$$ for all l=1,⋯,n $$l=1,\cdots ,n$$ . Moreover, our experiments do better than previous works by Jochemsz-May (Crypto 2007) and Herrmann-May (PKC 2010) when multiple exponents are used. For Takagi’s variant of RSA, assume that n key pairs (el,dl) $$(e_l,d_l)$$ for l=1,⋯,n $$l=1,\cdots ,n$$ are available for a common modulus N=prq $$N=p^rq$$ where r≥2 $$r\ge 2$$ . By solving several simultaneous modular univariate linear equations, we show that when dl<N(r-1r+1)n+1n $$d_l<N^{(\frac{r-1}{r+1})^{\frac{n+1}{n}}}$$ , for all l=1,⋯,n $$l=1,\cdots ,n$$ , one can factor the common modulus N.
Bibliography:Original Abstract: In this paper, we analyze the security of two variants of the RSA public key cryptosystem where multiple encryption and decryption exponents are used with a common modulus. For the most well known variant, CRT-RSA, assume that n encryption and decryption exponents (el,dpl,dql)\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$(e_l,d_{p_l},d_{q_l})$$\end{document}, where l=1,⋯,n\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$l=1,\cdots ,n$$\end{document}, are used with a common CRT-RSA modulus N. By utilizing a Minkowski sum based lattice construction and combining several modular equations which share a common variable, we prove that one can factor N when dpl,dql<N2n-38n+2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$d_{p_l},d_{q_l}<N^{\frac{2n-3}{8n+2}}$$\end{document} for all l=1,⋯,n\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$l=1,\cdots ,n$$\end{document}. We further improve this bound to dpl(ordql)<N9n-1424n+8\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$d_{p_l}(\mathrm {or}\,d_{q_l})<N^{\frac{9n-14}{24n+8}}$$\end{document} for all l=1,⋯,n\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$l=1,\cdots ,n$$\end{document}. Moreover, our experiments do better than previous works by Jochemsz-May (Crypto 2007) and Herrmann-May (PKC 2010) when multiple exponents are used. For Takagi’s variant of RSA, assume that n key pairs (el,dl)\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$(e_l,d_l)$$\end{document} for l=1,⋯,n\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$l=1,\cdots ,n$$\end{document} are available for a common modulus N=prq\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$N=p^rq$$\end{document} where r≥2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$r\ge 2$$\end{document}. By solving several simultaneous modular univariate linear equations, we show that when dl<N(r-1r+1)n+1n\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$d_l<N^{(\frac{r-1}{r+1})^{\frac{n+1}{n}}}$$\end{document}, for all l=1,⋯,n\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$l=1,\cdots ,n$$\end{document}, one can factor the common modulus N.
ISBN:3319266160
9783319266169
ISSN:0302-9743
1611-3349
DOI:10.1007/978-3-319-26617-6_6