Abstract
Recently, primitive sequences over \(\mathbf{Z}/(2^{32}-1)\) are shown to have many desirable properties, which makes them of potential interest for cryptographic applications. To further support the applications of this kind of sequences, in this paper, we consider the problem whether primitive sequences generated by two distinct primitive polynomials over \(\mathbf{Z} /(2^{32}-1)\) are pairwise distinct modulo 2. A sufficient condition is given for ensuring that the answer to this problem is positive.
Similar content being viewed by others
References
ETSI/SAGE Specification: Specification of the 3GPP confidentiality and integrity algorithms 128-EEA3 & 128-EIA3. Document 4: Design and Evaluation Report, Version: 2.0. Technicl Report, ETSI 2011. http://www.gsmworld.com/our-work/programmes-and-initiatives/fraud-and-security/gsm_security_algorithms.htm (2011). Accessed 9 Sept 2011.
Zhu X.Y., Qi W.F.: On the distinctness of modular reduction of maximal length modulo odd prime numbers. Math. Comput. 77(7), 1623–1637 (2008).
Zheng Q.X., Qi W.F., Tian T.: On the distinctness of modular reduction of primitive sequences over \({ Z}/(2^{32}-1)\). Des. Codes Cryptogr. 112(22), 872–875 (2012). doi:10.1007/s10623-012-9698-y.
Lidl R., Niedereiter H.: Finite Field. Addison-Wesley, Don Mills (1983).
Hall M.: An isomorphism between linear recurring sequences and algebraic rings. Trans. Am. Math. Soc. 44, 196–218 (1938).
Rueppel R.A.: Analysis and Design of Stream Ciphers. Springer, New York (1986).
Tian T., Qi W.F.: Typical primitive polynomials over integer residue rings. Finite Fields Appl. 15, 796–807 (2009).
Wang Y.K.: Residue-to-binary converters based on Chinese remainder theorems. IEEE Trans. Circut. Syst. II 40(3), 197–205 (2000).
Bugeaud Y., Corvaja P., Zannier U.: An upper bound for the G.C.D. of \(a^{n}-1\) and \(b^{n}-1\). Math. Z. 243, 79–84 (2003).
Acknowledgments
This research is supported by NSF of China under Grant No. (61272042, 61070178, 61100202).
Author information
Authors and Affiliations
Corresponding author
Additional information
Communicated by D. Panario.
Appendices
Appendix 1: Proof of Proposition 1
In this section, we always assume that \(p_{i}=2^{2^{i-1}}+1\) for \(1\le i\le 5\). Then it is clear that \(2^{32}-1=p_{1}p_{2}p_{3}p_{4}p_{5}\) is the canonical factorization of \(2^{32}-1\). To prove Proposition 1, we first give some necessary lemmas.
Lemma 6
For \(2\le j\le 5\) we have that
Proof
The result can be checked directly. \(\square \)
For any integer \(m\ge 2\), let \(v_{2}\left( m\right) \) denote the greatest nonnegative integer k such that \(2^{k}\) divides m. Then it follows from [7, Lemma 10] that
and so
Lemma 7
Let \(\xi _{f}\) be the associated element of a primitive polynomial \(f\left( x\right) \) of degree \(n\ge 2\) over \(\mathbf{Z} /(2^{32}-1)\). Then
where \(ord(\xi _{f},p_{i})\) is the multiplicative order of \(\xi _{f}\) mod \( p_{i}\).
Proof
It follows from 3 that
where \([(-1)^{n}f(0)]_{\,\mathrm{mod}\,{p_{i}}}\) is a primitive element of \(\mathbf{Z }/(p_{i})\) and \(\theta =\mathrm{lcm}\left( \frac{p_{1}^{n}-1}{p_{1}-1},\ldots , \frac{p_{5}^{n}-1}{p_{5}-1}\right) \). Then we get that
where
Applying 17 and 18 to 20 we obtain that
\(\square \)
Lemma 8
Let \(p\in \{5,17,257,65537\}\) and \(\xi \in \mathbf{Z}/(p)\) with \(ord(\xi ,p)=(p-1)/2\). If \(\underline{a}\in G(x-\xi ,p)\) with \( \underline{a}\ne \underline{0}\), then either 1 or 3 occurs in \( \underline{a}\).
Proof
Set \(\underline{b}=\left( [\xi ^{t}]_{\,\mathrm{mod}\,{p}}\right) _{t\ge 0}\). Then it is clear that \(\underline{b}\in G(x-\xi ,p)\) with \(per(\underline{b} )=(p-1)/2\). Since \(L^{k}\underline{b}\in G(x-\xi ,p)\) for any integer \(k\ge 0\), where \(L^{k}\underline{b}=(b(k+t))_{t\ge 0}\) is the \(k\)-shift of \( \underline{b}\), it implies that \(\underline{b},L\underline{b},\ldots ,L^{ \frac{p-1}{2}-1}\underline{b}\) are \(\frac{p-1}{2}\) distinct sequences in the set \(G(x-\xi ,p)\).
Note that three is a primitive element of \(\mathbf{Z}/(p)\), and so it does not occur in \(L^{k}\underline{b}\) for any integer \(0\le k\le \frac{p-1}{2}-1\) (since otherwise there exists an integer \(t^{*}\ge 0\) such that \([ \xi ^{t^{*}}] _{\,\mathrm{mod}\,{p}}=3\), which implies that \(\xi \) is a primitive element of \(\mathbf{Z}/(p)\), a contradiction to the assumption that \(ord(\xi ,p)=(p-1)/2\)). Set \(\underline{c}=\left( [3\cdot \xi ^{t}]_{ \,\mathrm{mod}\,{p}}\right) _{t\ge 0}\). Then \(\underline{c}\in G(x-\xi ,p)\) with \( per(\underline{c})=(p-1)/2\). Note that \(c\left( 0\right) =3\), and so we have that
Similar to the analyses above, \(\underline{c},L\underline{c},\ldots ,L^{ \frac{p-1}{2}-1}\underline{c}\) are \(\frac{p-1}{2}\) distinct sequences in the set \(G(x-\xi ,p)\backslash \{\underline{b},L\underline{b},\ldots ,L^{\frac{p-1}{2}-1}\underline{b}\}\).
Since there are only p distinct sequences in the set \(G(x-\xi ,p)\), we get that
Therefore either \(\underline{a}=L^{k}\underline{b}\) or \(\underline{a}=L^{k} \underline{c}\) for some \(0\le k\le \frac{p-1}{2}-1\). Note that \(b(0)=1\) and \(c(0)=3\), and so either 1 or 3 occurs in \(\underline{a}\). \(\square \)
Lemma 9
Let \(\xi _{f}\) be the associated element of a primitive polynomial \(f(x)\) of degree \(n\ge 2\) over \(\mathbf{Z}/(2^{32}-1)\) and \( \underline{a}\in G(x-\xi _{f},2^{32}-1)\). If \(\left[ \underline{a}\right] _{ \,\mathrm{mod}\,{p_{j}}}\ne \underline{0}\) for some \(j\in \{1,2,3,4,5\}\), then
where \(T=ord\left( \xi _{f},p_{j}\right) \). Moreover, if T is divisible by four, then
Proof
Note that \(ord(\xi _{f}^{T/2},p_{j})=2\) and \(ord(\xi _{f}^{T/4},p_{j})=4\) if \(4|T\), and so we get that
Then the result immediately follows by applying 22 to \(\underline{a} \). \(\square \)
Lemma 10
([8, Proposition 8]) Let p and q be two positive integers with \(\gcd (p,q)=1\). Then
is the unique integer in the half-open interval \([0,pq)\) such that
where \(p^{-1}\) is the multiplicative inverse of \(p\,\mathrm{mod}\,{q}\).
With the preparations above, now we are ready to prove Proposition 1.
Proof of Proposition 1
Let \(p_{j}\) be the largest prime divisor of \(2^{32}-1\) such that \([ \underline{a}]_{\,\mathrm{mod}\,{p_{j}}}\ne \underline{0}\). Put \( k=(2^{32}-1)/p_{1}\cdots p_{j}\). Then it is clear that
where \(\underline{c}\in G(x-[\xi _{f}]_{\,\mathrm{mod}\,{p_{1}\cdots p_{j}}},\,p_{1}\cdots p_{j})\) with \([\underline{c}]_{\,\mathrm{mod}\,{p_{j}}}\ne \underline{0}\). Note that \(k\) is odd, and so to prove the proposition, it suffices to show that there is an integer \(t^{*}\ge 0\) such that \( c(t^{*})\) is an odd integer. Since the result is obviously true if \(j=1\) (note that \([\xi _{f}]_{\,\mathrm{mod}\,{p_{1}}}\) is a primitive element of \(\mathbf{Z}/(p_{1})\), and so \(\underline{c}\) is a primitive sequence over \(\mathbf{Z} /(p_{1})\)), we only prove for the case \(j>1\).
Put \(q=p_{1}\cdots p_{j-1}\). Then by Lemma 10 together with 16, \(\underline{c}\) can be rewritten as
where
Denote \(T_{u}=per(\underline{u})\) and \(T_{v}=per(\underline{v})\). Note that \( per(\underline{u})=ord(\xi _{f},p_{j})\) and \(per(\underline{v})|ord(\xi _{f},p_{j-1})\), and so it follows from Lemma 7 that
and
It can be seen that
Case 1: \(n\) is odd.
Since, in this case, \(T_{u}=2^{2^{j-1}}=p_{j}-1\), it implies that \( \underline{u}\) is a primitive sequence of order 1 over \(\mathbf{Z}/(p_{j})\), and so 1 occurs in \(\underline{u}\). We assume without loss of generality that \(u(0)=1\). Then by Lemma 9 we have \(u(T_{u}/2)=p_{j}-1\). Therefore 23 yields that
Note that 24 and 15, respectively, imply that
Thus by applying 27 to 26 we get that
Then it can be seen from 25 and 28 that either \(c(0)\) or \(c(T_{u}/2)\) is an odd integer.
Case 2: \(n\) is even.
Case 2.1: \(n\) is even and \(j\ge 3\).
We note that in this case \(T_{v}\mid \frac{T_{u}}{4}\) and \( T_{u}=2^{2^{j-1}-1}=(p_{j}-1)/2\). If 1 occurs in \(\underline{u}\), then similar to Case 1, either \(c(0)\) or \(c(T_{u}/2)\) is an odd integer. If one does not occurs in \(\underline{u}\), then it follows from Lemma 8 that three must occur in \(\underline{u}\). Assume without loss of generality that \(u(0)=3\). Similar to the analyses in Case 1, we can get that
and so
If \([\frac{q+1}{2}\cdot v(0)]_{\,\mathrm{mod}\,{q}}\notin \{\frac{q-1}{2}, \frac{q+1}{2}\}\), then
In either case we have
which implies that either \(c(0)\) or \(c(T_{u}/2)\) is an odd integer.
If \([\frac{q+1}{2}\cdot v(0)]_{\,\mathrm{mod}\,{q}}=\frac{q+1}{2}\), then it is clear that \(v(0)=1\), and so
which implies that \(c(0)\) is an odd integer.
If \([\frac{q+1}{2}\cdot v(0)]_{\,\mathrm{mod}\,{q}}=\frac{q-1}{2}\), then it follows that \(v(0)=q-1\), and so by 24 we get that
Since \(u(T_{u}/4)=[\pm 2^{2^{j-2}}\cdot u(0)]_{\,\mathrm{mod}\,{p_{j}}}=[\pm 2^{2^{j-2}}\cdot 3]_{\,\mathrm{mod}\,{p_{j}}}\) by Lemma 9, it follows 23 that
It can be verified that \(c(T_{u}/4)\) is always an odd integer for any \(j\in \left\{ 3,4,5\right\} \).
Case 2.2: \(n\) is even and \(j=2\).
In this case, it is clear that \(p_{j}=5\), \(q=3\) and \(\underline{c}\in G(x-\left[ \xi _{f}\right] _{\,\mathrm{mod}\,{15}},15)\). Since by Lemma 7 we have that
it implies that \(\left[ \xi _{f}\right] _{\,\mathrm{mod}\,{3}}=2\) and \(\left[ \xi _{f}\right] _{\,\mathrm{mod}\,{5}}=4\), and so \(\left[ \xi _{f}\right] _{\,\mathrm{mod}\,{ 15}}=14\). It can be seen that \(c(1)=[14\cdot c(0)]_{\,\mathrm{mod}\,{15}}=15-c(0)\) and \(c(0)\ne 0\), thus either \(c(0)\) or \(c(1)\) is an odd integer. \(\square \)
Appendix 2
As a preparation, we first introduce a result of Bugeaud etc.
Lemma 11
([9, Theorem 1]) Let p and q be two distinct odd prime integers. Then for any given real number \(\varepsilon >0\), there exists a positive integer \(N_{\varepsilon }\) such that
Now we can make the result explicit in the following statement.
Lemma 12
There exists a positive integer \(N^{*}\) such that inequalities 5 always hold for all \(n>N^{*}\).
Proof
Given a real number \(\varepsilon >0\), for any \(1\le u<v\le 5\), it follows from Lemma 11 that there exists positive integers \( N_{\varepsilon }^{(u,v)}\) such that
Set \(N_{\varepsilon }=\max \{N_{\varepsilon }^{(u,v)}\mid 1\le u<v\le 5\}\). Then by 30 we have that
and so
Choose \(\varepsilon >0\) such that
then by 31 and 32 we obtain that
for \(1\le i\le 4\) and \(n>N_{\varepsilon }\). Thus \(N^{*}=N_{\varepsilon }\) is the desired integer. This completes the proof. \(\square \)
Rights and permissions
About this article
Cite this article
Yang, D., Qi, WF. & Zheng, QX. Further results on the distinctness of modulo 2 reductions of primitive sequences over \(\mathbf{Z}/(2^{32}-1)\) . Des. Codes Cryptogr. 74, 467–480 (2015). https://doi.org/10.1007/s10623-013-9871-y
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10623-013-9871-y