Correction to: The number of different true permutation polynomial based interleavers under Zhao and Fan sufficient conditions
- 146 Downloads
1 Correction to: Telecommun Syst (2016) 63:593–623 https://doi.org/10.1007/s11235-016-0144-8
In the original version of this article [1] unfortunately, there are mistakes in some formulas for determining the number of true different cubic, fourth degree, and fifth degree permutation polynomial based interleavers under Zhao and Fan sufficient conditions. The reason for these mistakes was considering all null polynomials. However under these sufficient conditions, the null polynomials have to be restricted in some cases. These particular cases are indicated, and the corresponding formulas are corrected.
In the original version, we have determined the number of true different cubic, fourth degree, and fifth degree permutation polynomial (PP) based interleavers under Zhao and Fan (ZF) sufficient conditions given in [2]. To determine the number of true different PPs of a certain degree n, we took into account the equivalence conditions for PPs considering null polynomials (NPs) of degree up to n. In [1], we have not considered the fact that by adding a NP to a PP fulfilling ZF sufficient conditions does not always result a PP that also fulfills ZF sufficient conditions. This omission generated a mistake in some formulas for the number of true different PPs of degree 3 to 5 and for the lengths leading this number to be equal to zero. In this paper, we will correct these formulas.
2 Null polynomials under Zhao and Fan sufficient conditions
Zhao and Fan sufficient conditions for coefficients \(q_1, q_2, \dots , q_n\) so that \(\pi (x)\) in (1) is a PP
1.a | \(p = 2\) | \(n_{N,2}=1\) | \((q_1 + q_2 + \dots + q_n) \ne 0 \ (\mathrm {mod}\ 2)\) |
1.b | \(n_{N,2}>1\) | \(q_1 \ne 0\), \(q_2+q_4+q_6+ \dots = 0 \ (\mathrm {mod}\ 2)\) and \(q_3+q_5+q_7+ \dots = 0 \ (\mathrm {mod}\ 2)\) | |
2 | \(p > 2\) | \(n_{N,p} \ge 1\) | \(q_1 \ne 0, q_2 = q_3 = \dots = q_n = 0 \ (\mathrm {mod}\ p)\) |
We have to impose additional constraints for \(z_{n=5}(x)\) to be a valid ZF NP only if \(3 \mid N\) and/or \(5 \mid N\) because for every prime \(p > 5\), with \(p \mid N\), the five coefficients of \(z_{n=5}(x)\) will be divisible by p, and thus, they fulfill conditions (4).
If \(n_{N,3} > 1\), we have \(3 \mid \frac{N}{\gcd (k!,N)}, \forall k = 3, 4, 5\) and from (6) it results that 3 divides all coefficients of \(z_k(x)\), thus fulfilling the conditions for prime \(p = 3\) for a valid ZF NP. If \(n_{N,5} > 1\), we have \(5 \mid \frac{N}{\gcd (5!,N)}\) and from (6) it results that 5 divides all coefficients of \(z_k(x)\), thus fulfilling the conditions for prime \(p = 5\) for a valid ZF NP.
From those above it results that we have to impose additional constraints for \(z_{n=5}(x)\) to be a valid ZF NP only if \(n_{N,3} = 1\) and/or \(n_{N,5} = 1\). We analyze these cases below.
As it results from (5), the coefficient of the maximum degree term of a NP, i.e., \(z_n\), is any multiple of \(\frac{N}{\gcd (n!, N)}\). Thus, as it is shown in Sect. 3 from [1], any PP modulo N is equivalent to a PP having the coefficient of the k degree term \(q_k < \frac{N}{\gcd (k!, N)}\), \(\forall k = \overline{2, n}\). However, for a PP under ZF sufficient conditions this condition is not always valid. We detail below these conditions for a NP of degree three, four, and five.
If \(n_{N,3} = 1\), then \(3 \not \mid \frac{N}{\gcd (k!,N)}\), \(\forall k = 3, 4, 5\). In this case for \(z_{n=k}(x)\), \(\forall k = 3, 4, 5\), to be a valid ZF NP, we have to impose the condition that \(3 \mid \tau _k\). Similarly, if \(n_{N,5} = 1\), then \(5 \not \mid \frac{N}{\gcd (5!,N)}\). In this case for \(z_{n=5}(x)\) to be a valid ZF NP we have to impose the condition that \(5 \mid \tau _5\).
Thus, when \(n_{N,3} = 1\) any PP modulo N under ZF sufficient conditions is equivalent to a PP under ZF sufficient conditions having the coefficient of the third degree term \(q_3 < 3 \cdot \frac{N}{\gcd (3!, N)}\).
For similar reasons when \(n_{N,3} = 1\), any PP modulo N under ZF sufficient conditions is equivalent to a PP under ZF sufficient conditions having the coefficient of the fourth degree term \(q_4 < 3 \cdot \frac{N}{\gcd (4!, N)}\).
When \(n_{N,3} = 1\) and \(n_{N,5} = 0\), any PP modulo N under ZF sufficient conditions is equivalent to a PP under ZF sufficient conditions having the coefficient of the fifth degree term \(q_5 < 3 \cdot \frac{N}{\gcd (5!, N)}\).
When \(n_{N,3} = 0\) and \(n_{N,5} = 1\), any PP modulo N under ZF sufficient conditions is equivalent to a PP under ZF sufficient conditions having the coefficient of the fifth degree term \(q_5 < 5 \cdot \frac{N}{\gcd (5!, N)}\).
When \(n_{N,3} = 1\) and \(n_{N,5} = 1\), any PP modulo N under ZF sufficient conditions is equivalent to a PP under ZF sufficient conditions having the coefficient of the fifth degree term \(q_5 < 15 \cdot \frac{N}{\gcd (5!, N)}\).
We have not to impose additional constraints for coefficients \(z_1\) and \(z_2\) of the NP \(z_{n=5}(x)\) because when \(3 \mid \tau _k \cdot \frac{N}{\gcd (k!,N)}\), \(\forall k = 3, 4, 5\) and \(5 \mid \tau _5 \cdot \frac{N}{\gcd (5!,N)}\), then \(p \mid z_1\) and \(p \mid z_2\), for \(p = 3 \text { or } 5\). Thus, \(z_2 = z_1 = 0 \ (\mathrm {mod}\ p)\), for \(p = 3 \text { or } 5\), and the conditions from (4) for primes 3 and 5 are fulfilled.
The analysis from Sect. 4 in [1] regarding the three types of prime factors from Table 1 remains valid. For primes 3 and 5, we can summarize the formula for the number of coefficients as follows.
For \(n=3\) and \(n_{N,3} \ge 1\), there are \(3^{n_{N,3}-1}\) coefficients \(q_{2,j}\) and \(3^{\max \{0; n_{N,3}-2\} }\) coefficients \(q_{3,j}\), out of which one is zero.
For \(n=4\) and \(n_{N,3} \ge 1\), there are \(3^{n_{N,3}-1}\) coefficients \(q_{2,j}\) and \(3^{\max \{0; n_{N,3}-2\} }\) coefficients \(q_{3,j}\) and \(q_{4,j}\), respectively, out of which one is zero.
For \(n=5\), if \(p_j=3\) (i.e., \(n_{N,3} \ge 1\)) there are \(3^{n_{N,3}-1}\) coefficients \(q_{2,j}\) and \(3^{\max \{0; n_{N,3}-2\} }\) coefficients \(q_{3,j}\), \(q_{4,j}\), and \(q_{5,j}\), respectively, out of which one is zero.
If \(p_j=5\) (i.e., \(n_{N,5} \ge 1\)), there are \(5^{n_{N,5}-1}\) coefficients \(q_{2,j}\), \(q_{3,j}\), and \(q_{4,j}\), respectively, and \(5^{\max \{0; n_{N,5}-2\} }\) coefficients \(q_{5,j}\), out of which one is zero.
3 Corrections for formulas in Sects. 4.1, 4.2, and 4.3 from [1] considering null polynomials under Zhao and Fan sufficient conditions
In the following, we correct the formulas in Sects. 4.1, 4.2, and 4.3 from [1] in cases when the power of 3 and/or 5 in the factorization of N is equal to one. As in some cases the formulas become simpler, we restate and proof the complete theorems.
3.1 Corrections for formulas in Sect. 4.1 from [1]
-
\(q_2 < N/2\) and \(q_3 < N/2\), when \(2 \mid N\) and \(3 \not \mid N\), or when \(6 \mid N\) and \(9 \not \mid N\);
-
\(q_3 < N/3\), when \(2 \not \mid N\) and \(9 \mid N\);
-
\(q_2 < N/2\) and \(q_3 < N/6\), when \(18 \mid N\).
Theorem 2.1
Proof
See Appendix A.1. \(\square \)
From (9), we see that the number of ZF CPPs is equal to 0, if N is three or nine times a product of prime numbers greater than three, each of them to the power of 1.
Theorem 2.2
Proof
See Appendix A.2. \(\square \)
From (10), we see that the number of ZF CPPs is equal to 0 if N is 6 or 18 times a product of prime numbers greater than three, each of them to the power of 1.
Theorem 2.3
Proof
See Appendix A.3. \(\square \)
From (11), we see that the number of ZF CPPs is equal to 0, if N is 12 or 36 times a product of prime numbers greater than three, each of them to the power of 1.
The conclusions from the end of Sect. 4.1 in [1] have to be corrected as follows.
The remark that the LTE standard contains 58 of all 188 lengths for which ZF sufficient conditions of CPPs become also necessary remains valid. However, for all lengths for which in their prime factorization there exists the prime 3 at power of 1, but no primes p of the form \((p = 2 \ (\mathrm {mod}\ 3)\) at power of 1, the number of all true different CPPs is equal to the number of true different CPPs under ZF sufficient conditions. There exist 25 such lengths among the LTE lengths (namely 48, 96, 168, 192, 312, 336, 384, 456, 624, 672, 768, 912, 1248, 1344, 1536, 1824, 2496, 2688, 3072, 3648, 4800, 4992, 5376, 5952, and 6144). Thus for 83 lengths from the LTE standard, the number of all true different CPPs is equal to the number of true different CPPs under ZF sufficient conditions. For another 105 lengths, the ratio (in percentage) between the number of true different CPPs under ZF sufficient conditions and the number of all true different CPPs is less than 50%. This is shown in Fig. 1 (the corrected Fig. 1 from [1]).
3.2 Corrections for formulas in Sect. 4.2 from [1]
-
\(q_2 < N/2\), \(q_3 < N/2\) and \(q_4 < N/2\), when \(2 \mid N\), \(3 \not \mid N\) and \(4 \not \mid N\), or when \(4 \not \mid N\), \(6 \mid N\), and \(9 \not \mid N\);
-
\(q_2 < N/2\), \(q_3 < N/2\) and \(q_4 < N/4\), when \(3 \not \mid N\), \(4 \mid N\) and \(8 \not \mid N\), or when \(8 \not \mid N\), \(9 \not \mid N\), and \(12 \mid N\);
-
\(q_2 < N/2\), \(q_3 < N/2\) and \(q_4 < N/8\), when \(3 \not \mid N\) and \(8 \mid N\), or when \(9 \not \mid N\) and \(24 \mid N\);
-
\(q_3 < N/3\) and \(q_4 < N/3\), when \(2 \not \mid N\) and \(9 \mid N\);
-
\(q_2 < N/2\), \(q_3 < N/6\) and \(q_4 < N/6\), when \(4 \not \mid N\) and \(18 \mid N\);
-
\(q_2 < N/2\), \(q_3 < N/6\) and \(q_4 < N/12\), when \(8 \not \mid N\) and \(36 \mid N\);
-
\(q_2 < N/2\), \(q_3 < N/6\) and \(q_4 < N/24\), when \(72 \mid N\).
Theorem 2.4
Proof
See Appendix B.1. \(\square \)
From (13), we see that the number of ZF 4-PPs is equal to 0 if N is three or nine times a product of prime numbers greater than three, each of them to the power of 1.
Theorem 2.5
Proof
See Appendix B.2. \(\square \)
From (14), we see that the number of ZF 4-PPs is equal to 0 if N is 6 or 18 times a product of prime numbers greater than three, each of them to the power of 1.
Theorem 2.6
Proof
See Appendix B.3. \(\square \)
From (15), we see that the number of ZF 4-PPs is equal to 0 if N is 12 or 36 times a product of prime numbers, greater than three, each of them to the power of 1.
Theorem 2.7
Proof
See Appendix B.4. \(\square \)
From (16), we see that the number of ZF 4-PPs is equal to 0 if N is 24 or 72 times a product of prime numbers greater than three, each of them to the power of 1.
The conclusions from the end of Sect. 4.2 have to be corrected as follows.
The ratio (in percentage) between the number of true different 4-PPs under Zhao and Fan sufficient conditions and the number of all true different 4-PPs, for all lengths used in the LTE standard
We compared the total number of true different 4-PPs, obtained with the algorithm from [5], for all lengths from the LTE standard with the number of true different 4-PPs under ZF sufficient conditions. Figure 2 shows the ratio (in percentage) between the number of true different 4-PPs under ZF sufficient conditions and the number of all true different 4-PPs for all the 188 lengths of the LTE standard. We note that for 23 from the 188 lengths, the number of all true different 4-PPs is zero (i.e., there are no true 4-PPs for these lengths), and thus, the percentage is 100%. For 84 from the 188 lengths (from which 23 lengths are those mentioned above), the ZF sufficient conditions are also necessary (i.e., this percentage is 100%). For other 4 lengths, this percentage is 0% (i.e., there are no true 4-PPs under ZF sufficient conditions, but there are true different 4-PPs, for these lengths, namely 56, 168, 280, and 504), for 12 lengths the percentage is equal to 50%, and for the other 88 lengths, the percentage is less than 20%.
3.3 Corrections for formulas in Sect. 4.3 from [1]
-
\(q_2 < N/2\), \(q_3 < N/2\), \(q_4 < N/2\), and \(q_5 < N/2\) when \(2 \mid N\), \(3 \not \mid N\), \(4 \not \mid N\), and \(5 \not \mid N\), or when \(4 \not \mid N\), \(5 \not \mid N\), \(6 \mid N\), and \(9 \not \mid N\), or when \(3 \not \mid N\), \(4 \not \mid N\), \(10 \mid N\), and \(25 \not \mid N\), or when \(4 \not \mid N\), \(9 \not \mid N\), \(25 \not \mid N\), and \(30 \mid N\);
-
\(q_2 < N/2\), \(q_3 < N/2\), \(q_4 < N/4\), and \(q_5 < N/4\) when \(3 \not \mid N\), \(4 \mid N\), \(5 \not \mid N\), and \(8 \not \mid N\), or when \(5 \not \mid N\), \(8 \not \mid N\), \(9 \not \mid N\), and \(12 \mid N\), or when \(3 \not \mid N\), \(8 \not \mid N\), \(20 \mid N\), and \(25 \not \mid N\), or when \(8 \not \mid N\), \(9 \not \mid N\), \(25 \not \mid N\), and \(60 \mid N\);
-
\(q_2 < N/2\), \(q_3 < N/2\), \(q_4 < N/8\), and \(q_5 < N/8\) when \(3 \not \mid N\), \(5 \not \mid N\), and \(8 \mid N\), or when \(5 \not \mid N\), \(9 \not \mid N\), and \(24 \mid N\), or when \(3 \not \mid N\), \(25 \not \mid N\), and \(40 \mid N\), or when \(9 \not \mid N\), \(25 \not \mid N\), and \(120 \mid N\);
-
\(q_3 < N/3\), \(q_4 < N/3\), and \(q_5 < N/3\) when \(2 \not \mid N\), \(5 \not \mid N\), and \(9 \mid N\), or when \(2 \not \mid N\), \(25 \not \mid N\), and \(45 \mid N\);
-
\(q_2 < N/2\), \(q_3 < N/6\), \(q_4 < N/6\), and \(q_5 < N/6\) when \(4 \not \mid N\), \(5 \not \mid N\), and \(18 \mid N\), or when \(4 \not \mid N\), \(25 \not \mid N\), and \(90 \mid N\);
-
\(q_5 < N/5\) when \(2 \not \mid N\), \(3 \not \mid N\), and \(25 \mid N\), or when \(2 \not \mid N\), \(9 \not \mid N\), and \(75 \mid N\);
-
\(q_2 < N/2\), \(q_3 < N/6\), \(q_4 < N/12\), and \(q_5 < N/12\) when \(5 \not \mid N\), \(8 \not \mid N\), and \(36 \mid N\), or when \(8 \not \mid N\), \(25 \not \mid N\), and \(180 \mid N\);
-
\(q_2 < N/2\), \(q_3 < N/2\), \(q_4 < N/2\), and \(q_5 < N/10\) when \(3 \not \mid N\), \(4 \not \mid N\), and \(50 \mid N\), or when \(4 \not \mid N\), \(9 \not \mid N\), and \(150 \mid N\);
-
\(q_2 < N/2\), \(q_3 < N/6\), \(q_4 < N/24\), and \(q_5 < N/24\) when \(5 \not \mid N\) and \(72 \mid N\), or when \(25 \not \mid N\) and \(360 \mid N\);
-
\(q_2 < N/2\), \(q_3 < N/2\), \(q_4 < N/4\), and \(q_5 < N/20\) when \(3 \not \mid N\), \(8 \not \mid N\), and \(100 \mid N\), or when \(8 \not \mid N\), \(9 \not \mid N\), and \(300 \mid N\);
-
\(q_2 < N/2\), \(q_3 < N/2\), \(q_4 < N/8\), and \(q_5 < N/40\) when \(3 \not \mid N\) and \(200 \mid N\), or when \(9 \not \mid N\) and \(600 \mid N\);
-
\(q_3 < N/3\), \(q_4 < N/3\), and \(q_5 < N/15\) when \(2 \not \mid N\) and \(225 \mid N\);
-
\(q_2 < N/2\), \(q_3 < N/6\), \(q_4 < N/6\), and \(q_5 < N/30\) when \(4 \not \mid N\) and \(450 \mid N\);
-
\(q_2 < N/2\), \(q_3 < N/6\), \(q_4 < N/12\), and \(q_5 < N/60\) when \(8 \not \mid N\) and \(900 \mid N\);
-
\(q_2 < N/2\), \(q_3 < N/6\), \(q_4 < N/24\), and \(q_5 < N/120\) when \(1800 \mid N\).
Theorem 2.8
Proof
See Appendix C.1. \(\square \)
From (18), we see that the number of ZF 5-PPs is equal to 0 if N is three or nine times a product of prime numbers greater than five, each of them to the power of 1.
Theorem 2.9
Proof
See Appendix C.2. \(\square \)
From (19), we see that the number of ZF 5-PPs is equal to 0 if N is 5 or 25 times a product of prime numbers greater than five, each of them to the power of 1.
Theorem 2.10
Proof
See Appendix C.3. \(\square \)
From (20), we see that the number of ZF 5-PPs is equal to 0 if N is 6 or 18 times a product of prime numbers greater than five, each of them to the power of 1.
Theorem 2.11
Proof
See Appendix C.4. \(\square \)
From (21), we see that the number of ZF 5-PPs is equal to 0 if N is 10 or 50 times a product of prime numbers greater than five, each of them to the power of 1.
Theorem 2.12
Proof
See Appendix C.5. \(\square \)
From (22), we see that the number of ZF 5-PPs is equal to 0 if N is 12 or 36 times a product of prime numbers greater than five, each of them to the power of 1.
Theorem 2.13
Proof
See Appendix C.6. \(\square \)
From (23), we see that the number of ZF 5-PPs is 0 if N is 15, 45, 75, or 225 times a product of prime numbers greater than five, each of them to the power of 1.
Theorem 2.14
Proof
See Appendix C.7. \(\square \)
From (24), we see that the number of ZF 5-PPs is equal to 0 if N is 20 or 100 times a product of prime numbers greater than five, each of them to the power of 1.
Theorem 2.15
Proof
See Appendix C.8. \(\square \)
From (25), we see that the number of ZF 5-PPs is equal to 0 if N is 24 or 72 times a product of prime numbers greater than five, each of them to the power of 1.
Theorem 2.16
Proof
See Appendix C.9. \(\square \)
From (26), we see that the number of ZF 5-PPs is 0 if N is 30, 90, 150, or 450 times a product of prime numbers greater than five, each of them to the power of 1.
Theorem 2.17
Proof
See Appendix C.10. \(\square \)
From (27), we see that the number of ZF 5-PPs is equal to 0 if N is 40 or 200 times a product of prime numbers greater than five, each of them to the power of 1.
Theorem 2.18
Proof
See Appendix C.11. \(\square \)
From (28), we see that the number of ZF 5-PPs is equal to 0 if N is 60, 180, 300, or 900 times a product of prime numbers greater than five, each of them to the power of 1.
Theorem 2.19
Proof
See Appendix C.12. \(\square \)
From (29), we see that the number of ZF 5-PPs is equal to 0 if N is 120, 360, 600, or 1800 times a product of prime numbers greater than five, each of them to the power of 1.
The conclusions from the end of Sect. 4.3 have to be corrected as follows.
The ratio (in percentage) between the number of true different 5-PPs under Zhao and Fan sufficient conditions and the number of all true different 5-PPs, for all lengths used in the LTE standard
We compared the total number of true different 5-PPs, obtained with the algorithm from [5], for all lengths from the LTE standard with the number of true different 5-PPs under ZF sufficient conditions. Figure 3 shows the ratio (in percentage) between the number of true different 5-PPs under ZF sufficient conditions and the number of all true different 5-PPs for all the 188 lengths of the LTE standard. We note that for 8 from the 188 lengths, the number of all true different 5-PPs is zero (i.e., there are no true 5-PPs for these lengths, namely 40, 88, 120, 248, 264, 328, 440, and 488), and thus, the percentage is 100%. For 31 from the 188 lengths (from which 8 lengths are those mentioned above), the ZF sufficient conditions are also necessary (i.e., this percentage is 100%). For other 20 lengths, this percentage is 0% (i.e., there are no true 5-PPs under ZF sufficient conditions, but there are true different 5-PPs, for these lengths) and for the other 137 lengths, the percentage is less than 20%.
The second example for Theorem 2.19 has to be corrected as in Appendix D.2.
References
- 1.Trifina, L., & Tarniceriu, D. (2016). The number of different true permutation polynomial based interleavers under Zhao and Fan sufficient conditions. Telecommunication Systems, 63(4), 593–623.CrossRefGoogle Scholar
- 2.Zhao, H., & Fan, P. (2007). Simple method for generating \(m\)th-order permutation polynomials over integer rings. IEE Electronics Letters, 43(8), 449–451.CrossRefGoogle Scholar
- 3.Ryu, J. (2007). Permutation polynomial based interleavers for turbo codes over integer rings, Ph.D. thesis, [Online] Available https://etd.ohiolink.edu/rws_etd/document/get/osu1181139404/inline.
- 4.3GPP TS 36.212 V8.3.0, 3rd Generation partnership project, multiplexing and channel coding (Release 8), 2008. [Online] http://www.etsi.org.
- 5.Trifina, L., & Tarniceriu, D. (2018). Determining the number of true different permutation polynomials of degrees up to five by Weng and Dong algorithm. Telecommunication Systems, 67(2), 211–215.CrossRefGoogle Scholar