Skip to main content
Log in

Efficient RNS Reverse Converters for Moduli Sets with Dynamic Ranges Up to \((10n+1)\)-bit

  • Short Paper
  • Published:
Circuits, Systems, and Signal Processing Aims and scope Submit manuscript

Abstract

This paper presents reverse converters for dynamic ranges (DRs) up to \(10n+1\) achieved with horizontal and vertical extensions of the traditional three-moduli set \(\{2^{n},2^n-1, 2^n+1\}\). The proposed approach results in increased parallelism with a smaller number of bits per channel. An arithmetic application is considered to show that the proposed converter scales better with an increasing number of residue operations per channel, confirming the practical applications of the moduli set and reverse converter. Experimental results suggest that the proposed architecture outperforms the best reverse converters in the related state of the art, with equivalent DR, presenting speedup of 2.09 times, and average reductions of 62.00 and 45.20% in area and power consumption, respectively.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Fig. 1
Fig. 2
Fig. 3

Similar content being viewed by others

References

  1. P. Ananda Mohan, A. Premkumar, RNS-to-binary converters for two four-moduli sets \(\{2^{n}-1,2^{n},2^{n}+1,2^{n+1}-1\}\) and \(\{2^{n}-1,2^{n},2^{n}+1,2^{n+1}+1\}\). IEEE Trans. Circuits Syst. I Regul. Pap. 54(6), 1245–1254 (2007). https://doi.org/10.1109/TCSI.2007.895515

    Article  MathSciNet  MATH  Google Scholar 

  2. B. Cao, C.H. Chang, T. Srikanthan, A residue-to-binary converter for a new five-moduli set. IEEE Trans. Circuits Syst. I Regul. Pap. 54(5), 1041–1049 (2007). https://doi.org/10.1109/TCSI.2007.890623

    Article  MathSciNet  MATH  Google Scholar 

  3. R. Chaves, L. Sousa, \(\{2^n+1\),\(2^{n+k}\),\(2^n-1\}\): a new RNS moduli set extension. in Digital System Design, 2004. DSD 2004. Euromicro Symposium on (2004), pp. 210–217. https://doi.org/10.1109/DSD.2004.1333279

  4. R. de Matos, R. Paludo, N. Chervyakov, P.A. Lyakhov, H. Pettenghi, Efficient implementation of modular multiplication by constants applied to rns reverse converters, in 2017 IEEE International Symposium on Circuits and Systems (ISCAS) (2017), pp. 1–4. https://doi.org/10.1109/ISCAS.2017.8050779

  5. G. Dimitrakopoulos, D. Nikolos, High-speed parallel-prefix VLSI Ling adders. IEEE Trans. Comput. 54(2), 225–231 (2005). https://doi.org/10.1109/TC.2005.26

    Article  Google Scholar 

  6. A. Hiasat, VLSI implementation of new arithmetic residue to binary decoders. IEEE Trans. Very Large Scale Integr. (VLSI) Syst. 13(1), 153–158 (2005). https://doi.org/10.1109/TVLSI.2004.840400

    Article  Google Scholar 

  7. A. Hiasat, A reverse converter and sign detectors for an extended rns five-moduli set. IEEE Trans. Circuits Syst. I Regul. Pap. 64(1), 111–121 (2017)

    Article  Google Scholar 

  8. M. Hosseinzadeh, A. Molahosseini, K. Navi, An improved reverse converter for the moduli set \(\{2^n-1,2^n,2^n+1, 2^{n+1}-1\}\). Trans. IEICE Electron. Express 5(17), 672–677 (2008)

    Article  Google Scholar 

  9. A. Molahosseini, K. Navi, C. Dadkhah, O. Kavehei, S. Timarchi, Efficient reverse converter designs for the new 4-moduli sets \(\{2^{n}-1,2^{n},2^{n}+1,2^{2n+1}-1\}\) and \(\{2^{n}-1,2^{n}+1,2^{2n},2^{2n}+1\}\) based on new CRTs. IEEE Trans. Circuits Syst. I Regul. Pap. 57(4), 823–835 (2010). https://doi.org/10.1109/TCSI.2009.2026681

    Article  MathSciNet  Google Scholar 

  10. H. Pettenghi, R. de Matos, A. Molahosseini, RNS reverse converters for moduli sets with dynamic ranges of \(9n\)-bit, in 2016 IEEE 7th Latin American Symposium on Circuits Systems (LASCAS) (2016), pp. 143–146. https://doi.org/10.1109/LASCAS.2016.7451030

  11. H. Pettenghi, R. Chaves, L. Sousa, Method to design general RNS reverse converters for extended moduli sets. IEEE Trans. Circuits Syst. II Express Briefs 60(12), 877–881 (2013). https://doi.org/10.1109/TCSII.2013.2286433

    Article  Google Scholar 

  12. H. Pettenghi, R. Chaves, L. Sousa, RNS reverse converters for moduli sets with dynamic ranges up to \((8n+1)\)-bit. IEEE Trans. Circuits Syst. I Regul. Pap. 60(6), 1487–1500 (2013). https://doi.org/10.1109/TCSI.2012.2220460

    Article  MathSciNet  Google Scholar 

  13. H. Pettenghi, R. Chaves, R.D. Matos, L. Sousa, Method for designing two levels RNS reverse converters for large dynamic ranges. Integr. VLSI J. 55, 22–29 (2016). https://doi.org/10.1016/j.vlsi.2016.02.004

    Article  Google Scholar 

  14. S. Piestrak, K. Berezowski, Design of residue multipliers-accumulators using periodicity, in Signals and Systems Conference, 208 (ISSC 2008). (IET Irish, 2008), pp. 380–385. https://doi.org/10.1049/cp:20080692

  15. A. Skavantzos, An efficient residue to weighted converter for a new residue number system, in VLSI, 1998. Proceedings of the 8th Great Lakes Symposium on (1998), pp. 185–191. https://doi.org/10.1109/GLSV.1998.665223

  16. L. Sousa, S. Antao, MRC-based RNS reverse converters for the four-moduli sets \(\{2^{n}+1, 2^{n}-1, 2^{n}, 2^{2n+1}-1\}\) and \(\{2^{n}+1, 2^{n}-1, 2^{2n}, 2^{2n+1}-1\}\). IEEE Trans. Circuits Syst. I Express Briefs 59(4), 244–248 (2012). https://doi.org/10.1109/TCSII.2012.2188456

    Article  Google Scholar 

  17. N. Szabó, R. Tanaka, Residue Arithmetic and Its Applications to Computer Technology, McGraw-Hill series in information processing and computers (McGraw-Hill, New York, 1967)

    MATH  Google Scholar 

  18. A. Tyagi, A reduced-area scheme for carry-select adders. IEEE Trans. Computers 42(10), 1163–1170 (1993). https://doi.org/10.1109/12.257703

    Article  Google Scholar 

  19. Virtual Silicon Technology, Inc, UMC high density standards cells library \(90n\)m CMOS process (2010)

  20. Y. Wang, Residue-to-binary converters based on new Chinese remainder theorems. IEEE Trans. Circuits Syst. II Analog Digit. Signal Process. 47(3), 197–205 (2000). https://doi.org/10.1109/82.826745

    Article  MATH  Google Scholar 

  21. R. Zimmermann, Efficient VLSI implementation of modulo \((2^n \pm 1)\) addition and multiplication, in Proceedings 14th IEEE Symposium on Computer Arithmetic (1999), pp. 158–167. https://doi.org/10.1109/ARITH.1999.762841

Download references

Acknowledgements

This work was supported in part by the Brazilian Federal Agency for Support and Evaluation of Graduate Education (CAPES) and by the Russian Federation President Grant \(SP-2245.2018.5\).

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Rogerio Paludo.

Appendices

Appendices

1.1 Proof of Lemmas in the Paper

Proof

In order to prove the value of multiplicative inverse of \((2^{9n}-2^{3n})\) modulo \(M_{b}=2^{n-1}-1\) for \(n=12k_1+6, k_1 \in {\mathbb {N}}\) given in Eq. (4), we need to distinguish two subcases \(n=6\) and \(n>6\). For \(n=6\) we have to show that:

$$\begin{aligned} \left| (2^{9n}-2^{3n}) \times \left| {(2^{9n}-2^{3n})}^{-1} \right| _{M_b} \right| _{M_b} = \left| (2^{9n}-2^{3n}) \times 2^{n-4}\right| _{M_b} = 1. \end{aligned}$$
(11)

Therefore substituting the value \(n=6\):

$$\begin{aligned} | (\overbrace{|2^{54}|_{31}}^{16}-\overbrace{|2^{18}|_{31}}^{8}) \times (2^{6-4})|_{31} = |8\times 4|_{31} = 1. \end{aligned}$$
(12)

For \(n>6\) we have to show that:

$$\begin{aligned}&\left| (2^{9n}-2^{3n}) \times \left| (2^{9n}-2^{3n})^{-1} \right| _{M_b} \right| _{M_b}\nonumber \\&\quad =\left| (2^{9n}-2^{3n}) \times \left( 2^{n-4}+\frac{2^{n}}{504}-\frac{64}{504}\right) \right| _{M_b} = 1, \end{aligned}$$
(13)

where \(| 2^{9n}-2^{3n} |_{M_b} = | \overbrace{|2^{9n}|_{M_b}}^{2^{9}}-\overbrace{|2^{3n}|_{M_b}}^{2^{3}} |_{M_b} = 504\). Therefore, Eq. (13) can be written as:

$$\begin{aligned}&\left| (2^{9}-2^{3}) \times \left( 2^{n-4}+\frac{2^{n}}{504}-\frac{64}{504} \right) \right| _{M_b} \nonumber \\&\quad = \left| 2^{n+5}-2^{n-1}+2^{n}-2^{6}\right| _{M_b}= \left| 2^{6}-1+2-2^{6}\right| _{M_b}=1. \end{aligned}$$
(14)

It is important to note that \(\left| (2^{9n}-2^{3n})^{-1} \right| _{M_b}\) is also expressed for \(n>6\) in terms of powers of two in (4). Thus, to finish the proof we have to show that:

$$\begin{aligned} \left| (2^{9n}-2^{3n})^{-1} \right| _{M_b} = 2^{n-4}+\displaystyle {\sum _{i=0}^{\frac{n-12}{6}}} 2^{6i+3}. \end{aligned}$$
(15)

By using the sum of n terms of a geometric series with common ratio r:

$$\begin{aligned} \sum _{i=m}^{n}r^{i}=\frac{r^{m}-r^{n+1}}{1-r} . \end{aligned}$$
(16)

we can obtain:

$$\begin{aligned} 2^{n-4}+\displaystyle {\sum _{i=0}^{\frac{n-12}{6}}} 2^{6i+3}= & {} 2^{n-4}+2^{3} \times \frac{1-2^{n-6}}{1-2^6} \nonumber \\= & {} 2^{n-4}+2^{3} \times \left( \frac{2^{n-6}}{63}- \frac{1}{63}\right) =2^{n-4}+\frac{2^{n}}{504}-\frac{64}{504}. \end{aligned}$$
(17)

Thus, Lemma 1 is proved. \(\square \)

Proof

In order to prove the value of the multiplicative inverse of \((2^{9n}-2^{3n})\) modulo \(M_{b}=2^{n-1}-1\) for \(n=12k_2+14, k_2 \in {\mathbb {N}}\) given in Eq. (5) we need to distinguish again two subcases \(n=14\) and \(n>14\). For \(n=14\) we have to show that:

$$\begin{aligned}&\left| (2^{9n}-2^{3n}) \times \left| {(2^{9n}-2^{3n})}^{-1} \right| _{M_b} \right| _{M_b} \nonumber \\&\quad = \left| (2^{9n}-2^{3n}) \times (2^{n-2}+2^{n-3}-2^{n-9}+2^{4}-1) \right| _{M_b} = 1. \end{aligned}$$
(18)

Therefore, substituting the value \(n=14\):

$$\begin{aligned} |(\overbrace{|2^{126}|_{8191}}^{512}-\overbrace{|2^{42}|_{8191}}^{8}) \times (6127)|_{8191} = |504 \times 6127|_{8191} = 1. \end{aligned}$$
(19)

For \(n>14\) we have to show that:

(20)

It is important to note that \(\left| (2^{9n}-2^{3n})^{-1} \right| _{M_b}\) is also expressed for \(n>14\) in terms of powers of two in Eq. (5). Thus, to finish the proof we have to show that:

$$\begin{aligned} \left| (2^{9n}-2^{3n})^{-1} \right| _{M_b} = 2^4 - 1 + 2^{n-2} + 2^{n-3} - 2^{n-9} + \sum _{i=0}^{\frac{n-20}{6}} \left( 2^{6i + 10} - 2^{6i + 5}\right) . \nonumber \\ \end{aligned}$$
(21)

Applying Eq. (16), we can obtain:

$$\begin{aligned}&2^4 - 1 + 2^{n-2} + 2^{n-3} - 2^{n-9} + \frac{2^{n-4} - 2^{10}}{2^6 - 1} - \frac{2^{n-9} - 2^{5}}{2^6 - 1}\nonumber \\&\quad =\frac{(2^{n-1}+376(2^{n-1}-1))}{(2^{9n}-2^{3n})}. \end{aligned}$$
(22)

Thus, Lemma 2 is proved. \(\square \)

Proof

In order to prove the value of the multiplicative inverse of \((2^{9n}-2^{3n})\) modulo \(M_{b}=2^{n+1}-1\) for \(n=12k_3+6, k_3 \in {\mathbb {N}}\) given in (6), we need to distinguish two subcases \(n=6\) and \(n>6\). For \(n=6\) we have to show that:

$$\begin{aligned} \left| (2^{9n}-2^{3n}) \times \left| {(2^{9n}-2^{3n})}^{-1} \right| _{M_b} \right| _{M_b} = \left| (2^{9n}-2^{3n}) \times 2^{3}\right| _{M_b} = 1. \end{aligned}$$
(23)

Therefore substituting the value \(n=6\):

$$\begin{aligned} \left| (\overbrace{|2^{54}|_{127}}^{32}-\overbrace{|2^{18}|_{127}}^{16}) \times (2^{3})\right| _{127} = |16 \times 8|_{127} = 1. \end{aligned}$$
(24)

For \(n>6\) we have to show that:

$$\begin{aligned} \left| (2^{9n}-2^{3n}) \times \left| (2^{9n}-2^{3n})^{-1} \right| _{M_b} \right| _{M_b} = \left| (2^{9n}-2^{3n}) \times \frac{8 \times 2^{n+1} - 520}{(2^{6}-1)} \right| _{M_b} = 1. \end{aligned}$$
(25)

where \(| 2^{9n}-2^{3n} |_{M_b} = | \overbrace{|2^{3n}|_{M_b}}^{2^{n-2}}(\overbrace{|2^{6n}|_{M_b}}^{2^{n-5}}-1) |_{M_b} = 2^{n-2} (2^{n-5}-1)\). Therefore, Eq. (25) can be written as:

(26)

Note that \(\left| (2^{9n}-2^{3n})^{-1} \right| _{M_b}\) is also expressed in terms of powers of two in (6). Thus, to finish the proof we have to show that:

$$\begin{aligned} \left| (2^{9n}-2^{3n})^{-1} \right| _{M_b} = 2^3 + \sum _{i=0}^{\frac{n-12}{6}}{2^{6i+10}}. \end{aligned}$$
(27)

Applying Eq. (16), we can obtain:

$$\begin{aligned} 2^3 + \left( \frac{2^{n-6}-1}{2^6-1} \right) \times 2^{10} = 2^3 + \left( \frac{2^{n+1}-2^7}{2^7 \times (2^6-1)} \right) \times 2^{10} = \frac{8 \times 2^{n+1} - 520}{(2^{6}-1)}.\nonumber \\ \end{aligned}$$
(28)

Thus, Lemma 3 is proved. \(\square \)

Proof

In order to prove the value of the multiplicative inverse of \((2^{9n}-2^{3n})\) modulo \(M_{b}=2^{n+1}-1\) for \(n=12k_4+10, k_4 \in {\mathbb {N}}\) given in (7), we need to distinguish two subcases \(n=10\) and \(n>10\). For \(n=10\) we have to show that:

$$\begin{aligned}&\left| (2^{9n}-2^{3n}) \times \left| {(2^{9n}-2^{3n})}^{-1} \right| _{M_b} \right| _{M_b} \nonumber \\&\quad =\left| (2^{9n}-2^{3n}) \times ( 2^{n} + (2^9 - 2^5) + (2^4 - 2^0))\right| _{M_b} = 1. \end{aligned}$$
(29)

Therefore substituting the value \(n=10\):

$$\begin{aligned}&|(\overbrace{|2^{90}|_{2047}}^{4}-\overbrace{|2^{30}|_{2047}}^{256}) \times \overbrace{(2^{10} + (2^9 - 2^5) + (2^4 - 2^0))}^{1519}|_{2047}\nonumber \\&\quad = |-252 \times 1519|_{2047} = 1. \end{aligned}$$
(30)

For \(n>10\) we have to show that:

$$\begin{aligned} \begin{aligned} \left| (2^{9n}-2^{3n}) \times \left| (2^{9n}-2^{3n})^{-1} \right| _{M_b} \right| _{M_b} {=} \left| (2^{9n}-2^{3n}) \times \frac{(47\times 2^{n+1}-559)}{(2^{6}-1)} \right| _{M_b} = 1. \end{aligned} \end{aligned}$$
(31)

where \(| 2^{9n}-2^{3n} |_{M_b} = | \overbrace{|2^{3n}|_{M_b}}^{2^{n-2}}(\overbrace{|2^{6n}|_{M_b}}^{2^{n-5}}-1) |_{M_b} = 2^{n-2} (2^{n-5}-1)\). Therefore, Eq. (31) can be written as:

(32)

It is important to note that \(\left| (2^{9n}-2^{3n})^{-1} \right| _{M_b}\) is also expressed in terms of powers of two in (5). Thus, to finish the proof we have to show that:

$$\begin{aligned} \left| (2^{9n}-2^{3n})^{-1} \right| _{M_b} = 2^n + \displaystyle {\sum _{i=0}^{\frac{n-16}{6}} \left( 2^{6i + 15} - 2^{6i + 10}\right) } + (2^9 - 2^5) + (2^4 - 2^0). \end{aligned}$$
(33)

Applying Eq. (16), we can obtain:

$$\begin{aligned}&2^n + \frac{2^{15} + 2^{15} \times 2^{n-16}}{1 - 2^6} - \frac{2^{10} + 2^{10} \times 2^{n-16}}{1 - 2^6} + (2^9 - 2^5) + (2^4 - 2^0) \nonumber \\&\quad = \frac{(47 \times 2^{n+1} - 559)}{63}. \end{aligned}$$
(34)

Thus, Lemma 4 is proved. \(\square \)

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Pettenghi, H., Paludo, R., Matos, R. et al. Efficient RNS Reverse Converters for Moduli Sets with Dynamic Ranges Up to \((10n+1)\)-bit. Circuits Syst Signal Process 37, 5178–5196 (2018). https://doi.org/10.1007/s00034-018-0815-7

Download citation

  • Received:

  • Revised:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s00034-018-0815-7

Keywords

Navigation