CN102929574A - Pulse multiplying unit design method on GF (Generator Field) (2163) domain - Google Patents
Pulse multiplying unit design method on GF (Generator Field) (2163) domain Download PDFInfo
- Publication number
- CN102929574A CN102929574A CN2012103952515A CN201210395251A CN102929574A CN 102929574 A CN102929574 A CN 102929574A CN 2012103952515 A CN2012103952515 A CN 2012103952515A CN 201210395251 A CN201210395251 A CN 201210395251A CN 102929574 A CN102929574 A CN 102929574A
- Authority
- CN
- China
- Prior art keywords
- module
- multiplier
- design method
- pulsation
- parallel
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Error Detection And Correction (AREA)
Abstract
The invention belongs to the technical field of digital integrated circuits and systems and particularly relates to a pulse multiplying unit design method on a domain. According to the pulse multiplying unit design method, a full-parallel 8*8 module is used as a basic unit to carry out 21-grade cascading; a combined logic distribution of s whole circuit is adjusted by adopting a retiming algorithm; and the highest clock frequency of the system is improved.
Description
Technical field
The invention belongs to digital integrated circuit and systems technology field, be specifically related to a kind of GF (2
163) pulsation Multiplier Design method on the territory.
Background technology
Theory of finite fields not only has important theoretical research to be worth in Modern Mathematics, has had widely aspect advanced information society a lot of and has used.Galois field multiplier is the elementary cell in a lot of the application, such as elliptic curve cipher, error correcting code, digital signal processing etc.So, design a efficient Galois field multiplier and have great significance for the lifting of system performance.
Tan Siwei in August, 2011 in " based on design and the realization of the restructural Galois field multiplier of RS code " literary composition that " calculating and application and software " is delivered, adopted naturally basic algorithm, realized the multiplying on the Galois field with simple logic gates, and proposed the reconfigurable iterative computation structure based on the RS code, can satisfy territory length and be 3~8 multiplier.The demand of design has been satisfied in this design to a certain extent, and especially it considers that the RS code has stronger error correcting capability, and the accuracy of multiplying is had certain guarantee.But its deficiency when realizing based on FPGA is: (1) does not implement the design philosophy of " bottom-up ", does not use well based on submodule to make up the multiplier that is applicable to large exponent number territory; (2) consideration of optimization is not done in its design to the delay of circuit, does not have the to the full extent maximum clock frequency of elevator system.
The present invention at first makes up the basic multiplication unit of 8*8bit, then builds based on the multiplier on the Galois field that is applicable to high exponent number 163 of elementary cell, and using for the digital signal processing of intensive provides possibility.And the thought that the present invention is based on when resetting is optimized the structure of integrated circuit, has promoted to a certain extent the maximum clock frequency of system, for high-speed digital design provides guarantee.
Summary of the invention
The object of the present invention is to provide the multiplier that a kind of area is little, performance is high
Pulsation multiplier on the territory, the maximum clock frequency that it can elevator system can satisfy in the digital information processing system of the accuracy control system of guarantee reliable data transmission, the encryption system that ensures information safety, intensive the requirement to multiplier.
Provided by the invention a kind of
Pulsation Multiplier Design method on the territory, whole system is carried out 21 grades of cascades take complete parallel 8*8 module as elementary cell, and adopts algorithm when resetting that the combinational logic of integrated circuit is distributed and adjust.
Adopted systolic structures to dividing to the multiplier architecture on the localization among the present invention.
Among the present invention, in Galois field
On, the full parallel organization elementary cell that makes up described 8*8 has adopted the multiplication algorithm of high priority (MSB-first), and specific algorithm is described below:
Output:?
Wherein,
Expression
Individual coefficient,
,
,
Expression
,
,
Individual coefficient, m=163.Among the present invention, when designing the full parallel organization elementary cell of described 8*8, adopted the scheme of coordinate transform, that is:
(1)
Wherein i represents the number of times of circuit iteration, and j represents polynomial exponential factor, thereby the two-way input between each module of X direction is mapped as the dependency graph of one dimension.
Among the present invention, complete parallel structural drawing is divided into N*N module, and L*L pulsating element arranged in each module, and algorithm was optimized the combinational logic distribution of integrated circuit when employing reset, and comprising:
Add the Ctrl=011 of a N+1 length ... 1 control signal, wherein A (x) and G (x) enter complete parallel 8*8 module with the pattern of serial, and B (x) enters complete parallel 8*8 module with parallel pattern;
Register and L the MUX of adding L 1 bit arrive the 8*8 module that each walks abreast entirely, are used for result of calculation
Be broadcast to the capable unit of each i;
Add the register of the individual delay unit of 3 (L-1) to each 8*8 module that entirely walks abreast, be used for preserving the data that pass to the unit of region k+1 leftmost from the unit of region k rightmost with diagonal manner;
Add again the register of L MUX and L 1 bit to each 8*8 module that entirely walks abreast, be used for allowing B (x) also enter complete parallel 8*8 module with the serial pattern, when the Ctrl signal is 0, B (x) is reprinted the 8*8 module that walks abreast into entirely, the circuit structure diagram after when obtaining resetting.
Among the present invention, a pulsation multiplier that contains the L bit digit width of N basic module can be in the territory
Upper take following sequence as list entries, wherein
:
………………………………………………………(4)
0≤i≤N-1 wherein, and d=NL-m, and in the situation for the j that exceeds [0, m-1] scope,
, the data sequence of output
For:
Beneficial effect of the present invention is: algorithm was adjusted the combinational logic distribution of integrated circuit when its method for designing employing reset, and had improved system's maximum clock frequency.
Description of drawings
Fig. 1 is basic processing unit in the circuit that obtains of the multiplication algorithm analysis of high priority;
Fig. 2 is
The structural drawing of the full Parallel Design that obtains of the multiplication algorithm of high priority;
Fig. 3 is the circuit basic processing unit after the coordinate transform;
Fig. 4 is the full Parallel Design structural drawing after the coordinate transform;
Fig. 5 is that m/L is the one dimension dependency graph that is obtained by Fig. 4 projection of integer;
Fig. 6 is the cut-away view of each basic module;
Fig. 7 is the cut-away view of each basic module after B (X) is loaded;
One dimension dependency graph after Fig. 8 attaches most importance to regularly;
Fig. 9 is the systolic structures figure of 21 grades of cascades;
Figure 10 is the 1*8 computing unit;
Figure 11 is the 8*8 computing unit;
Figure 12 is the 163*163bit multiplier with the 8*8bit structural generation;
Figure 13 is the coefficient of as a result each power of MATLAB emulation;
Figure 14 is for beginning Modelsim simulation result to the 4400ns;
Figure 15 is Modelsim simulation result in 4400ns~5us;
Figure 16 is Modelsim simulation result in 5us~5600ns;
Figure 17 is Modelsim simulation result in 5600ns~6us.
Embodiment
Further describe the present invention below in conjunction with accompanying drawing.
In Galois field
On, the present invention has adopted the multiplication algorithm of high priority (MSB-first), and specific algorithm is described below:
Output:?
Expression
Individual coefficient,
,
,
Expression
,
,
Individual coefficient, among the present invention, m=163.
With Galois field
The multiplication algorithm of high priority be example, the structural drawing of its full Parallel Design as shown in Figure 2, it has comprised 4*4 basic processing unit (i, j), wherein i represents the number of times of circuit iteration, and j represents polynomial exponential factor, multinomial coefficient all is parallel input, and the result is parallel output.For the Digit-serial structure, take L=2 as example, be about to complete parallel structural drawing and be divided into N*N module, L*L pulsating element, wherein N=m/L are arranged in each module, and then it is mapped as the dependency graph of one dimension, but can be found to have two-way input between each module of X direction by Fig. 2, this is so that we can not directly be mapped as the dependency graph of one dimension, to this, the present invention has adopted the scheme of coordinate transform to solve this problem, that is:
(1) the circuit basic processing unit through conversion becomes shown in Figure 3.Corresponding full Parallel Design structural drawing as shown in Figure 4, can find out the problem that does not have two-way input when carrying out Module Division for this structure, and the function that this circuit is realized with do not carry out coordinate transform before circuit identical, the direction that is input and output has changed.
If m/L is integer, Fig. 4 is projected as the one dimension dependency graph, as shown in Figure 5, wherein "
" delay unit of expression.The cut-away view of each basic module in order to control this module, has added the Ctrl=011 of a N+1 length as shown in Figure 6 ... 1 control signal.Wherein A (x) and G (x) enter module with the pattern of serial, and B (x) enters module with parallel pattern, and FACTOR P (x) will be exported L-bit output with a clock cycle as a result.Because result of calculation
Must broadcast to the capable unit of each i, so having added register and L MUX of L 1 bit, the present invention arrives each module, in order to preserve the data that pass to the unit of region k+1 leftmost from the unit of region k rightmost with diagonal manner, we have added the register of the individual delay units of 3 (L-1) to each module.
Add the register of L MUX and L 1 bit to each module again, can allow B (x) also enter module with the serial pattern, when the Ctrl signal was 0, B (x) was reprinted the progressive die piece, namely as shown in Figure 7.And then the circuit structure diagram after when obtaining resetting as shown in Figure 8, if can see input this moment continuously, this circuit structure can be after the time-delay of the clock period of a 3N+1 be exported result of calculation take every N clk as the cycle.
The present invention has designed the basic module of 8*8, carries out 21 grades cascade again, thereby has realized
The pulsation multiplier of 8bit on the territory, whole schematic diagram as shown in Figure 9.
The design of 8*8 computing unit at first utilizes 8 minimum units to be combined into the computing unit of 1*8, and as shown in figure 10, this unit is pure combinational circuit.Then be combined into as shown in figure 11 8*8 computing unit by the computing unit of 8 1*8, a pair of ordinate of alphabetical r mark is connected with a pair of ordinate of lastrow letter g mark among the figure, and g represents to exist a delay unit.Letter y represents data selector, when being used for processing this computing unit and beginning to calculate in the calculating formula
Value.The zero hour, control signal was that the output of 0, y is one section broken line signal of its top; The control signal of y is 1 afterwards, and output is exactly the T in left side
iSignal.This computing unit needs two clock period just can finish calculating, and the result is offered the next stage computing unit, so two all after dates of control signal time-delay are as the control signal of next computing unit.The rectangular expression that the palisade grid of below is filled among the figure be used for the delaying time register of a clock period, the existence of this register is in order to realize pipeline organization, to reduce the length of critical path.In fact consider the time-lag action that adds green block, in fact the Gout of its output and Aout are exactly the Gin of input and the time-delay in two cycles of Ain.
Can consist of 163*163 position multiplier as shown in figure 12 by the 8*8 unit, so owing to 163 can not be divided exactly by 8 and must process the non-problem that divides exactly.The computation structure of 163*163 is expanded to the computation structure of 168*168 for this reason, five unnecessary whole assignment are zero, because the initialization input when top and right-hand lowest order calculate in the computing all is zero, so the figure place of replenishing can not exert an influence to result of calculation.
Adopt the MATLAB calculation procedure to calculate one group of data, the data of theory calculating and the result of MODELSIM emulation are compared.163 of zero padding input data are not expressed as follows with the set mode:
(1) i|0≤i≤162 and a[i]=1={ 0,2,4,5,6,7,8,9,10,94,100,110,147,148,149,150,151,152,153,154 }
(2) i|0≤i≤162 and b[i]=1={ 0,8,9,10,11,12,13,14,15,52,53,54,55,56,57,58,59 }
(3) i|0≤i≤162 and g[i]=1={ 0,3,6,7 }.
The coefficient of as a result each power of emulation as shown in figure 13.Figure 14 to Figure 17 correspondence the simulation result of Modelsim, at first output is a high position among the result among Figure 14, last position of 8 bit data of first output effectively, later 8 in each clock period Output rusults, the front two of last 8 Output rusults is effective.Several bit data that most begin to export in Figure 14 simulation result are 1,10010100, and 11000001. is consistent with the result of calculation 1,10010100,11000001 of MATLAB, contrasts all 163 results, and both are identical.
Claims (4)
1. one kind
Pulsation Multiplier Design method on the territory, it is characterized in that: adopted systolic structures to dividing to the multiplier architecture on the localization, take complete parallel 8*8 module as elementary cell, carry out 21 grades of cascades, and adopt algorithm when resetting that the combinational logic of integrated circuit is distributed and adjust; Simultaneously in Galois field
On, having adopted the multiplication algorithm of high priority when making up the full parallel organization elementary cell of described 8*8, specific algorithm is described below:
2. pulsation Multiplier Design method according to claim 1 is characterized in that: design described 8*8 entirely also
During row structure elementary cell, adopted the scheme of coordinate transform, that is:
, wherein i represents the number of times of circuit iteration, and j represents polynomial exponential factor, thereby the two-way input between each module of X direction is mapped as the dependency graph of one dimension.
3. pulsation Multiplier Design method according to claim 1, it is characterized in that, described complete parallel structural drawing is divided into N*N module, and L*L pulsating element arranged in each module, algorithm is optimized the combinational logic distribution of integrated circuit when adopting described resetting, and comprising:
Add the Ctrl=011 of a N+1 length ... 1 control signal, wherein A (x) and G (x) enter complete parallel 8*8 module with the pattern of serial, and B (x) enters complete parallel 8*8 module with parallel pattern;
Register and L the MUX of adding L 1 bit arrive the 8*8 module that each walks abreast entirely, are used for result of calculation
Be broadcast to the capable unit of each i;
Add the register of the individual delay unit of 3 (L-1) to each 8*8 module that entirely walks abreast, be used for preserving the data that pass to the unit of region k+1 leftmost from the unit of region k rightmost with diagonal manner;
Add again the register of L MUX and L 1 bit to each 8*8 module that entirely walks abreast, be used for allowing B (x) also enter complete parallel 8*8 module with the serial pattern, when the Ctrl signal is 0, B (x) is reprinted the 8*8 module that walks abreast into entirely, the circuit structure diagram after when obtaining resetting.
4. pulsation Multiplier Design method according to claim 1 is characterized in that: for
When the territory has at first made up the module of full parallel organization of 8*8, because the exponent number 163 of Galois field can not be divided exactly by the pulsating element in the module several 8, therefore introduced " zero padding " thought: a pulsation multiplier that contains the L bit width of N basic module, can be in the territory
Upper take following sequence as list entries, wherein
:
…………………………………………………………(2)
0≤i≤N-1 wherein, and d=NL-m, and in the situation for the j that exceeds [0, m-1] scope,
, the data sequence of output
For:
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN2012103952515A CN102929574A (en) | 2012-10-18 | 2012-10-18 | Pulse multiplying unit design method on GF (Generator Field) (2163) domain |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN2012103952515A CN102929574A (en) | 2012-10-18 | 2012-10-18 | Pulse multiplying unit design method on GF (Generator Field) (2163) domain |
Publications (1)
Publication Number | Publication Date |
---|---|
CN102929574A true CN102929574A (en) | 2013-02-13 |
Family
ID=47644388
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN2012103952515A Pending CN102929574A (en) | 2012-10-18 | 2012-10-18 | Pulse multiplying unit design method on GF (Generator Field) (2163) domain |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN102929574A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103186360A (en) * | 2013-04-03 | 2013-07-03 | 哈尔滨工业大学深圳研究生院 | Fast arithmetic multi-bit serial pulse dual-base binary finite field multiplier |
CN104252332A (en) * | 2014-08-20 | 2014-12-31 | 哈尔滨工业大学深圳研究生院 | Multiplier and multiplier processing element for ellipse cipher apparatus |
CN106201433A (en) * | 2016-07-10 | 2016-12-07 | 北京工业大学 | A kind of Galois field multiplier based on RS code |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20030204546A1 (en) * | 2002-04-26 | 2003-10-30 | Rong-Liang Chiou | Systolic product-sum calculator over multi-basis in GF(2m) |
CN1459710A (en) * | 2002-05-17 | 2003-12-03 | 矽统科技股份有限公司 | Systoletype multiplier structure used in double basement finite region |
TW588280B (en) * | 2000-12-18 | 2004-05-21 | Ind Tech Res Inst | General finite-field multiplier |
CN1499358A (en) * | 2002-11-09 | 2004-05-26 | 三星电子株式会社 | Method and device for base transfer in finite extent |
KR20040055550A (en) * | 2002-12-20 | 2004-06-26 | 한국전자통신연구원 | Serial-Parallel Multiplier to Multiply Two Elements in the Finite Field |
CN1652075A (en) * | 2003-07-03 | 2005-08-10 | 诚致科技股份有限公司 | System and method for efficient VLSI architecture of finite fields |
CN101968732A (en) * | 2010-10-09 | 2011-02-09 | 中国人民解放军信息工程大学 | Bit parallel systolic array shifted polynomial basis multiplier with function of error detection |
CN102314330A (en) * | 2011-09-09 | 2012-01-11 | 华南理工大学 | Composite finite field multiplier |
-
2012
- 2012-10-18 CN CN2012103952515A patent/CN102929574A/en active Pending
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
TW588280B (en) * | 2000-12-18 | 2004-05-21 | Ind Tech Res Inst | General finite-field multiplier |
US20030204546A1 (en) * | 2002-04-26 | 2003-10-30 | Rong-Liang Chiou | Systolic product-sum calculator over multi-basis in GF(2m) |
CN1459710A (en) * | 2002-05-17 | 2003-12-03 | 矽统科技股份有限公司 | Systoletype multiplier structure used in double basement finite region |
CN1499358A (en) * | 2002-11-09 | 2004-05-26 | 三星电子株式会社 | Method and device for base transfer in finite extent |
KR20040055550A (en) * | 2002-12-20 | 2004-06-26 | 한국전자통신연구원 | Serial-Parallel Multiplier to Multiply Two Elements in the Finite Field |
CN1652075A (en) * | 2003-07-03 | 2005-08-10 | 诚致科技股份有限公司 | System and method for efficient VLSI architecture of finite fields |
CN101968732A (en) * | 2010-10-09 | 2011-02-09 | 中国人民解放军信息工程大学 | Bit parallel systolic array shifted polynomial basis multiplier with function of error detection |
CN102314330A (en) * | 2011-09-09 | 2012-01-11 | 华南理工大学 | Composite finite field multiplier |
Non-Patent Citations (6)
Title |
---|
C.-S.YEH,ET AL.,: "Systolic Multipliers for finite Fields GF(2 m)", 《IEEE TRANSACTIONS ON COMPUTERS》, vol. 33, no. 4, 30 April 1984 (1984-04-30), pages 357 - 360, XP000648847 * |
GUO J H, ET AL.,: "Digit-serial systolic multiplier for finite fields GF (2m)", 《IEE PROCEEDINGS-COMPUTERS AND DIGITAL TECHNIQUES》, vol. 145, no. 2, 31 March 1998 (1998-03-31), pages 143 - 148, XP006010827, DOI: doi:10.1049/ip-cdt:19981906 * |
KEE-WON KIM, ET AL.,: "A new digit-serial systolic multiplier for finite fields GF(2 m)", 《2001 INTERNATIONAL CONFERENCES ON INFO-TECH AND INFO-NET. PROCEEDINGS (CAT. NO.01EX479)》, vol. 5, 29 October 2001 (2001-10-29), pages 128 - 33, XP010576975, DOI: doi:10.1109/ICII.2001.983506 * |
SATZODA R K,ET AL.,: "VLSI performance evaluation and analysis of systolic and semisystolic finite field multipliers", 《ADVANCES IN COMPUTER SYSTEMS ARCHITECTURE》, 31 December 2005 (2005-12-31), pages 693 - 706, XP019022401 * |
王飞等: "可配置 GF (2 m) 域 Digit—Serial 乘法器", 《微电子学与计算机》, vol. 21, no. 1, 31 January 2004 (2004-01-31), pages 72 - 74 * |
陈弘毅等: "一种位级流水线乘法器的设计", 《电子学报》, no. 5, 31 May 1992 (1992-05-31), pages 39 - 46 * |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103186360A (en) * | 2013-04-03 | 2013-07-03 | 哈尔滨工业大学深圳研究生院 | Fast arithmetic multi-bit serial pulse dual-base binary finite field multiplier |
CN103186360B (en) * | 2013-04-03 | 2016-08-03 | 哈尔滨工业大学深圳研究生院 | Binary system Galois field multiplier at the bottom of rapid computations many bits series connection pulsation double-basis |
CN104252332A (en) * | 2014-08-20 | 2014-12-31 | 哈尔滨工业大学深圳研究生院 | Multiplier and multiplier processing element for ellipse cipher apparatus |
CN104252332B (en) * | 2014-08-20 | 2018-09-18 | 哈尔滨工业大学深圳研究生院 | A kind of multiplier processing unit and multiplier for elliptic curves cryptosystem device |
CN106201433A (en) * | 2016-07-10 | 2016-12-07 | 北京工业大学 | A kind of Galois field multiplier based on RS code |
CN106201433B (en) * | 2016-07-10 | 2019-01-11 | 北京工业大学 | A kind of Galois field multiplier based on RS code |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Kim et al. | A digit-serial multiplier for finite field GF (2/sup m/) | |
CN103092060B (en) | Time interval measuring system and time interval measuring method which are based on field programmable gate array (FPGA) | |
Meher et al. | Low-Latency, Low-Area, and Scalable Systolic-Like Modular Multipliers for $ GF (2^{m}) $ Based on Irreducible All-One Polynomials | |
CN103793199A (en) | Rapid RSA cryptography coprocessor capable of supporting dual domains | |
Xie et al. | High-throughput finite field multipliers using redundant basis for FPGA and ASIC implementations | |
CN102231101B (en) | Divider and division processing method | |
CN102929574A (en) | Pulse multiplying unit design method on GF (Generator Field) (2163) domain | |
CN102184161A (en) | Matrix inversion device and method based on residue number system | |
CN102184088B (en) | Method and device for realizing finite domain multiplication based on serial and parallel combination | |
CN101964664A (en) | Multi-mode Reed-Solomon decoder structure suitable for CMMB | |
CN107092462B (en) | 64-bit asynchronous multiplier based on FPGA | |
JP3659320B2 (en) | Multiplication module, multiplication inverse element operation circuit, multiplication inverse element operation control system, device using the multiplication inverse element operation, encryption device, error correction decoder | |
Ma et al. | Low-latency factorization architecture for algebraic soft-decision decoding of Reed–Solomon codes | |
Ibrahim et al. | Efficient parallel semi-systolic array structure for multiplication and squaring in GF (2m) | |
CN103176768B (en) | Calculate modular multiplication method and the scalable modular multiplier of classical modular multiplication | |
CN103605494A (en) | Square operation circuit | |
CN102315902A (en) | Universal addressing device and method of quasi-cyclic low-density parity check code | |
CN103186360A (en) | Fast arithmetic multi-bit serial pulse dual-base binary finite field multiplier | |
Alia et al. | On the lower bound to the VLSI complexity of number conversion from weighted to residue representation | |
CN116893800A (en) | Montgomery modular multiplier based on parallel structure and coding | |
Dake et al. | Implementation of high-throughput digit-serial redundant basis multiplier over finite field | |
Wu et al. | Optimized Design of ECC Point Multiplication Algorithm Over GF (2m) | |
CN102253923B (en) | Fast DFT (discrete Fourier Transform) system based on Turbo structure | |
Ram et al. | Delay Enhancement of Wallace Tree Multiplier with Binary to Excess-1 Converter | |
Karunakaran et al. | Exploration on Power Delay Product of various VLSI Multiplier Architectures |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20130213 |