TWI330333B - - Google Patents

Download PDF

Info

Publication number
TWI330333B
TWI330333B TW95143577A TW95143577A TWI330333B TW I330333 B TWI330333 B TW I330333B TW 95143577 A TW95143577 A TW 95143577A TW 95143577 A TW95143577 A TW 95143577A TW I330333 B TWI330333 B TW I330333B
Authority
TW
Taiwan
Prior art keywords
multiplier
gate
polynomial
parallel
montgomery
Prior art date
Application number
TW95143577A
Other languages
Chinese (zh)
Other versions
TW200710715A (en
Inventor
Chiou Yng Lee
Original Assignee
Univ Lunghwa Sci & Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Univ Lunghwa Sci & Technology filed Critical Univ Lunghwa Sci & Technology
Priority to TW095143577A priority Critical patent/TW200710715A/en
Publication of TW200710715A publication Critical patent/TW200710715A/en
Application granted granted Critical
Publication of TWI330333B publication Critical patent/TWI330333B/zh

Links

Landscapes

  • Complex Calculations (AREA)

Description

1330333 九、發明說明: 【發明所屬之技術領域】1330333 IX. Description of the invention: [Technical field to which the invention belongs]

^明係一前限場GF(2m)的特殊種以位元平行心臟收縮陣 哥^利乘法ϋ,尤指-種具魏賴和鱗路複雜, I 之乘法器速度的創新技術。 ^ GF(2 ) 【先前技術】 乘法器』相關發 關於先前技術,我國揭橥於中華民國專利公報中的 明專利技術,較相關者概可列舉如下:^Special species of the GF (2m) of the former line of the Ming Dynasty is a parallel convolution of the heart. The multiplication of the squad, especially the kind of innovation of the multiplier speed of the Wei Lai and the scale road complex. ^ GF(2) [Prior Art] Multiplier Related Issues Regarding the prior art, China has unveiled the patented technology in the Republic of China Patent Gazette, which can be enumerated as follows:

^⑽扁號第獅㈣『有限場G明的細胞陣列次方和電路』發明專 2、公告編號第440789號『乘法器』發明專利案。 3^告,號第况⑹號『⑽元半平行處理式格羅瓦場乘法器之 發明專利案。 』 4、 公告編號第360845號『陣列式乘法器架構及其方法』發明專利案。 5、 公告編號第405086號『快速正規乘法器架構』發明專利案。’、 其中有關有限場GF(2m)的說明如下: >、 在々有限場GF(2m)中’有效的代數運算(含加法、乘法、除法、及指數 等運算)廣泛職應將錯誤更正碼和料技術,軌二粒bch瑪 (Binary BCH Code)之解碼、RS 碼(Reed_s〇1〇_ c〇de)之編碼與解碼 及在安全通信(S謂e Communication)上數位信息的加密與解密(Encrypti〇n and Decryption)等’但儘管如此,GF(2m)的乘法及求反元素的運算仍然相 當複雜,因此針對GF(2m)中的乘法運算陸續有學者提出快速演算法及快速 電路’例如Itoh-Tsuju、Sunar-Koc、Hasan等;而這些乘法器架構均基於特 別的多項式,包括全一多項式(AOP) ’等距多項式(ESp)和三項多項式 等,此外,學者Guo和Wang並發展出二段式乘法器,這是使用一般乘法 單元和降-人方的處理單元所構成的。在有限域(FiniteFieid)(5F(2m)中包括2m 元素,其中m是一正整數,其能被表示成三種主要型態基底,包括:正規 化基底(Normal Basis,NB)、雙重基底(Dua] Basis,DB)和多項式基底 (Polynomial Basis ’ PB);其中若使用多項式基底(PB)表示式,場(Field)元 素火X)eGF(r)透過w向量(n2,...,a。)對於集合κ‘丨, •·5·χ2,χ51}表示為: 5 1330333^ (10) singular lion (four) "limited field G Ming cell array power and circuit" invention special 2, bulletin number 440789 "multiplier" invention patent case. 3^ 告, No. (6) No. (10) Yuan semi-parallel processing type Groovy field multiplier invention patent case. 』 4, Announcement No. 360845 "Array Multiplier Architecture and Method" invention patent case. 5. Announcement No. 405086 "Fast Regular Multiplier Architecture" invention patent case. ', the description of the finite field GF (2m) is as follows: >, in the finite field GF (2m) 'effective algebra operation (including addition, multiplication, division, and index operations) extensive job should correct the error Code and material technology, decoding of Binary BCH Code, encoding and decoding of RS code (Reed_s〇1〇_c〇de) and encryption of digital information in secure communication (S-e Communication) Decryption (Encrypti〇n and Decryption), etc. 'But nevertheless, the multiplication of GF(2m) and the operation of negating elements are still quite complicated. Therefore, for multiplication in GF(2m), scholars have proposed fast algorithms and fast circuits. 'Examples such as Itoh-Tsuju, Sunar-Koc, Hasan, etc.; and these multiplier architectures are based on special polynomials, including all-one polynomial (AOP) 'isometric polynomial (ESp) and trinomial polynomial, in addition, scholars Guo and Wang And developed a two-stage multiplier, which is composed of a general multiplication unit and a descending-human processing unit. In the Finite Field (FiniteFieid) (5F (2m) including 2m elements, where m is a positive integer, which can be represented as three main types of substrates, including: Normal Basis (NB), Double Substrate (Dua) Basis, DB) and Polynomial Basis 'PB; where the polynomial base (PB) representation is used, the field element fire X) eGF(r) is transmitted through the w vector (n2,..., a. ) For the set κ '丨, •······2, χ51} is expressed as: 5 1330333

A(x) = am_txm '+** + a2x2 + a}x + aQ f 中"是卿",)之生成S(generat〇r)且係是在基本域⑼egr〇und 在二因:’哪)的場元素,是唯一(—e)多項式的線性組合; H Γ丨Ϊ 已知此轉之表示式,執行多項式加法和模數作)多項 式孓法,則表示在場(域,Field)元素之階度為m l或更少。 ^者’在超大規模積體電路技術(VLSI)中,陣 ==路线,三個_輯:,_嫌(SystGii〇、、^早 ==)和官道(Pipeline);陣列越理器的優點是由基本細胞(臉 的現存收縮乘法器大多數 (LeaStSignifiC-b^-t) i , 個7C素之積;雖這些收縮乘法器適合於在錯誤更正碼方面應用,但是 縮ίί器對密碼應用而言’則具有很複雜的電路和較長的計算延遲, 去的乘法器的等待時間需要加脈波延遲、學者g⑽為g 的采法為的寺贿間需要2.5m脈波延遲。儘管上面所述低複雜乘法 於料技*,但其電路結構跡。敝縮路嗜 會產生很長轉魏搁。 _ m很錢 伽羅瓦域(GaloisField,GF)或稱有限場(FiniteField)算 Γιι公t共fff碼術系献重要的,尤其是兩個公共錄密碼^統,橢圓 Ξ進算法演算法通”要對多項式基底 列口 ’,:,使用一些廣為流傳的多項式,如全一多項 二 P〇lyn_ls ’ AQP狀三項式(TrinGmials)具有倾雜之電路。 在VLS^設計過程中,分佈於卿)之有限場(触6 &㈣心臟收縮 列式架,之算術運算基本上是取決於_電路,以快速執行計算,它們共 同的性質支持結構化的特性’例如:—致性、輸人/輸出平衡、簡單和有& 則,設計,*大多數的心敝縮_絲法器 二 理且通料列演算法被分類為^符 首先輸入(LSB-F_車列和最高符號位元首入m ^ ^ 些乘法器是繼鑛_咖如% itemi=^);; 6 1330333 結果且不完全糊原有的平行化;因此,它們在全料線浪費在需要一個 大的面積和潛伏延遲(Latency)。最近,又有一個學者李㈣使用定義如全 - (M-〇ne)和等距(EqUally_Spaced)多項式之内積運算去實現有效率的心臟 收,陣列式絲法H ’其具有低延料低複紐的親;但令人遺憾,不 可簡約全-多項式和不可簡約全等空間多項式非f稀有。例如:如 WS100,為全一多項式的階層w的值是不可簡約僅有2 ' 4、川、d、 28、36、52、58、60、66、82 及 1〇〇。 有關分佈於GF(2",)之傳統蒙哥馬利乘法,說明如下: 在1985年,蒙哥馬利乘法,實現有效率的整數模絲法;在1998年, 學者Koc和Acar使用分佈於卿Ί有限場乘法之蒙哥馬利技術,證明模 數乘法,乃由雄)5⑻/r切modPW所定義,其令作)產生㈣2„,)的場问邮 或稱域)且則、S(W〇;r1(x)eGF(n ’特別是,沪㈨表示元素阶)的乘法 反7L素。另學者吳(Wu)對於在GF(2",)位元-平行乘法,係藉由ρ(χ) =/, + 乂 + i 三項式-基底(trinomial-based)產生場(Field,域),且由選擇你)=?條件,則# 得分佈於GF(2",)低複雜性蒙哥馬利乘法器;且因為ρ(χ)與及⑷是相對質數X (relatively prime),及兩多項式與知)存在於科狀切+如巾特 性;因此,J(x)和5(x)在蒙哥馬利乘法可被定義如下: 步驟 1. Γ(χ) = 4(χ)β(χ) 步驟 2· t/(x) = TXx)/^)mod 7?(χ) 步驟 3, (Γ(χ) = (Γ〇) + (/(λ:)Ρ(λ:))/7?(χ)_}ρ(χ) 如前面提到,蒙哥馬利乘法是包含三步驟的複雜算術運算:傳統的乘 法模數乘法和除法。另學者Bajard等人建議分佈於之蒙哥馬利乘 法器,然而,那些乘法器是不規則設計且不適合實現心臟收縮陣列式的架 構。因此,如何對提供一種比其它心臟收縮陣列式乘法器顯示更低硬體複 雜性、更低潛伏延遲(Latency)及具有適合於VLSI系統的有規律相互連結、 模組化架構的乘法器,實為本發明人所欲解決之主要課題。 【發明内容】 本發明之目的係在提出一種分佈於有限場GF(2m)之特殊種類的低複雜 性之位元平行心臟收縮陣列式蒙哥馬利乘法器,其申在yLSI設計中,分佈 於有限場GF(2m)心臟收縮陣列式結構之算術運算,是依靠規則電路,二執 7 1330333 • 行快速計算,其共通性質提供結構化特性,例如一致性、輸入/輸出平衡、 簡單和有規則的設計;而本發明主要是由不可簡約全一多項式(Irredudble All One Polynomials)和不可簡約三項式(irredudble偷⑽恤)之轉換方法, 來貝現分佈於ίλΡ(2Π1)位元之平行心臟收缩^陣列式蒙哥馬利乘法器。 因此’為達上述之目的’本發明乘法器係由全一多項式(八丨丨_〇此 • Polynomial,A0P)和三項式(Trinomial)產生的有限域’以降低場(Field )乘法器的複雜性。上述兩者之表示式描述如下: (一) 全一多項式之表示式 多項式形式/>(1) = ?"+产|+.._ +尤+ 1被稱為全一多項式(八丨1〇1^ _ Polynomial, AOP) ’ 其是不可簡約(irreducible),若且唯若(if and oniy if )w + 1 疋i且2 w +1是本原模數(primitive m〇dui〇)。 (二) 三項式之表示式 一個階度為W的多項式/>(x),若稱是幾乎本原的(aim〇st primitive)或幾 乎不可簡約(almost irreducible) ’則/>(〇) * 〇且P(JC)有一階度r的本原因子 (primitive factor)或不可簡約因子(inducible factor),其中 〇sw — ,且我 們稱作)有指數(exponent) r且遞增(increase) w - r。例如,三項式;c16 + ? + j是 幾乎本原(almostprimitive)有指數13且遞增3,因為: xI6+x3 + 1=(x3+a;2+1)//)(jc) 其中Ζ)(λ:) = χ +x + λ:11+χ9 + χ6 + χ5+χ4+χ2+1是本原的從計 φ 算觀點,在環(r丨ng) Μ(2)/(χΆι)比在域(fleid) GF(2)/j0⑷是更有效率; 、 學者布儉特(Brent)和齊莫爾曼(Zimmermann)論證,三項式(trinomials)具有下 列特性: - i)令形式/ + / + 1的三項式型式,是幾乎本原的多項式(almost primitive polynomials),則有 GCD(w,《) = l。 2)令形式π + / + ι的三項式型式,是幾乎本原的多項式(ahli〇st primitive polynomials),則有 GCD(w,«)是奇數。 【實施方式】 本發明提出一種有限場GF(2m)之低複雜性的心臟收縮陣列式蒙哥馬利 乘法器,該乘法器係對有限場GF(2m)中提供兩種類型,係包括: 第一類型:為全一多項式(All-One Polynomial,AOP)基底乘法器(如第 8 1330333 一圖及第二圖所示); t 請參閱第一圖所示,為全一多項式(An_〇nep〇lynomia丨,AOp)基底乘法 器之電路架構圖’在第一圖中,電路為全一多項式基底乘法器以二項式約 簡實現低複雜性乘法器,其包含如第二圖之U細胞電路架構2丨,其由一個 2-輸入及閘22(AND gate)、一個2-輸入互斥或閘23 (XOR gate)和3個1-位 元栓閘24'25及26伽<:11)組成,因能計算^4〇)5(办-/^〇(1〇:”,+1),因此,第 一圖也可以稱為全一多項式基底(Aii-〇ne p〇iynom丨ai_Based)的蒙哥馬利乘 法器;該乘法器分佈於…⑺,在(χ + 1)ρ+1條件成立下,使用全一多 項式…+ χ + 1之約簡過程,是經常利用二項式 (bmomial)x'" + i進行;所以本發明將對分佈於之全一多項式基底乘 法态(AOP-based multiplier) ’ 使用二項式(binomial)jcm +1 之約簡多頊彳 1 實現低複雜性乘法器。 在有限場(Galois field) GF(2",-1)中,元素处)分佈於㈣2),其元素表示 式為雄)=α„,+ ,應用x,"+1之多項式乘法模數 (polynomial multiplication module)證明如以下定理。 定理1:假設分佈在GF(r-V兩元素是♦)〜〆-,+ ·.·切〆㈣與 = …+ + ¾,其中場(域,Field)由階度之不可簡乡^ 5 Α{χ)^ό 一多項式(irreducible All-One Polynomials ’ ΑΟΡ)所建構。因為?, 王 β(χ)兩者乘積為: m-\ Α(χ)Β{χ) = ^ /=*0 ΙΜ-1 m-1 <ia> \^i+js:even Σ « 7=0 < i+J-odd X' ⑴ 證明.因為= 1,所以j(x)和5(χ)乘積計算如下 m-1 /77-1 (2) Αχ)Β(χ)=ΣΣα<ί-Λχ' /=0 7=0 其中</>>表示ρ模(modulo)rn,且為—整數。 在以下兩種狀況,偶數/和奇數,·分別被討論; 狀況1 :偶數ί 假6又皮數Α:疋一偶數,其乾圍為j.被選擇如 9 1330333 m~\ 1 m-\ i + k :因此,以代入 Xt-vA式中,得 I /77-1A(x) = am_txm '+** + a2x2 + a}x + aQ f where " is qing",) generates S(generat〇r) and is in the basic domain (9) egr〇und in two causes: ' The field element of which is the linear combination of the unique (-e) polynomial; H Γ丨Ϊ is known to represent this transformation, and the polynomial addition and modulus are performed) polynomial method, which means presence (field) The gradation of the elements is ml or less. ^者' In the ultra-large-scale integrated circuit technology (VLSI), array == route, three _ series:, _ suspect (SystGii〇, , ^ early ==) and official (Pipeline); array of the processor The advantage is that the basic cell (the face of the existing shrink multiplier most (LeaStSignifiC-b^-t) i, a 7C product; although these shrink multipliers are suitable for error correction code, but the password is correct In terms of application, it has a very complicated circuit and a long calculation delay. The waiting time of the multiplier to be added requires a pulse delay, and the scholar g(10) is a method of g. The low complex multiplication described above is based on the material technology*, but its circuit structure traces. The shrinkage road will produce a very long turn. _ m very money Galois field (GF) or finite field (FiniteField) Γ ι The public t-fff code system is important, especially the two public recording passwords. The elliptical algorithm is called "to the polynomial base column", :, use some widely-distributed polynomials, such as one A plurality of two P〇lyn_ls 'AQP-like trinoms (TrinGmials) have a miscellaneous In the VLS^ design process, distributed in finite field (contact 6 & (four) systolic columnar, the arithmetic operation is basically dependent on the _ circuit to quickly perform calculations, their common properties support structure The characteristics of the 'such as: - sex, input / output balance, simple and have & then, design, * most of the heart contraction _ silk ruler and the algorithm is classified as ^ first Input (LSB-F_car column and highest symbol bit first into m ^ ^ some multipliers are relays_cafe such as % itemi=^);; 6 1330333 results and not completely paste the original parallelization; therefore, they are The whole line is wasted in need of a large area and latency (Latency). Recently, another scholar Li (4) used the inner product of definitions such as full- (M-〇ne) and equidistant (EqUally_Spaced) polynomial to achieve efficiency. The heart of the collection, the array of silk method H 'which has a low extension of the low complex of the pro; but unfortunately, can not be simple full polynomial and non-simple congruent space polynomial non-f rare. For example: such as WS100, for one The value of the class w of the polynomial is not simple and only 2 '4, Sichuan, d, 28, 36, 52, 58, 60, 66, 82 and 1 〇〇. The traditional Montgomery multiplication distributed in GF (2",) is as follows: In 1985, Montgomery multiplication, to achieve efficient integer modulus Silk method; in 1998, scholars Koc and Acar used the Montgomery technique distributed in the finite field multiplication of the Qing dynasty to prove that the modulus multiplication was defined by the male 5(8)/r modPW, which produced the field of (4) 2 „,) Ask the post or the domain) and S (W〇; r1 (x) eGF (n 'especially, Shanghai (nine) represents the elemental order) multiplication anti-7L. Another scholar Wu (Wu) for the GF (2 ",) bit-parallel multiplication, is generated by ρ(χ) = /, + 乂 + i trinomial-based field (Field, domain) ), and by selecting you) =? condition, then # is distributed over GF(2",) low complexity Montgomery multiplier; and because ρ(χ) and (4) are relative prime X, and two polynomials And knowledge) exists in the cut and the characteristics of the towel; therefore, the multiplication of J(x) and 5(x) in Montgomery can be defined as follows: Step 1. Γ(χ) = 4(χ)β(χ) Step 2 · t/(x) = TXx)/^)mod 7?(χ) Step 3, (Γ(χ) = (Γ〇) + (/(λ:)Ρ(λ:))/7?(χ) _}ρ(χ) As mentioned earlier, Montgomery multiplication is a complex arithmetic operation involving three steps: traditional multiplication modulus multiplication and division. Another scholar, Bajard et al., suggests distributing Montgomery multipliers, however, those multipliers are Irregularly designed and not suitable for systolic array architecture. Therefore, how to provide a lower hardware complexity, lower latency, and suitability for VLSI systems than other cardiac contraction array multipliers Regular phase The multiplier of the interconnected structure and the modular architecture is the main subject to be solved by the inventors. SUMMARY OF THE INVENTION The object of the present invention is to propose a special kind of low complexity distributed in a finite field GF (2m). The bit parallel parallel cardiac contraction array Montgomery multiplier, which is applied in the yLSI design, is distributed in the finite field GF (2m) cardiac contraction array structure arithmetic operation, is dependent on the rule circuit, the second implementation 7 1330333 • fast calculation, Its common nature provides structural features such as consistency, input/output balance, simple and regular design; while the invention is mainly composed of Irredudble All One Polynomials and Irregular Trinomial (irredudble steals) (10) conversion method, the comeback is distributed in the ίλΡ (2Π1) bit parallel heart contraction ^ array Montgomery multiplier. Therefore 'for the above purpose', the multiplier of the invention is composed of all polynomials (eight _〇• Polynomial, A0P) and the finite field generated by Trinomial to reduce the complexity of the field multiplier. Bottom: (1) The expression polynomial form of the all-one polynomial/>(1) = ?"+production|+.._ + especially + 1 is called the all-one polynomial (八丨1〇1^ _ Polynomial, AOP) 'It is irreducible, if and ifiy if w + 1 疋i and 2 w +1 is the primitive m〇dui〇. (ii) The expression of the trinomial is a polynomial of degree W/>(x), if it is said to be almost a primitive (aim〇st primitive) or almost irreducible (almost irreducible) then /> 〇) * P and P(JC) has a primitive factor or an inducible factor of r, where 〇sw — and we call it exponent r and increment (increase) ) w - r. For example, the trinomial; c16 + ? + j is almost primitive (almostprimitive) has an exponent of 13 and is incremented by 3 because: xI6+x3 + 1=(x3+a;2+1)//)(jc) where Ζ )(λ:) = χ +x + λ:11+χ9 + χ6 + χ5+χ4+χ2+1 is the calculation of the primitive φ, in the ring (r丨ng) Μ(2)/(χΆι) More efficient than in the domain (fleid) GF(2)/j0(4); scholars Brent and Zimmermann argue that trinomials have the following characteristics: - i) form The trinomial form of / + / + 1 is an almost primitive polynomials, and there is GCD(w, ") = l. 2) The trinomial form of the form π + / + ι is an almost primitive polynomial (ahli〇st primitive polynomials), and GCD(w,«) is an odd number. [Embodiment] The present invention proposes a low-complexity systolic array Montgomery multiplier with a finite field GF (2m), which provides two types in a finite field GF (2m), including: : All-One Polynomial (AOP) base multiplier (as shown in Figure 8 1330333 and Figure 2); t See the first figure, which is a full polynomial (An_〇nep〇 Lynomia丨, AOp) Circuit architecture diagram of the base multiplier 'In the first figure, the circuit is a full polynomial base multiplier to implement a low complexity multiplier with binomial reduction, which comprises a U cell circuit as shown in the second figure Architecture 2丨, consisting of a 2-input and gate 22 (AND gate), a 2-input mutex or gate 23 (XOR gate) and three 1-bit latches 24'25 and 26 gamma <:11) The composition, because it can calculate ^4〇)5 (do-/^〇(1〇:", +1), therefore, the first picture can also be called the all-one polynomial base (Aii-〇ne p〇iynom丨ai_Based) Montgomery multiplier; the multiplier is distributed in (7), and the (χ + 1)ρ+1 condition is established, using the all-one polynomial...+ χ + 1 reduction process is Using the binomial x'" + i; therefore, the present invention will use the binomial jcm +1 reduction for the AOP-based multiplier distributed over it.顼彳1 Implement a low complexity multiplier. In the finite field (Galois field) GF (2", -1), the element is distributed in (4) 2), whose element expression is male) = α„, + , apply x , " +1 polynomial multiplication module proves the following theorem. Theorem 1: Assume that the distribution is in GF (rV two elements are ♦)~〆-, + ···cutting (four) and = ...+ + 3⁄4, where field (field) is not simplistic by the degree ^ 5 Α{ χ)^ό A polynomial (irreducible All-One Polynomials ' ΑΟΡ) constructed. because? , the product of king β(χ) is: m-\ Α(χ)Β{χ) = ^ /=*0 ΙΜ-1 m-1 <ia> \^i+js:even Σ « 7=0 < i+J-odd X' (1) Proof. Because = 1, the product of j(x) and 5(χ) is calculated as m-1 /77-1 (2) Αχ)Β(χ)=ΣΣα<ί- Λχ' /=0 7=0 where </>> represents a modulo modul and is an integer. In the following two cases, even/and odd numbers are discussed separately; Condition 1: Even ί False 6 and skin number Α: 疋 an even number, whose circumference is j. Selected as 9 1330333 m~\ 1 m-\ i + k : Therefore, to substituting Xt-vA, get I /77-1

/Μ-I Σω<,-7Α = Σ ;接著,令變數灸是一奇數’其範圍為 k=0 k~evcn i — k — \ /u λ =<-> 代 入 2/Μ-I Σω<,-7Α = Σ; Next, let the variable moxibustion be an odd number' with a range of k=0 k~evcn i — k — \ /u λ =<->

I m m-\ Σ〜,Α+ Έα<'- ./=0 J> 式中,得 +]I m m-\ Σ~, Α+ Έα<'- ./=0 J> where, get +]

—I 2__ w-1 m-\ Σβ<',Α + Έα<'-Λ = Σa b /=0 I I /77-1 I k=odd ^^ ) > < ) > • 户疒―H+1 * - 狀況2 :奇數,· 足 假設變數A:是一偶數,其範圍為; 7•被選擇如 / + 1 m-\ • 1 <j<m-\ M 0<j< -——;因 jit ,以 =<^Azl> 滿 /-1 a _. 2 nj~\ nj-\ Σα<.-Λ + Σ^'-νΑ 式中,得Σβ<,·νΑ+ Σα<,-)Α = Σ',_\卜卜i.;接 y =0 _/+1 _ I m-\ I 7=0 广ί + 1+j^/j /77-1—I 2__ w-1 m-\ Σβ<',Α + Έα<'-Λ = Σa b /=0 II /77-1 I k=odd ^^ ) >< ) > • Household 疒H +1 * - Condition 2: Odd, · Foot hypothesis variable A: is an even number, the range is; 7• is selected as / + 1 m-\ • 1 <j<m-\ M 0<j<-; because jit, ==lt;^Azl> full /-1 a _. 2 nj~\ nj-\ Σα<.-Λ + Σ^'-νΑ where &β<,·νΑ+ Σα<, -)Α = Σ', _\卜卜i.; y =0 _/+1 _ I m-\ I 7=0 广ί + 1+j^/j /77-1

/+1 m-\ I 户了叶丁 I/+1 m-\ I have a leaf I

I m-Ι I k=even k=Q <-> <--> 著,令變數A:是一奇數,其範圍為o^hw-l,_/被選擇如滿足 / + 1 m-\ , nr i + k /+1 i + \ ^ m~\ j /77-1 Σα<,^Α= Σ" 6 ,+*如上述表明’ d和万oo的最後乘法能被表示如 i=/+l k=\ <1_> <^_> 一 7 k=odd 10 1330333 m-\ A{x)B{x) = Yj /»0 m-\ /M-1 Σ a^,b X· 、/+y.=mwI m-Ι I k=even k=Q <-><--> Let the variable A: be an odd number whose range is o^hw-l, _/ is selected as satisfying / + 1 M-\ , nr i + k /+1 i + \ ^ m~\ j /77-1 Σα<,^Α= Σ" 6 ,+* as indicated above, the final multiplication of 'd and oo oo can be expressed as i=/+lk=\ <1_><^_> A 7 k=odd 10 1330333 m-\ A{x)B{x) = Yj /»0 m-\ /M-1 Σ a^, b X· , /+y.=mw

sodJ g對每—y=h户午'1 VV,定義一個行向量Mumn她r) /+j=even (3) 為W'=[冰n W…W ί’sodJ g for every -y=h household lunch '1 VV, define a row vector Mumn her r) /+j=even (3) for W'=[冰n W...W ί’

L 〇,Ά,',,…(w_",'J 其中 若 z_ + y = even (偶數), (4) 若 z_ + 7’ = 〇dd (奇數),則', /«-1 在行向量W,所有項總合等於f a b ” u h B , y=0 Ά <ψ 7=f"啤人# 且 W,在 m X W 矩 f+j=evcn i+j=ocid 2 2 陣1[%]中,表示第,·行向量。其中w矩陣的結構表示如下: W: 1 X ... xm-X 'M;0,0 ^0,1 ^0,(/^-1) Wl,0 . · _W(m-”,0 w("卜丨),0 … vt^ (5) 不式中,»此的錄,分观被 Θ數),則在谈 簡的乘法,是規律的和簡單的,且藉由完全利广 乃利;以之方法; 令m是一正整數,且令 ^(0 -q + i(jn _ m〇(^ m (6) 1330333 其中lSMm-l、〇L“-l且>。假設若”的值被固定於 且GCO(«,m) = l ’則由,吨)被排列於完整殘餘集合 〇^1(11^561;)丨0,丨,2,_’,》?-1}。觀察在式(5)之矩陣评,第/行向量%,假設/被 固定於OUw-Ι,則 1) 若 / + y = even (偶數),則 << ΐ±ΐ > + < ίζΐ ·。 2 2 2) 若,+ y = 〇dd (奇數),則 «[ + j±} > + < Lzlzl .·。 2 2L 〇,Ά,',,...(w_",'J where z_ + y = even (even), (4) if z_ + 7' = 〇dd (odd), then ', /«-1 is in the line Vector W, the sum of all items is equal to fab ” uh B , y=0 Ά <ψ 7=f"Beer# and W, at m XW moment f+j=evcn i+j=ocid 2 2 array 1[% In the middle, it represents the first, and the row vector. The structure of the w matrix is expressed as follows: W: 1 X ... xm-X 'M; 0,0 ^0,1 ^0,(/^-1) Wl,0 · _W(m-", 0 w("卜丨),0 ... vt^ (5) In the formula, »this record, the number of points is counted), then the multiplication of the simple, is regular And simple, and by the full benefit of the method; to make m a positive integer, and let ^ (0 -q + i(jn _ m〇(^ m (6) 1330333 where lSMm-l, 〇L "-l and>. Assume that if the value of "is fixed" and GCO(«,m) = l ', then, ton) is arranged in the complete residual set 〇^1(11^561;)丨0,丨, 2, _', 》?-1}. Observe the matrix evaluation in equation (5), the line/row vector %, hypothesis/fixed to OUw-Ι, then 1) if / + y = even (even) , then << ΐ±ΐ > + < ίζΐ ·. 2 2 2) If, + y = 〇dd (odd), then «[ + j±} > + < Lzlzl .. 2 2

使用方程式(6),上述特性獲得如下·· 一] »-< q-l· π(ι) > ° i — j — \ 7:— »=< q + π(ί) > ο a) 若 i + _/ = even (偶數),|jj <π<ί±1>+π<ί 2 b) 若/ + _/ = odd (奇數),則 <π< 1 +Χί1 >+π< 2 c) 右 / =所一 1 ,貝1J < 分 + >= ζη-1。 因此,應用上述特性,方程式(1)能被重寫如下: m-]Using equation (6), the above characteristics are obtained as follows: one] »-< ql· π(ι) > ° i — j — \ 7: — »=< q + π(ί) > ο a) If i + _/ = even (even), |jj <π<ί±1>+π<ί 2 b) If / + _/ = odd (odd), then <π< 1 +Χί1 >+ π< 2 c) Right / = 1, Bay 1J < min + >= ζη-1. Therefore, applying the above characteristics, equation (1) can be rewritten as follows: m-]

Α(χ)Β(χ) = Σ (Σ \ l+Jx /=0 7=0 π(<—> /Τ(<—>) /+j=even ~ 其中妒 + Σ 7=0 <π{<· ^ . /+7 + 1 ^ /-/_] X >7(<—^―> /r(<-^L_L>) / + j-ndd "7-1 = LW<^n>x<q"{i)> /=0 /n-l w ’ <tf+/r(t)> =J »» /+j^even 2 ' m-1 y=0 / + js=Q(JJ ⑺ Σ ’現在,考慮蒙哥 馬利乘法’計算你)^;^-、。^"^)。 Β(χ)^ = Β{χ)χ~η mod(^ +1) = Σύ<^>χ' ⑻ 清楚地,β〇〇<-")是藉由SCc)向循環地移動《位置,即在元素β(^由代 <> + «>進入\的下標;因此,火;cW^fmodO^+l)基於方程式⑺能被求 得,如 12 1330333 A(x)B(x)x~" m〇cj(x»> += A(x)B(x)^n) m-1 m~\ =Σ ( Σ « b ^ y a b (9) 1+ J =0</</ 藉由兩元素移動排列計算⑺x-nmod(xm+l),其實施例可參閱例2 所述 第一#型·二項式(Trinomiai)基底乘法器,且具有GCD)=1條件所產 生的有限場,以實現低複雜性的乘法器。Α(χ)Β(χ) = Σ (Σ \ l+Jx /=0 7=0 π(<->/Τ(<->) /+j=even ~ where 妒+ Σ 7= 0 <π{<· ^ . /+7 + 1 ^ /-/_] X >7(<-^―>/r(<-^L_L>) / + j-ndd " 7-1 = LW<^n>x<q"{i)> /=0 /nl w ' <tf+/r(t)> =J »» /+j^even 2 ' m-1 y =0 / + js=Q(JJ (7) Σ 'Now, consider Montgomery multiplication' to calculate you)^;^-,.^"^). Β(χ)^ = Β{χ)χ~η mod(^ + 1) = Σύ<^>χ' (8) Clearly, β〇〇<-") is cyclically shifted by SCc), ie, in element β (^ by generation <> + «&gt Enter the subscript of \; therefore, fire; cW^fmodO^+l) can be obtained based on equation (7), such as 12 1330333 A(x)B(x)x~"m〇cj(x»> + = A(x)B(x)^n) m-1 m~\ =Σ ( Σ « b ^ yab (9) 1+ J =0</</ Calculated by two-element shift arrangement (7)x-nmod( Xm+l), the embodiment of which can refer to the first #-type binomia base multiplier described in Example 2, and has a finite field generated by the GCD)=1 condition to realize a low complexity multiplier .

有關不可簡約三項式x",+ χ» +】形式且具GCO(w,w)=丨之位元·平行心臟 收縮陣列式蒙哥馬利乘法器,學者布倫特與齊莫爾曼(Brent細 mmermann) 〇jt4 ^(almost primitive trinomials) xm +^ + 1 ^ ^ f ’滿足GC£»(m,”) = i的條件,且大多數不可簡約也滿足^^⑶⑼=i的條 ^ ’因=,本發明使用三項式作)=?,",+1形式且具⑽(叫=1,以減少 ,理過私’而可得到低複雜性位元·平行罐收、轉列式蒙哥馬利乘法器。 :雄)=〜-丨x ’+· · m %與離)=6",—〆,-丨+…+&+是分佈於GF(n ^ 4,其中場(域)從不可簡約三項式(irredu敲tri_ial)pw 且具GC£>(w,《) = 1建構。假設乘積叩)About the non-simple trinomial x", + χ» +] form with GCO(w,w)=丨位元·parallel systolic array Montgomery multiplier, scholar Brent and Zimmerman (Brent fine Immermann) 〇jt4 ^(almost primitive trinomials) xm +^ + 1 ^ ^ f 'Meet the condition of GC£»(m,") = i, and most of them are not simple and satisfy the ^^(3)(9)=i bar ^ ' =, the present invention uses the trinomial formula) =?, ", +1 form and (10) (call = 1, to reduce, over-private', and can obtain low complexity bits, parallel cans, and derivatives Montgomery multiplier. :Male)=~-丨x '+· · m % and off)=6",—〆,-丨+...+&+ is distributed over GF(n^4, where field (domain) From the incomprehensible trinomial (irredu knock tri_ial) pw and with GC£>(w, ") = 1 construct. Suppose the product 叩)

+ xn+l + V + r0是/⑻與5(χ)的一 〜,一产2 + . 般乘法, ' 其中+ xn+l + V + r0 is a / (8) and 5 (χ) one ~, one production 2 + . General multiplication, 'which

^m-\ " a0^m~\ + a\^m~2 ^----+ am 々Q tm = a^m-\ + a2^m~2 + '' * + 〇 A hm~2 ~ am-\^m~\ 考慮中間乘法分解如下0 T (x) = 7J (x) + T2 + (jc)jcm+,i 13 (10) 1330333 其中 G W = ’w+”HXm 丨 +. _ · +。丨I + (,^m-\ " a0^m~\ + a\^m~2 ^----+ am 々Q tm = a^m-\ + a2^m~2 + '' * + 〇A hm~2 ~ am-\^m~\ Consider the intermediate multiplication decomposition as follows: 0 T (x) = 7J (x) + T2 + (jc)jcm+, i 13 (10) 1330333 where GW = 'w+"HXm 丨+. _ · +丨I + (,

Ti(x) = t2m.2xm n 2 + · ·· + tni+ll+ix + tm+ll 給予中間乘積你)5⑺=+ 7Ά)Λ” + K,Koc與Acar兩位學者使 用全一多項式之約簡過程,利用二項式(binomial)xm+l ·,進行蒙哥馬利乘 法 d(x)5(x)x-” m〇d(x",+ 丨)計算,得式(11)如下 A(x)B(x)x~n m〇d(xm +1) = T' ^ + Τ^χ~ _+ 忑㈡(χ +” = Τ2{χ) + Τ^χ)χη,+T,{x)xm-n (11)Ti(x) = t2m.2xm n 2 + · ·· + tni+ll+ix + tm+ll Give the intermediate product you) 5(7)=+ 7Ά)Λ” + K, both Koc and Acar use the all-one polynomial The reduction process, using the binomial xm+l ·, is performed on the Montgomery multiplication d(x)5(x)x-" m〇d(x", + 丨), and the equation (11) is as follows: x)B(x)x~nm〇d(xm +1) = T' ^ + Τ^χ~ _+ 忑(2)(χ +" = Τ2{χ) + Τ^χ)χη,+T,{x )xm-n (11)

= T2(x) + T,(x) + Tt(x)xm-n 所以’若利用三項式x",+x"+l在蒙哥馬利乘法計算,能被表示如下: ♦)離K ” ηκκ1(χ”Ά D =独碰二 + ±^±i) =T2 (χ) + r3 (χ)χιη + Tx {x)xm~n + Τχ (x) (12) =(7; (x) + T,(x) + η (x)xm-/l) + (Τ, (χ) + τ3(χ)χη) = K(x) + G(x) 其中 (13) (14) Κ(χ) = Τ2 (χ) + Τ3(χ) + 7; (x)xm = /J(x)5(jcKnmod(xn,+l) G(x) = 7j(x) + Τ3(χ)χη 從Λ:⑺和G〇)關係’(?(χ)每一關係是包含在多項式/:(χ)中;換言之,多 項式G(x)能從尺(χ) =水x)S(x)x_”mod(xm+l)計算被選用;因此,由式(13): 火〇〇 = γ4(χ)5(;φ:_η mod(xm +1)及式(9). A(x)B(x)x~" mod(xm +1) m-\=y ( Σ a /-76 … ώ ;=〇 小了> <歸《了 /+j-evcn /77-1Σ- b y-〇 1+户 t-j~\ 1330333 κ(χ)=Σ (Σ α t'~Jb π(<—-> ‘ m-l /-0 7*0 i+j^ev&i w-1 o,+小力>,> > <η+π{< )> (15) Σ灸. (=0= T2(x) + T,(x) + Tt(x)xm-n So 'if the trinomial x", +x"+l is calculated in Montgomery multiplication, it can be expressed as follows: ♦) from K ” ηκκ1 (χ"Ά D = unique touch two + ±^±i) =T2 (χ) + r3 (χ)χιη + Tx {x)xm~n + Τχ (x) (12) =(7; (x) + T,(x) + η (x)xm-/l) + (Τ, (χ) + τ3(χ)χη) = K(x) + G(x) where (13) (14) Κ(χ) = Τ2 (χ) + Τ3(χ) + 7; (x)xm = /J(x)5(jcKnmod(xn,+l) G(x) = 7j(x) + Τ3(χ)χη From Λ: (7) and G〇) relationship '(?(χ) Each relationship is included in the polynomial /:(χ); in other words, the polynomial G(x) can be from the ruler (χ) = water x) S(x)x_”mod The (xm+l) calculation is chosen; therefore, from equation (13): fire 〇〇 = γ4(χ)5(; φ:_η mod(xm +1) and equation (9). A(x)B(x )x~" mod(xm +1) m-\=y ( Σ a /-76 ... ώ ;=〇小了>< returned to /+j-evcn /77-1Σ- b y-〇 1+ household tj~\ 1330333 κ(χ)=Σ (Σ α t'~Jb π(<--> ' ml /-0 7*0 i+j^ev&i w-1 o,+ small Force >, >><η+π{<)> (15) Acupuncture. (=0

X 因此,多項式尺0)被求得,如式(16): (16) 你)=<+痛夕+· + + …+ I·—,〆"。,-1), 同理G(x),可得式(17): 你)=痛> + g<f_⑴〆…(1 7)X Therefore, the polynomial ruler 0) is obtained, as in equation (16): (16) you) = < + pain eve + · + + ... + I·, 〆 ". , -1), the same as G(x), available formula (17): you) = pain > + g<f_(1)〆...(1 7)

而且’若0以-2 ’則在G(X)每一關係項〜+π(ω>,能由λ<</+冲+η>計算選 用;總之’由三項式(trinomials)建立位元-平行心臟收縮陣列式蒙哥馬利乘 法器,如以下步驟實施: 步驟1.因在方程式(11)函數Λ:(χ)由 尺0) = (AW + GO) + ” = X(x)s(x)x_” mod(xm +1)被定義,因此’第一 圖能被使用至實現尺(X)計算。 步驟2.分別在方程式(16)以及(17),從尺(X)和G(x)兩者的關係,第一圖 能被重新配置產生兩多項式尺⑺和G(x),顯示於第三圖之乘法單元。 步驟3.最後計算火(X)和G(x)之總和,完成總和步驟。換言之,這乘積,Moreover, if '0 to -2', then each relation term of G(X) is ~+π(ω>, which can be calculated by λ<</+rush+η>; in short, it is established by trinomials. The bit-parallel systolic array of Montgomery multipliers is implemented as follows: Step 1. Because of the function in equation (11) Λ: (χ) by ruler 0) = (AW + GO) + ” = X(x)s (x)x_" mod(xm +1) is defined, so 'the first graph can be used to implement the ruler (X) calculation. Step 2. In equations (16) and (17), respectively, from the relationship between the ruler (X) and G(x), the first map can be reconfigured to produce two polynomial rulers (7) and G(x), which are shown in The multiplication unit of the three figures. Step 3. Finally calculate the sum of fire (X) and G(x) and complete the sum step. In other words, this product,

D⑴=雄)5〇)χ-η mod(xm + / +1) ^<ι/+π(0)>Χ <"+/了(0)> + d <</十D(1)=male)5〇)χ-η mod(xm + / +1) ^<ι/+π(0)>Χ <"+/了(0)> + d <</ ten

X 則有 - A<g+;r<巾 +0<(/+;γ(φ mod2, 若0 < i < m - 2 則 <9+咖-丨>> = 所以三項式(trinomials)建立位元-平行心臟收縮陣列式蒙哥馬利乘法 器,其可參閱下述實施例3說明。綜上所述,本發明有三個實施例,敘述 如下: 例1 :令在(3F(24)兩元素,已知雄) = $>〆和= 。假設 ,=0 1=0 15 1330333 如上述表明,此二項式(作為全一多項式之約簡)的蒙哥馬利乘法能 被總結如下。 1) 置於位置(_/,0之關係項(term),如關係項(y,,·)指示,關係項(y,/)所有 係數是對角關係項(term)的係數。舉例,關係項(2,2)是α九、關係項(1,1)為6, 和(3,3)為响包含%係數。與此類似,關係項(3,1)為认、關係項(2,2)為响 和關係項(1,3)為包含匕係數(參考計算式(三))。 2) 全一多項式蒙哥馬利乘法,藉由伞)风x)x-"m〇d(x»,+1)計算,能被從 二項式-基底乘法(binomial-based multiplication)獲得。正如,第一圖也能提 供mod(x",+1)的乘法’由兩向量㈨外…凡,)和(6。,&|,·. A |)轉移成 兩向量(“聊〜”,...,‘_|>)和(6—。)>,6<_>,〜九_^ 從以上總結’在第一圖電路因能計算火m〇d(x",+丨);因此,第一 圖也被叫為全一多項式-基底(All-〇ne p〇lynomial_Based)的蒙哥馬利乘法 器。 例3 :我們將使用佈於<^(2)本原三項式(primitive trin〇mial)? + / + 〗去 說明此新奇佈於⑶⑺位元-平行心臟收縮陣列式蒙哥馬利乘法器。令兩元 素如)=«4χ4 + α3χ3 + α2χ2 + % + 與 5(x) = V4 + V3 + V2 + 4 + 6。是場(fldd,亦 稱域)GF(25)由分佈於奶2)之本原三項式(primitive trin〇miai) / + /+1產 生。蒙哥馬利乘法由計算尺(x) =雄)B(JC)x-2m〇d(x5+1)被表示如下: aib3 \^3\ a4bi a2b1 a2b4X has - A < g +; r < towel + 0; (/ +; γ (φ mod2, if 0 < i < m - 2 then < 9 + coffee - 丨 >> = so trinomial (trinomials) A bit-parallel systolic array Montgomery multiplier is established, which can be referred to the following description of Embodiment 3. In summary, the present invention has three embodiments, which are described as follows: Example 1: Let (3F(24) ) Two elements, known as male) = $>〆 and =. Assume, = 01 = 15 15 1330333 As indicated above, the Montgomery multiplication of this binomial (as a reduction of the all-one polynomial) can be summarized as follows. ) placed in the position (_ /, 0 relationship term (term), such as the relationship term (y,, ·) indicates that the relationship term (y, /) all coefficients are the coefficients of the diagonal relationship term (term). The term (2, 2) is α IX, the relationship term (1, 1) is 6, and (3, 3) is ringing containing the % coefficient. Similarly, the relationship term (3, 1) is the recognition and relationship term (2). 2) The ringing and relationship term (1, 3) is the inclusion factor (refer to the calculation formula (3)). 2) The all-one polynomial Montgomery multiplication, by umbrella) wind x) x-"m〇d(x The », +1) calculation can be obtained from binomial-based multiplication. Just as the first figure can also provide mod(x", +1) multiplication 'by two vectors (nine) outside... where,) and (6., &|, ·. A |) are transferred into two vectors ("Language~ ",...,'_|>) and (6-.)>,6<_>,~9_^ From the above summary, the circuit in the first figure can calculate the fire m〇d(x", +丨); Therefore, the first map is also called the Montgomery multiplier of the All-〇ne p〇lynomial_Based. Example 3: We will use the <^(2) primitive trin〇mial? + / + 〗 to illustrate this novelty in the (3) (7) bit-parallel systolic array Montgomery multiplier. Let the two elements be as follows ==4χ4 + α3χ3 + α2χ2 + % + and 5(x) = V4 + V3 + V2 + 4 + 6. The field (fldd, also known as the domain) GF(25) is produced by the primitive trin〇miai / + /+1 distributed in milk 2). Montgomery multiplication is represented by the slide rule (x) = male) B(JC)x-2m〇d(x5+1) as follows: aib3 \^3\ a4bi a2b1 a2b4

I 84^4 1II 84^4 1I

Sgbi 32^0 33〇b0 a2^2 [^3^4 I a〇b〇 I aib2 + 3gb^ 3(3^2 agb〇 3-)13〇 ^<1+/Γ(1)> ^<1+Λ·(2)> ^<1+Λ·(3)> ^<1+/γ(4)> 計算式(四) 1330333 • 從上述乘法’多項式G⑺基於方程式(17),能被表示如: G(x) = 〇〇b0 + (α,ό0 + a0bx)x + («3ό4 + α463)χ2 + aJ)Ax3 =(αφΑ + a4b3)x<U!!W> + + aAb,x<Unm> + (α,6〇 + a〇b,)x<Uir{i)> • 顯而易見,在GW中心丨+Λ(,)>每一項目能被發現且存在尺(χ)中々<ι+τ〇♦每 一項目0 其中使用第一圖的蒙哥馬利乘法器來實現三項式-基底(trin〇mial_Based) - 的蒙哥馬利乘法,如第三圖所示’其包括乘法單元31和總和單元32 ;其中 乘法單元31包含第四圖之V細胞33與第五圖之Q細胞34,如第四圖中所 φ 示,V細胞電路架構41之組成係由一個2_輸入及閘42(AND gate),二個2· 輸入的互斥或閘43、44(XOR gate)和四個1位元栓閘(Latch)45、46、47及 48 ;並且如第五圖中所示Q細胞電路架構51,其組成由一個2_輸入及閘 52(ANDgate)’ 一個2-輸入的互斥或閘53(X〇Rgate)和四個i位元栓閘54、 55、56及57(Latch)。所以在第三圖乘法單元31的結果,是基於兩個基本 的細胞(V細胞33及Q細胞34),產生兩多項式尺⑷和^⑺;而且,由 ,係數g<_'>是容易在第(,’+ι)行被產生;最後,在第三圖的總合單 兀U,係由m個W細胞35組合,以執行尺⑴和研々之加總和;如第六圖 顯示,每一w細胞電路結構61係包含一個2_輸入的互斥或閘62(x〇R职把) 卜個丨位元栓閘63(Latch)去完成、+ Φ 二圖使用三種細胞類型(V細胞33、Q細胞34及W細胞35)進行由三項 - 式(Tnnomial)之蒙哥馬利乘法。如上所述,本發明之平行心臟收縮陣列式蒙 哥馬利乘法II僅需m + l延遲,且最大的計算延遲在每細胞裡需要—個2·輸 入及閘52(ANDgate)和一個2-輸入互斥或閘53(x〇Rgate)延遲。 ..’τ'上所述,種有限% GF(2 )的特殊種類之位元平行心臟收縮陣列式 蒙哥馬利乘法器,其係對有限場GF(n中之特殊多項式m,產生簡單乘 法器’以應用於全一多項式(Α0Ρ)Ρ(χ) = χηι_ι+χΐη_2+ +χ + ι之約簡過程其 中所有元素的係數是等於0或丨,該乘法器包括一個乘積單元· 八 該乘積單元的電路結構是由(m+1)2相同_細胞所組成,以形成㈣ X (m+1)陣列,該每- u細胞包含有至少三個輸入信號線及至少三個輸出作 號線,及該每- U細胞包含一個2輸入及閘、一個2輸入互斥或閑和三個 1330333 1位元栓閘組成; 其中該乘法器的電路特性係對有限場GF(2m)中之_ _ 二元素B進行乘積運算,以得到__第三元素c,其 飞素A與-第 多項式基底(Κ...,α™-|)之表示式,該有限場gf〇為=、B及C是以 式X’’’ + 1所產生的,及α為該不可分解的多項式’、、可分解之多項 吵) = + y表示第y.項傘刺^的g乘積=產生 ^4(χ)β(Λ〇 之乘積為火x)jB(JC) = g /*=0 m~\ OT-| Σ w a a:‘之計算· 該計算式 f α 6 + f , 、 >=〇 午令户? 與定義有一個行向旦 I+J^ndd - 2 〜问里 vector)為 W,=K,Wi ;,…5MWi)/。 其中弟ί行向量w = [Μ,1,假士ο / ,. . 丁门里L/,J非又如,+户_ (偶數),則在 係數^>及,分別地被係數w 和 J> 5 2 <丁> ㈤丨).('+丨)的表示式確定。 其中第/行向量w = [〜],若…=〇 ,.八…… ^數)則在〜表示式中,係 \i+J=even (colu imn 數a <一>及气㈠_丨> ’分別地被係數>Vu洳 ^ + 2 丁 (厂1)(叫和的表示式確定„ 其中該乘法器的每一 U細胞是執行第 (偶數),則進行ίΓ , =ί/切, 弟y列其中右i + 7、even aJ+Ul "+«,W><(,+7>/2> 的計算,或 ^ , . even 則進行‘、+fl<(,+7+,)/2>Wi他的計算。k 4右…=〇dd (奇數), 其中該全一多項式基底/^)=χίΒ-, + γ_2 中約簡’以獲得雄)5(X)X-',mody + J)的計首。1 從一項多項式+ ( 其中該有限場GF(2m)的所有元夸θ:。 的,使絲法器輯算延龍f要㈣^^分解的全—多似所產生 其中該有限場GF(2m)的所有元素是尤。 生的,能提供♦)離)x-”m〇d(m)的”刀解的二項多項式〜所產 (WA-丨)轉移成兩向量(α_.α · a ,則由兩向量(a。,〜…气-,)和 其中該有限場GF(2m)的所右开本β』 <”+_>,—(丨)>,’九帅卜,〇。 有几素々由何分_三衫項式 所產生的’且GC£)(m,«) = l(相對最大公 數〜1 ) ’ 則雄)5(χ)χ-” mod(xm + ^+1 Η) 20 1330333 之计异,可由函數&(X)獲得,其中尺⑺二你)的及 作)=,2 (X) +,3 (X) + 7; (X)产。 其中該有限場GF(2m)的所有元素是由不可分解的三項多項式χ„, + / + i 所產生的’且GC’.l ’ g卩函數你)每—關係是包含在多項式你);亦 即多項式G〇)能從你)=♦冲小―"mQd(m)計算獲得。 其中该有限% GF(2m)的所有元素是由不可分解的三項多項式γ 所產生的’且GCZ)(W,„) = 1 ’由你)和咏)之完成總和計算;該乘積 Z)(x) = J(x)5(x)jT” mod(xm + χη +!)可得 £>(χ) = d.Sgbi 32^0 33〇b0 a2^2 [^3^4 I a〇b〇I aib2 + 3gb^ 3(3^2 agb〇3-)13〇^<1+/Γ(1)> ^ <1+Λ·(2)>^<1+Λ·(3)>^<1+/γ(4)> Calculation formula (4) 1330333 • From the above multiplication polynomial G(7) based on the equation (17), can be expressed as: G(x) = 〇〇b0 + (α, ό0 + a0bx) x + («3ό4 + α463) χ 2 + aJ) Ax3 = (αφΑ + a4b3)x<U!!W&gt + + aAb, x <Unm> + (α,6〇+ a〇b,)x<Uir{i)> • Obviously, every item in the GW center 丨+Λ(,)> can be found and There is a ruler (χ) 々<ι+τ〇♦Each item 0 where the Montgomery multiplier of the first figure is used to implement the Montgomery multiplication of the trinomial-based (trin〇mial_Based), as shown in the third figure' It comprises a multiplication unit 31 and a summation unit 32; wherein the multiplication unit 31 comprises the V cell 33 of the fourth figure and the Q cell 34 of the fifth figure, as shown in the fourth figure, the V cell circuit architecture 41 consists of a 2_AND AND gate 42 (AND gate), two 2· inputs of mutual exclusion or gate 43, 44 (XOR gate) and four 1-bit latches (Latch) 45, 46, 47 and 48; and as in the fifth Shown in the figure The Q cell circuit architecture 51 consists of a 2-input and gate 52 (ANDgate) a 2-input mutex or gate 53 (X〇Rgate) and four i-bit latches 54, 55, 56 and 57 ( Latch). Therefore, the result of the multiplication unit 31 in the third graph is based on two basic cells (V cell 33 and Q cell 34), producing two polynomial scales (4) and ^(7); and, by the coefficient g<_'> is easy In the first (, '+ι) line is generated; finally, in the third figure, the total unit 兀U is combined by m W cells 35 to perform the sum of the ruler (1) and the mortar; as shown in the sixth figure Each w cell circuit structure 61 includes a 2_input mutual exclusion or gate 62 (x〇R job). A bit of the gate locks 63 (Latch) to complete, + Φ two maps use three cell types (V Cell 33, Q cell 34 and W cell 35) were subjected to Montgomery multiplication by Tnnomial. As described above, the parallel cardiac contraction array Montgomery multiplication II of the present invention requires only m + l delay, and the maximum computational delay is required in each cell - 2 input and gate 52 (ANDgate) and a 2-input mutual exclusion Or gate 53 (x〇Rgate) is delayed. .. 'τ', a special type of finite element GF(2) parallel systolic array Montgomery multiplier, which is a pair of finite field GF (special polynomial m in n, yielding a simple multiplier' To apply to the all-one polynomial (Α0Ρ)Ρ(χ) = χηι_ι+χΐη_2+ +χ + ι reduction process in which the coefficients of all elements are equal to 0 or 丨, the multiplier includes a product unit · eight circuits of the product unit The structure is composed of (m+1)2 identical cells to form a (iv) X (m+1) array, the per-u cell comprising at least three input signal lines and at least three output lines, and Each U-cell consists of a 2-input and gate, a 2-input mutex or idle, and three 1330333 1-bit latches; the circuit characteristics of the multiplier are _ _ two elements B in the finite field GF(2m) Perform a product operation to obtain a representation of the __third element c, its singular A and the - polynomial base (Κ..., αTM-|), the finite field gf 〇 =, B and C are The equation X''' + 1 produces, and α is the indecomposable polynomial ', and the decomposable multiple arguments) = + y represents the y. The g product of the thorn ^ = ^4(χ)β (the product of Λ〇 is fire x) jB(JC) = g /*=0 m~\ OT-| Σ waa: 'calculation · The calculation formula f α 6 + f , , >=〇午令户? With a definition, there is a line to I+J^ndd - 2 ~ Ask the vector) as W, =K, Wi ;,...5MWi)/. Where the brother line vector w = [Μ, 1, 士士ο / ,. . Dingmeni L /, J is not like, + household _ (even), then in the coefficient ^ > and, respectively, by the coefficient w And J> 5 2 < Ding > (v) 丨). ('+丨) The expression is determined. Where the first/row vector w = [~], if...=〇,.eight... ^number) then in the ~ expression, the system \i+J=even (colu imn number a < one > and gas (a) _丨> 'Differently by coefficient> Vu洳^ + 2 □ (factory 1) (the expression of the sum is determined „ where each U cell of the multiplier is the execution (even), then ίΓ, = ί/切, brother y column where i + 7,even aJ+Ul "+«,W><(,+7>/2> is calculated, or ^ , . even is done with ', +fl<( , +7+,)/2>Wi his calculation. k 4 right...=〇dd (odd number), where the all-one polynomial base /^)=χίΒ-, + γ_2 in the reduction 'to get male' 5) X) X-', mody + J) is the first. 1 From a polynomial + (where the finite field GF (2m) of all the elements boast θ:., so that the silk ruler calculates the Yanlong f to (four) ^ ^Decomposition of all-multiple seems to be generated in which all elements of the finite field GF (2m) are special. The two-term polynomial of the knife solution that can provide ♦))) x-"m〇d(m)~ The produced (WA-丨) is transferred into two vectors (α_.α · a , then by two vectors (a., ~... gas-,) and the right-handed β of the finite field GF(2m) <" +_&g t;,—(丨)>, '九帅卜,〇. There are a few 々 々 何 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三 三1) 'Xi Xiong' 5(χ)χ-” mod(xm + ^+1 Η) 20 1330333 is calculated by the function &(X), where the ruler (7) is the same as you =) 2 ( X) +,3 (X) + 7; (X) where all elements of the finite field GF(2m) are produced by the indecomposable trinomial χ„, + / + i 'and GC'. l 'g卩 function you) Every-relationship is included in the polynomial you; that is, the polynomial G〇) can be calculated from your) = ♦ small "&" mQd (m) calculation. Where all elements of the finite % GF(2m) are calculated by the sum of the 'and GCZ' (W, „) = 1 'by you) and 咏) generated by the indecomposable trinomial γ; the product Z) (x) = J(x)5(x)jT" mod(xm + χη +!) is available for £>(χ) = d.

X <^+/τ(0)> «Ι+π(])> +----h d 則有 ;若〇仏m_2,則 其中該有限場GF(2m)的所有轉是由不可分解的三項多項^+x,,+1 所產生的、’ ^0^) = 1,即該乘法器包括—乘法單元及—總和單元; ,乘法單7G的電路結構是v細胞與Q細胞所組成,形成m χ m陣列. 母-細胞包含有至少三個輪入信號線及三個輸出信號線; V細胞包含-個2_輪人及閘、二個2_輸人的互斥或閘及四個 兀松閉, 元栓細胞包含一個2-輪入及閘、一個2_輸入的互斥或閘及四個1位 路結構是w細胞所組成,以形成1 x爪陣列; w田二%3^至少二個輸入信號線及一個輸出信號線;及 w<i+/T(〇> S<l+;r(/):> 十 Λ<1+·?Γ(〇> DT 'ΤΤ ο 算^中每—脈波週期的傳波延遲最大需要―個及間及-個互斥或問的計 其中該有限% GF(2m)的所有元素是由不可分解的 所產生的,該乘法器的計算延遲僅需要m+1脈波週期了、夕項式hx +1 前文係針對本發明之較佳實施例為本發明之技術特徵進行具體之說 21 隹滅悉此項技術之人士當可在不脫離本發明 明進行變更與修改,而該等變更與修改,皆應涵蓋 界定之範疇中。 >【圖式簡單說明】 第一圖:係本發明之分佈於GF(24)之A〇P位元-並行心臟收縮陣列式乘法器 片 之架構圖。 ,一圖··係本發明之U細胞詳細的電路圖。X <^+/τ(0)>«Ι+π(])> +----hd is there; if 〇仏m_2, then all the transitions of the finite field GF(2m) are not The decomposed trinomial ^+x,, +1 produces '^0^) = 1, that is, the multiplier includes - multiplication unit and - summation unit; , the circuit structure of multiplication single 7G is v cell and Q cell Composition, forming an array of m χ m. The mother-cell contains at least three round-in signal lines and three output signal lines; the V-cell contains - 2_ rounds of people and gates, and two 2_ inputs of mutual exclusion or The gate and the four sputum loose, the meta-stem cells contain a 2-wheel entry and gate, a 2_ input mutual exclusion or gate and four 1-position structures are composed of w cells to form a 1 x claw array; Two %3^ at least two input signal lines and one output signal line; and w<i+/T(〇>S<l+;r(/):>十Λ<1+·?Γ(〇> DT 'ΤΤ ο The maximum delay of the wave delay per pulse period is - and between and - a mutual exclusion or question, where all elements of the finite % GF (2m) are produced by indecomposable, The calculation delay of the multiplier only needs m+1 pulse period, and the evening term hx +1 The present invention has been described with respect to the preferred embodiments of the present invention. The present invention can be modified and modified without departing from the scope of the invention. Should be covered in the scope of definition. > [Simple diagram of the diagram] The first diagram: is the architecture diagram of the A〇P-bit-parallel cardiac contraction array multiplier slice distributed in GF(24) of the present invention. Fig. is a detailed circuit diagram of the U cell of the present invention.

第二圖:係本發明由x5+x2+l場產生之位元·並行心臟收縮蒙哥馬利乘法 器。 第四圖:係第三圖之詳細的v細胞電路。 第五圖:係第三圖之詳細的Q細胞電路。 第六圖:係第三圖之詳細的W細胞電路。The second figure is a bit-integrated parallel-contracted Montgomery multiplier produced by the x5+x2+l field of the present invention. Figure 4: The detailed v-cell circuit of the third figure. Figure 5: Detailed Q cell circuit in the third figure. Figure 6: Detailed W cell circuit in the third figure.

之精神與原則下對本發 於如下申請專利範圍所 【主要元件符號說明】 11 U細胞 12 一階-1位元拴閘 13 二階-1位元掩閘 14 三階-1位元掩閘 15 四階-1位元拾閘 21 U細胞電路架構 22 2輸入及閘 23 2輸入互斥或閘 24 1位元拾閘 25 1位元拴閘 26 1位元拾閘 31 乘法單元 32 加總單元 33 V細胞 34 Q細胞 35 W細胞 41 V細胞電路架構 42 2輸入及閘 43 2輸入互斥或閘 44 2輸入互斥或閘 45 1位元拴閘 46 1位元拾閘 47 1位元拴閘 48 1位7L拾間 51 Q細胞電路架構 52 2輸入及閘 53 2輸入互斥或間 54 1位元拾閘 55 1位元拴閘 56 1位元拾閘 22 1330333 57 1位元拴閘 2輸入互斥或閘 61 W細胞電路架構 62 63 1位元拴閘The spirit and principle of this application in the following patent application scope [main component symbol description] 11 U cell 12 first-order-1 bit gate 13 second-order-1 bit mask 14 third-order-1 bit mask 15 four Step-1 bit pick-up gate 21 U cell circuit architecture 22 2 input and gate 23 2 input mutual exclusion or gate 24 1 bit pick-up gate 25 1 bit gate 26 26-bit pick-up gate 31 multiplication unit 32 total unit 33 V cell 34 Q cell 35 W cell 41 V cell circuit architecture 42 2 input and gate 43 2 input mutual exclusion or gate 44 2 input mutual exclusion or gate 45 1 bit gate 46 46 1 bit gate 47 1 bit gate 48 1 bit 7L pick up 51 Q cell circuit architecture 52 2 input and gate 53 2 input mutual exclusion or between 54 1 bit pick up 55 55 1 bit gate 56 1 bit pick up gate 22 1330333 57 1 bit gate 2 Input Mutual Exclusion or Gate 61 W Cell Circuit Architecture 62 63 1 Bit Gate

23twenty three

Claims (1)

1330333 十、申請專利範圍: 乘、、=、—輯师、特殊麵钱元平行心谢_狀蒙哥弓利 =,:,GF(n中之特殊多項式〜,= 應用於全-多項式(AOP)作)=严|+ 裔以 素的係數衫㈣或卜該餘純括—縣積=壯’料所有元 , η 少—個輸入佗说線及至少三個輪屮作 &元栓細胞包含—個2輸人及間、—個2輸入互斥或閘和三個 其中該乘法器的電路特性係對有限場GF m ,. 二元素B進行乘積運算暑),以得與一第 c是以多項謝4,以,A、B及 所產生的,及。為該不可分:=== I Γ由 72 ',〜。表示第7•項你)和阶)的中間乘積。 。申凊專纖,項所述核場_陶殊翻之位元平行心 臟收料^蒙哥馬㈣其巾獅)之 / m~\ J(x)5(;c) = t /=:〇 /之計算 ^ m-1 a b + V a b 、/:。 予午々 <ψ> \,+J^n ,+ ί^οώ/ 請翻細第2項所述能場GF(n的特殊種類之位元平行心 臟收,縮陣列式蒙哥馬利乘法器,其中該 〒Άν; %户地 疋義有個行向量(column vector)為W,= i+J^even i^jf=〇dd [w..,w 4 ;如巾請專概圍第3酬述树場GF(2m)㈣殊翻之位元平行心 臟收轉列式蒙哥馬齡法&其中第ζ·行向量w 數),則在〜的表示式中,係數〜及 +户_(偶 ’分別地被係數 wu-η.(Ά的表示式確定 24 1330333 5、如申請專利範圍第3項所述有限場GF(2m)的特殊種類之位元平行心 臟收縮陣列式蒙哥馬利乘法器,其中第Μ亍向量w = [wJ,若(奇 數),則在'表示式中,係數α平及气,分別地被係數^)(,+1)和 你。-”.(卜丨)的表示式確定。 瞄申請專利範圍第1項所述有限場啊2"1)的特殊種類之位元平行心 ^收^列式蒙哥馬利乘法器,其中該乘法器的每,田胞是執筛行和 trii )’則進行〜,'=〜+α卿丨)八順的計算。 臟收範圍第1項所述有限場GF(2m)的特殊麵之料平行心 臟收鈿陣列式蒙哥馬利乘法器’其中該全一丁 ⑼w〜...+x+1能從二項多項式,,+1中約簡,、:二& 4(χ)β〇)ΛΓ;, mody +1)的計算。 k 件 8、如申請專利範圍第7項所述有限 分解的全-多項式所產生的,使該乘法器 si有^疋由不可 =的二,項式〜所產生的,能提供聊水有 由不可 由兩向里(α。外…八和队ν·.Α丨)轉移成兩向量& ,/法,則 d+鄭九+·,·"九+咖以)。 ’’叫),,β·-Ι))和 ω、如申糊瓣i撕梅G明鴨雜之位元平〜 心臟收縮陣列式蒙哥馬利乘法器,其中該有限場证⑺ 仃 可分解的三_式〜”+峨生的,_(_ (相對=林 …’則學K”m()d(〜”+1)之計算,可由函數你)獲得=數 伽触-”mod(x”,+1)及你)=沾)+版+伽产。又”中 1卜如申請專利範圍!! 10項所述有限 心臟收縮陣列式蒙哥馬利乘法器,其令該有限場(GF(2m)gj之位元平行 25 1330333 關係疋L 3在夕項式尤(x);亦即多項式 夂⑷=火x)5(;c);c-〃m〇d(;cm +!)計算獲得。 u 月匕攸 12、如申請專利範圍帛u項所述有限場GF(2m)的特殊種類之位元平行 心臟收縮相式蒙哥馬利乘法器,其巾該有限場啊2)的所有元素是由不 可分解的三項多項式?、”+1所產生的,且⑽(叫=1,由你)和%^ 完成總和計算;該乘積£)⑺=J⑻5(咖-"m〇d W +丨)可得 D(x) ~ d. <ί/+/Τ(0)> X <(/+π(0)> + d X <y+/T( I )> <^+/T(/?/-])> ^<ν+Λ·(/ϊϊ~1 )> 則有 ^/+,τ(,)> = k<c,+,ni)> + 9<?+^,)> mod2 , 、V 1 « WJJ> 13、 如申請專利範圍第12項所述有限場GF(2m)的特殊種類之位_ 心臟收縮陣列式蒙哥馬利乘法器,其中該有限場GF(2m)的所有_ 仃 可分解的三項多項式〜+1所產生的,且)= :去 -乘法料及-總和單元; 包括 該乘法單7L的電路結構是V細胞與Q細胞所組成,形成爪乂爪陣列· 每一細胞包含有至少三個輸入信號線及三個輸出信號線; , 每-V細胞包含-個2-輸入及閘、二個2_輸入的互斥或閉及 元栓閘; 1 每-Q細胞包含-個2-輸入及閘、-個2_輸入的互斥或閘及四個 元栓閘。 14、 如申請專利細第13項所述有限場GF(2m)的特殊翻之位元平行 心臟收縮陣列式蒙哥馬利乘法器’其中該總和單元的電路結構是^細 組成,以形成1X m陣列; 該每一 W細胞包含有至少二個輪入信號線及一個輸出信號線;及 s亥母一 W細胞包含一個2-輸入的互斥閘和—個1位元^全閘去完成 ^<1+Λ·(/)> ~ ^<l+ir(/)> ^<Ι + /τ(/)> ®Ί" 15、 如申請專利範圍第Η項所述有限場GF(2m)的特殊種類之位元平行 心臟收縮陣列式蒙哥馬利乘法器,其中每一脈波週期的傳波延遲最大需要 一個及閘及一個互斥或閘的計算時間。 而 26 1330333 . I6、如申請專利範圍第14項所述有限場GF(2m)的特殊種類之位元平行 心臟收縮陣列式蒙哥馬利乘法器,其中該有限場GF(2rn)的所有元素是由不 可分解的二項多項式X〃+X"+1所產生的,該乘法器的計算延遲僅需要 脈波週期。 17、 -種有限場GF(2m)的特殊種類之位元平行心臟收縮陣列式蒙哥馬 ‘ 利乘法器,使用於全一多項式(All-OnePolynomial,AOP)基底乘法器,包 含u細胞電路架構,其由一個2_輸入及閘(AND gate)、一個2_輸入互斥或 • 閘(x〇R gate)和3個丨_位元栓閘(丨atch)組成,因能計算 ♦_K”m〇d(x'"+1),亦為全一多項式基底(Au_〇nep〇丨卿mialBasecW^ φ 哥馬利乘法器;該乘法器分佈於GF(2),在(χ + 1)/>(χ) = χ",+1條件成立下,使 用全一多項式= + 之約簡過程,是經常利用二項式 (binomial) X"’ +1進行;所以將對分佈於丨)之全一多項式基底乘法器 (AOP-based multiplier),使用二項式(b—iv+i之約簡多項式,以實現 低複雜性乘法器。 ' ' 18、 一種有限場GF(2m)的特殊種類之位元平行心臟收縮陣列式蒙哥馬 利乘法益’其中使用蒙哥馬利乘法器來實現三項式_基底和丨11〇〇^1七3记(1)的 象哥馬利乘法,其包括乘法單元和總和單元;其中乘法單元包含V細胞與 Q細胞;乘法單元的結果’是基於兩個基本的細胞(v細胞及Q細胞),產 生兩多項式/:(=)和G(x);而且,由0化3,係數g小朴是容易在第〇 + 1)行 • 被產生;總合單元,係由m個w細胞組合,以執行尺(χβσ(7(χ)之加總和; , 每—W細胞電路結構完成計算;據此,使用三種細 胞類型(V細胞、Q細胞及W細胞)進行由三項式(Tri_ial)之蒙哥馬利 乘法者。 19、 如申請專利範圍第18項所述有限場GF(2m)的特殊種類之位元平行 心臟收縮陣列式料馬利乘法器,其巾該v細胞電路架構係由—個2_輸入 . 及閑(AND gate) ’二個2-輸入的互斥或閘(X〇R gate)和四個〖位元栓閘(Latch) 組成。 20、 如申請專利範圍第18項所述有限場GF(2m)的特殊種類之位元平行 心臟收縮陣列式料馬齡法器,其巾該Q細胞電路架構,由—個2輸入 及閘(AND gate)’ —個2_輸入的互斥或閘(x〇Rgate)和四個i位元检問㈣邮 27 1330333 組成。 2卜如申請專利範圍第18項所述有限場GF(2m)的特殊種類之位元平行 心臟收縮陣列式蒙哥馬利乘法器,其中該每一 W細胞電路結構係包含一個 2-輸入的互斥或閘(XOR gate)和一個1位元栓閘(Latch)。 Η—、圖式: 如次頁 281330333 X. The scope of application for patents: multiplication,, =, - division, special face money yuan parallel heart thank _ shape Mengge bow profit =,:, GF (n special polynomial ~, = applied to all-polynomial (AOP )作)=严|+ 系数 的 的 ( 四 四 四 四 四 四 四 四 四 四 四 四 县 县 县 县 县 县 县 县 县 县 县 县 县 县 县 县 县 县 县 县 县 县 县 县 县 县 县 县 县 县 县 县 县 县 县- 2 inputs and between, - 2 input mutual exclusion or gate and three of which the circuit characteristics of the multiplier are for the finite field GF m , the two element B is subjected to the product operation), so that a c is Thanks a lot of 4, to, A, B and the resulting, and. For the indivisible: === I Γ by 72 ', ~. Indicates the intermediate product of the 7th item and the order). . Shen Hao special fiber, the nuclear field of the item _ Tao Shu turned the bit parallel heart receiving ^ Montgomery (four) its towel lion) / m~\ J (x) 5 (; c) = t / =: 〇 / Calculate ^ m-1 ab + V ab , /:.午午々<ψ> \,+J^n ,+ ί^οώ/ Please refine the energy field GF of the second item (n of the special type of parallel parallel heart, shrink array Montgomery multiplier, where The 〒Άν; % household 疋 has a row vector (column vector) for W, = i + J ^ even i ^ jf = 〇 dd [w.., w 4 ; Tree field GF (2m) (four) special position of the parallel heart transfer column type Montgomery method & ζ · row vector w number), then in the expression of ~, the coefficient ~ and + household _ ( Evenly by the coefficient wu-η. (Ά 表示 确定 24 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 , , , , , , , , , , , , , , , Where the third vector w = [wJ, if (odd), then in the expression, the coefficient α is equal to the gas, respectively by the coefficient ^) (, +1) and you. - ". (di) The expression is determined. The special type of parallel line of the limited field 2 "1) mentioned in the scope of the patent application scope is applied to receive the Montgomery multiplier, wherein each of the multipliers is a sieve line. And trii)' then proceed ~, '=~+α卿丨) The calculation of the eight-shun. The visceral range of the first field of the finite field GF (2m) is parallel to the heart-receiving array of Montgomery multipliers, which is the whole one (9)w~. ..+x+1 can be calculated from the binomial polynomial, +1 reduction, and: 2 &4(χ)β〇)ΛΓ;, mody +1). k piece 8, as claimed in the patent scope The all-polynomial of the finite decomposition of the seven items is generated so that the multiplier si has a 二=2, which can be provided by the non-integral, and can provide water for the two sides (α. ...eight and team ν·.Α丨) transferred into two vectors & , / law, then d + Zheng Jiu + ·, · " nine + coffee to). ''call),, β·-Ι)) and ω Such as Shen paste flap i tearing Mei G Ming duck miscellaneous bit flat ~ heart contraction array Montgomery multiplier, which limited field certificate (7) 仃 decomposable three _ type ~ "+ twin, _ (_ (relative =林...'th learning K"m()d(~"+1) calculation, can be obtained by the function you) = number of gigabits - "mod (x", +1) and you) = dip) + version + gamma Production. Also, "Zhong 1 Bu as claimed patent scope!! 10 items of the limited cardiac contraction array Montgomery multiplier, which makes the finite field (GF (2m) gj bit parallel 25 1330333 relationship 疋 L 3 in the evening (x); that is, the polynomial 夂(4)=firex)5(;c);c-〃m〇d(;cm +!) is calculated. u 匕攸12, as described in the scope of patent application 帛u Field GF (2m) special kind of bit parallel systolic phase Montgomery multiplier, the towel of the limited field ah 2) all elements are produced by the indecomposable trinomial polynomial ?, "+1", and (10) (call = 1, by you) and %^ to complete the sum calculation; the product £)(7)=J(8)5(cafe-"m〇d W +丨) can get D(x) ~ d. <ί/+/Τ (0)> X <(/+π(0)> + d X <y+/T( I )><^+/T(/?/-])>^<ν+ Λ·(/ϊϊ~1 )> Then there are ^/+, τ(,)> = k<c,+,ni)> + 9<?+^,)> mod2 , , V 1 « WJJ&gt 13. A special type of finite field GF (2m) as described in claim 12 of the patent scope _ cardiac contraction array Montgomery multiplier, where all _ 仃 decomposable three of the finite field GF (2m) The polynomial ~ +1 is generated, and) =: de-multiply and - sum unit; the circuit structure including the multiplication single 7L is composed of V cells and Q cells, forming an array of claws and claws. Each cell contains at least Three input signal lines and three output signal lines; , each -V cell contains -2-input and gate, two 2_input mutually exclusive or closed and meta-lock; 1 per-Q cell contains -2 Input and gate, - 2_ input mutual exclusion or gate and four element lock gate. 14. A special tumbling parallel systolic array Montgomery multiplier of the limited field GF (2m) as described in claim 13 of the patent application, wherein the circuit structure of the summation unit is a fine composition to form a 1×m array; Each of the W cells includes at least two rounding signal lines and one output signal line; and the sigma-W cells comprise a 2-input mutual sluice gate and a 1-bit thyristor to complete ^<1+Λ·(/)> ~ ^<l+ir(/)>^<Ι + /τ(/)>®Ί" 15. The limited field GF as described in the scope of the patent application (2m) special type of parallel systolic array Montgomery multiplier, where the maximum wave delay per pulse period requires one gate and one mutex or gate calculation time. And 26 1330333. I6, a special type of parallel systolic array Montgomery multiplier of the limited field GF (2m) as described in claim 14 of the patent application, wherein all elements of the finite field GF (2rn) are not The decomposed binomial polynomial X〃+X"+1 produces that the multiplier's computational delay requires only the pulse period. 17, a finite field GF (2m) special type of parallel parallel systolic array Montgomery 'multiplier, used in all-One Polynomial (AOP) base multiplier, including u cell circuit architecture It consists of a 2_ input AND gate, a 2_ input mutual exclusion or • gate (x〇R gate) and 3 丨_bit locks (丨atch), because it can calculate ♦_K”m 〇d(x'"+1) is also the base of the all-one polynomial (Au_〇nep〇丨卿 mialBasecW^ φ Gomley multiplier; the multiplier is distributed in GF(2), at (χ + 1) />(χ) = χ", +1 condition is established, using the all-one polynomial = + reduction process, is often done using binomial X"' +1; so will be distributed on 丨) A polynomial multiplier (AOP-based multiplier) using a binomial (b-iv+i reduction polynomial to achieve a low complexity multiplier. ' ' 18. A special finite field GF(2m) Category of Parallel Heart Contraction Array Montgomery Multiplication Method 'Using the Montgomery Multiplier to Realize the Trinomial_Base and 丨11〇〇^17 3 (1) Like the Commeli multiplication, it includes a multiplication unit and a summation unit; wherein the multiplication unit contains V cells and Q cells; the result of the multiplication unit is based on two basic cells (v cells and Q cells), producing a two polynomial /: ( =) and G(x); and, by 0, 3, the coefficient g is simple, and is generated in the third + 1) line; the unit is composed of m w cells combined to perform the ruler (χβσ( The sum of 7 (χ); , the calculation of the circuit structure of each W cell; accordingly, the three cell types (V cells, Q cells, and W cells) are used to perform the Montgomery multiplication by Tri_ial. For example, the special type of bit-parallel systolic array material Marley multiplier of the limited field GF (2m) described in claim 18 of the patent scope, the v-cell circuit structure of the towel is composed of -2_ input. AND gate) 'Two 2-input mutexes or gates (X〇R gate) and four 〖bits of latches (Latch). 20. The finite field GF (2m) as described in claim 18 Special type of parallel parallel cardiac contraction array material ageing device, its towel Q cell circuit architecture, by - 2 input and gate (AND gate) - a 2_ input mutual exclusion or gate (x〇Rgate) and four i-bit inquiries (four) post 27 1330333. 2 Bu as claimed in the scope of claim 18 Field GF (2m) special type of parallel parallel cardiac contraction array Montgomery multiplier, where each W cell circuit structure contains a 2-input mutual exclusion gate (XOR gate) and a 1-bit padlock ( Latch). Η—, schema: as the next page 28
TW095143577A 2006-11-24 2006-11-24 Finite field GF(2m) specific bit-parallel systolic array type Montgomery multiplier TW200710715A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
TW095143577A TW200710715A (en) 2006-11-24 2006-11-24 Finite field GF(2m) specific bit-parallel systolic array type Montgomery multiplier

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
TW095143577A TW200710715A (en) 2006-11-24 2006-11-24 Finite field GF(2m) specific bit-parallel systolic array type Montgomery multiplier

Publications (2)

Publication Number Publication Date
TW200710715A TW200710715A (en) 2007-03-16
TWI330333B true TWI330333B (en) 2010-09-11

Family

ID=45074563

Family Applications (1)

Application Number Title Priority Date Filing Date
TW095143577A TW200710715A (en) 2006-11-24 2006-11-24 Finite field GF(2m) specific bit-parallel systolic array type Montgomery multiplier

Country Status (1)

Country Link
TW (1) TW200710715A (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108804075A (en) * 2018-06-12 2018-11-13 信阳师范学院 A kind of efficient Montgomery Multiplications building method based on special five formulas

Also Published As

Publication number Publication date
TW200710715A (en) 2007-03-16

Similar Documents

Publication Publication Date Title
Hossain et al. High‐performance elliptic curve cryptography processor over NIST prime fields
Lee Low complexity bit-parallel systolic multiplier over GF (2m) using irreducible trinomials
Fan et al. Subquadratic computational complexity schemes for extended binary field multiplication using optimal normal bases
Namin et al. A word-level finite field multiplier using normal basis
TW201135477A (en) Sequential Galois field multiplication architecture and method
JP2015512585A (en) Parallel encoding of non-binary linear block codes
Lee et al. Fast block inverse jacket transform
Imana LFSR-Based Bit-Serial $ GF (2^ m) $ G F (2 m) Multipliers Using Irreducible Trinomials
Chang et al. Low complexity bit-parallel multiplier for GF (2/sup m/) defined by all-one polynomials using redundant representation
TWI330333B (en)
Huang et al. Non-XOR approach for low-cost bit-parallel polynomial basis multiplier over GF (2m)
Lee Low-Latency Bit-Parallel Systolic Multiplier for Irreducible x m+ x n+ 1 with gcd (m, n)= 1
Chiou et al. Low-complexity Gaussian normal basis multiplier over GF (2m)
Agrawal et al. Quantum-proof lightweight McEliece cryptosystem co-processor design
CN103942027B (en) A kind of fast parallel multiplier of restructural
Lee et al. Low-complexity bit-parallel systolic architectures for computing A (x) B2 (x) over GF (2m)
Liu et al. A high speed VLSI implementation of 256-bit scalar point multiplier for ECC over GF (p)
de Gier et al. Brauer loops and the commuting variety
Chen et al. Scalable and systolic Montgomery multipliers over GF (2 m)
TW200841232A (en) Finite field Montgomery multiplier
Jeon et al. Low-power exponent architecture in finite fields
Chiou et al. Palindromic-like representation for Gaussian normal basis multiplier over GF (2m) with odd type t
Zhang et al. Low-complexity transformed encoder architectures for quasi-cyclic nonbinary LDPC codes over subfields
Raviv Asymptotically optimal regenerating codes over any field
Oliveira et al. On polynomials xn− 1 over binary fields whose irreducible factors are binomials and trinomials

Legal Events

Date Code Title Description
MM4A Annulment or lapse of patent due to non-payment of fees