Practical and Provable Security against Differential and Linear Cryptanalysis for Substitution-Permutation Networks
|
|
- Loren Hopkins
- 5 years ago
- Views:
Transcription
1 Practcal ad Provable Securty agast Dfferetal ad ear Cryptaalyss for Substtuto-Perutato Networks Ju-Sug Kag, Seokhe Hog, Sagj ee, Okyeo Y, Choosk Park, ad Jog We exae the dffuso layers of soe block cphers referred to as substtuto-perutato etworks. We vestgate the practcal ad provable securty of these dffuso layers agast dfferetal ad lear cryptaalyss. Frst, ters of practcal securty, we show that the u uber of dfferetally actve S-boxes ad that of learly actve S-boxes are geerally ot detcal ad propose soe specal codtos whch those are detcal. We also study the optal dffuso effect for soe dffuso layers accordg to ther costrats. Secod, we obta the results that the cosecutve two rouds of SPN structure provde provable securty agast dfferetal ad lear cryptaalyss,.e., we prove that the probablty of each dfferetal (resp. lear hull) of the cosecutve two rouds of SPN structure wth a axal dffuso layer s bouded by p (resp. q ) ad that of each dfferetal (resp. lear hull) of the SDS fucto wth a se-axal dffuso layer s bouded by p - (resp. q - ), where p ad q are axu dfferetal ad lear probabltes of the substtuto layer, respectvely. auscrpt receved ay 4, 2; revsed Aug. 7, 2 ad Oct. 9, 2. Ju-Sug Kag (phoe: , e-al: jskag@etr.re.kr) s wth Iforato Securty Research Dvso, ETRI, Daejeo, Korea. Seokhe Hog (e-al: hsh@cst.korea.ac.kr), Sagj ee (e-al: saj@tger.korea.ac.kr), ad Jog (e-al: jl@tger.korea.ac.kr) are wth Ceter for Iforato Securty Techologes, Korea Uversty, Seoul, Korea. Okyeo Y (e-al: oyy@ku.kook.ac.kr) s wth the Departet of atheatcs, Kook Uversty, Seoul, Korea. Choosk Park (e-al: csp@etr.re.kr) s wth Natoal Securty Research Isttute, ETRI, Daejeo, Korea. I. INTRODUCTION. Itroducto to SPN Structure ad Dffuso ayer Shao suggested that practcal ad secure product cphers aybe costructed by usg a xg trasforato cosstg of a uber of layers or rouds of cofuso ad dffuso []. The cofuso copoet s a olear substtuto o a sall subblock ad the dffuso copoet s a lear xg of the subblock coectos. The Substtuto- Perutato Networks (SPN) structure s drectly based o the cocepts of cofuso ad dffuso. Oe roud of a SPN structure geerally cossts of three layers of substtuto, perutato, ad key addto. A substtuto layer s ade up of sall olear substtutos referred to as S-boxes easly pleeted by table lookup for cofuso effect. A perutato layer s a lear trasforato that dffuses the cryptographc characterstcs of the substtuto layer. A key addto layer plats roud subkeys of the cpher ad the posto of ths layer s varable accordg to cphers. Due to eory requreets, ost block cpher desgers use sall S-boxes, e.g. wth 4 or 8 put bts. Thus, the dffuso of S-box outputs by a perutato layer plays a great role provdg resstace to varous attacks cludg dfferetal ad lear cryptaalyss. O the other had, perutato layers of ost oder block cphers are ot sple bt-wse posto perutatos or traspostos but lear trasforatos o soe vector spaces over varous fte felds. Hece ths paper, we refer to a perutato layer as a dffuso layer for the sake of clarty. ost dffuso layers have approprate atrx represetatos, sce they are lear trasforatos over 58 Ju-Sug Kag et al. ETRI Joural, Volue 23, Nuber 4, Deceber 2
2 soe fte felds ad have oe-to-oe correspodece to a approprate atrx. Wth these atrx represetatos, we study the practcal ad provable securty agast dfferetal ad lear cryptaalyss. 2. Related Works ad Our Results The ost well kow ethod of aalyzg block cphers today s the dfferetal cryptaalyss (DC), proposed by Bha ad Shar [2], [3] 99. DC s a chose platext attack whch the attacker chooses soe platexts of certa wellcosdered dffereces. Bha ad Shar used the oto of characterstc, whle a, assey ad urphy [4] showed that the oto dfferetal strctly reflects the stregth of a cpher agast DC. Roughly speakg, a dfferetal s a collecto of characterstcs. Aother ethod of aalyzg block cphers s the lear cryptaalyss (C) publshed by atsu [5] 993. The attacks based o C are kow platext attacks ad the attack o the DES s faster tha the attack by DC. The frst verso of C appled lear approxato to a attack of block cphers, but Nyberg [6] has cosdered a collecto of lear approxato, whch she called a lear hull for strct evaluato of the stregth agast C. Kada et al. [7] classfed four easures to evaluate the securty of a cpher agast DC ad C as follows: Precse easure: The axu average of dfferetal ad lear hull probabltes [4], [6]. Theoretcal easure: The upper bouds of the axu average of dfferetal ad lear hull probabltes [8]-[]. Heurstc easure: The axu average of dfferetal characterstc ad lear approxato probabltes [2], [3], [5]. Practcal easure: The upper bouds of the axu average of dfferetal characterstc ad lear approxato probabltes [2]-[4]. DC ad C are the ost powerful attacks to ost syetrc block cphers. Accordgly, t s a basc requste for the desger to evaluate the securty of ay ew proposed cpher agast DC ad C, ad to prove that t s suffcetly resstat agast the. I ths paper, we cosder a practcal easure ad theoretcal easure out of the above four easures. Nyberg ad Kudse [] stated that Festel cphers evaluated wth the theoretcal easure are provably secure agast DC ad C. Therefore, a block cpher s called to have provable securty agast DC ad C, where the upper bouds of the axu average of dfferetal ad lear hull probabltes are suffcetly sall. eawhle, Kudse [3] oted that Festel cphers evaluated wth the practcal easure are practcally secure agast DC ad C. Thus, a block cpher s called to have practcal securty agast DC ad C f the upper bouds of the axu average of dfferetal characterstc ad lear approxato probabltes are suffcetly sall. Frst, we show that ters of practcal securty, the u uber of dfferetally actve S-boxes ad that of learly actve S-boxes are geerally ot detcal ad propose soe specal codtos whch those are detcal. We also show that all dffuso layers of E2, Crypto ad Rjdael have acheved optal dffuso effects accordg to ther each costrat of usg operatos. Secod, the cosecutve two rouds of SPN structure are show to provde provable securty agast dfferetal ad lear cryptaalyss, where the dffuso layer has a axal or se-axal dffuso effect,.e., we prove that the probablty of each dfferetal (resp. lear hull) of the cosecutve two rouds of SPN structure wth a axal dffuso layer s bouded by p (resp. q ) ad that of each dfferetal (resp. lear hull) of the SDS fucto wth a seaxal dffuso layer s bouded by p - (resp. q - ), where p ad q are axu dfferetal ad lear probabltes of the substtuto layer, respectvely. Ths paper s the refed verso of [5] ad [6]. II. PREIINARIES. Basc Deftos et S be a S-box wth put ad output bts,.e., S : Z 2 Z 2. Dfferetal ad lear probabltes of S are defed as the followg defto. Defto For ay gve x, y; Γx; Γy Z 2, defe dfferetal ad lear probabltes of S by ad S ( x y) = (/2 ) ( # {x Z 2 : S(x) S(x x) = y }) P S (Γy Γx) = [ (/2 - ) ( # {x Z 2 : Γx x = Γy S(x) }) ] 2, respectvely, where a b deotes the party ( or ) of bt-wse product of a ad b. S ad P S for a strog S-box S should be sall eough for ay put dfferece x ad output ask value Γy. Therefore, we defe paraeters that represet resstace to DC ad C of a S-box ad each substtuto layer of a SPN structure as the followg defto. ETRI Joural, Volue 23, Nuber 4, Deceber 2 Ju-Sug Kag et al. 59
3 Defto 2 The axu dfferetal ad lear probabltes of S are defed by ad respectvely. S ax = ax x, y S ( x y) P S ax = ax Γx, Γy P S (Γy Γx), 2. Dfferetally ad early Actve S-Boxes Evaluato of securty for a block cpher of SPN structure by a practcal easure begs wth the cocept of a actve S-box. The followg fve deftos ad oe theore of ths subsecto are already wrtte soe prevous works [7], [4], [6], [7]. At ths pot, we slghtly revse soe deftos order to descrbe our results. Defto 3 A dfferetally actve S-box s defed as a S-box gve a o-zero put dfferece ad a learly actve S-box as a S-box gve a ozero output ask value. By coputg the u uber of dfferetally ad learly actve S-boxes, we ca evaluate securty of a block cpher ters of practcal securty agast DC ad C [2]- [4], [7]. We ca obta upper bouds of the axu dfferetal characterstc ad lear approxato probabltes fro the u uber of actve S-boxes. Thus, the case of a SPN structure, t s portat to aalyze the creasg aouts of u uber of actve S-boxes by cosderg a dffuso layer cosecutve two rouds. Note that we ca ot the key addto layer to copute the uber of actve S-boxes sce ths layer has o fluece uder the assupto that the key addto layer s perfored by bt-wse EXORs. Defe the SDS fucto wth three layers of substtuto-dffuso-substtuto for aalyzg the role of a dffuso layer to crease the uber of actve S-boxes cosecutve two rouds of a SPN structure (Fg. ). Throughout ths paper, we cosder the SDS fucto wth -bt put ad output values ad assue that all S-boxes the substtuto layer are ad bjectve. If a S-box s bjectve ad dfferetally/learly actve, the t has a o-zero output dfferece/put ask value []. Therefore, whe all S- boxes substtuto layer are bjectve, we ca defe the u uber of actve S-boxes of the SDS fucto. Deote the dffuso layer of SDS fucto as D, put dfferece of D as x = x x*, output dfferece as y = y y* = D(x) D(x*), ad put ad output ask value as Γx ad Γy, respectvely. Defto 4 The u uber of dfferetally ad learly actve S-boxes of the SDS fucto are defed by ad S S 2 S 3 S Substtuto(S-boxes) Dffuso layer S S 2 S 3 S Substtuto(S-boxes) Fg.. SDS fucto. β d (D) = x { H c ( x) + H c ( y) } β l (D) = Γy { H c (Γx) + H c (Γy) }, respectvely, where, for each x = (x, x 2,, x ) (Z 2 ) or GF(2 ), the copoet Hag weght of x s defed by H c (x) = # { : x }. β d (D) ad β l (D) are lower bouds for the uber of actve S-boxes two cosecutve rouds of a dfferetal characterstc ad lear approxato, respectvely. O the other had, fro the alty, we ca see that β d (D) ad β l (D) are at ost + by cosderg H c ( x) = H c (Γy) =. So a dffuso layer s called axal f the β d (D) ad β l (D) are +. Now we ca defe dfferetal characterstc ad lear approxato probabltes of the SDS fucto lke the deftos for S-boxes. See Fgs. 2 ad 3. Defto 5 For ay gve x, y, Γx, Γy Z 2, defe the dfferetal characterstc ad lear approxato probabltes of the SDS fucto by ad DCP SDS = ax AP SDS = ax w y) = ( Γy Γx) Γz = P S S ( Γy ) Γz ) P S S ( D( w ) y ) ( D ( Γz) Γx ), respectvely, where x = ( x,, x ) (Z 2 ) ad y, w, Γx, Γy, Γz are deoted the sae way as x.. Here, D( w) = (D( w),, D( w) ) ad D - (Γz) = (D - (Γz),, D - (Γz) ) deote the output dfferece ad put ask value of the dffuso layer D wth probablty, respectvely. Defto 6 The axu dfferetal characterstc ad l- 6 Ju-Sug Kag et al. ETRI Joural, Volue 23, Nuber 4, Deceber 2
4 ear approxato probabltes of the SDS fucto are defed by DCP SDS ax = ax x, y DCP SDS ( x y) ad AP SDS ax = ax Γx, Γy AP SDS (Γy Γx), respectvely. Defto 7 Assue that the substtuto layer of a SDS fucto cossts of S-boxes S, S 2,, S. The axu dfferetal ad lear probabltes of the substtuto layer are defed by p = ax S ax ad q = ax P S ax, respectvely. Theore The axu dfferetal characterstc ad lear approxato probabltes DCP SDS ax ad AP SDS ax of the SDS fucto hold for DCP SDS ax p β d (D) ad AP SDS ax q β l (D). The above theore s obtaed easly by the axalty of p(or q) ad the alty of β d (D) (or β l (D)). Evaluato of practcal securty agast DC ad C s based o ths theore. III. PRACTICA SECURITY AGAINST DC AND C. atrx Represetato of a Dffuso ayer ost dffuso layers of oder block cphers of a SPN structure are lear trasforatos o the vector space GF(2 ) ad have oe-to-oe correspodece to a approprate atrx. That s, ost dffuso layers have approprate atrx represetatos. If we use ths atrx represetato for a dffuso layer, the we obta the relatoshp betwee put ad output dffereces (or ask values). Throughout ths paper, we assue that the dffuso layer D of the SDS fucto ca be represeted by a atrx = ( j ), where j GF(2 ). Hece we oly eed to vestgate the atrx to aalyze the role of the dffuso layer D. To beg, we descrbe soe otatos ad deftos. Wthout loss of geeralty, we ay assue GF(2 ) = { a + a γ + a 2 γ a - γ - a {,}} for soe γ GF(2 ). I geeral, we ca regard the fte feld GF(2 ) as the -desoal vector space over GF(2) ad GF(2 ) as the -desoal vector space over GF(2). We wll use a otato a GF(2) as the colu vector corre- spodg to a GF(2 ) ad ξ GF(2) as the colu vector correspodg to ξ GF(2 ). By assupto of dffuso layer, we ca rewrte β d (D) as follows: β d (D) = x { H c ( x) + H c ( x) }. Now, we cosder β l (D). et's defe a ap φ fro GF(2 ) to GF(2) as follows: φ (a) = = - a, where a = Σ = - a γ, a {,}. ea There exsts a uque bary vertble atrx B so that φ(a b) = a t B b for all a, b GF(2 ), where t suffx deotes trasposto of a vector. Proof: et a = Σ = - a γ ad b = Σ = - b γ be two geeral eleets GF(2 ) ad c be a b = Σ = - c γ. The c k = (,j) I a - k b j for soe dex set I k ad φ(c) = = c = (,j) I a b j where I = - k= I k except ts eleets eed ot be dstct. et j be the repetto uber of (,j) I ad b j j od 2. Cosder a atrx B whose th row ad jth colu s b j. By defto of B, φ(a b) = a t B b. It reas to prove B s vertble. Suppose B s ot vertble the there exsts a ozero eleet a GF(2 ) so that a t B =. φ(a a - ) = a t B a - = but t s a cotradcto to the fact φ(a a - ) = φ() =. Hece B s vertble. et ξ = (ξ, ξ ) t, η = (η, η ) t GF(2 ). A scalar product o GF(2 ) over GF(2 ) ad over GF(2) are deoted by <, > ad <, >, respectvely ad defed by: <, > : GF(2 ) GF(2 ) GF(2 ) (ξ, η) ξ η + + ξ η <, > : GF(2 ) GF(2 ) GF(2) (ξ, η) ξ t η. As a atter of coveece we deote the feld eleet correspodg to Ba r GF(2) by Ba. et ~ η = (Bη, Bη ) GF(2 ) ad ηˆ = (B - η, B - η ) GF(2 ). By deftos of <, >, <, > ad φ, t < ξ, η > = ξ BB = φ = φ = < ( < ξ, ˆ η > ) t ( < ξ, ˆ η > ) η ~ t ξ, ˆ η > Hece we obta the followg lea.. t ξ BB η ETRI Joural, Volue 23, Nuber 4, Deceber 2 Ju-Sug Kag et al. 6
5 ea 2 et Γy be a output ask value of dffuso layer D ~ the the put ask value becoes t Γ ˆ y. It s dcated fro ea 2 that f s a bary atrx, Γx = t Γy, ad ths result s show [7]. Corollary The u uber of learly actve S- boxes s Γy { H c ( t Γy) + H c (Γy) }. Proof: Ths corollary follows fro ea 2 ad the fact that there exst oe-to-oe correspodeces betwee η, ηˆ, ad ηˆ ad H c (η) = H c (ηˆ ) = H c (η ~ ) for ay η GF(2 ). It s possble that we copute the u ubers of dfferetally ad learly actve S-boxes (β d (D) ad β l (D)) of the SDS fucto by usg the above atrx represetato. However, the u ubers of dfferetally ad learly actve S-boxes are ot detcal geeral. I the ext subsecto, we wll show that β d (D) β l (D) by proposg a couterexaple. O the other had, the u ubers of dfferetally ad learly actve S-boxes are detcal for the specal types of represetato atrx as the followg two theores. Theore 2 et the dffuso layer D of the SDS fucto be represeted as atrx. If s a syetrc or orthogoal atrx, the β d (D) = β l (D). Proof: Observe that β d (D) = x { H c ( x) + H c ( x) }, β l (D) = Γy { H c ( t Γy) + H c (Γy) }. Fro ths, we ca easly see that β d (D) = β l (D) f s a syetrc atrx where t =. eawhle, f s a orthogoal atrx that - = t, the Γx = t Γy ples that Γy = Γx, ad the codto Γy = Γx s detcal to Γx sce s a vertble atrx. Thus so β d (D) = β l (D). β l (D) = Γx { H c (Γx) + H c (Γx) }, Theore 3 If t s obtaed fro by applyg operatos of exchagg row or colu vectors, the β d (D) = β l (D). Proof: The operato of exchagg row vectors of results chagg the order of copoets of output dfferece y, ad ths operato does ot affect the copoet Hag weght H c ( y). O the other had, t s clear that H c ( y) s detered by colu vectors of but ot by ther locato. Thus, the operato of exchagg colu vectors of also does ot affect the copoet Hag weght H c ( y). Sce a row(colu) vector of s a colu(row) vector of t, operatos of exchagg row or colu vectors of does ot affect the copoet Hag weght. Therefore, f t s obtaed fro by those operatos, β d (D) = β l (D). It s easy to see that the dffuso layer of block cpher CRYPTON [8] s represeted as a syetrc atrx. Hece, we obta β d (D) = β l (D) by Theore 2, ths case [6]. O the other had, [6] authors also showed that for the dffuso layers of block cpher Rjdael [9] ad E2 [2], Theore 3 ca be appled. 2. Optal Dffuso Effects of Dffuso ayers uder Soe Costrats Assue that puts of the SDS fucto are learly trasfored to outputs per -bt ad the dffuso layer s costructed wth just btwse EXORs. The dffuso layer s represeted as a atrx where all etres are zero or oe as follows: y = j= µ j x j = µ j = x j, where x = (x, x 2,, x ) (Z 2 ) s a put, y = (y, y 2,, y ) s the output, ad = (µ j ). Kada et al. [7] studed dffuso propertes of the dffuso layer wth ths atrx represetato. Ther study was based o the relatoshp betwee the atrx for dfferetal characterstc ad lear approxato. However, they ade two cojectures to ufold ther theory. The Cojecture [7] s correct sce ths s a specal case of Theore 2, but the Cojecture 2 [7] s a wrog opo. We dsprove ths cojecture by proposg a couterexaple. Cojecture 2 of [7] I the SDS fucto, the u uber of dfferetally actve S-boxes s equal to the u uber of leally actve S-boxes. That s, β d (D) = β l (D), where s the represetato atrx of the dffuso layer D. Couterexaple for the Cojecture 2 of [7]: Suppose that the dffuso layer of SDS fucto wth =4 be represeted by the followg vertble atrx: =, t =. If H c ( x) =, the H c ( y) 2 sce H c ( y) s detered by a colu vector of ad Hag weght of each colu vector s at least 2. H c ( y) s detered by the EXORs betwee 62 Ju-Sug Kag et al. ETRI Joural, Volue 23, Nuber 4, Deceber 2
6 ay dfferet two colu vectors f H c ( x) = 2. Ay EXOR betwee two colu vectors has the Hag weght of at least. Thus, the u uber of dfferetally actve S-boxes s β d (D) =3. O the other had, ea 2, the relatoshp betwee output ad put ask values s represeted as the traspose atrx t of. Note that the Hag weght of the fourth colu vector of t s. Cosder the output ask value of the for b = (,,, b 4 ), b 4, ad =, 4 4 b b the correspodg put ask value a = (,, b 4, ). Fro ths we ca obta β l (D) = 2. Cosequetly we kow that β d (D) β l (D) for the above 4 4 atrx. I the block cpher E2, desgers cosdered the SDS fucto wth = 8. Kada et al. [7] suggested a ethod of deterg a 8 8 atrx yeldg the axu value of β d (D) usg the search algorth. Usg ths search algorth, they foud that there s o atrx wth β d (D) 6, ad that there are soe caddate atrces wth β d (D) = 5. Here, we gve a theoretcal proof for the fact that β d (D) = 5 s optal ad also that β l (D) 5, where s a 8 8 bary vertble atrx. Theore 4 Assue that the uber of S-boxes the substtuto layer of the SDS fucto s 8(=8). If the represetato atrx of the dffuso layer s a 8 8 bary vertble atrx, the β d (D), β l (D) 5. Proof: Sce s a 8 8 bary vertble atrx, eght colu vectors, 2,, 8 are learly depedet. Thus, the uber of colus wth the Hag weght 8 s at ost oe. Note that β d (D) s closely related to the Hag weghts of colu vectors of. We separate the proof to four cases. Here, the Hag weght H c ( j ) of a colu vector j s the uber of etres wth j. Case If j 8 H c ( j ) = 7, for ay two colu vectors j ad k, H c ( j k ) 2. By cosderg x such that H c ( x) = 2, we obta that β d (D) = 4. Case 2 Suppose that j 8 H c ( j ) = 6. If there exsts a colu vector wth Hag weght 8, the H c ( y) 2 for soe x such that H c ( x) = 2. If there exsts a colu vector wth Hag weght 7, the u value of H c ( y) s at ost 3, sce we ca cosder the EXORs betwee the colus wth Hag weght 6 ad 7, where H c ( x) = 2. Fally, f the Hag weght of all colu vectors s 6, the although soe dfferet four colu vectors clude etres dstct rows, the Hag weght of EXORs betwee oe of ths four colus ad aother ffth colu vector s 2. Cosequetly, we obta that β d (D) 5. Case 3 Assue that j 8 H c ( j ) = 6. If there exsts a colu vector wth the Hag weght 8, the H c ( y) 3 for soe x so that H c ( x) = 2. If there exsts a colu vector wth the Hag weght 7 or 6, by a slar aalyss as Case 2, we ca obta what we wat. I the case that the Hag weght of each colu vectors s 5, although etres of soe dfferet fve colu vectors are arraged optally, aother sxth colu vector ad oe of these fve colus have coo at least two etres at the sae rows. Thus, H c ( y) s at ost 2 where H c ( x) = 2. Therefore, β d (D) 5 also holds ths case. Case 4 Assue that j 8 H c ( j ) 4. Cosder oly x such that H c ( x) =. The, we easly obta β d (D) 5 sce there exsts a colu wth the Hag weght 4. By Cases to 4, we obta that β d (D) 5 always holds wheever s a 8 8 bary vertble atrx. O the other had, by Theore 2, β l (D) s related to t. Thus, we ca also obta the sae result for β l (D) by cosderg the Hag weght of row vectors stead of colu vectors of. The dffuso layer of block cpher CRYPTON [8] cossts of btwse EXOR ad AND logc. I ths case, we ca also theoretcally show that the optalty of the dffuso layer by the slar process of the proof of Theore 4. O the other had, the block cpher Rjdael, the axal dffuso layer s used. It was show that the axalty of ths dffuso layer was obtaed by usg a axal dstace separable code [4]. Ths fact also ca be show by the slar ethods used the proof of Theore 4. Sce the addtve operato of the fte feld GF(2 ) s the bt-wse EXOR, the Hag weghts of EXORs aog colu vectors of the atrx are reflected to copute β d. See [6] for the detals. IV. PROVABE SECURITY AGAINST DC AND C FOR THE AXIA DIFFUSION AYER I ths secto we wll gve a provable securty for the SDS fucto wth a axal dffuso layer agast DC ad C. Recall that a dffuso layer s called axal f β d (D) = β l (D) = +. By Theore, we kow that the practcal securty for the SDS fucto wth a axal dffuso layer ca be estated as DCP SDS ax p + ad AP SDS ax q +. However, ths does ot gve provable securty o the vewpot of theoretc easure. ETRI Joural, Volue 23, Nuber 4, Deceber 2 Ju-Sug Kag et al. 63
7 Now we cosder the provable securty agast DC ad C o the pot of vew of dfferetal ad lear hull. et us call ' a s k subatrx of f ' s of the followg for: ' = j 2 j k j j2 2 j2 k j2 O jk 2 jk. k jk The, we say that cotas ' as a s k subatrx. ea 3 et be the atrx represetg a dffuso layer D. The β d (D) = + f ad oly f the rak of each k k subatrx of s k for all k. Proof: Ths lea was prove [2]. Corollary 2 If β d (D) s equal to +, β l (D) s also + ad vce versa. Proof: Ths ca be draw by the fact that the rak of equals that of t for ay atrx. I [22], t was show how a axal dffuso layer over GF(2 ) ca be costructed fro a axu dstace separable code. If G e = [I B ] s the echelo for of the geerator atrx of (2,, +) RS-code, the D : GF(2 ) GF(2 ) x a Bx s a axal dffuso layer by ea 3. It s ot ecessary to fx the values of teredate dffereces whe we cosder dfferetals of SDS fucto. Therefore, the dfferetal characterstc of SDS fucto wth put dfferece x ad output dfferece y s defed by SDS y) = w,..., w [ = = S S ( z ) ] y ), () where w ca have ay output dfferece value the frst substtuto layer ad z s D( w). By slar arguet, we ca defe lear hull probablty wth put ask value Γx ad output ask value Γy as follows: P = SDS ( Γy Γx) S S [ = P ( Γy Γz ) = P ( Γw Γx )], Γz,..., Γz where Γz s every possble put ask value of secod substtuto layer ad Γw = D - (Γz). Δx = Δx Δx 2 Δx S S 2 Δw = Δw Δw 2 Δw Dffuso layer S Δz = Δz Δz 2 Δz S S 2 S Δy = Δy Δy 2 Δy Fg. 2. Dfferetal of SDS fucto. Гx = Гx Гx 2 Гx S S 2 Гw = Гw Гw 2 Гw Dffuso layer Fg. 3. ear hull of SDS fucto. S Гz = Гz Гz 2 Гz S S 2 S Гy = Гy Гy 2 Гy Defto 8 The axu dfferetal ad lear hull probabltes of the SDS fucto are defed by ad respectvely. SDS ax = ax x, y SDS ( x y) P SDS ax = ax Γx, Γy P SDS (Γy Γx), ea 4 et be the atrx represetg a dffuso layer D ad β d (D) = +. I Fg. 2, f H c ( x) = k ad H c ( y) = -s+ ( s k ), the there s a dex set {,, s- } such that x,, x s- ad { w,, w s- } are detered by the other w 's. Proof: Wthout loss of geeralty we ay assue y =,, y s- = (or equvaletly z =,, z s- = ). 64 Ju-Sug Kag et al. ETRI Joural, Volue 23, Nuber 4, Deceber 2
8 et w = ( w,, w k ) t be the collecto of all o-zero copoets w = ( w,, w ) t. That s, w j for all j k ad w = f t {,, k }. et = s O s s s s s s O k s k By the deftos of ad w ad the assupto o y, w =. et us dvde w to two parts, w I ad w II, ad to I ad II as followgs: ad w I = ( w,, w s- ) t, w II = ( w s,, w k ) t, I II = s = O s s s s s s O k s k Fro w =, we get I w I + II w II = (or equvaletly I w I = II w II ). Sce I s a vertble atrx by ea 3, we have the equato: w I = I II w II. Hece { w,, w s- } are detered by { w s,, w k }. ea 4 eas that the suato () s ot take for all w,, w but take for all w k j,, w jk-s+ for soe dex set {j,, j k-s+ } {,, k }. Now, we are ready to prove the followg theore. Theore 5 If β d (D) = +, SDS ax of the SDS fucto s bouded by p. Proof: Cosder the dfferetal as depcted Fg. 2. et H c ( x) = k ad H c ( y) = -s+ (s k), the wthout loss of geeralty we ay assue. x,, x k (2) (equvaletly, w,, w k ) ad y j,, y j-s+ (3) (equvaletly, z j,, z j-s+ ). The, = = = SDS = p y) w,..., w w,..., w k [ [ = k = w,..., wk s+ w,..., wk s+ s+ = [ S S [ k = w,..., wk s+ s+ t = k s+ l= [ ) ) S j S k s+ l= ( z Sl j S j ) ( z = y Sl t j )] y ( z )] ) p t j t S j s y )] p (4) (5) s+ ] )] p (6) Equato (4) follows fro assuptos (2) ad (3), (5) follows fro ea 4, ad the equalty (6) follows fro the defto of p. We apply slar arguet to C. Therefore, we ca coclude the followg theore. Theore 6 If β l (D) = +, P SDS ax of the SDS fucto s bouded by q. Proof: A proof of ths theore s very slar to that of Theore 5. V. PROVABE SECURITY AGAINST DC AND C FOR THE SEI-AXIA DIFFUSION AYER I ths secto, we show that the probablty of each dfferetal (resp. lear hull) s bouded by p - (resp. q - ) whe β d (D) (resp. β l (D)) s equal to. A dffuso layer s called seaxal wth respect to DC (resp. C) whe β d (D) (resp. β l (D)) equals. Also we say that a dffuso layer s seaxal f β d (D) ad β l (D) are equal to. ea 5 If β d (D) =, the the rak of each k k subatrx of s greater tha or equal to k- for all k ad there exsts at least oe s s subatx wth rak s- for soe s. Proof: et β d (D) = ad suppose that there exsts a k k subatrx k of whose rak s less tha k-. That s, there exst at least two depedet vectors v, w GF(2 ) k so that k v = k w =. We ca ake a vector x GF(2 ) k wth H c (x) k- ad k x = by a lear cobato of v ad w over GF(2 ). Fro x ad k, we ca get a vector X GF(2 ) such that H c (X) k- ad H c (X) -k. Ths cotradcts to the fact that β d (D) s equal to. Hece the rak of t ETRI Joural, Volue 23, Nuber 4, Deceber 2 Ju-Sug Kag et al. 65
9 each k k subatrx of s greater tha or equal to k- for all k. By ea 3, there exsts at least oe s s subatrx wth rak s-. We also gve a stateet slar to ea 4. et be the atrx represetg a dffuso layer D ad β d (D) =. I Fg. 2, f H c ( x) = k ad H c ( y) = -s (s k), there s a dex set {,, s- } such that { w,, w s- } are represeted by the other w s. A proof of ths stateet s slar to that of ea 4. Theore 7 If β d (D) =, SDS ax of the SDS fucto s bouded by p -. Proof: We use the sae otatos as used the proof of Theore 5. There s oly oe dfferece betwee the proof of Theore 5 ad that of ths theore; H c ( y) s ot -s+ but -s. Thus SDS ( x y) goes up by p -. Hece we have SDS ax p -. Corollary 3 If β l (D) =, P SDS ax of the SDS fucto s bouded by q -. We ca geeralze Theore 7 ad Corollary 3 ad get the followg theore. Theore 8 If β d (D) = -t (or β l (D) = -t), SDS ax (or P SDS ax ) of the SDS fucto s bouded by p -(t+) (or q -(t+) ). Sketch of Proof: Note that a u v atrx wth a rak w cotas a w w vertble subatrx. It ca be easly checked that f β d (D) = -t, the the rak of each k k subatrx of s greater tha or equal to k-t- for all t+ k. I Fg. 2, let H c ( x) = k ad H c ( y) = -s (s k). The, we ca prove there s a dex set {,, s- } such that { w,, w s- } are represeted by the other w s. By slar arguet to the proof of Theore 7, t ca be show that SDS ax p -(t+). VI. CONCUSION We exaed the dffuso layers of soe block cphers referred to as substtuto-perutato etworks. We vestgated the practcal securty of dffuso layers agast dfferetal ad lear cryptaalyss by usg the oto of actve S-boxes. We showed that the u uber of dfferetally actve S- boxes ad that of learly actve S-boxes were geerally ot detcal ad proposed soe specal codtos whch those were detcal. The optal dffuso effects for soe dffuso layers accordg to ther each costrat were also studed. I ters of provable securty, we proved that the probablty of each dfferetal (resp. lear hull) of the SDS fucto wth a axal dffuso layer was bouded by p (resp. q ) ad that of each dfferetal (resp. lear hull) of the SDS fucto wth a se-axal dffuso layer was bouded by p - (resp. q - ), where p ad q were axu dfferetal ad lear probabltes of the substtuto layer, respectvely. REFERENCES [] C.E. Shao, Coucato Theory of Secrecy Systes, Bell Syste Tech. J., 28, 949, pp [2] E. Bha ad A. Shar, Dfferetal Cryptaalyss of DES-ke Cryptosystes, Advaces Cryptology-CRYPTO'9, NCS 537, Sprger-Verlag, 99, pp [3] E. Bha ad A. Shar, Dfferetal Cryptaalyss of DES-ke Cryptosystes, J. of Cryptology, o. 4, 99, pp [4] X. a, J.. assey, ad S. urphy, arkov Cphers ad Dfferetal Cryptaalyss, Advaces Cryptology-Eurocrypt'9, NCS 547, Sprger-Verlag, 99, pp [5]. atsu, ear Cryptaalyss ethod for DES Cpher, Advaces Cryptology-Eurocrypt'93, NCS 765, Sprger-Verlag, 993, pp [6] K. Nyberg, ear Approxato of Block Cphers, Advaces Cryptology-Eurocrypt'94, NCS 95, Sprger-Verlag, 994, pp [7]. Kada, Y. Takasha, T. atsuoto, K. Aok, ad K. Ohta, A Strategy for Costructg Fast Roud Fuctos wth Practcal Securty agast Dfferetal ad ear Cryptaalyss, Selected Areas Cryptography, NCS 556, 999, pp [8] K. Aok ad K. Ohta, Strct Evaluato of the axu Average of Dfferetal Probablty ad the axu Average of ear Probablty, IEICE TRANS. FUNDAENTAS, o., 997, pp [9] Y. Kaeko, F. Sao, ad K. Sakura, O Provable Securty a- gast Dfferetal ad ear Cryptaalyss Geeralzed Festel Cphers wth ultple Rado Fuctos, Proc. of SAC'97, 997, pp []. atsu, New Structure of Block Cphers wth Provable Securty agast Dfferetal ad ear Cryptalayss, Fast Software Ecrypto, NCS 39, Sprger-Verlag, 996, pp [] K. Nyberg ad.r. Kudse, Provable Securty agast Dfferetal Cryptaalyss, J. of Cryptology, o. 8, (), 995, pp [2] J. Daee,.R. Kudse, ad V. Rje, The Block Cpher SQUARE, Fast Software Ecrypto, NCS 267, Sprger- Verlag, 997, pp [3].R. Kudse, Practcally Secure Festel Cphers, Fast Software Ecrypto, NCS 89, 994, pp [4] V. Rje, J. Daee, B. Preeel, A. Bosselaers, ad E.D. W, 66 Ju-Sug Kag et al. ETRI Joural, Volue 23, Nuber 4, Deceber 2
10 The Cpher SHARK, Fast Software Ecrypto, NCS 39, Sprger-Verlag, 996, pp [5] S.H. Hog, S.J. ee, J.I., J.C. Sug, ad D.H. Choe, Provable Securty agast Dfferetal ad ear Cryptaayss for the SPN structure, Proc. of FSE2, NCS Sprger- Verlag, 978, pp [6] J.S. Kag, C.S. Park, S.J. ee, ad J.I., O the Optal Dffuso ayer wth Practcal Securty agast Dfferetal ad ear Cryptaalyss, Proc. of ICISC'99, NCS 787, Sprger- Verlag, 999, pp [7]. Kada, Practcal Securty Evaluato agast Dfferetal ad ear Cryptaalyses for Festel Cphers wth SPN Roud Fucto, Selected Areas Cryptography, NCS 22, Sprger- Verlag, 2, pp [8] C.H., CRYPTON: A New 28-Bt Block Cpher, AES Proposal, 998. [9] J. Daee ad V. Rje, The Rjdael Block Cpher, AES Proposal, 998. [2] NTT-Nppo Telegraph ad Telephoe Corporato, E2: Effcet Ecrypto Algorth, AES Proposal, 998. [2] F.J. acwllas ad N.J.A. Sloa, The Theory of Error- Correctg Codes, NorthHollad, Asterda, 977. [22] J. Daee, R. Govaerts, ad J. Vadewalle, Correlato atrces, Fast Software Ecrypto, NCS 8, Sprger-Verlag, 994, pp Ju-Sug Kag receved the B.S.,.S., ad Ph.D. degrees atheatcs fro Korea Uversty, Seoul, Korea 989, 99, ad 996, respectvely. He joed ETRI 997, ad he s curretly wth Iforato Securty Dvso of ETRI. Hs curret research terests clude cryptographc algorths ad protocols. Seokhe Hog receved the B.S. ad.s. degrees atheatcs fro Korea Uversty. 995 ad 997, respectvely. He also receved the Ph.D. degree atheatcs fro Korea Uversty 2. Sce 2, he has bee wth Ceter for Iforato Securty Techologes (CIST) Korea Uversty. Hs curret research terests clude block cpher aalyss ad publc key cryptosyste. Sagj ee receved the B.S. ad.s. degrees atheatcs fro Korea Uversty 987 ad 989, respectvely. He also receved the Ph.D. degree atheatcs fro Korea Uversty 994. Fro 989 to 998, he was a Techcal Staff eber at ETRI. Fro 999 to 2, he was a faculty eber of atheatcs Departet at Korea Uversty, ad sce 2, he has bee a Assocate Professor of Graduate School of Iforato Securty. Hs curret research terests clude strea cpher ad block cpher. Okyeo Y receved the B.S. ad.s. degrees Korea Uversty, Seoul, Korea 988 ad 99, respectvely. He also receved the Ph.D. degree atheatcs fro Uversty of Ketucky, KY, USA 996. He s curretly teachg atheatcs ad cryptology at the Departet of atheatcs Kook Uversty, Seoul, Korea. Hs curret research terests clude ellptc curve cryptography ad forato securty IT-2. Choosk Park receved the B.S. degree fro Kwagwoo Uversty ad the.s. fro Hayag Uversty, Seoul, Korea 98 ad 983, respectvely, ad the Dr. Eg. degree electroc egeerg fro Tokyo Isttute of Techology, Tokyo, Japa 995. Sce jog Codg Techology ad Research Secto of ETRI 982, he has bee egaged research ad developet o forato securty. Hs research terests are forato securty ad cryptographc protocols. Jog receved the B.S. ad.s. degrees atheatcs fro Korea Uversty 98 ad 982, respectvely. He also receved the Ph.D. degree atheatcs fro Korea Uversty 986. Sce jog Korea Uversty 986, he was a Professor of atheatcs Departet utl 2, ad sce 2, he has bee a Professor of Graduate School of Iforato Securty. Hs curret research terests clude block cpher ad publc key cryptosyste. ETRI Joural, Volue 23, Nuber 4, Deceber 2 Ju-Sug Kag et al. 67
Some Different Perspectives on Linear Least Squares
Soe Dfferet Perspectves o Lear Least Squares A stadard proble statstcs s to easure a respose or depedet varable, y, at fed values of oe or ore depedet varables. Soetes there ests a deterstc odel y f (,,
More informationHongjun Wu, Feng Bao, Dingfeng Ye, and Robert H. Deng
Cryptaalyss of the Perutato Protecto Schees Hogu Wu, Feg Bao, gfeg Ye, ad Robert H. eg et Rdge gtal abs Heg Mu eg errace Sgpore 963 {hogu, baofeg, dfye, deg}@krdl.org.sg Abstract. Aderso ad uh have proposed
More informationA New Method for Solving Fuzzy Linear. Programming by Solving Linear Programming
ppled Matheatcal Sceces Vol 008 o 50 7-80 New Method for Solvg Fuzzy Lear Prograg by Solvg Lear Prograg S H Nasser a Departet of Matheatcs Faculty of Basc Sceces Mazadara Uversty Babolsar Ira b The Research
More informationKURODA S METHOD FOR CONSTRUCTING CONSISTENT INPUT-OUTPUT DATA SETS. Peter J. Wilcoxen. Impact Research Centre, University of Melbourne.
KURODA S METHOD FOR CONSTRUCTING CONSISTENT INPUT-OUTPUT DATA SETS by Peter J. Wlcoxe Ipact Research Cetre, Uversty of Melboure Aprl 1989 Ths paper descrbes a ethod that ca be used to resolve cossteces
More informationA Conventional Approach for the Solution of the Fifth Order Boundary Value Problems Using Sixth Degree Spline Functions
Appled Matheatcs, 1, 4, 8-88 http://d.do.org/1.4/a.1.448 Publshed Ole Aprl 1 (http://www.scrp.org/joural/a) A Covetoal Approach for the Soluto of the Ffth Order Boudary Value Probles Usg Sth Degree Sple
More informationA tighter lower bound on the circuit size of the hardest Boolean functions
Electroc Colloquum o Computatoal Complexty, Report No. 86 2011) A tghter lower boud o the crcut sze of the hardest Boolea fuctos Masak Yamamoto Abstract I [IPL2005], Fradse ad Mlterse mproved bouds o the
More information7.0 Equality Contraints: Lagrange Multipliers
Systes Optzato 7.0 Equalty Cotrats: Lagrage Multplers Cosder the zato of a o-lear fucto subject to equalty costrats: g f() R ( ) 0 ( ) (7.) where the g ( ) are possbly also olear fuctos, ad < otherwse
More informationCoherent Potential Approximation
Coheret Potetal Approxato Noveber 29, 2009 Gree-fucto atrces the TB forals I the tght bdg TB pcture the atrx of a Haltoa H s the for H = { H j}, where H j = δ j ε + γ j. 2 Sgle ad double uderles deote
More informationAlgorithms behind the Correlation Setting Window
Algorths behd the Correlato Settg Wdow Itroducto I ths report detaled forato about the correlato settg pop up wdow s gve. See Fgure. Ths wdow s obtaed b clckg o the rado butto labelled Kow dep the a scree
More informationChapter 9 Jordan Block Matrices
Chapter 9 Jorda Block atrces I ths chapter we wll solve the followg problem. Gve a lear operator T fd a bass R of F such that the matrx R (T) s as smple as possble. f course smple s a matter of taste.
More information2/20/2013. Topics. Power Flow Part 1 Text: Power Transmission. Power Transmission. Power Transmission. Power Transmission
/0/0 Topcs Power Flow Part Text: 0-0. Power Trassso Revsted Power Flow Equatos Power Flow Proble Stateet ECEGR 45 Power Systes Power Trassso Power Trassso Recall that for a short trassso le, the power
More informationStationary states of atoms and molecules
Statoary states of atos ad olecules I followg wees the geeral aspects of the eergy level structure of atos ad olecules that are essetal for the terpretato ad the aalyss of spectral postos the rotatoal
More informationA Penalty Function Algorithm with Objective Parameters and Constraint Penalty Parameter for Multi-Objective Programming
Aerca Joural of Operatos Research, 4, 4, 33-339 Publshed Ole Noveber 4 ScRes http://wwwscrporg/oural/aor http://ddoorg/436/aor4463 A Pealty Fucto Algorth wth Obectve Paraeters ad Costrat Pealty Paraeter
More information18.413: Error Correcting Codes Lab March 2, Lecture 8
18.413: Error Correctg Codes Lab March 2, 2004 Lecturer: Dael A. Spelma Lecture 8 8.1 Vector Spaces A set C {0, 1} s a vector space f for x all C ad y C, x + y C, where we take addto to be compoet wse
More informationCHAPTER 4 RADICAL EXPRESSIONS
6 CHAPTER RADICAL EXPRESSIONS. The th Root of a Real Number A real umber a s called the th root of a real umber b f Thus, for example: s a square root of sce. s also a square root of sce ( ). s a cube
More informationImpossible Differential Cryptanalysis on Lai-Massey Scheme
Impossble Dfferetal Cryptaalyss o La-assey Scheme Ru Guo Chehu J The La-assey scheme proposed by Vaudeay s a modfed structure the Iteratoal Data Ecrypto Algorthm cpher A famly of block cphers amed FOX
More informationSome results and conjectures about recurrence relations for certain sequences of binomial sums.
Soe results ad coectures about recurrece relatos for certa sequeces of boal sus Joha Cgler Faultät für Matheat Uverstät We A-9 We Nordbergstraße 5 Joha Cgler@uveacat Abstract I a prevous paper [] I have
More informationSUBCLASS OF HARMONIC UNIVALENT FUNCTIONS ASSOCIATED WITH SALAGEAN DERIVATIVE. Sayali S. Joshi
Faculty of Sceces ad Matheatcs, Uversty of Nš, Serba Avalable at: http://wwwpfacyu/float Float 3:3 (009), 303 309 DOI:098/FIL0903303J SUBCLASS OF ARMONIC UNIVALENT FUNCTIONS ASSOCIATED WIT SALAGEAN DERIVATIVE
More informationfor each of its columns. A quick calculation will verify that: thus m < dim(v). Then a basis of V with respect to which T has the form: A
Desty of dagoalzable square atrces Studet: Dael Cervoe; Metor: Saravaa Thyagaraa Uversty of Chcago VIGRE REU, Suer 7. For ths etre aer, we wll refer to V as a vector sace over ad L(V) as the set of lear
More informationSummary of the lecture in Biostatistics
Summary of the lecture Bostatstcs Probablty Desty Fucto For a cotuos radom varable, a probablty desty fucto s a fucto such that: 0 dx a b) b a dx A probablty desty fucto provdes a smple descrpto of the
More informationPRACTICAL CONSIDERATIONS IN HUMAN-INDUCED VIBRATION
PRACTICAL CONSIDERATIONS IN HUMAN-INDUCED VIBRATION Bars Erkus, 4 March 007 Itroducto Ths docuet provdes a revew of fudaetal cocepts structural dyacs ad soe applcatos hua-duced vbrato aalyss ad tgato of
More informationOn Probability of Undetected Error for Hamming Codes over Q-ary Symmetric Channel
Joural of Coucato ad Coputer 8 (2 259-263 O Probablty of Udetected Error for Hag Codes over Q-ary Syetrc Chael Mash Gupta, Jaskar Sgh Bhullar 2 ad O Parkash Vocha 3. D.A.V. College, Bathda 5, Ida 2. Malout
More informationPseudo-random Functions
Pseudo-radom Fuctos Debdeep Mukhopadhyay IIT Kharagpur We have see the costructo of PRG (pseudo-radom geerators) beg costructed from ay oe-way fuctos. Now we shall cosder a related cocept: Pseudo-radom
More informationStandard Deviation for PDG Mass Data
4 Dec 06 Stadard Devato for PDG Mass Data M. J. Gerusa Retred, 47 Clfde Road, Worghall, HP8 9JR, UK. gerusa@aol.co, phoe: +(44) 844 339754 Abstract Ths paper aalyses the data for the asses of eleetary
More informationFunctions of Random Variables
Fuctos of Radom Varables Chapter Fve Fuctos of Radom Varables 5. Itroducto A geeral egeerg aalyss model s show Fg. 5.. The model output (respose) cotas the performaces of a system or product, such as weght,
More informationThird handout: On the Gini Index
Thrd hadout: O the dex Corrado, a tala statstca, proposed (, 9, 96) to measure absolute equalt va the mea dfferece whch s defed as ( / ) where refers to the total umber of dvduals socet. Assume that. The
More informationKNOWLEDGE-BASED SCHEMA FOR S-BOX DESIGN
IJRRA 8 (3) epteber 0 www.arpapress.co/volues/vol8issue3/ijrra_8_3_04.pdf KNOWLEDGE-BAED CHEMA FOR -BOX DEIGN Gabrela Mose Petroleu-Gas Uversty of Ploest, o. 39 Blvd. Bucurest, Ploest, Roaa ABTRACT I cryptography,
More informationConstruction of Composite Indices in Presence of Outliers
Costructo of Coposte dces Presece of Outlers SK Mshra Dept. of Ecoocs North-Easter Hll Uversty Shllog (da). troducto: Oftetes we requre costructg coposte dces by a lear cobato of a uber of dcator varables.
More informationAn Innovative Algorithmic Approach for Solving Profit Maximization Problems
Matheatcs Letters 208; 4(: -5 http://www.scecepublshggroup.co/j/l do: 0.648/j.l.208040. ISSN: 2575-503X (Prt; ISSN: 2575-5056 (Ole A Iovatve Algorthc Approach for Solvg Proft Maxzato Probles Abul Kala
More informationAnalytical Study of Fractal Dimension Types in the Context of SPC Technical Paper. Noa Ruschin Rimini, Irad Ben-Gal and Oded Maimon
Aalytcal Study of Fractal Deso Types the Cotext of SPC Techcal Paper oa Rusch R, Irad Be-Gal ad Oded Mao Departet of Idustral Egeerg, Tel-Avv Uversty, Tel-Avv, Israel Ths paper provdes a aalytcal study
More informationJournal of Mathematical Analysis and Applications
J. Math. Aal. Appl. 365 200) 358 362 Cotets lsts avalable at SceceDrect Joural of Mathematcal Aalyss ad Applcatos www.elsever.com/locate/maa Asymptotc behavor of termedate pots the dfferetal mea value
More informationA Characterization of Jacobson Radical in Γ-Banach Algebras
Advaces Pure Matheatcs 43-48 http://dxdoorg/436/ap66 Publshed Ole Noveber (http://wwwscrporg/joural/ap) A Characterzato of Jacobso Radcal Γ-Baach Algebras Nlash Goswa Departet of Matheatcs Gauhat Uversty
More informationThe Number of the Two Dimensional Run Length Constrained Arrays
2009 Iteratoal Coferece o Mache Learg ad Coutg IPCSIT vol.3 (20) (20) IACSIT Press Sgaore The Nuber of the Two Desoal Ru Legth Costraed Arrays Tal Ataa Naohsa Otsua 2 Xuerog Yog 3 School of Scece ad Egeerg
More informationOn the Capacity of Bounded Rank Modulation for Flash Memories
O the Capacty of Bouded Rak Modulato for Flash Meores Zhyg Wag Electrcal Egeerg Departet Calfora Isttute of Techology Pasadea, CA 925, USA Eal: zhyg@paradsecaltechedu Axao (Adrew) Jag Coputer Scece Departet
More informationLecture 8 IEEE DCF Performance
Lecture 8 IEEE82. DCF Perforace IEEE82. DCF Basc Access Mechas A stato wth a ew packet to trast otors the chael actvty. If the chael s dle for a perod of te equal to a dstrbuted terfrae space (DIFS), the
More informationMaps on Triangular Matrix Algebras
Maps o ragular Matrx lgebras HMED RMZI SOUROUR Departmet of Mathematcs ad Statstcs Uversty of Vctora Vctora, BC V8W 3P4 CND sourour@mathuvcca bstract We surveys results about somorphsms, Jorda somorphsms,
More informationEstimation of Stress- Strength Reliability model using finite mixture of exponential distributions
Iteratoal Joural of Computatoal Egeerg Research Vol, 0 Issue, Estmato of Stress- Stregth Relablty model usg fte mxture of expoetal dstrbutos K.Sadhya, T.S.Umamaheswar Departmet of Mathematcs, Lal Bhadur
More informationEntropy ISSN by MDPI
Etropy 2003, 5, 233-238 Etropy ISSN 1099-4300 2003 by MDPI www.mdp.org/etropy O the Measure Etropy of Addtve Cellular Automata Hasa Aı Arts ad Sceces Faculty, Departmet of Mathematcs, Harra Uversty; 63100,
More informationHájek-Rényi Type Inequalities and Strong Law of Large Numbers for NOD Sequences
Appl Math If Sc 7, No 6, 59-53 03 59 Appled Matheatcs & Iforato Sceces A Iteratoal Joural http://dxdoorg/0785/as/070647 Háje-Réy Type Iequaltes ad Strog Law of Large Nuers for NOD Sequeces Ma Sogl Departet
More informationOn the Capacity of Bounded Rank Modulation for Flash Memories
O the Capacty of Bouded Rak Modulato for Flash Meores Zhyg Wag Electrcal Egeerg Departet Calfora Isttute of Techology Pasadea, CA 95, USA Eal: zhyg@paradsecaltechedu Axao (Adrew) Jag Coputer Scece Departet
More informationSymmetry of the Solution of Semidefinite Program by Using Primal-Dual Interior-Point Method
Syetry of the Soluto of Sedefte Progra by Usg Pral-Dual Iteror-Pot Method Yoshhro Kao Makoto Ohsak ad Naok Katoh Departet of Archtecture ad Archtectural Systes Kyoto Uversty Kyoto 66-85 Japa kao@s-jarchkyoto-uacjp
More informationMULTIOBJECTIVE NONLINEAR FRACTIONAL PROGRAMMING PROBLEMS INVOLVING GENERALIZED d - TYPE-I n -SET FUNCTIONS
THE PUBLIHING HOUE PROCEEDING OF THE ROMANIAN ACADEMY, eres A OF THE ROMANIAN ACADEMY Volue 8, Nuber /27,.- MULTIOBJECTIVE NONLINEAR FRACTIONAL PROGRAMMING PROBLEM INVOLVING GENERALIZED d - TYPE-I -ET
More informationThe Mathematical Appendix
The Mathematcal Appedx Defto A: If ( Λ, Ω, where ( λ λ λ whch the probablty dstrbutos,,..., Defto A. uppose that ( Λ,,..., s a expermet type, the σ-algebra o λ λ λ are defed s deoted by ( (,,...,, σ Ω.
More informationStrong Convergence of Weighted Averaged Approximants of Asymptotically Nonexpansive Mappings in Banach Spaces without Uniform Convexity
BULLETIN of the MALAYSIAN MATHEMATICAL SCIENCES SOCIETY Bull. Malays. Math. Sc. Soc. () 7 (004), 5 35 Strog Covergece of Weghted Averaged Appromats of Asymptotcally Noepasve Mappgs Baach Spaces wthout
More informationSolving the fuzzy shortest path problem on networks by a new algorithm
Proceedgs of the 0th WSEAS Iteratoal Coferece o FUZZY SYSTEMS Solvg the fuzzy shortest path proble o etworks by a ew algorth SADOAH EBRAHIMNEJAD a, ad REZA TAVAKOI-MOGHADDAM b a Departet of Idustral Egeerg,
More informationEconometric Methods. Review of Estimation
Ecoometrc Methods Revew of Estmato Estmatg the populato mea Radom samplg Pot ad terval estmators Lear estmators Ubased estmators Lear Ubased Estmators (LUEs) Effcecy (mmum varace) ad Best Lear Ubased Estmators
More informationThe Geometric Least Squares Fitting Of Ellipses
IOSR Joural of Matheatcs (IOSR-JM) e-issn: 78-578, p-issn: 39-765X. Volue 4, Issue 3 Ver.I (May - Jue 8), PP -8 www.osrourals.org Abdellatf Bettayeb Departet of Geeral Studes, Jubal Idustral College, Jubal
More informationPseudo-random Functions. PRG vs PRF
Pseudo-radom Fuctos Debdeep Muhopadhyay IIT Kharagpur PRG vs PRF We have see the costructo of PRG (pseudo-radom geerators) beg costructed from ay oe-way fuctos. Now we shall cosder a related cocept: Pseudo-radom
More informationIdeal multigrades with trigonometric coefficients
Ideal multgrades wth trgoometrc coeffcets Zarathustra Brady December 13, 010 1 The problem A (, k) multgrade s defed as a par of dstct sets of tegers such that (a 1,..., a ; b 1,..., b ) a j = =1 for all
More informationCHAPTER VI Statistical Analysis of Experimental Data
Chapter VI Statstcal Aalyss of Expermetal Data CHAPTER VI Statstcal Aalyss of Expermetal Data Measuremets do ot lead to a uque value. Ths s a result of the multtude of errors (maly radom errors) that ca
More informationSome Notes on the Probability Space of Statistical Surveys
Metodološk zvezk, Vol. 7, No., 200, 7-2 ome Notes o the Probablty pace of tatstcal urveys George Petrakos Abstract Ths paper troduces a formal presetato of samplg process usg prcples ad cocepts from Probablty
More informationOrder Nonlinear Vector Differential Equations
It. Joural of Math. Aalyss Vol. 3 9 o. 3 39-56 Coverget Power Seres Solutos of Hgher Order Nolear Vector Dfferetal Equatos I. E. Kougas Departet of Telecoucato Systes ad Networs Techologcal Educatoal Isttute
More informationD. L. Bricker, 2002 Dept of Mechanical & Industrial Engineering The University of Iowa. CPL/XD 12/10/2003 page 1
D. L. Brcker, 2002 Dept of Mechacal & Idustral Egeerg The Uversty of Iowa CPL/XD 2/0/2003 page Capactated Plat Locato Proble: Mze FY + C X subject to = = j= where Y = j= X D, j =, j X SY, =,... X 0, =,
More informationRademacher Complexity. Examples
Algorthmc Foudatos of Learg Lecture 3 Rademacher Complexty. Examples Lecturer: Patrck Rebesch Verso: October 16th 018 3.1 Itroducto I the last lecture we troduced the oto of Rademacher complexty ad showed
More information3D Reconstruction from Image Pairs. Reconstruction from Multiple Views. Computing Scene Point from Two Matching Image Points
D Recostructo fro Iage ars Recostructo fro ultple Ves Dael Deetho Fd terest pots atch terest pots Copute fudaetal atr F Copute caera atrces ad fro F For each atchg age pots ad copute pot scee Coputg Scee
More informationAnalysis of Lagrange Interpolation Formula
P IJISET - Iteratoal Joural of Iovatve Scece, Egeerg & Techology, Vol. Issue, December 4. www.jset.com ISS 348 7968 Aalyss of Lagrage Iterpolato Formula Vjay Dahya PDepartmet of MathematcsMaharaja Surajmal
More informationInvestigating Cellular Automata
Researcher: Taylor Dupuy Advsor: Aaro Wootto Semester: Fall 4 Ivestgatg Cellular Automata A Overvew of Cellular Automata: Cellular Automata are smple computer programs that geerate rows of black ad whte
More informationPRACTICAL BIJECTIVE S-BOX DESIGN
Proceegs o the 5th Asa Matheatcal Coerece, Malaysa 009 PRACTICAL BIJECTIVE S-BOX DESIG Aburash Maaolov, Hera Isa, Moesa Soehela Mohaa,, Iorato Securty Cluster, Malaysa Isttute o Mcroelectroc Systes, Techology
More informationAssignment 5/MATH 247/Winter Due: Friday, February 19 in class (!) (answers will be posted right after class)
Assgmet 5/MATH 7/Wter 00 Due: Frday, February 9 class (!) (aswers wll be posted rght after class) As usual, there are peces of text, before the questos [], [], themselves. Recall: For the quadratc form
More informationPTAS for Bin-Packing
CS 663: Patter Matchg Algorthms Scrbe: Che Jag /9/00. Itroducto PTAS for B-Packg The B-Packg problem s NP-hard. If we use approxmato algorthms, the B-Packg problem could be solved polyomal tme. For example,
More informationGlobal Optimization for Solving Linear Non-Quadratic Optimal Control Problems
Joural of Appled Matheatcs ad Physcs 06 4 859-869 http://wwwscrporg/joural/jap ISSN Ole: 37-4379 ISSN Prt: 37-435 Global Optzato for Solvg Lear No-Quadratc Optal Cotrol Probles Jghao Zhu Departet of Appled
More informationSolutions to problem set ); (, ) (
Solutos to proble set.. L = ( yp p ); L = ( p p ); y y L, L = yp p, p p = yp p, + p [, p ] y y y = yp + p = L y Here we use for eaple that yp, p = yp p p yp = yp, p = yp : factors that coute ca be treated
More informationBlock-Based Compact Thermal Modeling of Semiconductor Integrated Circuits
Block-Based Compact hermal Modelg of Semcoductor Itegrated Crcuts Master s hess Defese Caddate: Jg Ba Commttee Members: Dr. Mg-Cheg Cheg Dr. Daqg Hou Dr. Robert Schllg July 27, 2009 Outle Itroducto Backgroud
More informationL5 Polynomial / Spline Curves
L5 Polyomal / Sple Curves Cotets Coc sectos Polyomal Curves Hermte Curves Bezer Curves B-Sples No-Uform Ratoal B-Sples (NURBS) Mapulato ad Represetato of Curves Types of Curve Equatos Implct: Descrbe a
More informationA Study on Generalized Generalized Quasi hyperbolic Kac Moody algebra QHGGH of rank 10
Global Joural of Mathematcal Sceces: Theory ad Practcal. ISSN 974-3 Volume 9, Number 3 (7), pp. 43-4 Iteratoal Research Publcato House http://www.rphouse.com A Study o Geeralzed Geeralzed Quas (9) hyperbolc
More informationMATH 247/Winter Notes on the adjoint and on normal operators.
MATH 47/Wter 00 Notes o the adjot ad o ormal operators I these otes, V s a fte dmesoal er product space over, wth gve er * product uv, T, S, T, are lear operators o V U, W are subspaces of V Whe we say
More informationSolving Constrained Flow-Shop Scheduling. Problems with Three Machines
It J Cotemp Math Sceces, Vol 5, 2010, o 19, 921-929 Solvg Costraed Flow-Shop Schedulg Problems wth Three Maches P Pada ad P Rajedra Departmet of Mathematcs, School of Advaced Sceces, VIT Uversty, Vellore-632
More informationDerivation of 3-Point Block Method Formula for Solving First Order Stiff Ordinary Differential Equations
Dervato of -Pot Block Method Formula for Solvg Frst Order Stff Ordary Dfferetal Equatos Kharul Hamd Kharul Auar, Kharl Iskadar Othma, Zara Bb Ibrahm Abstract Dervato of pot block method formula wth costat
More informationBounds on the expected entropy and KL-divergence of sampled multinomial distributions. Brandon C. Roy
Bouds o the expected etropy ad KL-dvergece of sampled multomal dstrbutos Brado C. Roy bcroy@meda.mt.edu Orgal: May 18, 2011 Revsed: Jue 6, 2011 Abstract Iformato theoretc quattes calculated from a sampled
More informationChapter 5 Properties of a Random Sample
Lecture 6 o BST 63: Statstcal Theory I Ku Zhag, /0/008 Revew for the prevous lecture Cocepts: t-dstrbuto, F-dstrbuto Theorems: Dstrbutos of sample mea ad sample varace, relatoshp betwee sample mea ad sample
More informationPolyphase Filters. Section 12.4 Porat
Polyphase Flters Secto.4 Porat .4 Polyphase Flters Polyphase s a way of dog saplg-rate coverso that leads to very effcet pleetatos. But ore tha that, t leads to very geeral vewpots that are useful buldg
More informationChapter 4 Multiple Random Variables
Revew for the prevous lecture: Theorems ad Examples: How to obta the pmf (pdf) of U = g (, Y) ad V = g (, Y) Chapter 4 Multple Radom Varables Chapter 44 Herarchcal Models ad Mxture Dstrbutos Examples:
More informationDescriptive Statistics
Page Techcal Math II Descrptve Statstcs Descrptve Statstcs Descrptve statstcs s the body of methods used to represet ad summarze sets of data. A descrpto of how a set of measuremets (for eample, people
More informationA Family of Non-Self Maps Satisfying i -Contractive Condition and Having Unique Common Fixed Point in Metrically Convex Spaces *
Advaces Pure Matheatcs 0 80-84 htt://dxdoorg/0436/a04036 Publshed Ole July 0 (htt://wwwscrporg/oural/a) A Faly of No-Self Mas Satsfyg -Cotractve Codto ad Havg Uque Coo Fxed Pot Metrcally Covex Saces *
More informationDiscrete Mathematics and Probability Theory Fall 2016 Seshia and Walrand DIS 10b
CS 70 Dscrete Mathematcs ad Probablty Theory Fall 206 Sesha ad Walrad DIS 0b. Wll I Get My Package? Seaky delvery guy of some compay s out delverg packages to customers. Not oly does he had a radom package
More informationA note on An efficient certificateless aggregate signature with constant pairing computations
A ote o A effcet certfcateless aggregate sgature wth costat parg computatos Debao He Maomao Ta Jahua Che School of Mathematcs ad Statstcs Wuha Uversty Wuha Cha School of Computer Scece ad Techology Uversty
More informationGrowth of a Class of Plurisubharmonic Function in a Unit Polydisc I
Issue, Volue, 7 5 Growth of a Class of Plursubharoc Fucto a Ut Polydsc I AITASU SINHA Abstract The Growth of a o- costat aalytc fucto of several coplex varables s a very classcal cocept, but for a fte
More informationPROJECTION PROBLEM FOR REGULAR POLYGONS
Joural of Mathematcal Sceces: Advaces ad Applcatos Volume, Number, 008, Pages 95-50 PROJECTION PROBLEM FOR REGULAR POLYGONS College of Scece Bejg Forestry Uversty Bejg 0008 P. R. Cha e-mal: sl@bjfu.edu.c
More informationNon-degenerate Perturbation Theory
No-degeerate Perturbato Theory Proble : H E ca't solve exactly. But wth H H H' H" L H E Uperturbed egevalue proble. Ca solve exactly. E Therefore, kow ad. H ' H" called perturbatos Copyrght Mchael D. Fayer,
More information( ) ( ) ( ( )) ( ) ( ) ( ) ( ) ( ) = ( ) ( ) + ( ) ( ) = ( ( )) ( ) + ( ( )) ( ) Review. Second Derivatives for f : y R. Let A be an m n matrix.
Revew + v, + y = v, + v, + y, + y, Cato! v, + y, + v, + y geeral Let A be a atr Let f,g : Ω R ( ) ( ) R y R Ω R h( ) f ( ) g ( ) ( ) ( ) ( ( )) ( ) dh = f dg + g df A, y y A Ay = = r= c= =, : Ω R he Proof
More informationThe internal structure of natural numbers, one method for the definition of large prime numbers, and a factorization test
Fal verso The teral structure of atural umbers oe method for the defto of large prme umbers ad a factorzato test Emmaul Maousos APM Isttute for the Advacemet of Physcs ad Mathematcs 3 Poulou str. 53 Athes
More informationA CHARACTERIZATION OF THE CLIFFORD TORUS
PROCEEDINGS OF THE AERICAN ATHEATICAL SOCIETY Volue 17, Nuber 3, arch 1999, Pages 819 88 S 000-9939(99)05088-1 A CHARACTERIZATION OF THE CLIFFORD TORUS QING-ING CHENG AND SUSUU ISHIKAWA (Coucated by Chrstopher
More informationTHE TRUNCATED RANDIĆ-TYPE INDICES
Kragujeac J Sc 3 (00 47-5 UDC 547:54 THE TUNCATED ANDIĆ-TYPE INDICES odjtaba horba, a ohaad Al Hossezadeh, b Ia uta c a Departet of atheatcs, Faculty of Scece, Shahd ajae Teacher Trag Uersty, Tehra, 785-3,
More informationQueueing Networks. γ 3
Queueg Networks Systes odeled by queueg etworks ca roughly be grouped to four categores. Ope etworks Custoers arrve fro outsde the syste are served ad the depart. Exaple: acket swtched data etwork. γ µ
More informationSimple Linear Regression
Statstcal Methods I (EST 75) Page 139 Smple Lear Regresso Smple regresso applcatos are used to ft a model descrbg a lear relatoshp betwee two varables. The aspects of least squares regresso ad correlato
More informationAN UPPER BOUND FOR THE PERMANENT VERSUS DETERMINANT PROBLEM BRUNO GRENET
AN UPPER BOUND FOR THE PERMANENT VERSUS DETERMINANT PROBLEM BRUNO GRENET Abstract. The Permaet versus Determat problem s the followg: Gve a matrx X of determates over a feld of characterstc dfferet from
More informationStatistics Descriptive and Inferential Statistics. Instructor: Daisuke Nagakura
Statstcs Descrptve ad Iferetal Statstcs Istructor: Dasuke Nagakura (agakura@z7.keo.jp) 1 Today s topc Today, I talk about two categores of statstcal aalyses, descrptve statstcs ad feretal statstcs, ad
More informationParallelized methods for solving polynomial equations
IOSR Joural of Matheatcs (IOSR-JM) e-issn: 2278-5728, p-issn: 239-765X. Volue 2, Issue 4 Ver. II (Jul. - Aug.206), PP 75-79 www.osrourals.org Paralleled ethods for solvg polyoal equatos Rela Kapçu, Fatr
More information1 Onto functions and bijections Applications to Counting
1 Oto fuctos ad bectos Applcatos to Coutg Now we move o to a ew topc. Defto 1.1 (Surecto. A fucto f : A B s sad to be surectve or oto f for each b B there s some a A so that f(a B. What are examples of
More informationConnective Eccentricity Index of Some Thorny Graphs
Aals of ure ad Appled Matheatcs Vol. 7, No., 04, 59-64 IN: 79-087X (), 79-0888(ole) ublshed o 9 epteber 04 www.researchathsc.org Aals of oectve Eccetrcty Idex of oe Thory raphs Nlaja De, k. Md. Abu Nayee
More informationROOT-LOCUS ANALYSIS. Lecture 11: Root Locus Plot. Consider a general feedback control system with a variable gain K. Y ( s ) ( ) K
ROOT-LOCUS ANALYSIS Coder a geeral feedback cotrol yte wth a varable ga. R( Y( G( + H( Root-Locu a plot of the loc of the pole of the cloed-loop trafer fucto whe oe of the yte paraeter ( vared. Root locu
More information1 Lyapunov Stability Theory
Lyapuov Stablty heory I ths secto we cosder proofs of stablty of equlbra of autoomous systems. hs s stadard theory for olear systems, ad oe of the most mportat tools the aalyss of olear systems. It may
More informationComplete Convergence and Some Maximal Inequalities for Weighted Sums of Random Variables
Joural of Sceces, Islamc Republc of Ira 8(4): -6 (007) Uversty of Tehra, ISSN 06-04 http://sceces.ut.ac.r Complete Covergece ad Some Maxmal Iequaltes for Weghted Sums of Radom Varables M. Am,,* H.R. Nl
More information{ }{ ( )} (, ) = ( ) ( ) ( ) Chapter 14 Exercises in Sampling Theory. Exercise 1 (Simple random sampling): Solution:
Chapter 4 Exercses Samplg Theory Exercse (Smple radom samplg: Let there be two correlated radom varables X ad A sample of sze s draw from a populato by smple radom samplg wthout replacemet The observed
More information1 Mixed Quantum State. 2 Density Matrix. CS Density Matrices, von Neumann Entropy 3/7/07 Spring 2007 Lecture 13. ψ = α x x. ρ = p i ψ i ψ i.
CS 94- Desty Matrces, vo Neuma Etropy 3/7/07 Sprg 007 Lecture 3 I ths lecture, we wll dscuss the bascs of quatum formato theory I partcular, we wll dscuss mxed quatum states, desty matrces, vo Neuma etropy
More informationLecture 3 Probability review (cont d)
STATS 00: Itroducto to Statstcal Iferece Autum 06 Lecture 3 Probablty revew (cot d) 3. Jot dstrbutos If radom varables X,..., X k are depedet, the ther dstrbuto may be specfed by specfyg the dvdual dstrbuto
More informationIntroduction to Matrices and Matrix Approach to Simple Linear Regression
Itroducto to Matrces ad Matrx Approach to Smple Lear Regresso Matrces Defto: A matrx s a rectagular array of umbers or symbolc elemets I may applcatos, the rows of a matrx wll represet dvduals cases (people,
More informationOn Hilbert Kunz Functions of Some Hypersurfaces
JOURNAL OF ALGEBRA 199, 499527 1998 ARTICLE NO. JA977206 O HlbertKuz Fuctos of Soe Hypersurfaces L Chag* Departet of Matheatcs, Natoal Tawa Uersty, Tape, Tawa ad Yu-Chg Hug Departet of Matheatcs, Natoal
More informationNP!= P. By Liu Ran. Table of Contents. The P versus NP problem is a major unsolved problem in computer
NP!= P By Lu Ra Table of Cotets. Itroduce 2. Prelmary theorem 3. Proof 4. Expla 5. Cocluso. Itroduce The P versus NP problem s a major usolved problem computer scece. Iformally, t asks whether a computer
More informationbest estimate (mean) for X uncertainty or error in the measurement (systematic, random or statistical) best
Error Aalyss Preamble Wheever a measuremet s made, the result followg from that measuremet s always subject to ucertaty The ucertaty ca be reduced by makg several measuremets of the same quatty or by mprovg
More information