Top Banner

of 96

AES Candidates

Jul 04, 2018

Download

Documents

robotden
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
  • 8/15/2019 AES Candidates

    1/96

     

    Kris Gaj

    Electrical and Computer Engineering

    George Mason University

    Towards secure cryptographic transformationsefficient in both software and hardware: 

    case for synergy among

    math! computing! and engineering

    http:""ece#gmu#edu"crypto$te%t#htm

  • 8/15/2019 AES Candidates

    2/96

     

    Motivation

  • 8/15/2019 AES Candidates

    3/96

     

    Criteria used to evaluate cryptographic

    transformations

    &ecurity

    &oftware

    Efficiency 

    'ardware

    Efficiency 

    (le%ibility

  • 8/15/2019 AES Candidates

    4/96

     

    (le%ibility

    • dditional )ey$si*es and bloc)$si*es

    • Ability to function efficiently and securely in a wide

      variety of platforms and applications

      low-end smartcards, wireless: small memory requirements  IPSec, ATM small )ey setup time in !ardware

      "-IS#$, satellite communication large encryption speed

  • 8/15/2019 AES Candidates

    5/96

     

    dvanced Encryption &tandard +E&, Contest

    -../$011-

    -2 Candidates from %SA, &anada, "el'ium,

    (rance, )ermany, $orway, %*, Israel,

    *orea, +apan, Australia, &osta ica

    3une -..4

    ugust -...

    5ctober 0111

    - winner: 6ijndael7elgium

    2 final candidates

    Mars, &, i.ndael, Serpent, Twofis!

    6ound -

    6ound 0

    &ecurity

    &oftware efficiency

    (le%ibility

    &ecurity

    'ardware efficiency

  • 8/15/2019 AES Candidates

    6/96

     

    8E&&9E roject8ew European &chemes for &ignatures!9ntegrity! and Encryption

    0111$0110

    C6;T6EC roject0111$0110

    /urope

    +apan

  • 8/15/2019 AES Candidates

    7/96

     

    Multiple types of transformations:

    #evelopment of met!odolo'y of a fair evaluation andcomparison of al'orit!ms belon'in' to t!e same class,

    includin'

      software and hardware efficiency

    8E&&9E! C6;T6EC

    • &ymmetric$)ey bloc) ciphers• &tream ciphers• 'ash functions

    • MCs

    • symmetric encryption schemes• symmetric digital signature schemes

    • symmetric identification schemes

  • 8/15/2019 AES Candidates

    8/96

     0

    10

    200

    210

    300

    310

    400

    410

    500

    510100

    Serpent i.ndael Twofis! & Mars

    &peed of the final E& candidates in hardware

     &peed

  • 8/15/2019 AES Candidates

    9/96

     0

    20

    3040

    50

    10

    0

    60

    70

    80200

    Serpenti.ndael Twofis! & Mars

    &urvey filled by ->/ participants of

    the Third E& Conference! pril 0111? votes

  • 8/15/2019 AES Candidates

    10/96

  • 8/15/2019 AES Candidates

    11/96

     0

    1

    20

    21

    30

    31

    40

    Serpenti.ndael Twofis!& Mars

    Efficiency in software: 89&T$specified platform

    237-bit 9ey283-bit 9ey

    31-bit 9ey

    200 MHz Pentiu Pro, !orland C""

    &peed

  • 8/15/2019 AES Candidates

    12/96

  • 8/15/2019 AES Candidates

    13/96

     

    &ecurity: Theoretical attac)s better

    than e%haustive )ey search

    0 1 20 21 30 31 40 41

    Twofish

    &erpent

    6ijndael

    6C>

    Mars without -> mi%ing rounds

    ? of rounds in the attac)"total ? of rounds

    > ->

    A0.

    / -1

    -2 01

    ->--

    0A

    -1

    2

    A

    2

  • 8/15/2019 AES Candidates

    14/96

     0 20 30 40 50 10 0 60 70 80 200

    Twofish

    &erpent

    6ijndael

    6C>

    Mars

    &ecurity: Theoretical attac)s better

    than e%haustive )ey search

    ? of rounds in the attac)"total ? of rounds -11

    04 /0

    A4 >0

    >. A-

    /1 A1

    /2 02

  • 8/15/2019 AES Candidates

    15/96

     

    0200

    300

    400500

    100

    00600

    A2.

    >-1

    &peed in hardware

  • 8/15/2019 AES Candidates

    16/96

  • 8/15/2019 AES Candidates

    17/96

     

    'istorical view

    &ecret$)ey ciphers 'ash functions

    time

    -./1

    -.41

    -..1

    0111

    #/S optimied for hardware

    (ast Software /ncryption:

    cip!ers optimied for software:e;';, &1, "lowfis!, &5

    A/S optimied for

    software and hardware

    M#5-family

    optimied primarily

    for software

    #/S-based !as! functions

      optimied for hardware

  • 8/15/2019 AES Candidates

    18/96

     

    &oftware or hardwareF

    &5(TD6E '6D6E

    security of data

    during transmission

    fle%ibility

  • 8/15/2019 AES Candidates

    19/96

     

    Efficiency indicators

  • 8/15/2019 AES Candidates

    20/96

     

    Memory

    ower

    consumption

    rimary efficiency indicators

    &oftware 'ardware

    &peed Memory &peed rea

  • 8/15/2019 AES Candidates

    21/96

     

    Efficiency parameters

    Hatency Throughput I &peed

    /ncryption>decryption

    Time to

    encrypt"decrypta single bloc)

    of data

    Mi

    &i

    8umber of bits

    encrypted"decrypted

    in a unit of time

    Encryption"

    decryption

    Mi

    Mi?2

    Mi?3

    &i

    &i?2

    &i?3

    Throughput I7loc)Jsi*e 8umberJofJbloc)sJprocessedJsimultaneously

    Hatency

  • 8/15/2019 AES Candidates

    22/96

  • 8/15/2019 AES Candidates

    23/96

     

    8on$(eedbac) Cipher ModesEC7! counter

    C i f f db ) i h d

  • 8/15/2019 AES Candidates

    24/96

     

    Comparison for non$feedbac) cipher modes! e#g#

    Counter Mode $ CT6 

    M0 M2 M3

    /

    &i @ Mi ⊕

     /

  • 8/15/2019 AES Candidates

    25/96

     

    9ncreasing speed by parallel processing

    /ncryption>

    decryption

    unit

    /ncryption>

    decryption

    unit

    /ncryption>

    decryption

    unit

    /ncryption>

    decryption

    unit

    /ncryption>

    decryption

    unit

    /ncryption>

    decryption

    unit

    9 i d i i li i

  • 8/15/2019 AES Candidates

    26/96

     

    9ncreasing speed using pipelining

    Cipher - Cipher 0

    round -round -

    round 0

    round -1

    # # #

    round ->

    # # #

    &peed I

    targetJcloc)Jperiod

    bloc) si*e

    target

    cloc)

    period!

    e#g#! 01 ns

  • 8/15/2019 AES Candidates

    27/96

     

    ipelined operation of the encryption unit

    "2

    cloc) 

    cycle2

    "3

    3

    "2

    "4

    4

    "3

    "2

    7@

    5

    "4

    "3

    7-

    72

    1

    "5

    "4

    70

    7>

    "1

    "5

    7A

    7/

    6

    "

    "1

    7@

    74

    "6

    "

    72

    7

    7-A

    "5

    "4

    7-1

    7-@

    "1

    "5

    7--

    7-2

    "

    "1

    7-0

    7->

    "6

    "

    7-A

    7.

    "7

    "6

    7>

    7-1

    "8

    "7

    7/

    7--

    "20

    "8

    74

    7-0

    "4

    "3

    7.

    cloc) 

    cycle8 20 22 23 24 25 21 2

    E ti i f db ) d +EC7 t ,

  • 8/15/2019 AES Candidates

    28/96

     

    0

    2000

    3000

    4000

    5000

    1000

    000

    6000

    0 20000 30000 40000 50000 10000 0000

    rea

    6ijndael

    Mars

    5 6 lt ( ll i d i li i

  • 8/15/2019 AES Candidates

    29/96

     

    1

    0

    @

    >

    4

    -1

    -0

    -@

    ->

    -4

    5ur 6esults: (ull mi%ed pipelining

    Throughput

  • 8/15/2019 AES Candidates

    30/96

     0

    1000

    20000

    21000

    30000

    31000

    40000

    41000

    50000

    5100010000

    Serpent i.ndaelTwofis! &

    rea !.11

    -0!>11

    41 6Ms

    dedicated memory bloc9s, AMs

    5ur 6esults: (ull mi%ed pipelining

    89&T 6 t GMU 6 t

  • 8/15/2019 AES Candidates

    31/96

     

    89&T 6eport GMU 6eport:'ardware Efficiency

    8on$feedbac) cipher modes: EC7! CT6 

    &peed

    rea

    'igh

    How

    &mall

    6ijndael

    &erpent

    Twofish

    6C>

    Mars

    Medium

    Medium Harge

  • 8/15/2019 AES Candidates

    32/96

     

    (eedbac) cipher modesC7C! C(7! 5(7

    (eedbac) cipher modes C7C

  • 8/15/2019 AES Candidates

    33/96

     

    (eedbac) cipher modes $ C7C

    M2 M3 M4

    /

    I

    C ) @ /

  • 8/15/2019 AES Candidates

    34/96

     

    Initial transformation

    (inal transformation

    ?rounds

    times

    ound *eyBiC

    i:@i?2

    ound *eyB0C

    i:@2

    iDEroundsF

    &ip!er ound

    ound *eyBErounds?2C

    Typical (low iagram of

    a &ecret$Key 7loc) Cipher

    7 i it ti hit t

  • 8/15/2019 AES Candidates

    35/96

     

    re'ister 

    combinational

    lo'icone round

    multipleGer 

    7asic iterative architecture

  • 8/15/2019 AES Candidates

    36/96

    GMU 6esults: Encryption in cipher feedbac) modes

  • 8/15/2019 AES Candidates

    37/96

     

    GMU 6esults: Encryption in cipher feedbac) modes

    +C7C! C(7! 5(7, $ Lirte% (GThroughput

  • 8/15/2019 AES Candidates

    38/96

     

    8& 6esults: Encryption in cipher feedbac) modes

    +C7C! C(7! 5(7, $ &9C! 1#2 m CM5&Throughput

  • 8/15/2019 AES Candidates

    39/96

     

    ecreasing area by resource sharing

    ( (

    1 -

    1 -

    (

    1 -

    1 -

    multipleGer 

    7efore fter

    re'ister  re'ister 

    6esource sharing: &peed vs rea

  • 8/15/2019 AES Candidates

    40/96

     

    Throughput

    rea

     basic arc!itecture

    6esource sharing: &peed vs# rea

    - basic arc!itecture

    - resource s!arin'

    resource s!arin'

    89&T 6eport GMU 6eport:

  • 8/15/2019 AES Candidates

    41/96

     

    89&T 6eport GMU 6eport:'ardware Efficiency

    (eedbac) cipher modes: C7C! C(7

    &peed

    rea

    'igh

    How

    &mall

    6ijndael

    M6&

    &erpent

    Twofish

    6C>Medium

    Medium Harge

  • 8/15/2019 AES Candidates

    42/96

     

    rent software and hardware

    optimi*ations eBuivalentF

    Efficiency in software: 89&T$specified platform

  • 8/15/2019 AES Candidates

    43/96

     

    0

    1

    20

    21

    30

    31

    40

    &erpenti.ndael Twofis!& Mars

    Efficiency in software: 89&T$specified platform

    237-bit 9ey

    283-bit 9ey

    31-bit 9ey

    200 MHz Pentiu Pro, !orland C""

    &peed

  • 8/15/2019 AES Candidates

    44/96

     

    0

    10

    200210

    300

    310

    400

    410

    500

    510

    100

    &erpent i.ndael Twofis! & Mars

    5ur 6esults: 7asic architecture $ &peed

    Throughput

  • 8/15/2019 AES Candidates

    45/96

    tomic operations used in @- most popular

  • 8/15/2019 AES Candidates

    46/96

     

    tomic operations used in @- most popular

    secret$)ey ciphers +-,

     !. Chetw$nd, MS -he&i&, P* 

    Considered ciphers:

    "lowfis!, &AST, &AST-237, &AST-31, &HPT$,

    &S-&ip!er, #/AJ, #/S, #(&, /3,(/AJ, (), )ST, Kasty Puddin', I&/,

    I#/A, *!afre, *!ufu, J*I82, J*I86,

    Jucifer, Mac)uffin, MA)/$TA, MAS, MISTH2,

    MISTH3, MM", &3, &1, &,i.ndael, SA(/ *, SA(/?, Serpent, SL%A/,

    SKA*, S9ip.ac9, T/A, Twofis!, A*/,

    idera9e

    Major atomic operations used in @- most popular

  • 8/15/2019 AES Candidates

    47/96

     

    Major atomic operations used in @- most popular

    secret$)ey ciphers +0,

     !. Chetw$nd, MS -he&i&, P* 

    0

    1

    20

    21

    30

    31

    40

    41

    50

    A1

    -1

    / /-

    &$bo% Lariable

    rotation

    Modular

    multi$

    plication

    G(+0n,

    multi$

    plication

    Modular

    inversion

    u%iliary atomic operations used in @- most popular

  • 8/15/2019 AES Candidates

    48/96

     

    u%iliary atomic operations used in @- most popular

    secret$)ey ciphers +A,

     !. Chetw$nd, MS -he&i&, P* 

    7oolean

    +N56! 8! 56!

    etc#,

    (i%ed

    rotation

    Modular

    addition

    O subtraction

    ermutation0

    1

    20

    21

    30

    31

    40

    41

    50@1

    02

    01

    F

    Major cipher operations +-, $ &$bo%

  • 8/15/2019 AES Candidates

    49/96

     

    Major cipher operations +-, & bo%

    &$bo% n % m65M

    &oftware 'ardware

    C

    &M

    # SB2DDnC@

    N 0G34, 0G45, 0G1

    ; ; ; ; ; ; ; ; ; ; ; ; ; ;O

    S # 34K, 45K,

    1K

      ;;

    direct logic

    n

    3n words

    n bit address

     bit output

     ; 

     ; 

     ; 

    G2G3

    Gn

     ; 

     ; 

     ; 

    y2y3

    ym

    &

    3n⋅ 

    /it&

    &$bo%: Memory in hardware

  • 8/15/2019 AES Candidates

    50/96

     

    & bo%: Memory in hardware43 G 5 @ 237 bits

    &

    5

    5

    &

    5

    5

    &

    5

    5

    &

    5

    5

    &

    5

    5

    &

    5

    5

    &

    5

    5

    # # #

    Memory @ 43 ⋅ 35  ⋅ 5 bits @ 0 )bit 

    &

    7

    7

    &

    7

    7

    &

    7

    7

    &

    7

    7

    # # #

    2 G 7 @ 237 bits

    Memory @ 2 ⋅ 37  ⋅ 7 bits @ A0 )bit I -> ⋅ 3 9bit

    &$bo%: Memory in software

  • 8/15/2019 AES Candidates

    51/96

     

    & bo%: Memory in software43 G 5 @ 237 bits

    &

    5

    5

    &

    5

    5

    &

    5

    5

    &

    5

    5

    &

    5

    5

    &

    5

    5

    &

    5

    5

    # # #

    Memory @ 35  ⋅ 5 bits @ >@ bit 

    &

    7

    7

    &

    7

    7

    &

    7

    7

    &

    7

    7

    # # #

    2 G 7 @ 237 bits

    Memory @ 37 ⋅ 7 bits @ 0 )bit @ A0 ⋅ 5 bits

    Major cipher operations +0, P Lariable 6otation

  • 8/15/2019 AES Candidates

    52/96

     

    variable rotation

      65HA0

    Mu%$based shifter

    'igh$speed cloc) 

    C

    &M

    Major cipher operations +0, Lariable 6otation

    QQQ 7

    J A, "

    & @

  • 8/15/2019 AES Candidates

    53/96

     

    CI7 mod 0n  'alf$Multiplier

    &M

    C

    Major cipher operations +A, P Modular Multiplication

    'ardware&oftware

    & @ AU"

    M%J

    n n

    MUH

    n

    n n

    n

    unsi'ned lon' A, ", &

    A "

    &

    n@43, 2

  • 8/15/2019 AES Candidates

    54/96

    u%iliary cipher operations +-, $ ermutation

  • 8/15/2019 AES Candidates

    55/96

     

    ermutation

    C

    order of wires

    u%iliary cipher operations +-, ermutation

    'ardware&oftware

    &M

    compleG

    sequence of 

    instructionsDD, Q, X

    compleGsequence of 

    instructions

    J, , A$#

    n

    n

    G2 G3 G4 GnGn-2; ; ;

    y2 y3 y4 ynyn-2

    ; ; ;

    u%iliary cipher operations +0, $ (i%ed rotation

  • 8/15/2019 AES Candidates

    56/96

     

    C

    order of wires

    u%iliary cipher operations +0, (i%ed rotation

    'ardware&oftware

    &M

    J A, n

    G2 G3 G4 GnGn-2; ; ;

    y2 y3 y4 ynyn-2

    ; ; ;

    & @

  • 8/15/2019 AES Candidates

    57/96

    u%iliary cipher operations +@,

  • 8/15/2019 AES Candidates

    58/96

     

    CI7 mod 0ndder"subtractor

    &M

    C

    y p p + ,

    ddition"subtraction

    'ardware&oftware

    & @ A?"

    A##

    n n

    n

    n n

    n

    unsi'ned lon' A, ", &

    A "

    &

    n@43, 2

  • 8/15/2019 AES Candidates

    59/96

  • 8/15/2019 AES Candidates

    60/96

    7asic operations

  • 8/15/2019 AES Candidates

    61/96

     

    addition

    multiplication

    7oolean

    permutation

    fi%ed rotation

    G(+0n,

    multiplication

    variable rotation

    elay and area in &5(TD6E

    p

    elay

    Memory

    &$bo%

    @%@

    &$bo%

    4%4

    &$bo%

    .%A0

    modular inverse

    Major operations of E& finalists

  • 8/15/2019 AES Candidates

    62/96

     

    MarsTwofish&erpent 6C>6ijndael

    Major operations of E& finalists

    &$bo%es

    9nteger

    multiplication

    Lariable

    rotation

    Multiplication

    in G(+0m,

    u%iliary operations of E& finalists

  • 8/15/2019 AES Candidates

    63/96

     

    MarsTwofish&erpent 6C>6ijndael

    y p

    7oolean

    ddition"

    subtraction

    ermutation

    (i%ed rotation

    M6& P !M team

  • 8/15/2019 AES Candidates

    64/96

     

    elay

    rea

    modular

    multiplication

    7oolean

    permutation

    variable

    rotationG(+0n,

    multiplication

    fi%ed rotation

    elay and area in '6D6E

    addition +CH,

    addition +6C,

    &$bo%

    @%@

    &$bo%

    4%4

    &$bo%.%A0

    modular

    inverse

    &erpent P "# Anderson$ E# !iham$ %# &nudsen

  • 8/15/2019 AES Candidates

    65/96

     

    elay

    rea

    modular

    multiplication

    7oolean

    permutation

    variable

    rotationG(+0n,

    multiplication

    fi%ed rotation

    elay and area in '6D6E

    p

    addition +CH,

    addition +6C,

    &$bo%

    @%@

    &$bo%

    4%4

    &$bo%

    .%A0

    modular

    inverse

    6ijndael P  '# "ijmen$ (# )aemen

  • 8/15/2019 AES Candidates

    66/96

     

    elay

    rea

    modular

    multiplication

    7oolean

    permutation

    variable

    rotationG(+0n,

    multiplication

    fi%ed rotation

    elay and area in '6D6E

    j j

    addition +CH,

    addition +6C,

    &$bo%

    @%@

    &$bo%

    4%4

    &$bo%

    .%A0

    modular

    inverse

    M6& P !M team

  • 8/15/2019 AES Candidates

    67/96

     

    addition

    multiplication

    7oolean

    permutation

    fi%ed rotation

    G(+0n,

    multiplication

    variable rotation

    elay and area in &5(TD6E

    elay

    Memory

    &$bo%

    @%@

    &$bo%

    4%4

    &$bo%

    .%A0

    modular inverse

    5perations efficient in both software and hardware

  • 8/15/2019 AES Candidates

    68/96

     (ast O compact &low O big

    &oftware

    (ast Ocompact

    &low O

    big

    permutation

    addition

    G(+0n, multiply

    multiplication

    &$bo%

    7oolean

    fi%ed rotation

    variable rotation

    &ummary

    &low or

    big

    &low or big 'ardware

    modular inverse

  • 8/15/2019 AES Candidates

    69/96

    E&: Types of candidate algorithms

  • 8/15/2019 AES Candidates

    70/96

     

    *eistel +et,orks Modiied *eistel 

     +et,ork 

     Su.stitution-

     %inear /ransormation

     +et,orks

    thers

    yp g

    Twofish

    /3

    #(&

    #eal

    J*I86

    Ma'enta

    6C>

    M6&

    &AST-31

    6ijndael

    &erpent

    Safer?

    &rypton

    (ro'KP&

    (eistel 8etwor): &ingle 6ound of Twofish

  • 8/15/2019 AES Candidates

    71/96

     

    DDD 2

    RRR 2

    ( $ function

    (eistel 8etwor): &ingle 6ound of Twofish 

    #B4C #B3C #B2C #B0C

    #B4C #B3C #B2C #B0C

    * 3r?7 * 3r?8

    - units shared .et,een encr1ption and decr1ption

    Modified (eistel 8etwor): &ingle 6ound of M6& 

  • 8/15/2019 AES Candidates

    72/96

     

  • 8/15/2019 AES Candidates

    73/96

     

    &ingle 6ound of &erpent

    S-boGes

    Jinear Transformation

    237

    237

    *BiC

    - units shared .et,een encr1ption and decr1ption

    237

    &ubstitution$Hinear Transformation 8etwor):

    & i

  • 8/15/2019 AES Candidates

    74/96

     

    initial permutation

    encryption

     bloc9 

    decryption

     bloc9 

    final permutation

    237

    237

    237237

    237237

    237

    237

    *0, ;;; , *6, *43 *43, ;;; , *6, *0

    &erpent in 'ardware

    &ubstitution$Hinear Transformation 8etwor):

    6ij d l i ' d

  • 8/15/2019 AES Candidates

    75/96

     

    9nversion in G(+04,

    affine

    transformation

    inversed affine

    transformation

    S!iftow

    MiG&olumn

    sub9ey

    InvS!iftow

    sub9ey

    InvMiG&olumn

    encr1ption decr1ption

    6ijndael in 'ardware- units shared .et,een encr1ption and decr1ption

  • 8/15/2019 AES Candidates

    76/96

     

    8umber and comple%ity of rounds

    8umber vs# comple%ity of a round

  • 8/15/2019 AES Candidates

    77/96

     

    8umber of rounds

    Comple%ity of a round

    Triple E&

    E&

    &erpent

    6ijndael

    Mars

    6C>

    Twofish

    20

    30

    40

    50

    10

    Comple%ity of the cipher round in hardwareTime in hardware

  • 8/15/2019 AES Candidates

    78/96

     

    &erpent

    6ijndael

    Twofish

    6C>

    Mars

    &$bo% @%@ N56/

    &$bo% 4%4 N56> N562 V5

    > &$bo%es @%@0 A0 V1 V58 V3

    &R6A0 0 A0 65TA0

    MUHA0 5 M%V3

    5 M%V3

    3 M%V3

    M%V3

    3 M%V3

    regular round

    0 30 50 0 70 200

    Time in hardware

  • 8/15/2019 AES Candidates

    79/96

  • 8/15/2019 AES Candidates

    80/96

     

    Ma)ing all rounds identical

  • 8/15/2019 AES Candidates

    81/96

    &erpent P 'ardware rchitecture 9-

  • 8/15/2019 AES Candidates

    82/96

     

    237-bit re'ister 

    43 G S-boG 0

    *i re'ular Serpent round

    43 G S-boG 6

    linear transformation*43

    output

    237

    237

    237

    43 G S-boG 2

    7-to-2 237-bit multipleGer 

    237 237 237

    237 237 237

    GMU 6esults: Encryption in cipher feedbac) mode+C7C C(7 5(7, Lirte% (G

  • 8/15/2019 AES Candidates

    83/96

     

    +C7C! C(7! 5(7, $ Lirte% (GThroughput

  • 8/15/2019 AES Candidates

    84/96

     

    arallelism

    arallelism in &'$-

  • 8/15/2019 AES Candidates

    85/96

     

     A

    B

    D

    C

    E

    ROTL5

    f t

    ROTL30

    + + ++

    Kt

      Wt

     A

    B

    D

    C

    E

    32

    32

    32

    32

    32

     A

    B

    C

    ROTL5

    f t

    ROTL30

    + + ++

    Kt

      Wt

     A

    B

    D

    C

    E

    32

    32

    32

    32

    32

    5perations from two different steps that can be performed

    in parallel

    E%ecuting &'$- on a /$way superscalar processorA !o&&elaer& 1 Go(aert& +andewalle )4

  • 8/15/2019 AES Candidates

    86/96

     J1

    J2

    J40

    J2

    J1 J40

    J2

    J1 J40

    J2

    J1 J40

    J40

    J2

    J2

    J40

    step n

    step n3

    step n2

    step n"3

    step n"

     A. !o&&elaer&, 1. Go(aert&, . +andewalle, )4 

    8umber of operations that can be

    t d i ll l

  • 8/15/2019 AES Candidates

    87/96

     

    e%ecuted in parallel

    for various hash functions

    1

    -

    0

    A

    @

    2

    >

    /

    4

    &'$- 69EM

    ->1

    69EM

    -04

    69EM M2 M@

     A. !o&&elaer&, 1. Go(aert&, . +andewalle, )4 

  • 8/15/2019 AES Candidates

    88/96

     

    5ptimi*ation tric)s

    6ijndael round: Table$loo)up implementation

  • 8/15/2019 AES Candidates

    89/96

     

    a0,0 a0,2 a0,3 a0,4

    a2,0 a2,2 a2,3 a2,4

    a3,0 a3,2 a3,3 a3,4

    a4,0 a4,2 a4,3 a4,4

     b0  b2  b3  b4

    T0

    T2

    T3

    T4

    @9 3 G4,3 G3,3 G2,3 G0,3  b3

    &peed$up in software:  Z -11 times

    &peed$up in hardware:  Z 01

    &erpent: 7it$slice implementation43 G 5 @ 237 bits

  • 8/15/2019 AES Candidates

    90/96

     

    &

    G2

  • 8/15/2019 AES Candidates

    91/96

     

    The proposed approach

    Cipher design methodology +-,

  • 8/15/2019 AES Candidates

    92/96

     

    2; &!oose one or maGimum two major operations efficient in bot! software and !ardware

    best choice:  &$bo% @%@! G(+0n, multiplication

    3; &!oose one or maGimum two au%iliary operations efficient in

     bot! software and !ardwarebest choice:  7oolean! fi%ed rotation

    4; &!oose cipher type t!at enables maGimum s!arin'

    amon' encryption and decryption

    best choice:  (eistel networ)! modified (eistel networ) 

    Cipher design methodology +0,

  • 8/15/2019 AES Candidates

    93/96

     

    5; #esi'n a round ta9in' into account a trade-off amon'• round comple%ity

    • number of rounds necessary to 'uarantee  sufficient security mar'in

    1; Ma9e each round BpossiblyC identical

    negative e%amples:  &erpent! Mars

    ; Joo9 for parallelism wit!in a round and amon' consecutive

    rounds

    positive e%ample:  &'$-

    6; Joo9 for optimi*ation tric)s

    positive e%amples: 

    table$loo)$up in 6ijndael

    bit$slice implementation in &erpent

  • 8/15/2019 AES Candidates

    94/96

     

    Mathematicians

    Computer

    scientists

    Computer

    Engineers

    &ecurity

    &oftware

    efficiency

    'ardware

    efficiency

    (le%ibility

    S-11 Challenges

  • 8/15/2019 AES Candidates

    95/96

     

    g

    (or mathematicians:

    rove or disprove that &erpent with

    • all &$bo%es identical• -> rounds

    is at least as secure as 6ijndael

    (or computer scientists:

    9s there a way of using instruction level parallelism

    to speed$up software implementation of

  • 8/15/2019 AES Candidates

    96/96

    g

    (or computer scientists:

    Dhat is a level of parallelism present in

    &'$02>! &'$A4@! &'$2-0F

    (or mathematicians:

    9s there a way of changing &erpent into

    a modified (eistel networ) cipher

    without loosing its security propertiesF