Page 87 - ISCI’2017
P. 87

To estimate probabilistic properties of cyclic keys  K  ()x   sequences we summarize the positions of
                                                                  рк
            the lemma proved above for random, equiprobable and independent values  K  () x  , K () x  ,..., K () x  . Let us
                                                                                       1    2      t
                                                                                                  k
            estimate the average  number of different sequences  K  ()x   that is generated using all  2  values of
                                                                   рк
            master keys  K  ()x  . The following theorem is true.

               Theorem. The average number of different cyclic keys sequences  K  () x  = {K 1 () x  ,K 2 () x  ,...,K t () x  }, that
                                                                                 рк
                           k
            is formed by  2  independent implementations of random  homogeneous substitution  in i-th point,
             i∈ {1,2,..., }t , is defined by:

                                                                                 (
                                                                         −
                                                                    1  2 k tl           k tl
                                                                                            −
                                                         k
                                                        2
                                  ( , , ) =
                                                      tl
                                              −−
                                            l
                                                     −
                                N k lt    2 (1 (1 2 ) ) ≈    2 1   tl      −  ≈  2 1− (0,37 ) 2  )  .               (8)
                                                                                tl
                                                                     e   
                                                                          
               Proof  is a generalization of the  lemma`s results in the case of  K  ()x  = {K 1 ()x  ,K 2 ()x  ,...,K t ()x  }
                                                                                     рк
            sequences formation. Indeed, in accordance with the assumption of  K i ()x   cyclic keys generating by
            the independent implementations of random  homogeneous substitution  in  i -th point, for each
                                                                            Y
             i∈ {1,2,..., }t  the probability of  K  ()x   does not depend on  y ∈ = { , yy  ,..., y  }  or  K  ()x  . This
                                               i                         i        1  2     2 l
                                                        l −
            probability  is equal to  ( '( )Ps  y =  K  () x  ) =  2 .  The  joint probability  of  independent  events  is  the
                                        x  i     i
            product of the probabilities of these events, i. e.:
                                                                t
                                                                                       tl
                                  ( P K  () x  = {K 1 () x  , K () x  ,..., K t () x  }) = ∏ Ps  x  y i  K i () x  ) 2 .
                                                                    ( ' ( ) =
                                                                                   =
                                                                                      −
                                                 2
                                     рк
                                                                i= 1
               Master keys  K  ()x   are selected  independently  from each other and the corresponding events
             K () x  = {K 1 () x  ,K 2 () x  ,...,K t () x  } are independent too. Therefore, using the Bernoulli formula just as in the
               рк
            lemma proof, we obtain the expression
                                                                         −
                                                                          tl M
                                           PN      , )t =  C N M (1 2 )−  −  tl N M−  (2 ) ,
                                             ( ,M
            which specifies the probability of the case that in  N  independent implementations the sequence  K ()x
                                                                                                          рк
            would appear exactly  M  times.

               The value

                                                  2 k           k                                   k
                                        k
                                                                     tl i
                                                                                 k
                                                                    −
                                    P (2 , 0, )t>  = ∑ C i k  (1 2 )−  −  tl  2 −  i (2 ) = 1 P−  (2 ,0, ) 1 (1 2 )t = −−  −  tl  2          (9)
                                                 i= 1  2
            gives the probability of the case when in  2  independent tests the specific sequence  K  ()x   is formed
                                                      k
                                                                                                 рк
            at least once.


                                                                                                          87
   82   83   84   85   86   87   88   89   90   91   92