Information Technology Reference
In-Depth Information
V ( l ) corresponding to the l th layer of the P n/m and to the (2 m / n - l +1)th layer of the P -
1
n/m . For example, Fig. 1,d represents the CP-box operation P -1 4/4 , which is the inverse
of the CP-box operation P 4/4 (Fig. 1,c). According to [3] the minimum number of
layers required to implement the CP box of the order h = 1, 2, 4, …, n /4 is
s = log 2 n + log 2 h = log 2 ( nh ). To implement the P n/m -box, where for n = 2 k , of the
maximum order one has to use (2log 2 n - 1) layers.
2.1
DDP Architectures
In the next figures (Fig. 2-4) the architectures of the DDP boxes used in SPECTR-
H64 and CIKS-1 are illustrated. The basic building block of all these boxes is the P 2/1
box. For the hardware implementation of the P 2/1 two different approaches can be
followed: a 2-bit multiplexer and a logic gate chain. This box is designed with a chain
of logic gates, instead of a multiplexer block (Fig. 2). This selection has been done,
because the proposed logic-gate chain is faster compared with the 2-input multi-
plexer. The proposed P 2/1 box has time delay equal to 0.47 nsec and it is 21% less than
the time delay of the multiplexer design (0.6 nsec). In the following figures the archi-
tectures of different CP boxes are given. In addition, the architecture of the inverse
box of each CP box is also presented.
Fig. 2. DDP Boxes: P 8/12 and P -1 8/12
Search WWH ::




Custom Search