
198 c N A p~1r E R 4 Multiplication
an addition. As also done for the sequential multiplication, to avoid this multi-
ple it is possible to recode the multiplier into a signed-digit set. To optimize the
recoding we consider two cases:
1. The bit-array is added by a linear array of adders. As discussed further in
the next section, in this case each adder in the array has as operands the
partial sum (of the previous additions) and one multiple. Consequently, the
recoding can be done as in the sequential multiplication case (digit set
{-1, 0, 1, 2}). This has the advantage of requiring only four values per
digit. 5
2. The bit-array is