The weighting of past history in affecting the Exponentially
Weighted Moving Average function that calculates the current
average queue depth. The equation uses
diffServRandomDropWeight/65536 as the coefficient for the new
sample in the equation, and (65536 -
diffServRandomDropWeight)/65536 as the coefficient of the old
value.
Implementations may limit the values of diffServRandomDropWeight
to a subset of the possible range of values, such as powers of
two. Doing this would facilitate implementation of the
Exponentially Weighted Moving Average using shift instructions or
registers.