R help - Aug 2010 - semiparametric fractional autoregressive model

folks,
 
does anyone know if the SEMIFAR model has been implemented in R? i see that
there's a S-FinMetrics function SEMIFAR() that does the job, but I have no
access to that software. essentially, this semiparametric fractional
autoregressive model introduces a deterministic trend to the FARIMA(p,d,0) model
(which, as i understand it, takes care of the random trend and short and long
memory).
 
if not, are there any suggestions for how to estimate the model:
 
phi(L) (1 - L)^d [y(t)(1 - L) - g(t/T)] = epsilon(t)
 
for t = 1, ...., T, and where -0.5 < d < 0.5, phi(L) is the lag
polynomial, and g() is a smooth trend function on [0, 1].
 
what would be a good nonparametric kernel estimator of g()? 
 
i'm new at this so any pointers would be helpful.
 
thanks,
 
murali