UKF_MCDA_SMOOTH RTS Smoothing of particles in UKF-RBMCDA algorithm
Syntax:
[S,SM] = UKF_MCDA_SMOOTH(S,a,Q,param,same_p)
In:
S - NxNP cell array containing NP particle structures for N time steps.
a - Dynamical model as cell array
of size TxN, for each target in each particle,
or inline function or name of function in
form a(x,i,param), where i is the
index of the target x. (optional, default A(x,i)*X)
Q - Process noise of discrete model as matrix
if common for all targets, or as cell array
of size Tx1 for all targets separately. (optional, default zero)
param - Parameters of a.
same_p - 1 if the same parameters should be
used on every time step (optional, default 1)
Out:
S - NxNP cell array containing the smoothed particles for each time step.
SM - 1xT cell array containing smoothed means for each target as a
matrix DxN.
Description:
Perform Extended Kalman Filter prediction step for each target
and each association hypothesis particle. The model is
x_i[k] = a_i(x_i[k-1], q , param), q ~ N(0,Q_i)
for each target i. Dynamics a_i() for each target
are assumed to have known statistics.
See also:
EKF_MCDA_UPDATE, EKF_PREDICT, LTI_DISC, KF_PREDICT