Biomedical Engineering Reference
In-Depth Information
Switching probability
Target
estimation
Grouping
data
Measurement
data
< Grouping >
< IMME >
Geometric relationships
Fig. 3.5
General block diagram for the proposed MC-IMME
Fig. 3.6 System design for distributed body sensors has two stages, At the first stage, all the
distributed sensors are partitioned into the groups that have a tracking relationship with each
other. At the second stage, the interactive tracking estimate is performed for distributed groups
3.4.1 MC Mixed Initial Condition and the Associated
Covariance
Starting with an initial x a (k - 1) for each channel a in a group y, new filter state is
computed to estimate the mixed initial condition and Kalman filter covariance
matrices ( 3.10 ) according to the relationships
Þ X
r
Þ½ l ab
x 0y
a
x a
ð
k 1
ð
k 1
ð
k 1
Þ;
a ¼ 1
ð 3 : 10 Þ
Þ X
r
½ l ab k 1
Þþ DP ab k 1
P 0y
a
Þ½ P a k 1
ð
k 1
ð
ð
ð
Þ
a ¼ 1
where l ab is a switching probability presenting the relationship between channel
a and channel b within the same group y. As shown in Fig. 3.6 , we have added the
blue line indicating how the difference (D(k)) in Stage 1 would be used for Stage 2.
We denote r as the channel number of the group and DP ab k ð Þ as an increment
to the covariance matrix to account for the difference in the state estimates from
channels a and b, expressed by ½ k 1
ð Þ k ð Þ T .
Note that the initial states of IMME are extended into Eq. ( 3.10 ) incorporating
with the switching probability and D(k - 1). We have adopted the results of
ð
Þ k 1 j
ð
Þ ½ k 1
Search WWH ::




Custom Search