Biomedical Engineering Reference
In-Depth Information
Module 2
Module 2
Defect
Module 1
Module 3
Module 1
Module 3
Figure 2.4
Example of a partial reconfiguration.
Assume that each cell in the microfluidic array has an independent failure
probability
p
. The DTI
D
(
G
) value for a layout
G
can be estimated by multi-
plying the survival probabilities of all the modules, as follows [50]:
D
(
G
) ≈ ∏
P
s
(
M
i
) = ∏ (1 −
f
1
(M
i
) +
f
1
(
M
i
) ×
f
2
(
M
i
))
where
M
i
,
i
= 1, …,
N
, is a microfluidic module (e.g., mixer) contained in a
given layout
G
, and
P
s
(
M
i
) is the survival probability of module
M
. Note that
f
1
(
M
i
) is the probability that module
M
i
is faulty. It is determined by the equa-
tion
f
1
(M
i
) = 1 −
p
·
A
(
M
i
), where
A
(
M
i
) is the total number of cells contained
in
M
i
. Finally,
f
2
(
M
i
) is the probability that
M
i
can be successfully reconfig-
ured if it becomes faulty [28].
Now we incorporate DTI into the PRSA-based unified synthesis method.
We first define layout vulnerability by
V
= 1 −
D
. Layouts with low vulner-
ability are likely to provide high probability of successful partial reconfigu-
ration. To find such designs, we combine vulnerability with time and area
cost to derive a new fitness function to control the PRSA-based procedure.
Candidate designs with low survivability are discarded during evolution.
Thus, the synthesis procedure anticipates defect occurrences and selects
designs that allow reconfiguration of large number of modules, while meet-
ing constraints on array size and bioassay processing time.
2.3.2.2 Partial Reconfiguration and Partial Resynthesis
Next, we discuss how defects can be bypassed after manufacture. In the
defect-oblivious approach described in Section 2.2, defect tolerance is
achieved by complete resynthesis, which can be very time consuming. Here,
we propose an efficient method to achieve defect tolerance without the need
for resynthesis. This method is based on the concept of partial reconfigura-
tion, which was introduced in Section 2.3.2. If the number of defective cells
