Hardware Reference
In-Depth Information
Tabl e 5. 3 Comparison of the results derived from the proposed Approach 1, Approach 2, and the
baseline algorithm
Approach 1
Approach 2
Baseline algorithm
CPU
Size of the
CPU
Size of the CPU
Size of the
Benchmark .N r ;N d ;N h / time (s) biochip
time (s) biochip
time (s) biochip
1
(10, 2, 1)
3,720
16
11
832
16
11
0
25
25
Computationally
2
(18, 2, 1)
NA
impractical
5,797
19
17
0
30
30
Computationally
3
(23, 2, 1)
NA
impractical
11,147
20
19
0
35
40
*N r , N d ,andN h represent the number of reservoirs, DEs, and heaters, respectively
5.6.3
Defect Tolerance of Layouts for PCR Biochips
Because of manufacturing imperfections and degradation of electrodes, physical
defects may occur on DMFBs. These defects can be classified into two categories
based on their locations on the biochip. If a defect disconnects a droplet path into
two “isolated parts” (i.e., droplets cannot be moved from one part to the other part)
or the defect overlaps with a reservoir/DE/heater on the biochip, then it is defined as
a “catastrophic defect”. All other defects are defined as “non-catastrophic defects”.
If a catastrophic defect occurs, the PCR biochip cannot be used further. On
the layout designed for Bioassay 1, the positions of electrodes where defects are
catastrophic are shown in Fig. 5.15 a. For each of these three PCR biochips designed
for Bioassays 1 3, the total number of electrodes where defects are catastrophic
is shown in Table 5.4 .
If a non-catastrophic defect occurs, we can re-route the droplets and send them
to their destinations, as shown in Fig. 5.15 b. It is important to note that re-routing
the droplet may increase the length of the transportation path, and reduce the degree
of parallelism of fluid-handling operations. Therefore, compared to the defect-free
biochip, the execution time of the bioassay on a biochip with a non-catastrophic
defect is higher.
For each PCR biochip, we randomly insert a non-catastrophic defect into the
biochip, then calculate the execution time of the bioassay. The defect-insertion
simulations are executed for all the possible non-catastrophic defects on the layout.
The average value and standard deviation for the execution time of running
bioassays on biochips with defects are shown in Table 5.4 .Thevalueoft m is set
as 0.1. From the table, we find that the percentage of electrodes where defects are
catastrophic is in the range from 33.3 % to 52.3 %. With the presence of a single
non-catastrophic defect, the PCR biochip can be used via graceful degradation with
27.2 67.6 % average increase in the execution time of bioassay.
 
Search WWH ::




Custom Search