Biomedical Engineering Reference
In-Depth Information
Design
specifications
Input:
Sequencing graph
of bioassay
Digital microfluidic
module library
Mixing components
Maximum array area
Amax
:
20×20 array
Maximum number of
optical detectors:
4
Store
O1
O2
Area
Time
10 s
6 s
3 s
5 s
Mix
2×2-array mixer
2×3-array mixer
2×4-array mixer
1×4-array mixer
Detectors
LED+Photodiode
4 cells
6 cells
8 cells
4 cells
O4
O3
Store
Mix
Number of reservoirs:
3
O5
Mix
Detection
1 cell
30 s
Maximum bioassay
completion time
Tmax
:
50 seconds
O6
Unified Synthesis of Digital Microfluidic Biochip
Output:
Resource binding
Schedule
Placement
0
1
2
3
4
5
6
7
Operation
Resource
O1
O2
O1
O2
O3
O4
O5
O6
2×3-array mixer
O2
O1
Storage unit (1 cell)
O3
2×4-array mixer
O4
Storage unit (1 cell)
O3
1×4-array mixer
O4
O6
LED+Photodiode
O5
Biochip design results:
Array area:
8×8 array
Bioassay completion time:
25 seconds
Figure 1.3
An example illustrating system-level synthesis [15].
This method allows users to describe bioassays at a high level of abstraction,
and it automatically maps behavioral descriptions to the underlying micro-
fluidic array.
The design flow is illustrated in Figure 1.3. First, the different bioassay
operations (e.g., mixing and dilution) and their mutual dependences are
represented using a sequencing graph. Next, a combination of simulated
annealing and genetic algorithms are used for unified resource binding,
operation scheduling, and module placement. A chromosome is used to
represent each candidate solution, that is, a design point. In each chromo-
some, operations are randomly bound to resources. Based on the binding
results, list scheduling is used to determine the start times of operations;
that is, each operation starts with a random latency after its scheduled time.
Finally, a module placement is derived based on the resource binding and
the schedule of fluidic operations. A weighted sum of area and time cost is
used to evaluate the quality of the design. The design is improved through a
series of genetic evolutions based on PRSA. It generates an optimized sched-
ule of bioassay operations, the binding of assay operations to resources, and
a layout of the microfluidic biochip.
The top-down synthesis flow described earlier unifies architecture-level
design with physical-level module placement. However, it suffers from two
drawbacks. For operation scheduling, it is assumed that the time cost for drop-
let routing is negligible, which implies that droplet routing has no influence
