Biomedical Engineering Reference
In-Depth Information
FIgURE 1.2:
Comparison of system specifications for current technologies and those needed for
BMIs.
Typical numbers for the transmission per channel are around ~144 kbits/channel (12.5 kHz ×
12 bits). For a 16-electrode probe, this data rate is around 2.3 Mbits/sec. To compute with the
electrophysiological signals at line level, front-end amplifiers (40 dB gain) used with neural probes
take about 100 µW of power per channel. If one wants to transmit the data via a wireless implanted
device, the total power budget can range between 2.5 and 5 mW. The area of such devices is roughly
0.5-5 mm
2
, depending on the number of channels that is absolutely overshadowed by the biological
size/power factor [
11
,
12
]. To extend operation times of the BMI over days to weeks using hundreds
of neural inputs, the power budgets need to significantly decrease. For the envisioned applications,
the systems will have computational demands that exceed by orders of magnitude the capacity of a
single powerful workstation. Depending on the scope of the system, it may be necessary to use hun-
dreds or even thousands of processors to provide the needed computational power. Grid-computing
infrastructures can potentially deliver these necessary resources on demand. However, the systems
will also have stringent real-time requirements, as a result of the need for low latency between brain
signaling and sensory feedback.
Improvements using the present approaches are very likely to be evolutionary instead of
revolutionary. We have not yet discussed two fundamental and very difficult issues: how to reliably
interface with the neural tissue and how to model appropriately the interactions among neurons
