Transcript

0:11
Data storage is the hard disk drives that all computers have. So, in the ’50s, you had disk
0:20
drives that were maybe the size of this room. So we looked at how to make the technology
0:29
leapfrog maybe by a factor of 10, which means that the disks instead of maybe being five
0:36
feet in diameter could shrink to maybe two or three inches. But in the early 2000s, you
0:43
could actually put lots, record lots of bits on a disk. The problem was now, how
0:51
do you recover accurately, how could you read these bits. The possibility of cramming all
0:59
these bits in the disk that allowed you to make the space where you record the bit very,
1:05
very small. What that meant is that once you were able to do it, you’d still want to
1:11
recover those bits accurately, okay, and that’s where my work comes in. I brought my signal
1:17
processing to develop what we call detectors. Okay, detectors means that you need algorithms
1:23
that when the head of the disk reads back, reads from the disk, it’s an electrical signal,
1:30
it converts that signal into a zero or one, decides this is a zero, this is a one. So,
1:36
we designed the detector, and that detector turned out to be quite successful, and that
1:43
is what led to being able to shrink the size of computers to your current laptops. In the
1:50
early 2000s, it became a must-have technology. 60% at least of all computers that were built
1:59
in the 2000s up to now essentially incorporate this type of technology, this type of detector.
2:06
But nowadays, everything gets digitized and stored in a computer. What I do now is actually
2:13
look at these data from many different applications or failures in systems, very, very large scale
2:23
critical infrastructures like the power grid, you have the block outs. How can you predict,
2:29
or how can you from the data that you collect, from these very large distributed systems,
2:36
how can you figure what is relevant? What are the behaviors? What is expected in terms
2:42
of how these systems are going to behave? We live in very interesting times. The amount
2:47
of data that is being collected is tremendous. And it’s from all sorts of potential
2:55
different applications. So, one domain I’m very interested in is developing the algorithms
3:01
that allow the urban planners, for example, to rethink they way they imagine their own cities.
3:09
So this is an example of where signal processing and my students have significantly, I think,
3:16
contributed to the advance of technology.

By Admin

Leave a Reply

Your email address will not be published. Required fields are marked *