Signal Processing - May 2017 - 13

The atoms are imaged by shining
a second, weak laser beam through the
cloud. Because atoms in different energy
states absorb light of different frequencies, the final energy state of the atoms
can be detected. The resulting images
show interference bands of atom populations in the two different energy states.
The rotation rate and axis are measured by
analyzing the spacing and direction of the
interference bands across the atom cloud.
Acceleration is deduced from changes in
the central band. The interferometer is sensitive to acceleration along the direction of
the light and sensitive to rotations perpendicular to the light.
"The signal processing challenge in
our experiment is to take images of the
transition probability and estimate the
wavelength of the fringe pattern," Hoth
says. "To use the system as a gyroscope,
you would use the wavelength of the
fringe pattern to infer the unknown rotation rate." In experiments to date, the
researchers' goal has been to quantify
the relationship between the rotation rate
and the wavelength of the fringe pattern
so they can compare that relationship to
theoretical predictions. "We do that by
applying a known rotation rate and measuring the wavelength of the observed
fringe pattern," Hoth says.
The process requires three images.
"Each image has a fringe pattern with the
same wavelength, but we vary the phase
so that we see different parts of the fringe
pattern," Hoth says. "By combining the

From the eDitor

∆Φ
6
5
4
3
2
1

Figure 3. NIST's compact gyroscope measures
rotation by analyzing patterns of interfering
matter waves in an expanding cloud of atoms
transitioning between two energy states. Each
atom's speed determines both its final position
in the cloud and the size of the rotational signal
that shifts the interference patterns. Thus,
rotations generate interfering bands of atoms
across images of the final cloud. The color
coding indicates how much the interference
patterns shift in radians, the standard unit of
angular measure. The orientation of the interfering bands (horizontal in the image) indicates
the rotation axis. The rotation rate, determined
by an analysis of the band spacing, is
44 milliradians/s. (Figure used courtesy of NIST.)

three images, we can get the fourth image,
which shows the spatial variation of the
interferometer phase." The fringe pattern
is equivalent to a slope or gradient in the
interferometer phase. "By calculating the
spatial phase, we solve both of our problems," Hoth continues. "The unwanted
structure is suppressed and the hard problem of estimating a fringe wavelength has
turned into the easy problem of estimating
a best fit slope."

"The basic idea is that the part of
the signal we're interested in changes
when we modulate the phase, but the
parts that we want to suppress stay the
same," Hoth says. "So, by combining
multiple images, we can separate the
signal we want from the structure that
we don't want."
Hoth goes on to say that the researchers are still experimenting with different
ways of implementing the signal processing strategy. "It's mostly variations on the
idea of modulating the phase and looking
at the response," he says.
Although Stanford researchers were
the first to demonstrate the technique of
using an expanding cloud of laser-cooled
atoms, they presented it in a 10-m-tall
"atomic fountain" that was designed to
be the world's most sensitive accelerometer. "In contrast to their work, our system
was designed to be compact to open up
the possibility of portable applications,"
Hoth says. The current experimental system is tabletop sized, but the researchers
plan to eventually shrink the apparatus
down to a portable cube approximately
the size of a mini refrigerator. "There's a
lot of very exciting work on atom interferometry being done all over the world,"
Hoth says.

Author
John Edwards (jedwards@johnedwards
media.com) is a technology writer based
in the Phoenix, Arizona, area.
SP

(continued from page 3)

myself included, argued for a strategy
that interacts with arXiv to put signal
processing under a more appropriate
topic branch, which may also need to
be created. Thanks to the efforts led
by SPS Vice President for Membership
Dr. Nicholas Sidiropoulos, we learned
that the arXiv scientific board has

agreed to work toward creating an
electrical engineering topic branch
under which signal processing, information theory, and control theory will
be hosted. This would be a wonderful
development as SPS enriches its content ecosystem. Ultimately, our hope
is that you, our readers and members,
IEEE Signal Processing Magazine

May 2017

will find the content ecosystem beneficial to your professional development and be a regular contributor to
the ecosystem.

SP
13

http://www.media.com

Table of Contents for the Digital Edition of Signal Processing - May 2017