Interference Cancellation: Better Receivers for a New Wireless MAC
Daniel Halperin, Josie Ammer, Thomas Anderson, and David Wetherall
(University of Washington)

Summary: One of the fundamental problems in shared channel simultaneous access
has interference.  The CSMA was designed for single contenteous AP, but today
the deployments are dense and in 3-D.

The key idea is that exploit structure of interference to overcome its
effect; seperate it from noise; noise is random but interference has a
structure.  The presenter showed how the different phase, phase-drift
and amplitude to recover the packets that would otherwise be lost to interference

Q: What would happen when we use denser packing constellation, such as QAM

A. Even in that case we should be able to of separate the
constellations.

Q:  Why do the real systems have a slack; why do the standards just
use the slack for better rates.

A: They cannot predict the scenarios that the standardized equipement would be
deployed.

Q: How it is related to analog coding.

A: In analog coding case we need to know one or the other symbol,
which is not available in infrastructure-mode.

Q: When is it likely to happen, and when not; in particular, multiple
senders, multipath fading/interference, etc.

A:  We need more work to evaluate those scenarios.  But there are some
obvious cases where it will not work, e.g., when one of the receiver
is close to noise margin.

Q: Do you think that the AP will operate in this mode all the time?

A One possible way is that a AP just sits and listen and only recovers when
there is a collision.

Q:  How does is scale with the number of simultaneous speakers.

A: It blows up the search space, but one thing is that you can switch
to CSMA when there are more speakers.