Life (and routing) on the Wireless Manifold
Varun Kanade and Santosh Vempala (Georgia Tech)
Speaker is Santosh Vempala.
Q: Doesn’t wireless manifold just boil down to graph routing instead of disk routing? The way it is done today is almost equivalent to the manifold.
A: Yes, at a high level, but Manifold represent the terrain, not the connectivity.
Q: Can we claim that routing depends on geography if the sender and receiver perspective on power are different? Does the description explain asymmetric signal?
A: One might use directed edges on the manifold.
Hari: Q: Did you do experiments with active interference and congestion? Would it be better if we measure bit/error rates instead of signal levels?
A: All measurements were made in Orbit Lab with the active interference. We probably should measure bit/error rates too.
Q: GR: How often do we need to compute the manifold?
A: Depends on whether the terrain changes; weather, and the mass mobility.
Q: GR: Do you need global knowledge in proposed scheme?
A: SV: Yes once in a while one has to share the manifold.
Q: How many measurements the proposed scheme needs to recover a complex manifold?
A: There is no clear answer to this, but we presume that it shall be linear to the size of the grid.
Q: Does the manifold ever get into holes that has no path out?
A: One will not get into the hole, if starting point is exactly on the manifold nodes. Otherwise there is slight probability of a hole, that you can get round by using the constraint routing.