Now let’s talk about interference Previous slide covered non-bursty weak interference Make this not look like directional antennas (broadcasts, waves) Distance, trees attenuate signal Delivery probability depends in part on signal-to-noise ratio P(delivery) = f(S/N) Perhaps many Roofnet node pairs have S/N corresponding to intermediate delivery probabilities Mention S/N measurements calibrated in the lab Replace hypothesis slides w/ pictures Mention s/n reported by the card Don’t #2 make it sound so definite Qualify the second point Mention ‘textbook model of ber’
We’ve eliminated the possibility that we’re losing bursts of packets We’ve narrowed the possibilities enough that we know we’re losing packets individiually “ the important special case” Start the explanation of this hypothesis here Why two phases? Mention it’s all beacons, constant, 8 APs XXX Roofnet shares its 802.11 channel with the rest of Cambridge XXX Other 802.11 traffic could corrupt our traffic at a time-scale of individual packets, without affecting our S/N measurements (xxx what’s the experiment?)