We discovered early on that it was best to construct and train the network in receive only mode. It was a result of our quad-symbol hack-fest. It had been quite some time, many random sessions had passed, and now, filtering through the network, were odd bits of information that were associated with the patterns that we learned in that incident. There was so much tied to that symbol that dared not inject it into the network again.
Instead, we concentrated on making sure that the electron network and the central network did not interfere. Using the active times to localize our communications in bands and channels that were not interfered with, we had to make sure that these channels did not inadvertently inject events into the random association periods.
We waited for a small random association event and recorded both the input and output, while running silently. Then, while communicating on a select set of channels and frequencies, we monitored for spurious activity in the association engine. If we found any, we cycled through the frequencies until we found the inadvertent input and eliminated it from the allocation list. Once we had a clean set of channels, we re-introduced the original input for the small event to make sure that the association was not affected by our resumed communications.
July 12, 2010
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment