Distributed Average Consensus
27 views
Skip to first unread message
Michael
Oct 15, 2013, 9:59:12 AM10/15/13
to swarm-...@googlegroups.com
Have a lengthier update coming either tonight or tomorrow, but something I found in wired of all places:
http://www.stanford.edu/~boyd/papers/pdf/lmsc_mtns06.pdf
I have to finish reading the paper but it looks like nodes in a weighted undirected graph are measured for their mean square deviation from the average arbitrary values of their neighbors. This creates a distributed load balancing system. Something to consider.
"In this model, each node updates its local variable with a weighted average of its neighbors’ values, and each new value is corrupted by an additive noise with zero mean. The quality of consensus can be measured by the total mean-square deviation of the individual variables from their average, which converges to a steady-state value. We consider the problem of finding the (symmetric) edge weights that result in the least mean-square deviation in steady state"
http://www.stanford.edu/~boyd/papers/pdf/lmsc_mtns06.pdf
I have to finish reading the paper but it looks like nodes in a weighted undirected graph are measured for their mean square deviation from the average arbitrary values of their neighbors. This creates a distributed load balancing system. Something to consider.
"In this model, each node updates its local variable with a weighted average of its neighbors’ values, and each new value is corrupted by an additive noise with zero mean. The quality of consensus can be measured by the total mean-square deviation of the individual variables from their average, which converges to a steady-state value. We consider the problem of finding the (symmetric) edge weights that result in the least mean-square deviation in steady state"
Reply all
Reply to author
Forward
0 new messages