Cooperation Gain In Steady-State Performance Of Diffusion Adaptive Networks With Noisy Links
Azam Khalili, Amir Rastegarnia, Saeid Sanei, Wael Bazzi

Recently, adaptive networks have been widely considered as a
powerful solution to distributed information processing problems.
In this paper, we provide some insights into the steadystate
performance of diffusion least-mean-squares (D-LMS)
adaptive network with noisy links. To this end, we define the
concept of cooperation gain in adaptive networks. Then, we
show that when the connecting links are ideal (error-free), the
cooperation between nodes in adaptive networks always leads
to a smaller steady-state error than what can be achieved by
a non-cooperative scheme. While, with noisy links, cooperation
between the nodes does not always provide better result
and only in specific conditions leads to better results. We
demonstrate the performance of the system by performing a
number of simulations.