Power constrained distributed estimation over noisy channels in WSNs

We consider distributed estimation of a random scalar parameter in a wireless sensor network (WSN), where, by employing a simple uncoded analog amplify-and-forward transmission scheme, the measurements are sent from the sensors to the fusion center (FC) over noisy wireless channels. Under a total transmission power constraint, we study the power allocation among the sensors to minimize the estimation mean-square error (MSE) at the FC. We present an effective solution which allows for correlated sensor observations. Numerical results are presented to illustrate the effectiveness of the proposed algorithm.