Amostragem Adaptativa para Redes de Difusão com Ajuste de Parâmetros em Tempo Real

Recently, we proposed a sampling algorithm for diffusion networks, which adapts locally the number of sampled nodes based on the estimation error. It reduces energy consumption and the computational cost. Its performance depends on the choice of the parameter that penalizes sampling, which is a function of the noise power. Inadequate choices of this parameter affect its tracking capability. In this paper, this parameter is automatically adjusted based on an adaptive estimation of the noise power. Although this modification slightly increases the computational cost, it improves the tracking capability, reduces the influence of noisy nodes and is advantageous in the presence of impulsive noise.