This paper deals with a sensor scheduling algrithm considering estimation error variance and communication energy in sensor networked feedback systems. We propose a novel decentralized estimation algorithm with unknown inputs in each sensor node. Most existing works deal with the sensor network systems as sensing systems and it is difficult to apply them to the real physical feedback control systems Then some local estimates are merged and the merged estimates can be optimized in the proposed method and the estimation error covariance has a unique positive definite solution under some assumptions. Next, we propose a novel sensor scheduling algorithm in which each sensor node transmits information. A sensor node uses energy by communication between other sensor node or the plant. The proposed algorithm achieves a sub-optimal network topology with minimum energy and a desired variance. Experimental results show an effectiveness of the proposed method.