In this paper, we address the problem of active noise cancellation in a spatial region in presence of outliers in secondary paths and impulse-like noises encountered in the error microphones due to malfunctioning or failure of the sensors, errors in measurement and movement of microphone or loudspeaker. In conventional multiple error filtered x least mean square (MEFxLMS) algorithm based multi-channel active noise control (MANC) system, such disturbances result in increase in residual noise leading to divergence of the weights of the controller. Furthermore, the traditional MANC techniques are not scalable and computationally complex when the noise cancellation is desired in a wide area. With an objective to mitigate noise in such an environment, acoustic nodes equipped with ANC modules are deployed in the area. A robust distributed active noise control system (RDANC) is developed based on incremental cooperative learning among these sensor nodes and R-estimator based robust cost function. The performance of the proposed system is compared with conventional block-MANC system and incremental block-DANC system for different strengths of outliers and disturbances. Evaluation results show that the proposed system is robust to disturbances in secondary paths and error sensors as compared to block-MANC and block-DANC systems. © 2017 IEEE.