Virtual sensing with active noise control reduces noise in a zone of quiet (ZoQ) where a physical error microphone is not present. Conventional virtual sensing is applied to the adaptive control of sound through a window. However, it is difficult to acoustically separate one room with many reflected waves into a quiet space and a space with noise sources. To achieve acoustic space separation, virtual sensing is required to estimate the time-varying primary path from the noise source to the ZoQ without a preliminary identification stage and to track time fluctuations. Therefore, in this paper, an acoustic path model is proposed, that can be used to observe the ZoQ from a physical error microphone, and the structure of virtual sensing is designed to cope with environments in which many reflected waves are present. The simulation results show that the proposed system can reduce noise without preliminary estimation of the primary path while tracking the fluctuation of the primary path.