Wireless networks in difficult conditions of signal receiving are characterized by a high level of burst data losses, at which a large number of data fragments can be lost in a row. In this case, to recover the lost data, the use of forward error correction methods (FEC) in most cases does not give a sufficient effect. The use of standard data loss recovery methods based on automatic retransmission request (ARQ) at the data link and transport layers of the OSI model can lead to significant delays, which is often unacceptable for real-time streaming services. In such a case, it may be preferable to skip the piece of data rather than delay waiting for the piece to be delivered on retransmissions. The use of ARQ-based techniques on application layer of OSI model for data streaming allows for a more efficient recovery of lost data chunks in wireless networks with a high level of burst losses. The known models of a discrete channel for wireless networks allow for analytically assessing the probability of data loss, however, they do not take into account cases with retransmission of lost data. The study proposes a mathematical model of data transmission in a wireless communication channel based on the Gilbert model, which takes into account the loss recovery by the ARQ method and allows you to calculate the data loss ratio. To check the adequacy of the proposed model, a software was developed that ensures the transmission of data streaming in a wireless communication network with recovery of fragment losses at the application level, and a corresponding experimental study was carried out. It is shown that the mathematical model takes into account the burstiness of transmitted data losses and their recovery by the ARQ method.