retransmission

Retransmission is the process of sending data again after the original transmission failed to be correctly received. It is used to improve reliability over imperfect communication channels, where errors, interference, or packet loss can occur. Retransmission can be triggered by error detection codes such as cyclic redundancy checks that indicate corruption, or by missing acknowledgments in acknowledged communication protocols.

In data communication, retransmission is a core mechanism of automatic repeat request (ARQ) schemes. In stop-and-wait ARQ, the sender transmits a single frame and waits for an acknowledgment; on timeout or a negative acknowledgment, the frame is resent. In go-back-N ARQ, multiple frames may be in flight, but if an error is detected in a frame, all subsequent frames are retransmitted. In selective repeat ARQ, only the erroneous frames are resent. Piggybacking acknowledgments can combine acknowledgments with data sent in the opposite direction to improve efficiency.

In TCP, retransmission operates with a timeout-based mechanism and fast retransmit, aided by congestion control. If the sender does not receive an acknowledgment within a calculated retransmission timeout, it resends the unacknowledged segment. Fast retransmit may trigger after several duplicate acknowledgments. Retransmissions interact with network congestion control and can affect overall throughput and latency.

Hybrid ARQ (HARQ) combines forward error correction with ARQ, allowing partial corrections to be applied before deciding to retransmit. This approach is common in wireless and satellite links, where it can reduce latency and the number of retransmissions by leveraging redundancy and incremental redundancy.

In broadcasting and content distribution, retransmission can also refer to resending content on a different channel or at a later time to reach audiences who missed the original transmission. This usage relates to content availability rather than protocol-level data integrity.