SREJ

Selective Rejection, abbreviated SREJ, is an error-control mechanism used in some data-link and transport protocols to retransmit only the data segments that were lost or corrupted. It is a form of ARQ (automatic repeat request) that aims to improve efficiency by avoiding unnecessary retransmissions of data that the receiver has already correctly received.

How it works: The sender transmits a sequence of data blocks and maintains them until delivery is confirmed. When the receiver detects missing blocks in the sequence, it buffers any in-order data and sends a selective reject message detailing the gaps (often by specifying the range or ranges of missing sequence numbers). The sender then retransmits only the indicated blocks. This approach contrasts with Go-Back-N, where a single loss can trigger retransmission of all subsequent blocks, regardless of their receipt status.

Design and trade-offs: Implementing SREJ requires additional buffering and state tracking at both ends and careful handling of out-of-order arrivals. It is most advantageous on links with occasional, bursty losses or long round-trip times, where selective retransmission saves bandwidth. However, it increases protocol complexity and may introduce more intricate timing and reassembly logic.

Applications: SREJ has been employed in various specialized data-link and transport protocols, particularly in high-latency or lossy environments such as some satellite and wireless communications. The exact encoding of SREJ messages and how many missing ranges can be reported vary by protocol.

See also: ARQ, Selective Repeat, selective negative acknowledgments (SNACKs), selective acknowledgment (SACK).