How is RTT estimated in TCP and how does it influence retransmission timers?

• A. RTT is estimated from sample RTTs; smoothed RTT and deviation determine the RTO, affecting retransmission timing.
• B. RTT is measured only at connection setup.
• C. RTT does not affect retransmission timers.
• D. RTT is used to order streams in multiplexing.

Answer: (A)

TCP estimates RTT from actual sample measurements each time an acknowledgment arrives. Those samples are used to update a smoothed RTT, which tracks the typical delay, and a deviation term that captures how much the RTT varies. The retransmission timeout (RTO) is derived from these two values, commonly as RTO = SRTT + 4 * RTTVAR. This dynamic timer adapts to changing network conditions: when delays grow, the deviation increases and the RTO lengthens to avoid premature retransmissions; when delays shrink, the timer tightens to recover quickly. RTT measurements continue throughout the connection, not just at setup, and they directly influence when unacknowledged data is retransmitted. RTT is not used to order streams in multiplexing; that function lies elsewhere in the protocol.