Retransmission Strategies — Architecture & Tradeoffs

Overview

Retransmission strategies define how and when a communication system resends lost or corrupted packets. They operate on top of error‑detection mechanisms (typically CRC) and interact closely with ARQ, sliding windows, and flow control. The choice of retransmission strategy has a direct impact on throughput, latency, energy efficiency, and overall protocol behavior.

Modern systems use a combination of timers, acknowledgments, selective feedback, and congestion signals to determine the optimal retransmission policy for a given channel.

Architectural Principles

Error Detection as the Trigger

Retransmissions are always triggered by error detection, not error correction. The receiver identifies a corrupted or missing packet using:

CRC mismatch
sequence number gaps
missing ACKs
explicit NACKs

Once an error is detected, the retransmission strategy determines the next step.

Feedback‑Driven Behavior

Retransmission relies on feedback from the receiver or the network:

ACKs confirm successful delivery
NACKs request retransmission
timeouts detect missing feedback
selective acknowledgments identify specific missing packets

The feedback granularity shapes the retransmission policy.

Sender Buffering

The sender must retain copies of unacknowledged packets until:

they are acknowledged
they are retransmitted
they expire (in some protocols)

Buffer size and retention time influence memory requirements and latency.

Core Retransmission Strategies

Timeout‑Based Retransmission

The sender starts a timer for each transmitted packet or window. If no ACK arrives before the timeout expires, the packet is retransmitted.

Pros: simple, robust
Cons: sensitive to timeout tuning; may cause unnecessary retransmissions

Timeouts must be carefully calibrated to the round‑trip time (RTT) and channel variability.

NACK‑Driven Retransmission

The receiver explicitly signals which packet was lost or corrupted.

Pros: fast error recovery
Cons: requires reliable NACK delivery and additional control logic

Used in protocols where feedback overhead is acceptable.

Cumulative ACK‑Driven Retransmission

The sender infers loss when cumulative ACKs stop advancing.

Example:
If ACK 37 is received repeatedly, packet 38 is assumed lost.

Pros: low overhead
Cons: slow to detect isolated losses; may trigger burst retransmissions

Widely used in TCP.

Selective Retransmission

The receiver identifies exactly which packets are missing using:

selective ACKs (SACK)
bitmap acknowledgments
explicit sequence number lists

The sender retransmits only the missing packets.

Pros: highly efficient, minimal redundancy
Cons: requires more complex receiver state and larger buffers

Used in modern wireless systems and TCP SACK.

Hybrid Retransmission Strategies

Fast Retransmit

Triggered by duplicate ACKs rather than timeouts.

Example:
Receiving three duplicate ACKs for the same sequence number implies a missing packet.

Pros: faster recovery than timeout
Cons: may misinterpret reordering as loss

Used in TCP to reduce latency.

Adaptive Retransmission

The sender dynamically adjusts retransmission behavior based on:

RTT estimates
jitter
congestion signals
error rate
link quality

Adaptive strategies are essential in wireless and mobile environments.

HARQ (Hybrid ARQ)

Combines FEC with retransmission:

Type I: independent FEC + ARQ
Type II/III: incremental redundancy (IR), where retransmissions add new parity bits

HARQ provides excellent performance in noisy channels and is used in LTE, 5G NR, and Wi‑Fi 6/7.

Performance Considerations

Latency

Retransmissions add delay due to:

timeout expiration
feedback propagation
reordering and buffering

Fast retransmit and selective retransmission reduce latency significantly.

Throughput

Throughput is affected by:

retransmission frequency
window size
congestion control
channel error rate

Selective retransmission maximizes throughput in high‑error environments.

Efficiency

Efficiency depends on avoiding:

unnecessary retransmissions
burst retransmissions
retransmission storms during congestion

Adaptive strategies help maintain efficiency across varying conditions.

Integration with Protocol Layers

Link Layer

Implements:

simple ARQ
Go‑Back‑N
Selective Repeat
HARQ (in wireless systems)

Link‑layer retransmissions hide errors from upper layers.

Transport Layer

Implements:

sliding windows
congestion control
selective acknowledgments
fast retransmit

TCP is the canonical example.

Application Layer

Some applications implement their own reliability mechanisms (e.g., QUIC), often combining selective retransmission with congestion‑aware logic.

Comparison of Retransmission Strategies

Strategy	Trigger	Efficiency	Latency	Complexity	Typical Use
Timeout	Timer expiry	Low	High	Low	Simple ARQ
NACK	Explicit feedback	Medium	Medium	Medium	Link‑layer ARQ
Cumulative ACK	ACK stalling	Medium	Medium-High	Low	TCP
Selective	SACK/bitmap	High	Low	High	Wireless, TCP SACK
Fast Retransmit	Duplicate ACKs	Medium-High	Low	Medium	TCP
HARQ	FEC + ARQ	Very High	Low-Medium	High	LTE, 5G

Design Tradeoffs

Feedback overhead vs efficiency — selective retransmission requires richer feedback.
Buffer size vs performance — selective strategies need larger buffers.
Latency vs robustness — timeout‑based strategies are robust but slow.
Congestion vs error recovery — retransmissions must not be confused with congestion signals.
Energy consumption — critical in wireless systems where retransmissions cost power.

Summary

Retransmission strategies define how communication systems recover from errors using feedback, timers, and selective mechanisms. They are central to reliability and performance across link‑layer ARQ systems, transport protocols like TCP, and advanced wireless technologies. The choice of strategy shapes latency, throughput, and efficiency, and must be tuned to the characteristics of the underlying channel.

DexterLab