Adaptive coding

Adaptive coding is the dynamic adjustment of a communication system's code rate, modulation format, or error-correction strength in response to real-time estimates of channel quality. Unlike fixed-rate coding, which assumes a worst-case channel and wastes capacity when conditions are favorable, adaptive coding matches the protection level to the instantaneous noise environment. When the channel is clean, it sends more data bits per symbol; when the channel degrades, it increases redundancy to maintain reliability.

The technique requires a feedback loop: the receiver estimates the channel state — through signal-to-noise ratio measurements, bit error rate monitoring, or pilot tone analysis — and communicates this estimate back to the transmitter. The transmitter then selects a code from a pre-defined family, typically a rate-compatible punctured convolutional code or a set of LDPC codes with different rates. The latency of this feedback loop is a critical design parameter; in fast-fading wireless channels, the adaptation must occur faster than the coherence time of the fading process.

Adaptive coding bridges the gap between Shannon's static capacity theorem and the time-varying reality of physical channels. It is the engineering response to the observation that the binary symmetric channel with fixed crossover probability is a useful mathematical abstraction but a poor model of actual communication media.

Adaptive coding is not an optimization technique. It is an admission that the world is not stationary — that the channel capacity itself is a random variable, and that the best code is not the one that maximizes throughput in the worst case, but the one that tracks the moving target of actual channel capacity. It is the difference between building a seawall and learning to sail.