Rate adaptation using long range channel prediction based on discrete prolate spheroidal sequences

Rate adaptation which adjusts the transmission rate based on channel quality, plays a key role in the performance of 802.11 networks and is critical for achieving high throughput. Traditional statistics-based methods for rate adaptation are unsuited for mobility scenarios because of the delay involved in statistics gathering. Methods based on channel state information (CSI) perform better but still fall short of optimal performance in high mobility. In this paper, we consider adaptive modulation based on Slepian channel prediction as a basis for rate adaptation in high mobility scenarios. Our proposed method utilizes low-complexity projection on a subspace spanned by discrete prolate spheroidal (DPS) sequences. These sequences are simultaneously bandlimited and highly energy concentrated, and they can be used to obtain a minimum energy bandlimited extension of a finite sequence. Using the predicted channel coefficients, we select the modulation scheme resulting in the highest expected throughput. Unlike Wiener prediction, the proposed method does not require detailed knowledge of the channel correlation, but only of the Doppler bandwidth. Our numerical results show that adaptive modulation based on the low-complexity Slepian prediction is substantially better than using outdated CSI and performs very close to Wiener prediction.