Link Adaptation Mechanisms Based on Logistic Regression Modeling

Abstract

In this letter, we obtain the optimal signal-to-noise ratio regions for the adaptive modulation and coding (AMC) over a set of correlated subcarriers. We use logistic regression to model the block error rate (BLER) with the aim of maximizing the throughput while keeping the BLER below a predefined target. We also study the robustness of a suboptimal optimization procedure, which maintains the relative distance among thresholds to that preserving the instantaneous BLER under the target. Moreover, we bring a new understanding of the outer loop link adaptation (OLLA) mechanism in terms of the AMC. We show that OLLA is able to converge (in average) to the thresholds designed by this suboptimal procedure. We also study the unstable behavior of OLLA implementations that update one offset per modulation and coding scheme, bringing out the need for new more flexible link adaptation mechanisms.