CQI Prediction via Hidden Markov Model for Link Adaptation in Ultra Reliable Low Latency Communications

The ultra reliable and low latency communications (URLLC) building blocks are among the main drivers of future wireless networks. Precise link adaptation (LA) procedure is a necessity to achieve a highly reliable communications link with low latency. We propose an algorithm to enhance the performance of LA by predicting the next channel quality indicator (CQI) at the base station.We use the hidden Markov model (HMM) to capture the latent intrinsic randomness of the mobile cellular networks. This method does not use ACK/NACK feedback to improve the CQI estimation. Instead, we observe the previous instances of CQI reports and fit an HMM model to output a probability distribution on the next CQI values. In order to combat the non-stationary wireless channel, we introduce a forgetting factor to only keep the most relevant samples. We then sample from this probability density function to achieve the predicted CQI value that ensures a target BLER. The simulations are ran over the data obtained in a fully 3GPP compliant system simulator. The results indicate that the proposed method outperforms the state-of-the-art outer loop link adaptation (OLLA) algorithm in predicting the SINR sequence.