Adaptive signaling based on statistical characterizations of outdated feedback in wireless communications

Wireless links form a critical component of communication systems that aim to provide ubiquitous access to information. However, the time-varying characteristics (or "state") of wireless channels caused by the mobility of transmitters, receivers, and objects in the environment make it difficult to achieve reliable communication. Adaptive signaling exploits any channel state information (CSI) available at the transmitter to provide the potential to significantly increase the throughput of wireless links and/or greatly reduce the receiver complexity. As such, adaptive signaling has attracted significant research interest in the last decade and has found application in numerous commercial wireless systems, ranging from cellular data systems to wireless local area networks (WLANS). However, one of the great challenges of wireless communications is that it is difficult to obtain perfect CSI due to the inherently noisy and outdated nature of CSI available at the transmitter. Over the last decade, we have championed the idea of choosing the appropriate transmitted signal based on statistical models for the current channel state conditioned on the channel measurements. In this semi-tutorial paper, we first review how this class of methods has been developed for single-antenna systems, and then present novel recent designs for multiple-antenna systems. Key to the development in each case is the development of the error characterization given the outdated estimates and the use of such to allocate data rate and power over time and possibly space. in general, the focus is on rate allocation, while power allocation is done through a pruning method. Numerical results will demonstrate in both the single-antenna and multiple-antenna cases that such an approach provides a robust method for improving system data rate versus the standard practice of employing link margin to compensate for such uncertainties.