Achievable Rates of Opportunistic Cognitive Radio Systems Using Reconfigurable Antennas with Imperfect Sensing and Channel Estimation

TL;DR

This paper analyzes the achievable data rates in opportunistic cognitive radio systems with reconfigurable antennas, considering imperfect sensing and channel estimation, and proposes an optimization framework for system parameters.

Contribution

It introduces a lower bound on achievable rates accounting for errors and formulates an optimization problem to maximize these rates by tuning sensing and training durations.

Findings

Optimizing channel training duration yields higher rate improvements.

The lower bound accounts for sensing, estimation, and detection errors.

Numerical results highlight the importance of training over sensing duration.

Abstract

We consider an opportunistic cognitive radio (CR) system in which secondary transmitter (SUtx) is equipped with a reconfigurable antenna (RA). Utilizing the beam steering capability of the RA, we regard a design framework for integrated sector-based spectrum sensing and data communication. In this framework, SUtx senses the spectrum and detects the beam corresponding to active primary user's (PU) location. SUtx also sends training symbols (prior to data symbols), to enable channel estimation at secondary receiver (SUrx) and selection of the strongest beam between SUtx-SUrx for data transmission. We establish a lower bound on the achievable rates of SUtx-SUrx link, in the presence of spectrum sensing and channel estimation errors, and errors due to incorrect detection of the beam corresponding to PU's location and incorrect selection of the strongest beam for data transmission. We formulate a novel constrained optimization problem, aiming at maximizing the derived achievable rate lower bound subject to average transmit and interference power constraints. We optimize the durations of spatial spectrum sensing and channel training as well as data symbol transmission power. Our numerical results demonstrate that between optimizing spectrum sensing and channel training durations, the latter is more important for providing higher achievable rates.