An experimental study on digital predistortion for radio-over-fiber links

Radio-over-fiber (RoF) has been proposed as an enabling technology for broadband networks, such as WiMAX and WiFi. Besides the inherent high bandwidth and reliability of RoF systems, they also allow the reduction of installation and maintenance cost of the remote antenna units (RAUs) and improvement in the coverage area of the base station/access point. However, the nonlinear distortion of the optical link, which stems mainly from the laser diode, may impose serious limitations on the system performance, especially when high PAPR, wideband signals are used. Thus, distortion compensation is a key issue in order to facilitate the application of the RoF technology for broadband networks. In this work, digital predistortion for directly modulated RoF links is experimentally investigated. A memory-polynomial-based predistorter model is identified from measurements of the baseband OFDM input-output signals and through the use of an indirect learning architecture. Then, a predistorted signal is generated and fed to the RoF link for comparing its output with that of the non-predistorted one. As a result of this compensation technique, an improvement of the static link linearity and the output constellation diagram has been found, with an EVM reduction of 1.73 %.