Widely linear equalization and blind channel identification for interference-contaminated multicarrier systems

• Authors: D. Darsena;G. Gelli;L. Paura;F. Verde
• Venue: IEEE Transactions on Signal Processing
• Year: 2005

This work addresses the problem of designing efficient detection techniques for multicarrier transmission systems operating in the presence of narrowband interference (NBI). In this case, conventional linear receivers, such as the zero-forcing (ZF) or the minimum-mean square error (MMSE) ones, usually perform poorly since they are not capable of suppressing satisfactorily the NBI. To synthesize interference-resistant detection algorithms, we resort to widely linear (WL) filtering, which allows one to exploit the noncircularity property of the desired signal constellation by jointly processing the received signal and its complex-conjugate version. In particular, we synthesize new WL-ZF receivers for multicarrier systems, which mitigate, in the minimum output-energy (MOE) sense, the NBI contribution at the receiver output, without requiring knowledge of the NBI statistics. By exploiting the noncircularity property, we also propose a new subspace-based blind channel identification algorithm and derive the channel identifiability condition. Blind identification can be performed satisfactorily also in the presence of NBI, requiring only an approximate rank determination of the NBI autocorrelation matrix. The performance analysis shows that the proposed MOE WL-ZF receiver, even when implemented blindly, assures a substantial improvement over the conventional linear ZF and MMSE ones, particularly when the NBI bandwidth is very small in comparison with the intercarrier spacing and the NBI is not exactly located on a subcarrier.