Radio Front-end Imperfections

One of the major trends in the transceiver design has been to keep the evolution of digital signal processing domain separate from analogue (RF) front-end. However, physical implementations of these low cost and energy efficient RF front-ends suffer from sensitivity to a number of imperfections and if not handled carefully, could become a limiting factor in the overall system performance. These impairments are predominantly a manifestation of the fabrication process variations, which cannot be predicted or controlled and hence differ from chip to chip.

In order to design appropriate solutions, a better understanding of the underlying phenomena for imperfect behavior of the transceiver chain is required. A realistic performance evaluation of any practical system design must take dominant RF impairments into consideration. Furthermore, by understanding these complex impairment processes, we can design efficient algorithm to compensate their effects at transmitter through pre-distortion and/or at receiver through post-distortion utilizing sophisticated digital signal processing algorithms. Integration of these advanced techniques in digital paradigm could leverage some system performance or conversely relax the stringent performance specifications on RF circuitry, leading to a much more cost effective solution.

While OFDM-based MIMO techniques have received a lot of interest in research community, their RF front end implementation, its implication on the system performance and the effectiveness of the DSP-based mitigation techniques have not been thoroughly investigated. Most essential RF impairments typically encountered in design and implementation include limited image rejection due to I/Q imbalance, non-linear distortion due to PA at the transmitter, LNA and mixer at the receiver, and  phase noise originating from random fluctuations or instability of the oscillators. This project deals with analysis, modeling and low complexity suppression of RF non-idealities using novel digital signal processing algorithms.

Contact


Aamir Ishaque