Phase noise estimation via adapted interpolation

Abstract

Wideband communication systems operating at high SNR's, like many point to point communication systems, generally suffer from poor oscillator phase noise characteristics. When QPSK pilot symbols are inserted periodically into a QAM data symbol stream, it is possible to obtain phase noise estimates for these time instants. Wiener interpolation is used to generate linear minimum mean-squared error (LMMSE) estimates for the time instants between those values. Finally, an adaptive scheme based on the well known least mean-squares (LMS) approach is presented. Using this adaptive method, the estimation algorithm converges to its optimum independently of the phase noise characteristic of a specific oscillator, or slowly varying temperatures.