engleski

Signal Processing Algorithms in Multicarrier Systems

Multicarrier modulation is a technique in which a spectrally shaped channel with colored noise is partitioned into a set of parallel orthogonal subchannels, each with its own carrier. The work presented in this dissertation focuses on multicarrier systems whose basis vectors are inverse discrete Fourier transform and discrete Fourier transform vectors. To be able to reliably transmit and receive digital data at the highest possible data rate, every component of the communication system must be optimized. This dissertation addresses the following problems: optimal energy allocation (the bit-loading problem), frequency error estimation and correction, and impulse noise cancellation. A number of practical bit-loading algorithms designed for communication systems with power spectral density constraint are proposed and analyzed. Special attention is given to the problem of noise margin maximization with power spectral density constraint and bit rate constraint. A major drawback of multicarrier systems is their sensitivity to synchronization errors. The influence of sampling frequency offset is discussed and methods for its estimation and correction are proposed. A novel sampling frequency offset correction structure is developed, which interpolates the twiddle factors instead of input samples. Impulse noise is known to be highly damaging to the performance of a multicarrier system. In this dissertation, a selection of impulse noise detection algorithms is presented, and their performance is investigated. Consequently, a novel impulse noise cancellation algorithm that combines time-domain and frequency-domain analysis is proposed. Finally, computationally efficient receiver structures, for systems employing modified frequency hopping modulation are presented.

multicarrier modulation; bit-loading algorithm; sampling frequency offset estimation and correction; impulse noise detection and cancellation

nije evidentirano