GENERALIZED MINIMUM SYMBOL ERROR RATE FOR SYMBOL-RATE AND FRACTIONALLY-SPACED EQUALIZATIONS

摘要

Using a generalized formula for the decision error probability is derived for an M-ary PAM system. The gradient method, which we termed GMSER, is derived to minimize the symbol error rate for linear equalization. Comparisons made between the GMSER algorithm and the MMSE criterion show that the minimum symbol error rate criterion outperforms the minimum mean-squared error criterion. Then, comparisons between GMSER and AMBER as proposed by Yeh and Barry is also made. Fiannlly, an extension of the GMSER algorithm employing fractionally-spaced equalization is presented with mathematical formulations.