A non-data-aided decision-directed feedforward symbol timing estimation with various spacing

A non-data-aided decision-directed feed forward symbol timing estimation with various spacing is proposed. It uses four samples per symbol. Based on a look-up table, the computation is simple but the performance is excellent at small roll-off factors and is comparable to that of ML-based LOGN. Computer simulation using QPSK signals shows that the performance difference between various spacings in DD is small, but DD with spacing of 3T/4 is marginally better than the others at high S/N and small roll-off factors.