Citation
Abstract
The tracking (mean-squared timing-error) performance of a simplified datatransition tracking loop (SDTTL) suggested in the conference literature [1] about 10 years ago for symbol synchronization of a nonreturn-to-zero (NRZ) data stream is presented. In the original paper that suggested this simplification of the datatransition tracking loop (DTTL), which itself dates back about 30 years [2,3], the statement was made that [1, p. 34.8.1] "the SDTTL has exactly the same S-curve and noise performance as compared with the DTTL." Hence, the simplified version, which had the advantage of not requiring a transition detector in the in-phase arm and no delay in the quadrature-phase arm, would have represented the obvious choice for the system designer. Unfortunately, the authors arrived at these sweeping equivalence statements after examining only the noise-free S-curves, which were found to be exactly alike for the two implementations. In this article, we show that while indeed the noise-free S-curves are exactly alike, the similarity in performance stops there; that is, the performances of the original DTTL and the SDTTL are quite different in the presence of noise. In fact, asymptotically for a large symbol signal-to-noise ratio (SNR), the mean-squared timing error (jitter) of the SDTTL is 3-dB worse than that of the DTTL, which considerably negates its implementation simplification. Even at a low SNR, where neither scheme would be motivated by maximum a posteriori (MAP) estimation of symbol synchronization considerations, the DTTL still outperforms the SDTTL.
Details
- Volume
- 42-131
- Published
- November 15, 1997
- Pages
- 1–9
- File Size
- 240.4 KB