Low-Complexity Blind Timing Recovery in Time-Dispersive Channels

The problem addressed in this thesis is how to perform symbol synchronization in the presence of channel dispersion. The motivation for this work was to design a synchronization system suitable for an ATSC terrestrial digital television receiver. While such receivers present some specific challenges, the work here is generally applicable to digital receivers working in the presence of dispersive channels. We seek a general efficient low-complexity solution for timing recovery in time-dispersive channels that does not rely on timing signals or sequences of known data symbols to be sent by the transmitter. This problem is a difficult one and is especially important since a satisfactory solution is currently lacking. Our first contribution is in terms of developing a general design philosophy in Chapter 3. This lays the foundations for some of the ideas for explicit designs in later chapters. Toward the end of improving receiver acquisition in terms of speed and performance, we develop a general design philosophy which is a general strategy to improve receiver performance. The primary application is for timing recovery in time-dispersive channels but the design philosophy is also illustrated by other examples such as CMA blind equalization. The “chicken and egg problem”, whereby timing recovery requires the benefit of equalization and equalization requires the benefit of timing recovery is also explored. It is shown how our design philosophy can break this inter-dependency and lead to a new design strategy which utilizes a pre-equalizer prior to timing recovery. This pre-equalizer does not fully equalize the channel output signal but acts in a way to assist timing recovery. Later chapters explore this strategy in depth. Finally, to further inform the design process we make a list of design attributes in terms of desirable properties for the v receiver structure and the algorithms employed. Chapter 4 utilizes the design philosophy established in the previous chapter to introduce a blind pre-equalization stage before timing recovery. The form of the pre-equalizer (that is, its structure and the rules governing its adaptation) is developed via a strategy of partitioning both the channel and the equalization portion of the receiver. The particular method that is used to partition the equalizer allows the identification of an all-pass blind magnitude-only equalizing portion. We demonstrate this portion to be a suitable choice for pre-equalization as it helps the receiver to meet the list of desirable properties for a receiver that is outlined in Chapter 3. It is further demonstrated that pre-equalization can be retro-fitted to existing receiver designs, broadening the class of dispersive channels that they can successfully deal with. Finally, it is shown that the new receiver structure utilizing pre-equalization can improve receiver performance, acquisition and robustness in a number of difficult time-dispersive channels. The ideas of the previous chapters, namely joint timing recovery and equalization, and structural variations in the receiver architecture are used in Chapter 5 to derive an equalizer-based timing recovery scheme. This novel timing recovery scheme obtains its timing information from the movement of the taps of an equalizer away from their center and introduces the idea of using a short and fast “timing-equalizer”. Two different methods of detecting the timing information from the equalizer taps are also introduced and their performance is evaluated for their suitability to ATSC digital television reception.