Soft-decision detection and synchronization of continuous phase modulated signals
Abstract
Continuous phase modulation (CPM) is a constant envelope modulation and can be made to give both good bit error probability (BEP) and bandwidth performance by varying certain parameters. These characteristics are important, especially in mobile communication systems. However, the receiver is in general more complex and sequence detection via the Viterbi algorithm is commonly used. But in some applications which include coding with interleaving, symbol-by-symbol detection is preferred since it offers superior BEP performance. In this thesis, first an optimal K lag symbol-by-symbol algorithm for CPM signals in additive white Gaussian noise (AWGN) channels is developed. Since this is computationally complex for some important CPM signals, some sub-optimal techniques are proposed. These techniques can offer a range of computational complexity/BEP performance trade-offs. These algorithms are then modified to also jointly achieve carrier synchronization. In narrowband mobile communication systems the channel is often modeled as a frequency flat, Rayleigh fading channel. A soft-output algorithm for detecting CPM signals transmitted over this channel is developed. Some theoretical bounds for the performance of the algorithms for both the AWGN and the Rayleigh fading channel are developed. The performance of the algorithms in a system with coding and interleaving is characterized extensively by simulations. The superiority of symbol-by-symbol detection compared to sequence detection is demonstrated. In some communication systems digital phase lock loops (DPLL) are used to achieve carrier synchronization. The second order DPLL can achieve phase lock even in the presence of a frequency offset, whereas the first order loop cannot do so. A rigorous analysis of the second order DPLL using Markov analysis techniques, taking into account the effects of quantization is performed. It is shown that this can be used as design criterion for determining the number of quantization levels required for satisfactory performance.
Degree
Ph.D.
Advisors
Fitz, Purdue University.
Subject Area
Electrical engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.