Blind acquisition characteristics of PSP-based sequence detectors

The blind acquisition characteristics of binary sequence detection algorithms based on per-survivor processing (PSP) are demonstrated via computer simulations. These capabilities are impressive as compared to those of traditional blind equalizers. It is demonstrated that the short-term acquisition performance of PSP-based algorithms is dominated by the poor performance obtained when certain sequences are transmitted. These sequences are those that cannot be distinguished by the joint maximum likelihood (ML) channel and sequence estimator or do not allow for a complete identification of the channel impulse response. Such sequences are defined and analytically characterized, resulting in asymptotic results on performance for joint ML channel and sequence estimators. Typical misacquisition conditions, the effects of initialization, and the impact of increased tree search complexity are also all characterized by simulations and motivated by the analytical results.