The per survivor processing (PSP) algorithm consists a general frame work for the maximum likelihood sequence estimation (MLSE) algorithms whenever we are trouble in Viterbi decoding because of the phase offset. The main idea of PSP is combining data-aided estimation techniques into the structure of the Viterbi algorithm itself. However, the complexity of PSP is much higher than the convectional tentative decision (TD) algorithm. In this paper, we propose the new algorithm which uses the adaptive number of estimators, depending on channel condition, not only to reduce complexity but also to improve the performance.