Enhanced Forced Symbol Method for improving the error rate performance of turbo codes

Abstract

The Forced Symbol Method (FSM) lowers significantly the error rate, especially in the error floor region. The core of the FSM is the application of repeated decoding, with one or more least reliable symbols being forced to certain values, when an error is detected. The main drawback of the FSM method is its peak complexity which can be high. This paper presents a new approach for reducing the peak and average complexity of the FSM, called Enhanced Forced Symbol Method (EFSM). The main core of the EFSM relies on the evaluation of a set of decisions, when decoding of a frame fails, to determine positions of symbols in decoded frame that are likely to be in error. The efficiency of the EFSM is demonstrated by its ability to reduce significantly both the peak and the average complexity by a factor of 3 to 5.7 when compared to the standard FSM, while achieving essentially the same error rate performance of the standard FSM.