An improved fast Viterbi algorithm for track-before-detect

This paper addresses the problem of moving dim target detection in radar systems using track-before-detect approaches (TBD), which are implemented by a fast Viterbi algorithm (FVA). Traditional Viterbi algorithms (VA) traverse all the resolution cells including many noise cells in every frame, whereas FVA improves the searching method by combining with a Constant False Alarm Rate (CFAR) approach to expel many noise cells, and provides the same performance with much less computational cost. What´s more, it avoids target extend with no post-processing due to a novel maximization operation, and consequently reduces the complexity of the proposed algorithm. Simulation results prove the performance of the new algorithm.