A comparison of approximate Viterbi techniques and particle filtering for data estimation in digital communications

We consider trellis-based algorithms for data estimation in digital communication systems. We present a general framework which includes approximate Viterbi algorithms like the M-algorithm and the T-algorithm as well as marginal and joined particle-filtering algorithms. The algorithmic concepts are very close, since the difference only is the choice of the norm in the weights calculation. The general framework yields hence a new interpretation of these algorithms and may give rise to a series of new algorithms by using general selection schemes or a different choice for the norm. We show the potential improvements using different selection schemes for the EMVA.

• Type: Conference
• Year: 2010
• Where: ICASSP
• Authors: Steffen Barembruch
• Keywords: Algorithms, Approximate Viterbi Algorithms, General Framework, ICASSP 2010, Signal Processing