2016/05/10 by Giorgio M. Vitetta, Emilio Sirignano, Vitetta, Giorgio M. +5 · 1 citation
Computer Science · #Bayesian Modeling and Causal Inference #Error Correcting Code Techniques #FOS: Electrical engineering #FOS: Mathematics #Statistics Theory (math.ST) #Systems and Control (eess.SY) #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.1605.03017
openalex publication_date 2016/05/10 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
In this manuscript a factor graph approach is employed to investigate the\nrecursive filtering problem for a mixed linear/nonlinear state-space model,\ni.e. for a model whose state vector can be partitioned in a linear state\nvariable (characterized by conditionally linear dynamics) and a non linear\nstate variable. Our approach allows us to show that: a) the factor graph\ncharacterizing the considered filtering problem is not cycle free; b) in the\ncase of conditionally linear Gaussian systems, applying the sum-product rule,\ntogether with different scheduling procedures for message passing, to this\ngraph results in both known and novel filtering techniques. In particular, it\nis proved that, on the one hand, adopting a specific message scheduling for\nforward only message passing leads to marginalized particle filtering in a\nnatural fashion; on the other hand, if iterative strategies for message passing\nare employed, novel filtering methods, dubbed turbo filters for their\nconceptual resemblance to the turbo decoding methods devised for concatenated\nchannel codes, can be developed.\n