ABSTRACT
This chapter provides a tutorial introduction to factor graphs and associated message passing algorithms. It explains the global multivariate function factorizes into a product of simpler "local" functions and these functions by factor nodes. The chapter examines the basic terms associated with factor graphs and some rules that help in constructing factor graphs with the help of illustrative examples. It discusses the modeling systems using factor graphs where two specific modeling regimes, behavioral modeling and probabilistic modeling, are described. The chapter presents the message passing algorithm for inference in factor graphs. It describes the sum-product algorithm and the max-product algorithm. The chapter also discusses Factor graphs with cycles and some general comments. It also describes an application of the sum-product algorithm on factor graphs to jointly estimate the carrier frequency offset and phase noise in orthogonal frequency division multiplexing transmission. The chapter concludes with some directions for future research in factor graphs.
