论文信息 - Comments on Why Generalized BP Serves So Remarkably in 2-D Channels

Comments on Why Generalized BP Serves So Remarkably in 2-D Channels

Generalized belief propagation (GBP) algorithm has been shown recently to infer the a-posteriori probabilities of finite-state input two-dimensional (2D) Gaussian channels with memory in a practically accurate manner, thus enabling near-optimal estimation of the transmitted symbols and the Shannon-theoretic information rates. In this note, a rationalization of this excellent performance of GBP is addressed.

[1] Anthony J. Weiss,et al. Generalized belief propagation receiver for near-optimal detection of two-dimensional channels with memory , 2004, Information Theory Workshop.

[2] Shlomo Shamai,et al. On the achievable information rates of finite-state input two-dimensional channels with memory , 2005, Proceedings. International Symposium on Information Theory, 2005. ISIT 2005..

[3] Shlomo Shamai,et al. Discrete-Input Two-Dimensional Gaussian Channels With Memory: Estimation and Information Rates Via Graphical Models and Statistical Mechanics , 2008, IEEE Transactions on Information Theory.