Date Added: Dec 2010
An Expectation-Maximization (EM) based scheme is presented in the context of compressed image transmission for blind Multiple-Input Multiple-Output (MIMO) wireless channel estimation. This cross-layer content-aware algorithm employs a combination of Hard and Soft (HS) symbol information from the header blocks of the scalable JPEG compressed image to estimate the complex baseband MIMO wireless channel. The proposed HS-EM scheme can significantly reduce the overhead for communication by avoiding the transmission of dedicated pilot symbols which constitute a considerable overhead in a MIMO wireless system. Further, the unique blend of HS information makes the algorithm ideally suited for scalable image transmission in the fast-fading wireless environment, while it simultaneously avoids the convergence problems typically associated with conventional blind estimation algorithms.