A semi-blind multiuser projection receiver based on conjugate gradient and subspace tracking

A semi-blind multiuser projection receiver is proposed. Using interference cancellation of the known users' signal and subspace tracking for the unknown users' signal subspace, the orthogonal projection matrix of the unknown users' signal subspace is obtained. Then, using this matrix, we transform the detection problem of known users' data into a minimization problem of a quadratic convex function in K variables and solve it using a conjugate gradient (CG) algorithm. The initial estimation of the known users' data also serves as the initial value of the CG to speed convergence.