VBLAST detection based on quadratic zero-one programming

Maximum likelihood detection for VBLAST architectures is equivalent to an integer quadratic optimization problem. Branch and bound (BB) algorithm is a simple method for solving quadratic problem, though other methods exist. In this paper, we use a depth-first BB algorithm for VBLAST detection. For reducing detection complexity, we use a preprocessing stage before running the BB algorithm. We also show that the stage can be improved with a simple modification which leads to lower detection complexity.