A Constrained Optimization Approach to Nonlinear Interference Cancellation
Non-linear interference cancellation has been successful as a low-complexity iterative detection strategy for multiple-input, multiple-output channels. However, such strategies are generally difficult to analyze. Here, the authors develop a connection to constrained optimization through the use of penalty functions. The resulting framework provides a means of analysis, and includes known nonlinear cancellation schemes. Optimal multiuser detection is prohibitively complex for a large number of active users. Lower complexity solutions can be provided by iterative or multi-stage detection structures corresponding to serial or parallel interference cancellation. This class of detectors forms new decisions for each user by subtracting estimates of other user interference obtained from previous iterations.