Within the LOEWE Priority Program Cocoon (Cooperative Sensor Communication) the utilization of mixed-integer Nonlinear optimization in wireless telecommunication networks is explored. We focus on applications that can be modeled as (nonconvex) quadratically constrained quadratic problems (QCQP) featuring "on/off" constraints. Solution strategies for solving the underlying QCQPs include the consideration of the semidefinite programming relaxation as well as sequential second-order cone programming, and a local rank reduction heuristic. The "on/off"-constraints are dealt with within a Branch-and-Bound framework. As an interesting application we consider the successive interference cancellation model. We assume a downlink communication system comprising one multi-antenna base station and single-antenna users which have multiuser detection receivers. That is, a receiver is capable of decoding and cancelling an interfering signal before decoding the desired signal, if it is received strongly enough. The goal is to jointly find a transmission strategy at the base station and an interference cancellation order for each receiver which minimizes the total transmission power while insuring signal-to-interference-plus-noise ratio requirements at each user. We conclude by presenting some first numerical results.