We consider the downlink of a wireless cellular network where the base stations are equipped with multiple antennas and operate in the same frequency band. Due to temporal multi-user scheduling, the spatial transmit signal processing changes with each time slot and base stations require non-causal information about future scheduling and precoding decisions of neighboring base stations in order to encode their data accurately. This can, in theory, be accomplished by a high-capacity backhaul network through which the base stations can exchange channel state information (CSI) and other control signaling. In reality, the temporal granularity of the scheduler does not allow for timely distribution of CSI among base stations. We propose a two-phase scheduler which optimizes the precoding in the first phase and allows the users to feed back their instantaneous interference power in the second phase. If the scheduling is synchronized among base stations, additional infrastructure is not required and base stations operate independently. For single-user transmissions we propose a proportional-fair two-phase scheduler and compare its performance to multi-user two-phase scheduling with dirty paper coding and to algorithms that share CSI among base stations. Simulation results unveil that two-phase scheduling is a viable and technically feasible solution to deal with non-stationary intercell interference. We propose a precoding scheme that explicitly makes use of two-phase scheduling by maximizing the signal-to-leakage-plus-noise ratio in the first phase. It can be implemented for both instantaneous and average CSI and works with and without coordination among base stations. Lastly, we analyze the impact of non-stationary intercell interference on heterogeneous cellular architectures where traditional networks are overlaid with additional low-power base stations. We show that open-access picocells can be deployed as is whereas closed-access femtocells require coordination between the legacy macro-layer and the novel femto-layer. We propose a probabilistic power control algorithm that computes the femto-layer transmit power for a given cell at the corresponding macro base station which distributes it to the femto base stations. The algorithm is solely based on measurements readily available and no additional information exchange from the femto-layer to the macro-layer is required.