Circumventing base station cooperation through Kalman prediction of intercell interference

We consider the downlink of a cellular multiple input single output (MISO) system with multiple users per cell. Fast scheduling and spatial signal processing at the base stations result in unpredictable non-stationary intercell interference when the base stations do not cooperate. We show how the per cell sum-rate can be increased when Kalman filters are employed to forecast the interference power for the next transmission. We compare these results for dirty-paper coding and beamforming to systems where the intercell interference powers are perfectly known through feedback channels or cooperating base stations and to systems where outdated information is used.