Adaptive Feedback Partitions in Dynamic Zero-Forcing Beamforming Based on Stochastic Geometry

In this paper, we analyze the performance of dynamic zero-forcing beamforming with limited feedback and propose adaptive feedback partition strategies. Assuming a cellular network model based on stochastic geometry, we derive analytical expressions for the complementary cumulative distribution function (CCDF) of the instantaneous signal-to-interference ratio (SIR) and the ergodic spectral efficiency as functions of the relevant system parameters: the cluster size, the number of feedback bits, the intra-cluster geometry, and the path-loss exponent. By leveraging these expressions, we propose two feedback partition strategies, each of which is applied for known intra-cluster geometry and unknown intra- cluster geometry, respectively. In the simulations, we verify the accurateness of our analysis and demonstrate the proposed strategies by comparing to the equal feedback partition. Our results show that by adaptively partitioning the feedback bits, the intra- cluster interference is efficiently managed whether the intra-cluster geometry is known or not.