Beamforming Design for Wireless Coded Caching with Different Cache Sizes

This paper studies the performance of wireless coded caching over multiple-input and single-output (MISO) channels in a finite signal-to-noise power ratio (SNR) region when every user has a different cache memory size. We first propose multicast beamforming for the network with the conventional coded caching based on quadratic transform (QT) and then point out the non-optimality of the caching scheme when the spatial degree of freedom (DoF) is exploited. We hence formulate a new optimization problem to enhance the caching gain by minimizing the difference between the generated codewords. Numerical results confirm the non-optimality of the conventional coded caching in terms of the average transmission rate and the improvement of our proposed caching.