Exploiting the initial and the final conditions for the alternating minimization algorithm

In this paper, we study in depth the alternating minimization algorithm (AMA) proposed by S. W. Peters and R. W. Heath Jr. Among iterative interference alignment (IIA) schemes, AMA is considered more feasible for real applications than other IIA schemes. However, no research has investigated how AMA is affected by some real system characteristics. Motivated by this, we look into the effects when taking into account the initial conditions and the final conditions of AMA in the MIMO interference networks. We found the sum rate performance of AMA can be improved by well exploiting its initial conditions. On the other hand, by considering the interference leakage residue (ILR) which is the final condition of AMA, we derived a more realistic formulation which fast calculates the smallest possible codebook for precoders.