“…By assuming fixed arrays consisting of non-uniformly spaced antennas at the Tx and/or Rx, reduced capacity fluctuations and, hence, improved performance can be achieved over a range of transmit distances. However, due to the complexity of the LoS MIMO array design problem, which is highly non-convex, most prior works rely on brute force exhaustive search (ES) [6], [7], stochastic optimization techniques, such as genetic algorithms (GAs) [8], [9], and on numerical results. Specifically, by applying ES and assuming identical antenna locations at the Tx and Rx, non-uniform linear arrays (NULAs) are derived for 4 × 4 systems to maximize the range of transmit distances over which capacity remains above a threshold [6] and the minimum capacity over a given range of transmit distances [7].…”