Lower Bound on Expected Complexity of Depth-First Tree Search with Multiple Radii

Abstract

Depth-first tree search with multiple radii (DFTS-MR) algorithm attains significant complexity reduction over DFTS with a single radius (DFTS-SR) for solving integer least-squares (ILS) problems. Herein, we derive the lower bound on the expected complexity of DFTS-MR under i.i.d. complex Gaussian environments. Currently, the upper bound on the expected DFTS-MR complexity is known. Our analytical result shows the computational dependence on the statistics of the channel, the noise, and the transmitted symbols. It also reflects the use of multiple radii, which is one of the main characteristics of DFTS-MR. The resultant lower bound provides an efficient means to better understand the complexity behavior of DFTS-MR, along with the (known) upper bound.