1：Experimental Design and Method Selection:

The study uses a sequential hypothesis test based on Generalized Likelihood Ratio (GLR) statistics to adaptively select the best beam in massive MIMO systems. It involves comparing stochastic trajectories of GLR statistics for each beam against a fixed threshold to terminate the test when one beam surpasses it.

2：Sample Selection and Data Sources:

Simulations are conducted with a single path channel model, where the angle of arrival (AoA) is uniformly distributed in [-90°, 90°], and signal-to-noise ratio (SNR) is varied. Data is generated for a uniform linear array with a Butler matrix codebook of 16 beams.

3：List of Experimental Equipment and Materials:

No specific physical equipment is mentioned; the study is simulation-based using mathematical models and algorithms.

4：Experimental Procedures and Operational Workflow:

For each simulation run, observations are generated as y_i[n] = A_i s[n] + w_i[n], where s[n] is a pseudo-random training sequence, A_i is the signal amplitude for beam i, and w_i[n] is white Gaussian noise. The sequential test involves computing GLR statistics γ_i(n) for each beam and comparing them to a threshold γ_th derived from a target probability of false alarm (P_FA). The test continues until one γ_i(n) exceeds γ_th, indicating the selected beam.

5：Data Analysis Methods:

Performance is evaluated in terms of average captured relative magnitude loss (ˉl) and average test length (ˉn). Monte Carlo simulations with 10^4 runs per SNR point are used to estimate these metrics. Statistical analysis involves Q-functions and χ^2 distributions for hypothesis testing.