[B-17] スマート無線

[B-17-1] Initialization of FCME Algorithm for Noise Floor Estimation

^○Yau Hong Leow¹, Kenta Umebayashi¹, Janne Lehtomäki², Miguel López-Benítez³ (1.Tokyo Univ. of Agriculture and Tech., 2.Univ. of Oulu, 3.Univ. of Liverpool)

この講演は本会「学術奨励賞受賞候補者」の資格対象です。

キーワード：Noise floor estimation、Initialization、Constant false-alarm rate (CFAR)

In this paper, we investigate a real-time noise power estimation with Forward Consecutive Mean Excision (FCME) algorithm. In the real-time noise power estimation, the observed samples may include both noise-only samples (H0) and noise-plus-signal samples (H1). The FCME algorithm extracts noise-only samples with an iterative process in which a set of estimated noise-only samples is updated iteratively. We investigate an appropriate size for the initial noise samples set, denoted by |Q0|, in FCME algorithm while |Q0| is typically assumed to be 10% smallest samples from the total observed samples (N). In this paper, the appropriate size |Q0| is obtained analytically based on a relationship between false alarm rate (PF A) and |Q0|.