Computer Science > Information Theory
[Submitted on 19 May 2014 (v1), revised 26 Oct 2015 (this version, v14), latest version 26 Nov 2019 (v22)]
Title:Signal-noise search RMT estimator with adaptive decision criterion for estimating the number of signals based on random matrix theory
View PDFAbstract:Estimating the number of signals is a fundamental problem in many scientific and engineering fields. As a well-known estimator based on the random matrix theory (RMT), the RMT estimator estimates the number of signals via detecting the largest noise eigenvalues. Since the RMT estimator does not consider the interaction among eigenvalues, it tends to down-estimate the number of signals when there are weak signals. In order to overcome this problem, we propose signal search enhanced RMT estimators by considering the interaction among the sample eigenvalues. The main contributions are as follows: (a) We firstly derive a novel decision statistics for signal number estimation by utilizing the distribution of the sample signal eigenvalues. Then, based on this decision statistics, a signal search enhanced RMT estimator is proposed to estimate the number of signals via sequentially testing whether the decision statistics is above a threshold to determine whether an eigenvalue is from a signal or noise; (b) Secondly, the analysis shows that this signal search enhanced RMT estimator has better detection performance than the RMT estimator for weak signals, but has larger over-estimation probability than the RMT estimator for the cases when there is no signal or when all signals are strong signals. In order to overcome the individual drawbacks of these two RMT-based estimators, we further propose a signal-noise search RMT estimator whose decision criterion can be adaptively modified to avoid both the large over-estimation probability of the signal search enhanced RMT estimator and the large down-estimation probability of the RMT estimator. Finally, simulation results show that the proposed signal-noise search RMT estimator significantly outperforms the existing estimators including the RMT estimator, the classic AIC and MDL estimators, and the modified AIC estimator.
Submission history
From: Huiyue Yi [view email][v1] Mon, 19 May 2014 13:18:47 UTC (227 KB)
[v2] Fri, 27 Jun 2014 11:18:28 UTC (195 KB)
[v3] Wed, 13 Aug 2014 13:19:06 UTC (207 KB)
[v4] Wed, 20 Aug 2014 12:29:16 UTC (203 KB)
[v5] Mon, 6 Oct 2014 08:34:21 UTC (202 KB)
[v6] Thu, 6 Nov 2014 14:42:54 UTC (208 KB)
[v7] Wed, 7 Jan 2015 13:20:51 UTC (208 KB)
[v8] Wed, 4 Mar 2015 13:22:46 UTC (219 KB)
[v9] Thu, 19 Mar 2015 14:16:23 UTC (228 KB)
[v10] Mon, 30 Mar 2015 14:02:26 UTC (237 KB)
[v11] Wed, 22 Apr 2015 11:38:17 UTC (247 KB)
[v12] Mon, 31 Aug 2015 12:24:00 UTC (285 KB)
[v13] Thu, 10 Sep 2015 14:41:24 UTC (279 KB)
[v14] Mon, 26 Oct 2015 11:51:49 UTC (289 KB)
[v15] Tue, 17 Nov 2015 13:25:27 UTC (296 KB)
[v16] Mon, 1 Feb 2016 14:07:46 UTC (292 KB)
[v17] Thu, 16 Jun 2016 15:10:52 UTC (399 KB)
[v18] Thu, 6 Oct 2016 15:05:27 UTC (428 KB)
[v19] Thu, 2 Nov 2017 14:31:06 UTC (470 KB)
[v20] Thu, 28 Jun 2018 12:56:41 UTC (478 KB)
[v21] Sat, 13 Apr 2019 16:07:50 UTC (413 KB)
[v22] Tue, 26 Nov 2019 15:02:25 UTC (443 KB)
Current browse context:
cs.IT
References & Citations
Bibliographic and Citation Tools
Bibliographic Explorer (What is the Explorer?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)
Code, Data and Media Associated with this Article
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Papers with Code (What is Papers with Code?)
ScienceCast (What is ScienceCast?)
Demos
Recommenders and Search Tools
Influence Flower (What are Influence Flowers?)
Connected Papers (What is Connected Papers?)
CORE Recommender (What is CORE?)
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.