Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Infrastructure-less Localization from Indoor Environmental Sounds Based on Spectral Decomposition and Spatial Likelihood Model

Satoki Ogiso, Yoshiaki Bando, Takeshi Kurata, Takashi Okuma

TL;DR

A microphone localization method using supervised spectral decomposition and spatial likelihood to solve problems of low signal-to-noise-ratio (SNR) condition and non-uniqueness of sound over the coverage area was proposed.

Abstract

Human and/or asset tracking using an attached sensor units helps understand their activities. Most common indoor localization methods for human tracking technologies require expensive infrastructures, deployment and maintenance. To overcome this problem, environmental sounds have been used for infrastructure-free localization. While they achieve room-level classification, they suffer from two problems: low signal-to-noise-ratio (SNR) condition and non-uniqueness of sound over the coverage area. A microphone localization method was proposed using supervised spectral decomposition and spatial likelihood to solve these problems. The proposed method was evaluated with actual recordings in an experimental room with a size of 12 x 30 m. The results showed that the proposed method with supervised NMF was robust under low-SNR condition compared to a simple feature (mel frequency cepstrum coefficient: MFCC). Additionally, the proposed method could be easily integrated with prior distribution, which is available from other Bayesian localizations. The proposed method can be used to evaluate the spatial likelihood from environmental sounds.

Infrastructure-less Localization from Indoor Environmental Sounds Based on Spectral Decomposition and Spatial Likelihood Model

TL;DR

A microphone localization method using supervised spectral decomposition and spatial likelihood to solve problems of low signal-to-noise-ratio (SNR) condition and non-uniqueness of sound over the coverage area was proposed.

Abstract

Human and/or asset tracking using an attached sensor units helps understand their activities. Most common indoor localization methods for human tracking technologies require expensive infrastructures, deployment and maintenance. To overcome this problem, environmental sounds have been used for infrastructure-free localization. While they achieve room-level classification, they suffer from two problems: low signal-to-noise-ratio (SNR) condition and non-uniqueness of sound over the coverage area. A microphone localization method was proposed using supervised spectral decomposition and spatial likelihood to solve these problems. The proposed method was evaluated with actual recordings in an experimental room with a size of 12 x 30 m. The results showed that the proposed method with supervised NMF was robust under low-SNR condition compared to a simple feature (mel frequency cepstrum coefficient: MFCC). Additionally, the proposed method could be easily integrated with prior distribution, which is available from other Bayesian localizations. The proposed method can be used to evaluate the spatial likelihood from environmental sounds.
Paper Structure (14 sections, 12 equations, 6 figures, 1 table)

This paper contains 14 sections, 12 equations, 6 figures, 1 table.

Table of Contents

  1. Introduction
  2. Related Work
  3. Infrastructure-assisted methods
  4. Infrastructure-less methods
  5. Environmental sound-based methods
  6. Proposed Method
  7. Feature extraction with noise-aware supervised NMF
  8. Spatial likelihood-based localization with GP regression
  9. Experimental Evaluation
  10. Experimental setting
  11. Sound source separation and Gaussian regression
  12. Localization performance
  13. Integration with other localization method
  14. Conclusions

Figures (6)

  • Figure 1: Comparison between naïve direct regression-based localization and our spatial likelihood-based localization.
  • Figure 2: Architecture of the proposed localization method.
  • Figure 3: Target area for localization. (a) lists landmark sound sources used in our experiment, (b) shows the top view of the area, and (c) shows a panoramic view of the area.
  • Figure 4: Examples of GP regression models with separated source RMS for each location. Figures correspond to the sound sources described in Fig. \ref{['experiment_map']}: (a) a projector, (b) PCs, (c) air conditioners, (d) server room, and (e) ventilation fans, respectively.
  • Figure 5: Localization performance. (a) Localization with proposed spatial likelihood in the experiment, (b) effect of SNR on localization error (CEP).
  • ...and 1 more figures