Towards Compressive Sampling and Fine-Grained Indoor Fingerprint Positioning

Yongle Chen; Dan Yu

Location based services (LBS) enable various mobile application and improve the human life significantly. GPS has largely solved the problem for outdoor positioning. However, the accuracy and usability of positioning algorithm remains a grand challenge for indoor scenarios. WiFi-based fingerprint positioning has attracted lots of research attentions, for the advantage of leveraging existing indoor WLAN infrastructure. But the positioning accuracy and fingerprint sampling are both challenge for WiFi-based fingerprint positioning. This paper proposed a trade-off method between accurate positioning and tedious sampling. It first utilizes the Gaussian process model to complete compressive sampling instead of the whole-cell sampling, and then proposed an improved K-Means method based on the similarity threshold to increase the performance accuracy. Our experiment shows our improve method can get better accuracy with 2m median error which approximates results of the real raw data. Meanwhile, our Gaussian process model also largely reduces the tedious sampling.

Select Volume / Issues:

• Volume 6 Issue 1, Jan. 2015

Year:

2015

Type of Publication:

Article

Keywords:

Compressive Sampling; Fine-Grained; Fingerprint Positioning

Journal:

IJECCE

Volume:

6

Number:

1

Pages:

158-162

Month:

Jan.-Feb.

ISSN:

2249-071X