| Indoor Positioning System Using Multiple Services Set Identifiers-Based Fingerprints<ch> Version 0 |
|
| 👤 Author: by as17744 2018-01-07 00:30:26 |
Tracking people and localizing objects within indoor environments have become a necessity and thus motivated many researchers to tackle this challenge. Indoor positioning systems (IPSs) have been proposed using several technologies and efforts are honing down on mitigating the positioning error. The Received Signal Strength (RSS)-fingerprint based IPS has been recognized as one of the possible most promising techniques. In this paper, we propose a new technique which uses the general framework that is based on fingerprinting of multiple service set identifiers (SSIDs) configured on the same access point. A voting scenario is also proposed as a tool to enhance IPS performance. We implemented the system inside the College of
Engineering and Applied Sciences (CEAS) at Western Michigan University, and compared its performance with some of the conventional techniques such as K-Nearest Neighbors (KNN) and support vector machine (SVM). The results also demonstrate the effect of redundancy when using a random selection technique of access points and its impact on IPS performance.
All large buildings such as malls, hospitals, airports, schools, and museums have hundreds of Wi-Fi access points which are installed to provide WLAN infrastructure service. Wi-Fi signals based positioning techniques exploit the existing WLANs devices to provide accurate indoor positioning without any additional hardware. Currently, Global Positioning System (GPS) is the most widely-used location-sensing technique. However, GPS signals; which are transmitted from satellites; require line of sight (LOS) to work properly inside the building. The Wi-Fi based fingerprinting technique is a simple method to deploy inside the buildings compared to other techniques such as angle of arrival (AOA) and time difference of arrival (TDOA). Instead of relying on these systems which require additional hardware to estimate the location, fingerprinting-based location systems use RSS features to compute mobile system (MS) position without any additional infrastructure. The number of APs, time and spatial variance of RSS, and time-varying RSS affect the performance of IPS greatly. Many researches try to figure out the relationship between these measures and the performance of IPS with different hypothesizes.
Daisuke Taniuchi et al. proposed ensemble learning technique by composing multiple weak estimators to improve the accuracy of IPS. In order to reduce the dimensionality of finger prints (FPs), each weak estimator used a random set of APs to estimate the user position. Furthermore, weak estimators have different weights based on their usefulness. The results exhibit a great performance comparing with the existing methods.
Ayah Abusara et al. used the modified fast orthogonal search (mFOS) algorithm to reduce the FPs cardinality. Two considerations were taken, they are: the amount of achieved reduction and the positioning accuracy. The results show that the mFOS is better than the original FOS algorithm.
Mora-Becerra introduced a multiple communication channel RSS as a frequency diversity method to mitigate multipath propagation effects. Three methods for RSS fingerprinting have been used and implemented on smart watch to improve the accuracy of IPS; they are: KNN, Neural Networks model, and Particle Filter. The results show that the performance of the particle filter is the best as compared to other methods.
Zhou et al. utilized fuzzy C-mean off-line clustering to mitigate the online computational cost. The system is evaluated on the tested of underground parking. The results exhibit a slight enhancement in the accuracy of the proposed system comparing with traditional methods.
Feng et al. proposed Compressive Sensing (CS) technique based IPS. The proposed algorithm consists of two stages: coarse positioning, which is achieved by affinity propagation clustering, and fine positioning, which used CS theory to recover user position. In order to apply the CS theory; i.e. satisfy the sparsity and incoherence conditions; certain scenarios have been used to select APs; they are: strongest APs and Fisher criterion. Experimental results show that CS theory is very effective with IPS due to the sparsity of the positioning problem.
Generally, the aforementioned research works overlook the functionality of the access points, that is, they assume as if each access point is transmitting one signal. In this work, we propose a new technique through changing the setup of access points itself and designing a framework that allows for multiple signals at the same time from each access point by setting up multiple SSIDs. The proposed framework shows a remarkable improvement in the IPS accuracy performance.
The paper is organized as follow, fingerprinting-based IPS is reviewed through section II. In section III, we describe the configuration of access points to work with multiple SSIDs and discuss the statistical features of multiple SSID signals as well. Section IV discusses the proposed system based on fingerprinting for multiple SSIDs and voting technique which is used to improve the system performance. Section V reviews the procedure of measurements. Finally, the simulation results are illustrated in section VI. The conclusion and future work is shown in section VII.
> Computer and Education