Lightweight indoor localisation and linguistic location authority /
Indoor positioning and navigation unlike outdoor positioning requires different techniques apart from the classical geometric based approached utilizing satellite communications. This is due to the fact that satellite signal reception is poor in indoor environment. Approaches to indoor localization...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
Kuala Lumpur :
Kulliyyah of Information and Communication Technology, International Islamic University Malaysia,
2015
|
| Subjects: | |
| Online Access: | http://studentrepo.iium.edu.my/handle/123456789/5461 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Indoor positioning and navigation unlike outdoor positioning requires different techniques apart from the classical geometric based approached utilizing satellite communications. This is due to the fact that satellite signal reception is poor in indoor environment. Approaches to indoor localization using Received Signal Strengths (RSS) are generally based on signal propagation models or location fingerprinting methods, using different algorithms. All algorithms whether applied on signal propagation models or location fingerprinting can be classified as heavyweight or lightweight algorithms. Heavyweight algorithms generally have better accuracies but require rigorous and complex computations thereby place critical strain on processing power of mobile devices and suffer from location response delay due to the complexity of the computation and extended time requirement. Lightweight algorithms are less complex and do not require extensive time or processing power compare to the heavyweight algorithms, however they perform relatively poorer in accuracy. Lightweight algorithms have been investigated in this thesis for near heavyweight accuracy and sufficiently accurate for indoor environments. The two novel algorithms proposed achieve 95% room level accuracy and a maximum update time of 2 seconds reducing update time considerably. The first one is Fuzzy Weighted Aggregation of Received Signal Strengths of Wi-Fi signals with Compensated Weighted Attenuation Factor (CWAF) in the form of fuzzy weighted signal quality and noise while the second is lightweight localization approach based on the extreme learning algorithm (ELM), a single hidden layer neural network. For every location based system requires the representation of the location in effective and efficient scheme. In order to provide suitable location authority for indoor positioning approaches proposed, this work introduced a perception-based linguistic approach to locations relative to landmarks to extend present location authority with a view to making it more user-friendly. The idea is due to the realisation that people respond to the question “where are you?” naturally in linguistic forms such as “I am close to Lab A” rather than “I am 5m to Lab A” etc., which is what entails in most positioning & navigation devices such as GPS. Therefore, it is argued that positioning and navigation systems should incorporate linguistic description of distances rather than the present quantitative distances, such as 5m to Lab A. Three fuzzy schemes based on α-cut, Gaussian and enhanced interval type-2 (EIA T2) have been proposed. The first two gave above 80% accuracy while the third gave around 85% accuracy, given the subjective validation data elicited from groups of subjects taken from ordinary mobile users, experts and blind subjects. The two sets of algorithms compared favourably with other traditional models such as Bayesian, Decision Tree, and ANFIS Type-1 & Type-2. The EIA shows the best results in terms of accuracy though it requires more processing power due to complexity than that of α-cuts and Gaussian models which are less accurate but more efficient computationally. |
|---|---|
| Physical Description: | xvii, 202 leaves : ill. ; 30cm. |
| Bibliography: | Includes bibliographical references (leaves 165-181). |