Tools
Tools
Pandya, BH, 2023. A distributed architecture for fuzzy logic systems and its application in human activity recognition. PhD, Nottingham Trent University.
Preview |
Text
Bhavesh_Pandya_2024.pdf - Published version Download (33MB) | Preview |
Abstract
Fuzzy Logic Systems (FLS) have the full potential in handling imprecise and uncertain data due to the inherent advantages of the Fuzzy Inference System (FIS). Traditionally, fuzzy logic systems are linked to specific hardware or software systems. The literature review reveals that dispersed and distributed architectures of FLS are in high demand due to their capability to handle the complexities of fuzzy logic computations. However, the absence of best practices and standard methodologies prevents widespread adoption. As a result, some specific requirements, such as web communications and Service-Oriented Architecture (SOA), which can be found in many modern systems, are rarely adapted for FLSs. Sharing FLSs accessibility as web services (called Fuzzy-as-a-Service alias FaaS), in which the service is developed independently from a specific client platform, allows for autonomy, openness, load balancing, efficient resource allocation and eventually cost-effective, particularly for computationally intense FLSs.
The proposed novel architectural solution (FaaS) is a web-based service that distributes the main services for FLS on more than one client and servers nodes that can reach multiple users. By extending the IEEE-1855 (2016) standard in terms of system definition and data exchange, this research offers a standard solution for building FaaS as a novel method of implementing fuzzy logic systems by means of a cloud-based collecting, processing, and examining data over the web. Recent advances in standardising Fuzzy Mark-up Language (IEEE 1855-2016) and its associated software libraries (such as JFML and Simpful) have made this achievable. Two different cloud service providers and software libraries (Amazon Web Services using JFML as a java-based library and Azure Web Services using Simpful as a python-based library) are exploited to realise the FaaS on the cloud.
As a case study to establish the efficacy of the proposed FaaS, Human Activity Recognition (HAR) that plays a pivotal role in monitoring the health status of the Persons Under Observation (PUO)has been taken under consideration. In order to monitor the data related to HAR and physiological data, which are imprecise and uncertain in nature, various previous researchers have developed a good number of machine learning tools. However, such monitoring systems suffer from certain limitations due to the nature and amount of data being analysed.
A number of experiments are carried out in order to showcase and evaluate FaaS performance in different HAR scenarios. The first scenario has been a real-time walking/running detection. Secondly, a fall detection system via FaaS is designed based on IEEE 1855-2016 and JFML. In view, the pandemic caused due to COVID-19, the third application dealt with developing a system to determine the health status of individuals by remotely monitoring their Oxygen saturation and heartbeat rate using wearable sensors. Finally, a performance comparison between a stand-alone fuzzy system and a FaaS solution for fall detection is performed on two different cloud services, namely AWS and Azure. Research findings exhibit that while the proposed algorithm can keep the same accuracy as a stand-alone fuzzy system (90%), it can significantly improve the processing time, e.g., reducing the processing time for 10K data samples from 179 to 45 seconds (78% improvement).
Towards the end of this PhD project, the new IEEE 1855 extension is taken as a proposal into the consideration of the IEEE standards committee and is currently in the process of final approval in 2023.
| Item Type: | Thesis |
|---|---|
| Creators: | Pandya, B.H. |
| Contributors: | Name Role NTU ID ORCID |
| Date: | March 2023 |
| Divisions: | Schools > School of Science and Technology |
| Record created by: | Laura Ward |
| Date Added: | 12 Aug 2024 11:05 |
| Last Modified: | 12 Aug 2024 11:05 |
| URI: | https://irep.ntu.ac.uk/id/eprint/51969 |
https://orcid.org/0000-0001-7737-1393Actions (login required)
 |
Edit View |
Statistics
Views
Views per month over past year
Downloads
Downloads per month over past year
