Date of Award

2-5-2024

Document Type

Thesis

School

School of Computing

Programme

Ph.D.-Doctoral of Philosophy

First Advisor

Dr.V.S.Shankar Sriram

Keywords

Computational Offloading, Mobile Cloud, Fog Computing, Trust Node Offloading, Service Composition, Cache Management

Abstract

Nowadays, mobile applications have drawn a lot of attention as they bring computational and storage resources close to consumers globally through high-speed networks. Applications such as the medical microscope, 2D barcode reader, environmental sensor(s), mobile security and authenticator, vehicle remote controller, and IoT-based synchronizer come pre- applications are resource-intensive, as it performs computation(s) by utilizing diverse services like location, app-tracking, networking, camera, calendar, contacts, Bluetooth, etc. for each user activity. Parallel execution of such intense services might utilize the utmost memory and CPU of the mobile device which in turn degrades the overall Quality of Experience (QoE). This ultimately ends up creating functional discrepancies during complex application execution in such mobile devices with limited computational resources. To overcome such challenges identified in mobile application user experience, intense computational processes that demand diverse application services can be offloaded to high-performance cloud cluster(s) and enhance the overall QoE.

According to the report from International Data Corporation (IDC), mobile devices are expected to generate 175 Zettabytes of data by 2025. Thereby, processing such voluminous data is both inevitable and a laborious task. Moreover, when information is accessed from a central data center, it takes longer to reach the end user. This access delay can cause a significant decline in the QoE of mobile users. Therefore, to achieve a minimal response time, caching techniques can be imposed to minimize the content delivery delay experienced by mobile users. For addressing the above-mentioned issues, this thesis aims develop adaptive algorithms for effectively utilizing the remote resources thereby improving the overall QoE of mobile users The significant contributions of this thesis are as follows:

1. Offloading intensive modules to optimum multi-site resource-rich server(s) is considered to be the conventional method for reducing the energy utilization required to execute intensive application modules on mobile devices. In order to identify highly configured nodes from a large number of available heterogeneous natured cloud nodes and to decrease the computational overhead of mobile

applications, a fuzzy logic based node classification framework is proposed. The proposed framework incorporates the Simplified Swarm Optimization technique for task integration and decomposition to lower the weighted total cost of intensive applications. The proposed framework is validated based on the following metrics such as least weighted total cost, processing time, and energy usage incurred during execution.

2. The standard protocol for offloading delay-sensitive applications to an unknown fog environment is to compute and validate each highly configured fog node(s) with a trust parameter. Hence, this work proposes a beta distribution-based algorithm for computing a trust score for each fog node. In addition to trust computation, the identified fog nodes are even made to incorporate load balancing. By imposing

parallelism in a fog environment, services are provided with faster response time. The result is validated by the latency and execution time.

3. Distributed services have to be combined and provided as a single output to the service requester. Fog node capacity directly influences the number of sub-services it can host. It is assumed that each fog device can support a single sub-service. To fully satisfy the other quality requirements of the mobile user, the system should logically identify a group of reliable fog services. The functionality of several fog services may be the same, but their QoS vary. In order to achieve the best outcomes, an algorithm is proposed in this thesis for the effective selection of the appropriate fog services for each subtask. The service composition is first modelled as a Multi-Dimensional Multi-Choice Knapsack Problem (MMKP), which is then solved using a Teaching Learning based Optimization method to identify the most reliable fog services. The proposed work is validated in terms of selecting highly trusted services with satisfied user preferences.

4. Caching popular material in a nearby fog node to reduce access time delay is the conventional method for mobile users to obtain popular content from remote servers. In order to discover popular content and an effective cache node to retain it, this work provides a framework that results in faster access time. A collaborative filtering-based recommendation system is employed to determine the popular content that should be placed. An overlay network is built over the physical fog network to identify the caching nodes in the fog environment using unique graph theory-based notions like semigraph and dominating set. Selected popular content is located in the identified nearby fog nodes to the mobile user, and it is accessible to mobile users, resulting in the reduction of access time and improvement in the QoE of mobile users. The proposed work is validated in terms of throughput and latency when accessing remote content.

The algorithms proposed in this thesis are implemented and compared with the state-ofthe- art algorithms using the generated synthetic dataset. The proposed work's energy consumption, overall execution time, weighted total cost, end-to-end delay, and throughput are examined for various computationally-intensive applications. The outcomes demonstrate that the proposed algorithms are capable of executing intensive, trustworthy mobile applications with reduced energy consumption and accessing remote content with reduced latency, which in turn improves the QoE of mobile users.

Download

Share

COinS