Dynamic Offloading Technique for Latency Sensitive Iot Applications Using Fog Computing

Title

Dynamic Offloading Technique for Latency Sensitive Iot Applications Using Fog Computing

Creator

Priya, Thomas

Contributor

Deepa, V Jose.

Description

The Internet of Things (IoT) has evolved as one of the most popular technological newlineinnovations that offers processing power to different types of entities connected to it. IoT has made traditional applications smarter and easier to use. IoT offers reliable service to different sectors such as healthcare, industrial control, agriculture, autonomous vehicles, traffic management etc. IoT nodes are generally energy-constrained and hence depend on cloud platforms for storage and analytics of generated data. The cloud provides required services for the newlineconnected applications based on pay per use policy. But cloud datacenter being at remote location fails to accommodate the time requirements of delay-sensitive IoT newlineapplications. Edge/fog computing was designed to address the demands of timesensitive IoT applications. The IoT-Fog-Cloud architecture reduces the delay and response time incurred by the IoT-Cloud model. The fog layer in the three-tier architecture is distributed in nature. Hence the latency depends on how well the underlying offloading algorithms distribute the tasks among available fog nodes. Different offloading policies are mentioned in the literature to address this issue. This work initially tries to solve the offloading problem using one of the novel newlineoffloading approaches Flamingo Search Algorithm (FSA). Later, the results obtained from FSA are fine-tuned using another metaheuristic algorithm, the Honey Badger Algorithm (HBA). Finally, both FSA and HBA are hybridized to generate the HB-FS algorithm which effectively solves the task offloading problem. The performance evaluation of the proposed approach is done with different existing metaheuristic algorithms and the evaluations show that the newlineproposed work outperforms the existing algorithms in terms of latency, average newlineresponse time and execution time. The methodology also offers a lesser degree of newlineimbalance and standard deviation than the compared approaches.

Source

Author's Submission

Date

2024-01-01

Publisher

Christ(Deemed to be University)

Subject

Computer Science

Rights

Open Access

Relation

61000426

Format

PDF

Language

English

Type

PhD

Identifier

http://hdl.handle.net/10603/618217