An effective technique to schedule priority aware tasks to offload data on edge and cloud servers. (April 2023)

• Record Type: Journal Article
• Authors: Bali, Malvinder Singh; Gupta, Kamali; Gupta, Deepali; Srivastava, Gautam; Juneja, Sapna; Nauman, Ali
• Is Part Of: Measurement. Volume 26(2023)
• Publication Date: 2023-04
• Subjects: Queue model -- Data offloading -- Edge computing -- Scheduling; Detectors -- Periodicals; Measurement -- Periodicals; 530.7
• DOI: 10.1016/j.measen.2023.100670
• Languages: English
• ISSNs: 2665-9174

Abstract

Abstract: Recent advancements in the Internet of Things (IoT) have enhanced the quality of life globally. Billions of devices are brought under the ambit of IoT to make them smarter. IoT-based applications are generating voluminous data and managing this widespread amount of data in real-time through Cloud Technology, which offers high computational and storage facilities. However, sending all data to the cloud can bring serious concerns for applications, which are critical and require instant action without any delay. Edge computing has recently emerged as an effective technology to handle the instant processing of tasks of IoT-based applications locally. Additionally, an important concern in IoT networks is response to emergency tasks on time to increase the performance of large-scale IoT systems. As such, scheduling of tasks becomes vital, where emergency and non-emergency tasks can be prioritized to offload data to the nearby edge and cloud servers respectively and enhance Quality of Service (QoS). The execution order of tasks and allocating resources for computation to avoid delays are two of the most important factors that must be addressed during task scheduling in Edge Computing. With the aforementioned issues, we design a Priority aware Task Scheduling (PaTS) algorithm for sensor networks to schedule priority aware tasks to offload data on edge and cloud servers. The problem is formulated as a multi-objective function and the efficiency of the proposed algorithm is evaluated using the Bio-inspired NSGA-2 technique. The overall improvement for average queue delay, computation time, and energy obtained for 200 tasks is 17.2%, 7.08% and 11.4%, respectively. The results obtained show significant improvement when compared with the benchmark algorithms demonstrating the effectiveness of the proposed solution. Similarly, comparative results for tasks when increased from 200 to 1000 tasks also shows subsequent improvements.