Dynamic Task Offloading Algorithm for optimising IoT network quality of service in the Mobile-Fog-Cloud System

The application of the Internet of Things (IoT) is increasing to almost all aspects of human endevour. IoT aims at getting everything (wearable, smart cameras, home appliances, vehicles, and hospital equipment) connected to the Internet. These devices continuously generate a massive amount of data on the network. The IoT (mobile) devices that generate these data are limited in terms of processing capacity and energy, because of these limitations of the mobile devices, they cannot process all generated tasks in the IoT application environment. Cloud computing and Fog computing are introduced to assist mobile devices to respond to environmental demand. Most times, the approach of relying on cloud infrastructure for IoT application analysis may be inefficient in terms of the limited battery life of the mobile devices, resource allocation algorithm delay, and computational offloading processes that sometimes increases the response time. Furthermore, many IoT applications are time sensitive such as health monitory systems, augmented reality services, agriculture, pest control, online natural language processing, smart home applications, smart cities, safe driving, waste management, emergency response systems, and traffic control systems. Therefore, offloading a massive amount of data from mobile devices to the fog or cloud introduces another problem of delay in choosing the optimal resources for processing the tasks resulting in incurring delay by the resource allocation algorithms. This problem sometimes makes the application of IoT inefficient in sensitive cases that require low response time. However, the problem of offloading large data sizes for analysis at the remote processing layer (fog or cloud) and efficient scheduling of tasks and resources is addressed in this study. Therefore, an Energy-Efficient Canonical Polyadic Decomposition (EECPD) scheduling algorithm to minimize the mobile device energy consumption in the system is proposed. Secondly, a hybrid Genetic Algorithm and Enhanced Inertia Weight Particle Swarm Optimization (GAEIWPSO) algorithm for optimal resource allocation to minimize the delay is proposed. Finally, a Dynamic Task Offloading Algorithm (DTOA) based on rank accuracy estimation model to efficiently schedule tasks and resources in the Mobile-Fog-Cloud system is proposed. The proposed algorithms achieved minimized data reduction ratio, number of deployed tasks, energy consumption, delay; and in addition, increased throughput, and better resource utilization, which in all enhanced the overall network quality of service. The attribute reduction technique is implemented with Matlab. The EECPD and GAEIWPSO algorithms are implemented with Python and Networkx simulators while DTOA algorithm is implemented with iFogSim to demonstrate the efficiency of the proposed scheme. The results proved that the proposed scheme performed better than the benchmark results.

Text 113155.pdf Download (1MB)

Actions (login required)

View Item