TY - JOUR
AU1 - Xia, HaoJi
AU2 - Cao, Jinyan
AB - Internet of Things (IoT) is the connection of devices to the Internet, where devices such as cameras, sensors, light bulbs, and smoke alarms are able to communicate with each other and their users. IoT allows devices to assist daily routines, such as cars, which can be synced with calendars for appointment or meeting tracking to plan the best routes. The IoT brings new potential to humans’ everyday life activities by enabling applications with stringent communication demands, including Mobile IoT (MIoT) applications. However, since most apps in MIoT have quality of service demands, running them on mobile devices with limited resources raises several difficulties. Offloading the part of the apps’ services to nearby substitute devices is one way to handle increasingly demanding applications. Also, with the increasing demands of big data applications in the IoT, various applications have emerged to provide mobile services for IoT devices. However, due to the increasing number of IoT devices, transmission delay, energy consumption, and heavy load have become new challenges for offloading tasks. Therefore, this article introduces an energy-aware task offloading method in MIoT to determine the optimal offloading strategy. Due to the NP-hard nature of this problem and to enhance the efficiency of the state-of-the-art techniques, this paper used a neural network-based evolutionary optimization algorithm. The usage of multi-layer perceptron neural networks with an error back-propagation training method is one of the effective solutions in this paper for enhancing load-discharge performance. The potential for trapping at local minimum points is one of the flaws of the error back-propagation training process. Meta-heuristic algorithms are able to get out of the trap of local minima and find global minima. Thus, in this paper, the combination of the locust search optimization algorithm with the error back-propagation training algorithm has been used to eliminate the weakness of the neural network training algorithm in improving the load discharge performance. Results from simulations demonstrated that the suggested strategy outperforms alternatives in terms of energy use, cost, and task completion time. The findings of the current publication are significant for academics and offer insights into potential areas of future research in this discipline.
TI - An energy-aware technique for solving the offloading problem of mobile internet of things using an enhanced neural network-based optimization algorithm
JF - Cluster Computing
DO - 10.1007/s10586-024-04986-3
DA - 2025-08-01
UR - https://www.deepdyve.com/lp/springer-journals/an-energy-aware-technique-for-solving-the-offloading-problem-of-mobile-0sPn0rOVKw
VL - 28
IS - 4
DP - DeepDyve
ER -