Final published version
Licence: CC BY: Creative Commons Attribution 4.0 International License
Research output: Contribution to Journal/Magazine › Journal article › peer-review
Research output: Contribution to Journal/Magazine › Journal article › peer-review
}
TY - JOUR
T1 - PHyPO: Priority-based Hybrid task Partitioning and Offloading in mobile computing using automated machine learning
AU - Bano, Shehr
AU - Abbas, Ghulam
AU - Bilal, Muhammad
AU - Abbas, Ziaul Haq
AU - Ali, Zaiwar
AU - Waqas, Muhammad
A2 - Dao, Nhu-Ngoc
PY - 2024/12/12
Y1 - 2024/12/12
N2 - With the increasing demand for mobile computing, the requirement for intelligent resource management has also increased. Cloud computing lessens the energy consumption of user equipment, but it increases the latency of the system. Whereas edge computing reduces the latency along with the energy consumption, it has limited resources and cannot process bigger tasks. To resolve these issues, a Priority-based Hybrid task Partitioning and Offloading (PHyPO) scheme is introduced in this paper, which prioritizes the tasks with high time sensitivity and offloads them intelligently. It also calculates the optimal number of partitions a task can be divided into. The utility of resources is maximized along with increasing the processing capability of the model by using a hybrid architecture, consisting of mobile devices, edge servers, and cloud servers. Automated machine learning is used to identify the optimal classification models, along with tuning their hyper-parameters, which results in adaptive boosting ensemble learning-based models to reduce the time complexity of the system to O(1). The results of the proposed algorithm show a significant improvement over benchmark techniques along with achieving an accuracy of 96.1% for the optimal partitioning model and 94.3% for the optimal offloading model, with both the results being achieved in significantly less or equal time as compared to the benchmark techniques.
AB - With the increasing demand for mobile computing, the requirement for intelligent resource management has also increased. Cloud computing lessens the energy consumption of user equipment, but it increases the latency of the system. Whereas edge computing reduces the latency along with the energy consumption, it has limited resources and cannot process bigger tasks. To resolve these issues, a Priority-based Hybrid task Partitioning and Offloading (PHyPO) scheme is introduced in this paper, which prioritizes the tasks with high time sensitivity and offloads them intelligently. It also calculates the optimal number of partitions a task can be divided into. The utility of resources is maximized along with increasing the processing capability of the model by using a hybrid architecture, consisting of mobile devices, edge servers, and cloud servers. Automated machine learning is used to identify the optimal classification models, along with tuning their hyper-parameters, which results in adaptive boosting ensemble learning-based models to reduce the time complexity of the system to O(1). The results of the proposed algorithm show a significant improvement over benchmark techniques along with achieving an accuracy of 96.1% for the optimal partitioning model and 94.3% for the optimal offloading model, with both the results being achieved in significantly less or equal time as compared to the benchmark techniques.
U2 - 10.1371/journal.pone.0314198
DO - 10.1371/journal.pone.0314198
M3 - Journal article
VL - 19
JO - PLoS One
JF - PLoS One
SN - 1932-6203
IS - 12
M1 - e0314198
ER -