DRL-OS: A Deep Reinforcement Learning-Based Offloading Scheduler in Mobile Edge Computing
Hardware bottlenecks can throttle smart device (SD) performance when executing computation-intensive and delay-sensitive applications. Hence, task offloading can be used to transfer computation-intensive tasks to an external server or processor in Mobile Edge Computing. However, in this approach, th...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-11-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1424-8220/22/23/9212 |
| _version_ | 1827642512360079360 |
|---|---|
| author | Ducsun Lim Wooyeob Lee Won-Tae Kim Inwhee Joe |
| author_facet | Ducsun Lim Wooyeob Lee Won-Tae Kim Inwhee Joe |
| author_sort | Ducsun Lim |
| collection | DOAJ |
| description | Hardware bottlenecks can throttle smart device (SD) performance when executing computation-intensive and delay-sensitive applications. Hence, task offloading can be used to transfer computation-intensive tasks to an external server or processor in Mobile Edge Computing. However, in this approach, the offloaded task can be useless when a process is significantly delayed or a deadline has expired. Due to the uncertain task processing via offloading, it is challenging for each SD to determine its offloading decision (whether to local or remote and drop). This study proposes a deep-reinforcement-learning-based offloading scheduler (DRL-OS) that considers the energy balance in selecting the method for performing a task, such as local computing, offloading, or dropping. The proposed DRL-OS is based on the double dueling deep Q-network (D3QN) and selects an appropriate action by learning the task size, deadline, queue, and residual battery charge. The average battery level, drop rate, and average latency of the DRL-OS were measured in simulations to analyze the scheduler performance. The DRL-OS exhibits a lower average battery level (up to 54%) and lower drop rate (up to 42.5%) than existing schemes. The scheduler also achieves a lower average latency of 0.01 to >0.25 s, despite subtle case-wise differences in the average latency. |
| first_indexed | 2024-03-09T17:33:18Z |
| format | Article |
| id | doaj.art-dcd99e070a934efb9fbe16204079b46b |
| institution | Directory Open Access Journal |
| issn | 1424-8220 |
| language | English |
| last_indexed | 2024-03-09T17:33:18Z |
| publishDate | 2022-11-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Sensors |
| spelling | doaj.art-dcd99e070a934efb9fbe16204079b46b2023-11-24T12:10:24ZengMDPI AGSensors1424-82202022-11-012223921210.3390/s22239212DRL-OS: A Deep Reinforcement Learning-Based Offloading Scheduler in Mobile Edge ComputingDucsun Lim0Wooyeob Lee1Won-Tae Kim2Inwhee Joe3The Department of Computer and Software, Hanyang University, 222 Wangsimni-ro, Seoul 04763, Republic of KoreaThe Department of Computer and Software, Hanyang University, 222 Wangsimni-ro, Seoul 04763, Republic of KoreaThe Department of Computer Science and Engineering, Korea University of Technology and Education, Cheonan-si 31253, Republic of KoreaThe Department of Computer and Software, Hanyang University, 222 Wangsimni-ro, Seoul 04763, Republic of KoreaHardware bottlenecks can throttle smart device (SD) performance when executing computation-intensive and delay-sensitive applications. Hence, task offloading can be used to transfer computation-intensive tasks to an external server or processor in Mobile Edge Computing. However, in this approach, the offloaded task can be useless when a process is significantly delayed or a deadline has expired. Due to the uncertain task processing via offloading, it is challenging for each SD to determine its offloading decision (whether to local or remote and drop). This study proposes a deep-reinforcement-learning-based offloading scheduler (DRL-OS) that considers the energy balance in selecting the method for performing a task, such as local computing, offloading, or dropping. The proposed DRL-OS is based on the double dueling deep Q-network (D3QN) and selects an appropriate action by learning the task size, deadline, queue, and residual battery charge. The average battery level, drop rate, and average latency of the DRL-OS were measured in simulations to analyze the scheduler performance. The DRL-OS exhibits a lower average battery level (up to 54%) and lower drop rate (up to 42.5%) than existing schemes. The scheduler also achieves a lower average latency of 0.01 to >0.25 s, despite subtle case-wise differences in the average latency.https://www.mdpi.com/1424-8220/22/23/9212computation offloadingdouble dueling deep Q-networkenergy consumptionmobile edge computing (MEC)resource managementreinforcement learning |
| spellingShingle | Ducsun Lim Wooyeob Lee Won-Tae Kim Inwhee Joe DRL-OS: A Deep Reinforcement Learning-Based Offloading Scheduler in Mobile Edge Computing Sensors computation offloading double dueling deep Q-network energy consumption mobile edge computing (MEC) resource management reinforcement learning |
| title | DRL-OS: A Deep Reinforcement Learning-Based Offloading Scheduler in Mobile Edge Computing |
| title_full | DRL-OS: A Deep Reinforcement Learning-Based Offloading Scheduler in Mobile Edge Computing |
| title_fullStr | DRL-OS: A Deep Reinforcement Learning-Based Offloading Scheduler in Mobile Edge Computing |
| title_full_unstemmed | DRL-OS: A Deep Reinforcement Learning-Based Offloading Scheduler in Mobile Edge Computing |
| title_short | DRL-OS: A Deep Reinforcement Learning-Based Offloading Scheduler in Mobile Edge Computing |
| title_sort | drl os a deep reinforcement learning based offloading scheduler in mobile edge computing |
| topic | computation offloading double dueling deep Q-network energy consumption mobile edge computing (MEC) resource management reinforcement learning |
| url | https://www.mdpi.com/1424-8220/22/23/9212 |
| work_keys_str_mv | AT ducsunlim drlosadeepreinforcementlearningbasedoffloadingschedulerinmobileedgecomputing AT wooyeoblee drlosadeepreinforcementlearningbasedoffloadingschedulerinmobileedgecomputing AT wontaekim drlosadeepreinforcementlearningbasedoffloadingschedulerinmobileedgecomputing AT inwheejoe drlosadeepreinforcementlearningbasedoffloadingschedulerinmobileedgecomputing |