A Knowledge Distillation-Based Lightweight Task Offloading Algorithm for Mobile Edge Computing

Xinyu Wang; Yimai Wang; Can Ti; Bo Cui

doi:10.6919/ICJE.202604_12(4).0042

Authors

Xinyu Wang
Yimai Wang
Can Ti
Bo Cui

DOI:

https://doi.org/10.6919/ICJE.202604_12(4).0042

Keywords:

Mobile Edge Computing; Task Offloading; Deep Reinforcement Learning; Knowledge Distillation; Lightweight Models; D3QN.

Abstract

To address the issues of complex models, high inference latency, and difficulties in deploying traditional deep reinforcement learning offloading strategies on resource-constrained devices in mobile edge computing, this paper proposes a lightweight D3QN computation offloading method based on knowledge distillation. This method constructs a teacher-student network architecture, transferring the decision-making capabilities of the teacher model to a lightweight student network via offline knowledge distillation. Concurrently, an adaptive exploration strategy incorporating heuristic rules is designed to accelerate convergence, whilst LSTM temporal awareness and a dynamic energy consumption normalisation mechanism are introduced to optimise QoE. Experimental results demonstrate that, at a compression ratio of 0.4, the number of parameters in the student model is reduced by 77.61%, inference latency is reduced by 26.32%, and QoE reaches 829.07, approaching the performance of the teacher model, thereby providing an efficient and feasible lightweight solution for edge intelligence offloading.

Downloads

Download data is not yet available.

References

[1] M. Patel et al., “Mobile-edge computing introductory technical white paper,” White Paper, Mobile-Edge Comput. (MEC), vol. 29, pp. 854-864, Sep. 2014.

[2] S. Lu, S. Liu, Y. Zhu, W. Liang, K. Li, and Y. Lu, “A DRL-based decentralized computation offloading method: An example of an intelligent manufacturing scenario,” IEEE Trans. Cogn. Commun. Netw., vol. 9, no. 4, pp. 1042-1055, Aug. 2023.

[3] L. Ale, S. A. King, N. Zhang, A. R. Sattar, and J. Skandaramayan, “D3PG: Dirichlet DDPG for task partitioning and offloading with constrained hybrid action space in mobile-edge computing,” IEEE Internet Things J., vol. 9, no. 19, pp. 19260-19272, Oct. 2022.

[4] L. Ale, N. Zhang, X. Fang, X. Chen, S. Wu, and L. Li, “Delay-aware and energy-efficient computation offloading in mobile-edge computing using deep reinforcement learning,” IEEE Trans. Cogn. Commun. Netw., vol. 7, no. 3, pp. 881-892, Sep. 2021.

[5] J. Li, Q. Jiang, V. C. M. Leung, Z. Ma, and K. K. Abrokwa, “Deep-reinforcement-learning-based joint optimization of task migration and resource allocation for mobile-edge computing,” IEEE Internet Things J., vol. 12, no. 13, pp. 24428-24441, 1 Jul. 2025.

[6] X. Qiu, W. Zhang, W. Chen, and Z. Zheng, “Distributed and collective deep reinforcement learning for computation offloading: A practical perspective,” IEEE Trans. Parallel Distrib. Syst., vol. 32, no. 5, pp. 1085-1101, May 2021.

[7] N. Zhao, Z. Ye, Y. Pei, Y.-C. Liang, and D. Niyato, “Multi-agent deep reinforcement learning for task offloading in UAV-assisted mobile edge computing,” IEEE Trans. Wireless Commun., vol. 21, no. 9, pp. 6949-6962, Sep. 2022.

[8] L. Ale, N. Zhang, X. Fang, X. Chen, S. Wu, and L. Li, “Delay-aware and energy-efficient computation offloading in mobile-edge computing using deep reinforcement learning,” IEEE Trans. Cogn. Commun. Netw., vol. 7, no. 3, pp. 881-892, Sep. 2021.

[9] S. Lu, S. Liu, Y. Zhu, W. Liang, K. Li, and Y. Lu, “A DRL-based decentralized computation offloading method: An example of an intelligent manufacturing scenario,” IEEE Trans. Cogn. Commun. Netw., vol. 9, no. 4, pp. 1042-1055, Aug. 2023.

[10] L. Ale, S. A. King, N. Zhang, A. R. Sattar, and J. Skandaramayan, “D3PG: Dirichlet DDPG for task partitioning and offloading with constrained hybrid action space in mobile-edge computing,” IEEE Internet Things J., vol. 9, no. 19, pp. 19260-19272, Oct. 2022.

[11] J. Li, Q. Jiang, V. C. M. Leung, Z. Ma, and K. K. Abrokwa, “Deep-reinforcement-learning-based joint optimization of task migration and resource allocation for mobile-edge computing,” IEEE Internet Things J., vol. 12, no. 13, pp. 24428-24441, 1 Jul. 2025.

[12] S. Chouikhi, M. Essgehir, and L. Merghem-Boulahia, “Energy-efficient computation offloading based on multiagent deep reinforcement learning for industrial Internet of Things systems,” IEEE Internet Things J., vol. 11, no. 7, pp. 12228-12239, Apr. 2024.

[13] N. Zhao, Z. Ye, Y. Pei, Y.-C. Liang, and D. Niyato, “Multi-agent deep reinforcement learning for task offloading in UAV-assisted mobile edge computing,” IEEE Trans. Wireless Commun., vol. 21, no. 9, pp. 6949-6962, Sep. 2022.

[14] X. Qiu, W. Zhang, W. Chen, and Z. Zheng, “Distributed and collective deep reinforcement learning for computation offloading: A practical perspective,” IEEE Trans. Parallel Distrib. Syst., vol. 32, no. 5, pp. 1085-1101, May 2021.

[15] J. Hao, L. Wang, M. Odiathevar, W. K. G. Seah, G. Xu, B. Huang, and Y. Gao, “Prune-based deep reinforcement learning offloading algorithm for mobile edge computing,” IEEE Trans. Cogn. Commun. Netw., vol. 12, pp. 2876-2889, 2026.