DDLO

Distributed Deep Learning-based Offloading for Mobile Edge Computing Networks

Python code to reproduce our works on Deep Learning-based Offloading for Mobile-Edge Computing Networks [1], where multiple parallel Deep Neural Networks (DNNs) are used to efficiently generate near-optimal binary offloading decisions. This project includes:

• memory.py: the DNN structure for DDLO, inclduing training structure and test structure

• data: all data are stored in this subdirectory, includes:

  • MUMT_data_3X3.mat: training and testing data sets, where 3X3 means that the user number is 3, and each has 3 tasks.

• main.py: run this file, inclduing setting system parameters

• MUMT.py: compute system utility Q, provided with the size of all tasks and offloading decision

Cite this work

1. Liang Huang, Xu Feng, Anqi Feng, Yupin Huang, and Li Ping Qian, “Distributed Deep Learning-based Offloading for Mobile Edge Computing Networks,” in Mobile Networks and Applications, 2018, DOI:10.1007/s11036-018-1177-x.

Required packages

• Tensorflow 1.x.

  • If you want to run DDLO on Tensorflow 2 or PyTorh, please find a clue from the memoryTF2.py or memoryPyTorch.py file in the DROO project

• numpy

• scipy

How the code works

run the file, main.py

Contacts

If you have any questions related to the codes, please feel free to contact Liang Huang (lianghuang AT zjut.edu.cn)

Related works

For deep reinforcement learning-based offloading for a simple MEC structure, please refer to our recent DROO project with much cleaner and well-commented source codes:

2. L. Huang, S. Bi, and Y. J. Zhang, “Deep reinforcement learning for online computation offloading in wireless powered mobile-edge computing networks,” IEEE Trans. Mobile Compt., vol. 19, no. 11, pp. 2581-2593, November 2020.