Learning to Continuously Optimize Wireless Resource in Episodically Dynamic Environment

@article{Sun2021LearningTC,
  title={Learning to Continuously Optimize Wireless Resource in Episodically Dynamic Environment},
  author={Haoran Sun and Wenqiang Pu and Minghe Zhu and Xiao Fu and Tsung-Hui Chang and Mingyi Hong},
  journal={ICASSP 2021 - 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
  year={2021},
  pages={4945-4949}
}
  • Haoran Sun, Wenqiang Pu, +3 authors Mingyi Hong
  • Published 16 November 2020
  • Computer Science, Engineering, Mathematics
  • ICASSP 2021 - 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)
There has been a growing interest in developing data-driven, in particular deep neural network (DNN) based methods for modern communication tasks. For a few popular tasks such as power control, beamforming, and MIMO detection, these methods achieve state-of-the-art performance while requiring less computational efforts, less channel state information (CSI), etc. However, it is often challenging for these approaches to learn in a dynamic environment where parameters such as CSIs keep changing… Expand
Learning to Continuously Optimize Wireless Resource in a Dynamic Environment: A Bilevel Optimization Perspective
TLDR
This work proposes to build the notion of continual learning (CL) into wireless system design, so that the learning model can incrementally adapt to the new episodes, without forgetting knowledge learned from the previous episodes. Expand
AI Empowered Resource Management for Future Wireless Networks
TLDR
For K-user interference management problem, it is theoretically show that graph neural networks (GNNs) are superior to multi-layer perceptrons (MLPs), and the performance gap between these two methods grows with √ K. Expand
Adapting to Dynamic LEO-B5G Systems: Meta-Critic Learning Based Efficient Resource Scheduling
TLDR
An enhanced meta-critic learning algorithm (EMCL) is proposed, where a hybrid neural network for parameterization and the Wolpertinger policy for action mapping are designed in EMCL, which demonstrates EMCL’s effectiveness and fast-response capabilities in over-loaded systems and in adapting to dynamic environments compare to previous actor-Critic and meta-learning methods. Expand
Algorithm Unrolling for Massive Access via Deep Neural Network with Theoretical Guarantee
TLDR
This paper proposes a novel algorithm unrolling framework based on the deep neural network to simultaneously achieve low computational complexity and high robustness for solving the JADCE problem, and maps the original iterative shrinkage thresholding algorithm into an unrolled recurrent neural network (RNN). Expand
ML-aided power allocation for Tactical MIMO
TLDR
This work presents a UWMMSE framework for power allocation in (multipleinput multiple-output) MIMO interference networks and illustrates the superiority of this approach in comparison to WMMSE and also analyzes its robustness to changes in channel conditions and network size. Expand
Edge Artificial Intelligence for 6G: Vision, Enabling Technologies, and Applications
  • K. Letaief, Yuanming Shi, Jianmin Lu, Jianhua Lu
  • Computer Science, Engineering
  • IEEE Journal on Selected Areas in Communications
  • 2021
TLDR
The vision for scalable and trustworthy edge AI systems with integrated design of wireless communication strategies and decentralized machine learning models is provided and new design principles of wireless networks, service-driven resource allocation optimization methods, as well as a holistic end-to-end system architecture to support edge AI will be described. Expand

References

SHOWING 1-10 OF 71 REFERENCES
LORM: Learning to Optimize for Resource Management in Wireless Networks With Few Training Samples
TLDR
A transfer learning method via self-imitation in LORM can quickly adapt a pre-trained machine learning model to the new task with only a few additional unlabeled training samples, and achieves comparable performance with the model trained from scratch with sufficient labeled samples. Expand
Multi-Agent Deep Reinforcement Learning for Dynamic Power Allocation in Wireless Networks
TLDR
The proposed algorithm is shown to achieve near-optimal power allocation in real time based on delayed CSI measurements available to the agents and is especially suitable for practical scenarios where the system model is inaccurate and CSI delay is non-negligible. Expand
Transfer Learning and Meta Learning-Based Fast Downlink Beamforming Adaptation
TLDR
This article proposes two offline adaptive algorithms based on deep transfer learning and meta-learning, which are able to achieve fast adaptation with the limited new labelled data when the testing wireless environment changes and an online algorithm to enhance the adaptation capability of the offline meta algorithm in realistic non-stationary environments. Expand
Learning to Optimize: Training Deep Neural Networks for Interference Management
Numerical optimization has played a central role in addressing key signal processing (SP) problems. Highly effective methods have been developed for a large variety of SP applications such asExpand
To Supervise or Not to Supervise: How to Effectively Learn Wireless Interference Management Models?
TLDR
The first set of theoretical results trying to understand different training approaches in learning-based wireless communication system design show that for some special power control problem, the unsupervised learning can perform much worse than its counterpart, because it is more likely to get stuck at some low-quality local solutions. Expand
Spatial Deep Learning for Wireless Scheduling
TLDR
It is shown that it is possible to bypass the channel estimation stage altogether and to use a deep neural network to produce a near optimal schedule based solely on geographic locations of the transmitters and receivers in the network. Expand
Towards Optimal Power Control via Ensembling Deep Neural Networks
TLDR
Simulation results show that for the standard symmetric $K$ -user Gaussian interference channel, the proposed methods can outperform state-of-the-art power control solutions under a variety of system configurations. Expand
Lifelong Learning with Dynamically Expandable Networks
TLDR
The obtained network fine-tuned on all tasks obtained significantly better performance over the batch models, which shows that it can be used to estimate the optimal network structure even when all tasks are available in the first place. Expand
Deep Learning Based Communication Over the Air
TLDR
This paper builds, train, and run a complete communications system solely composed of NNs using unsynchronized off-the-shelf software-defined radios and open-source deep learning software libraries, and proposes a two-step learning procedure based on the idea of transfer learning that circumvents the challenges of training such a system over actual channels. Expand
Learning Optimal Resource Allocations in Wireless Systems
TLDR
DNNs are trained here with a model-free primal-dual method that simultaneously learns a DNN parameterization of the resource allocation policy and optimizes the primal and dual variables. Expand
...
1
2
3
4
5
...