Optimal Wireless Resource Allocation With Random Edge Graph Neural Networks

@article{Eisen2019OptimalWR,
  title={Optimal Wireless Resource Allocation With Random Edge Graph Neural Networks},
  author={Mark Eisen and Alejandro Ribeiro},
  journal={IEEE Transactions on Signal Processing},
  year={2019},
  volume={68},
  pages={2977-2991}
}
We consider the problem of optimally allocating resources across a set of transmitters and receivers in a wireless network. The resulting optimization problem takes the form of constrained statistical learning, in which solutions can be found in a model-free manner by parameterizing the resource allocation policy. Convolutional neural networks architectures are an attractive option for parameterization, as their dimensionality is small and does not scale with network size. We introduce the… 

Transferable Policies for Large Scale Wireless Networks with Graph Neural Networks

  • Mark EisenAlejandro Ribeiro
  • Computer Science
    ICASSP 2020 - 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)
  • 2020
Such REG-NNs are shown to exhibit an essential permutation invariance property for the power allocation problem that suggest transference capabilities, and validated by showing how REGNNs trained on a single ad-hoc network outperform baselines in new randomly drawn networks of growing size.

Resource Allocation via Graph Neural Networks in Free Space Optical Fronthaul Networks

The primal-dual learning algorithm is developed to train the graph neural network in a model-free manner, where the knowledge of system models is not required, and the GNN is shown to retain the permutation equivariance that matches with the permutations equivariances of resource allocation policy in networks.

Efficient Power Allocation Using Graph Neural Networks and Deep Algorithm Unfolding

A hybrid neural architecture inspired by the algorithmic unfolding of the iterative weighted minimum mean squared error method is proposed, that is denote as unfolded WMMSE (UWMMSE), that achieves performance comparable to that of WMM SE while significantly reducing the computational complexity.

Unfolding WMMSE Using Graph Neural Networks for Efficient Power Allocation

This work puts forth a neural network architecture inspired by the algorithmic unfolding of the iterative weighted minimum mean squared error (WMMSE) method, that is permutation equivariant, thus facilitating generalizability across network topologies.

Unsupervised Learning for Asynchronous Resource Allocation In Ad-Hoc Wireless Networks

An unsupervised learning method based on Aggregation Graph Neural Networks that can be learned globally and asynchronously while implemented locally and proposes a permutation invariance property which indicates the transferability of the trained Agg-GNN.

Distributed Scheduling Using Graph Neural Networks

This work proposes a distributed MWIS solver based on graph convolutional networks (GCNs) that learns topology-aware node embeddings that are combined with the network weights before calling a greedy solver with good generalizability across graphs and minimal increase in complexity.

Graph Neural Networks for Scalable Radio Resource Management: Architecture Design and Theoretical Analysis

This paper demonstrates that radio resource management problems can be formulated as graph optimization problems that enjoy a universal permutation equivariance property, and identifies a family of neural networks, named message passing graph neural networks (MPGNNs), which can generalize to large-scale problems, while enjoying a high computational efficiency.

Wireless Power Control via Counterfactual Optimization of Graph Neural Networks

It is shown how the counterfactual optimization technique allows us to guarantee a minimum rate constraint, which adapts to the network size, hence achieving the right balance between average and 5th percentile user rates throughout a range of network configurations.

Decentralized Inference with Graph Neural Networks in Wireless Communication Systems

This paper develops a methodology to verify whether the predictions are robust, and analyzes and enhances the robustness of the decentralized GNN during the inference stage in different wireless communication systems, and proposes novel retransmission mechanisms.

Model-Free Learning of Optimal Ergodic Policies in Wireless Systems

This article constructs and exploits smoothed surrogates of constrained ergodic resource allocation problems, the gradients of the former being representable exactly as averages of finite differences that can be obtained through limited system probing, and develops a new model-free primal-dual algorithm for learning optimal ergodics resource allocations.
...

References

SHOWING 1-10 OF 34 REFERENCES

Large Scale Wireless Power Allocation with Graph Neural Networks

  • Mark EisenAlejandro Ribeiro
  • Computer Science
    2019 IEEE 20th International Workshop on Signal Processing Advances in Wireless Communications (SPAWC)
  • 2019
This work proposes using a convolutional neural network architecture, specifically one that employs a graph convolution over the randomly varying fading links of the wireless network, and shows strong performance of a primal-dual learning method in training an REGNN parameterization relative to existing heuristic benchmarks.

Learning Optimal Resource Allocations in Wireless Systems

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.

Spatial Deep Learning for Wireless Scheduling

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.

A Graph Neural Network Approach for Scalable Wireless Power Control

This paper proposes an interference graph convolutional neural network (IGCNet) to learn the optimal power control in an unsupervised manner and shows that one- layer IGCNet is a universal approximator to continuous set functions, which well matches the permutation invariance property of interference channels and it is robust to imperfect channel state information (CSI).

Learning to optimize: Training deep neural networks for wireless resource management

This work first characterize a class of ‘learnable algorithms’ and then design DNNs to approximate some algorithms of interest in wireless communications, demonstrating the superior ability ofDNNs for approximating two considerably complex algorithms that are designed for power allocation in wireless transmit signal design, while giving orders of magnitude speedup in computational time.

Graph Embedding-Based Wireless Link Scheduling With Few Training Samples

This article proposes a novel graph embedding based method for link scheduling in D2D networks that is competitive in terms of scalability and generalizability to more complicated scenarios and develops a K-nearest neighbor graph representation method to reduce the computational complexity.

Optimal resource allocation in wireless communication and networking

The article discusses the problem simplifications that arise by working in the dual domain and reviews algorithms that can determine optimal operating points with relatively lightweight computations.

Power Allocation in Multi-User Cellular Networks: Deep Reinforcement Learning Approaches

Policy-based REINFORCE, value-based deep Q-learning (DQL), actor-critic deep deterministic policy gradient (DDPG) algorithms are proposed for this sum-rate problem, and simulation results show that the data-driven approaches outperform the state-of-art model-based methods on sum- rate performance.

Round-robin power control for the weighted sum rate maximisation of wireless networks over multiple interfering links

A Gauss–Seidel type iterative power control algorithm for wireless networks consisting of multiple source–destination pairs that has the favourable properties that only simple operations are needed in each iteration and the convergence is fast.

Towards Optimal Power Control via Ensembling Deep Neural Networks

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.