Unsupervised Deep Learning for Optimizing Wireless Systems with Instantaneous and Statistic Constraints
This work addresses optimization problems in wireless communications, particularly for ultra-reliable and low-latency scenarios, but it is incremental as it extends existing unsupervised methods to handle more constraint types.
The paper tackles the challenge of applying unsupervised deep learning to wireless system optimization problems that involve both instantaneous and statistical constraints, by establishing a unified framework that converts variable optimizations into functional ones and uses DNNs to approximate Lagrange multipliers. Simulation results show that this approach outperforms supervised learning in QoS violation probability and policy approximation accuracy, with rapid convergence when pre-trained.
Deep neural networks (DNNs) have been introduced for designing wireless policies by approximating the mappings from environmental parameters to solutions of optimization problems. Considering that labeled training samples are hard to obtain, unsupervised deep learning has been proposed to solve functional optimization problems with statistical constraints recently. However, most existing problems in wireless communications are variable optimizations, and many problems are with instantaneous constraints. In this paper, we establish a unified framework of using unsupervised deep learning to solve both kinds of problems with both instantaneous and statistic constraints. For a constrained variable optimization, we first convert it into an equivalent functional optimization problem with instantaneous constraints. Then, to ensure the instantaneous constraints in the functional optimization problems, we use DNN to approximate the Lagrange multiplier functions, which is trained together with a DNN to approximate the policy. We take two resource allocation problems in ultra-reliable and low-latency communications as examples to illustrate how to guarantee the complex and stringent quality-of-service (QoS) constraints with the framework. Simulation results show that unsupervised learning outperforms supervised learning in terms of QoS violation probability and approximation accuracy of the optimal policy, and can converge rapidly with pre-training.