[Paper Review] Risk-Based Optimization of Virtual Reality over Terahertz Reconfigurable Intelligent Surfaces

Christina Chaccour, Mehdi Naderi Soorki+3 more|arXiv (Cornell University)|Jan 1, 2020

Advanced Wireless Communication TechnologiesEngineering19 references79 citations

TL;DR

The paper proposes a risk-aware user association and scheduling framework for VR over terahertz reconfigurable intelligent surfaces, using entropic value-at-risk and Lyapunov optimization with a deep policy learner to handle stochastic channels.

ABSTRACT

In this paper, the problem of associating reconfigurable intelligent surfaces (RISs) to virtual reality (VR) users is studied for a wireless VR network. In particular, this problem is considered within a cellular network that employs terahertz (THz) operated RISs acting as base stations. To provide a seamless VR experience, high data rates and reliable low latency need to be continuously guaranteed. To address these challenges, a novel risk-based framework based on the entropic value-at-risk is proposed for rate optimization and reliability performance. Furthermore, a Lyapunov optimization technique is used to reformulate the problem as a linear weighted function, while ensuring that higher order statistics of the queue length are maintained under a threshold. To address this problem, given the stochastic nature of the channel, a policy-based reinforcement learning (RL) algorithm is proposed. Since the state space is extremely large, the policy is learned through a deep-RL algorithm. In particular, a recurrent neural network (RNN) RL framework is proposed to capture the dynamic channel behavior and improve the speed of conventional RL policy-search algorithms. Simulation results demonstrate that the maximal queue length resulting from the proposed approach is only within 1% of the optimal solution. The results show a high accuracy and fast convergence for the RNN with a validation accuracy of 91.92%.

Motivation & Objective

• Motivate the challenge of delivering high-rate, low-latency VR over THz RIS-enabled networks.
• Propose a risk-based rate and reliability optimization framework using entropic value-at-risk (EVaR).
• Ensure queue stability and control delay statistics via Lyapunov optimization.
• Develop a policy-based algorithm with deep learning to learn user association in a large state space.
• Demonstrate near-optimal performance and fast convergence through simulations.

Proposed method

• Formulate a downlink VR service model where RISs reflect signals to serve mobile users.
• Introduce a risk measure based on entropic value-at-risk to capture higher-order delay statistics.
• Apply Lyapunov optimization to convert the problem into a linear weighted objective with queue stability guarantees.
• Propose a policy-based algorithm to perform user association under stochastic channels.
• Use a deep-learning framework to learn the policy due to the large state space and dynamic channel behavior.
• Demonstrate framework efficiency and convergence in simulations.

Experimental results

Research questions

• RQ1 How to maximize rate and reliability for VR over RIS-enabled networks under stochastic channels?
• RQ2 Can EVaR effectively capture higher-order delay statistics in RIS-aided VR systems?
• RQ3 Can a Lyapunov-based reformulation and a deep policy-learning approach achieve near-optimal user association with scalable convergence?

Key findings

• Maximal queue length under the proposed approach is within 1% of the optimal solution.
• Validation accuracy of 91.92% reported for the learned policy.
• The framework achieves high accuracy and fast convergence in simulations.