Abstract
The proliferation of the Internet of Video Things (IoVT) necessitates high-throughput and low-latency communication, which often conflicts with the limited battery life of IoVT devices. This paper proposes a framework to minimize long-term power consumption in multi-IRS aided IoVT systems. By employing Lyapunov optimization and fractional programming, we jointly optimize device transmit power and IRS phase shifts. Simulation results demonstrate that the proposed scheme achieves up to 38% power savings while strictly meeting delay requirements, highlighting the efficacy of multi-IRS coordination in dynamic wireless environments.
1. Introduction
The Internet of Video Things (IoVT) involves massive data generation from high-resolution cameras for applications like traffic monitoring and public safety. These systems face a "double bind": the need for high throughput and low latency versus the rapid battery depletion of mobile devices. Intelligent Reconfigurable Surfaces (IRS) offer a promising solution by passively steering radio waves to enhance signal strength. However, managing multiple IRS units in dynamic environments with unpredictable channel conditions and data arrivals remains a significant challenge. This work addresses the multi-IRS coordination problem over long durations to ensure average queuing delays remain below strict thresholds.
2. System Model
We consider a multi-IRS aided IoVT network consisting of a single-antenna Base Station (BS), multiple IoVT devices, and $M$ IRS units. The devices upload video data to the BS via both direct links and reflected links provided by the IRS units.
The environment is modeled in 3D with Rician fading channels to account for line-of-sight and scattered signals. Video traffic arrivals are modeled as independent and identically distributed (i.i.d.) processes.
3. Methodology
The optimization problem is formulated to minimize long-term average power consumption subject to stability and delay constraints. We decouple this complex problem using several mathematical frameworks:
- Lyapunov Optimization: We transform long-term delay constraints into virtual queue stability problems, allowing for per-slot decision-making without future channel knowledge.
- Drift-Plus-Penalty Method: This technique enables a tunable trade-off between power minimization and queue length (delay).
- Alternating Optimization (AO) and Fractional Programming (FP): The non-convex per-slot problem is solved by iteratively optimizing transmit power and IRS phase shifts. FP is used to linearize complex signal-to-noise ratio expressions.
- Discrete Phase Shifts: The model incorporates realistic hardware constraints by considering a finite set of discrete phase values for the IRS elements.
4. Performance Evaluation
Numerical simulations were conducted in a 100m radius deployment. Key findings include:
- Multi-IRS Advantage: Multiple IRS units provide spatial diversity, ensuring reliable reflections even when specific paths are obstructed.
- Power Efficiency: The proposed algorithm achieves 15% to 38% lower power consumption compared to benchmarks such as random phase shifts or non-IRS systems.
- Convergence: The AO algorithm typically converges within 4 to 6 iterations, making it suitable for real-time applications.
4.1 Delay-Power Trade-off
A critical aspect of the system is the trade-off between latency and energy consumption, controlled by the parameter $V$.
As shown in Figure 2, increasing $V$ prioritizes power savings at the expense of longer queues (higher delay). Conversely, lower $V$ values ensure minimal delay but require higher transmit power. This flexibility allows operators to adapt the system to specific application requirements, such as ultra-low latency for emergency services or maximum battery life for routine monitoring.
5. Conclusion
This paper presents a robust framework for power optimization in multi-IRS IoVT systems. By integrating Lyapunov optimization with advanced beamforming techniques, we provide a scalable solution that balances energy efficiency and latency. Future work will address challenges related to imperfect channel state information and the deployment of mobile IRS units.
Summary updated