Abstract
This paper investigates spectral efficiency (SE) in massive multiple-input multiple-output (MIMO) systems where terminals lack fixed power supplies and rely on radio frequency (RF) energy harvesting. By introducing a hybrid protocol that combines time-switching (TS) and power-splitting (PS) techniques, the authors develop a unified framework for joint uplink and downlink optimization. An iterative algorithm based on geometric programming is proposed to solve the resulting non-convex problem, demonstrating significant SE improvements over traditional single-protocol approaches.
The Challenge of Powering Smart Terminals
As the Internet of Things (IoT) expands into medical implants, industrial monitoring, and remote sensing, powering these "smart terminals" remains a critical hurdle. Battery replacement is often impractical or impossible in these environments. Wireless Energy Harvesting (EH) offers a sustainable solution, allowing devices to replenish their energy from the RF signals transmitted by a base station (BS).
Existing research typically treats uplink (terminal to BS) and downlink (BS to terminal) as separate optimization problems. Furthermore, most systems rely on either Time-Switching (TS)—alternating between data reception and energy harvesting—or Power-Splitting (PS)—dividing a single signal for simultaneous data and energy processing. This paper addresses these gaps by proposing a "Hybrid" protocol designed to maximize joint Spectral Efficiency (SE) across both directions.
A Unified Hybrid Protocol
The proposed system model utilizes a massive MIMO base station to serve multiple single-antenna terminals. The core innovation is the hybrid protocol, which acts as a flexible framework. By tuning specific parameters, the system can operate as a pure TS protocol, a pure PS protocol, or a combination of both, allowing it to adapt dynamically to varying channel conditions.
The Four-Phase Frame Structure
To facilitate joint optimization, the communication process is organized into a four-phase frame structure:
- Uplink Pilots: Terminals transmit pilot signals, enabling the BS to estimate the channel state.
- Downlink Wireless Power Transfer (WPT): The BS transmits pure energy to the terminals (TS-style harvesting).
- Downlink SWIPT: The BS transmits data and energy simultaneously; terminals use power-splitting to decode data while harvesting energy.
- Uplink Data: Terminals utilize the total harvested energy to transmit data back to the BS.
Optimization Methodology
The resulting optimization problem is non-convex, making it mathematically complex to solve directly. The authors developed a two-step iterative algorithm:
- One-Dimensional Search: Used to determine the optimal time durations for channel estimation and energy transfer.
- Successive Approximation via Geometric Programming (GP): Used to optimize power allocation coefficients, power-splitting ratios, and energy allocation fractions.
Numerical simulations using i.i.d. Rayleigh fading channels validated the approach. The algorithm typically converges within 31 iterations, providing a computationally efficient solution for real-time applications.
Key Insights and Results
- Protocol Superiority: The hybrid protocol consistently outperforms pure TS and matches or exceeds pure PS performance, proving the value of flexibility in EH systems.
- Resource Prioritization: A key finding is that time allocation (harvesting vs. transmitting) has a more significant impact on SE than the absolute amount of energy harvested.
- Impact of Massive MIMO: Increasing the number of BS antennas ($M$) significantly boosts SE. For instance, a configuration with 300 antennas and 3 terminals achieved a sum SE of approximately 18.8 bit/s/Hz.
- Channel Hardening: Due to the channel hardening effect in massive MIMO, the system benefits from shorter channel estimation phases, leaving more time for data transmission.
Conclusion and Future Directions
This research provides a robust framework for optimizing sustainable wireless networks. While the current model assumes linear energy harvesting and Rayleigh fading, future work will explore non-linear EH models, Rician fading (Line-of-Sight components), and the impact of hardware impairments. These additions will further bridge the gap between theoretical optimization and real-world deployment in next-generation IoT ecosystems.
Summary updated