TL;DR
This paper tackles the challenge of minimizing energy consumption in IoT networks by jointly optimizing power and time allocation for Machine-to-Machine (M2M) communications. It compares two access strategies, NOMA and TDMA, considering non-linear energy harvesting and circuit power consumption, revealing that the optimal strategy depends on the MTCD's circuit power.
What Problem Does This Paper Address?
The Internet of Things (IoT) relies heavily on Machine-to-Machine (M2M) communications, which are characterized by massive device connectivity, bursty traffic, and extremely low power budgets. A critical challenge is ensuring energy efficiency to prolong device lifetimes and network operation. This paper focuses on optimizing resource allocation (power and time) in M2M networks that utilize energy harvesting (EH) to power devices. Specifically, it addresses the complexities introduced by non-linear EH models and the non-negligible circuit power consumption of devices, which traditional models often overlook. The authors aim to minimize the total network energy consumption while meeting minimum throughput requirements and energy causality constraints.
Methods
The paper proposes joint power control and time allocation algorithms for two multiple access schemes: Non-Orthogonal Multiple Access (NOMA) and Time Division Multiple Access (TDMA).
- NOMA Strategy: The authors first derive optimal conditions, showing that it's optimal for each device to transmit at its minimum required throughput. They then develop an iterative algorithm (IPCTA-NOMA) to solve the non-convex optimization problem. This involves transforming non-smooth EH functions into continuous ones and converting non-convex constraints into convex ones through variable transformations and iterative updates of device EH capabilities.
- TDMA Strategy: Similar optimal conditions are derived for TDMA. The problem is then transformed into a tractable, equivalent problem that can be solved iteratively using a similar approach (IPCTA-TDMA).
Both algorithms leverage convex optimization techniques and iterative updates to find suboptimal solutions.
Datasets
Not discussed in the paper. The paper relies on theoretical analysis and simulation-based evaluation using parameters derived from existing literature and practical measurements.
Key Insights and Ideas
- Circuit Power Matters: The inclusion of circuit power consumption significantly impacts optimal transmission time. Unlike scenarios without circuit power, where longer transmission times always save energy, here, excessively long transmission times can increase total energy consumption due to circuit power overhead.
- NOMA vs. TDMA Trade-off: The choice between NOMA and TDMA for energy efficiency is not universal. NOMA is superior at low circuit power regimes for MTCDs, while TDMA becomes more energy-efficient at high circuit power regimes. This is because NOMA's simultaneous transmission can lead to higher circuit power consumption due to longer active times.
- Minimal Throughput is Optimal: For energy minimization, it's optimal for devices to transmit at their minimum required throughput, rather than exceeding it. This conserves power by reducing transmission time and/or power.
- Non-linear EH Complexity: The non-linear nature of energy harvesting, particularly the receiver sensitivity threshold, adds significant complexity to resource allocation, requiring careful modeling and algorithmic solutions.
Key Results
- Algorithm Convergence: The proposed iterative algorithms (IPCTA-NOMA and IPCTA-TDMA) are shown to converge to suboptimal solutions.
- Performance Comparison:
- At low MTCD circuit power (e.g., < 4 mW), NOMA demonstrates lower total energy consumption than TDMA.
- At high MTCD circuit power (e.g., ≥ 5 mW), TDMA outperforms NOMA in terms of total energy efficiency.
- Parameter Sensitivity: Total energy consumption decreases with increasing maximal transmission power of MTCGs and MTCDs. The sensitivity of NOMA to MTCG transmission power is higher than TDMA, especially at high circuit power.
- Payload Impact: Increased payload requirements directly lead to higher total energy consumption for both NOMA and TDMA.
Limitations and Future Work
- Cluster Formation: The paper assumes MTCDs are already associated with MTCGs. Joint optimization of cluster formation and resource allocation is suggested as future work.
- Model Simplifications: While considering non-linear EH and circuit power, the paper assumes ideal SIC for NOMA and channel reciprocity.
- Generalization: The analysis is primarily for uplink communications. Extending to downlink or full-duplex scenarios could be explored.
- Dynamic Environments: The current work assumes static channel conditions. Future research could investigate resource allocation in dynamic or time-varying environments.
Summary updated