Resource Allocation in Full-Duplex Mobile-Edge Computing Systems with NOMA and Energy Harvesting

Zhaohui Yang, Zhaohui Yang 0001, Jiancao Hou, Mohammad Shikh-Bahaei

TL;DR

This paper tackles the challenge of minimizing energy consumption in mobile-edge computing (MEC) systems by integrating full-duplex (FD) communication, non-orthogonal multiple access (NOMA), and energy harvesting (EH). They propose an iterative algorithm that efficiently allocates resources like power, time, and computation capacity, achieving better energy efficiency than existing methods.

What Problem Does This Paper Address?

Modern mobile devices are increasingly running computationally intensive applications like augmented reality. Offloading these tasks to edge servers (MEC) can reduce device energy consumption and latency. However, wireless systems face limitations due to finite battery power. Energy harvesting (EH) offers a solution by allowing devices to collect energy from wireless signals, but it can consume valuable bandwidth. Full-duplex (FD) communication allows simultaneous transmission and reception, potentially improving spectral efficiency. Non-orthogonal multiple access (NOMA) further enhances spectral efficiency by allowing multiple users to share the same frequency band.

This paper addresses the complex problem of optimizing resource allocation in a system that combines all these advanced technologies: FD communication for simultaneous data reception and energy broadcasting, NOMA for efficient uplink data offloading, and EH for powering user devices. The core challenge is to minimize the total energy consumed by the system (both users and the base station) while satisfying various constraints, such as data offloading requirements, energy causality (harvested energy must meet or exceed consumed energy), and computation latency. Existing research has explored these technologies individually or in pairs, but a comprehensive approach integrating all three for resource allocation was missing.

Methods

The authors formulate a total energy minimization problem for the described FD-MEC system with NOMA and EH. The problem involves optimizing several variables: user transmission power ($p_j$), base station broadcasting power ($q_i$), time allocation for different user groups ($t_i$), and data offloading amounts ($d_j$).

To tackle the inherent nonconvexity of the problem, they employ a multi-step approach:

  1. Problem Transformation: They first transform the original problem into an equivalent one with fewer variables. This is achieved by using the uplink achievable rate and data offloading constraints to express user transmission power as a function of other variables.
  2. Convexity Analysis: They prove that the transformed problem is convex with respect to each variable (power, time, or data offloading) when the other two are fixed. This property is crucial for designing an efficient optimization algorithm.
  3. Iterative Algorithm: Leveraging the convexity property, they design an iterative algorithm. In each iteration, the algorithm optimizes one variable while keeping the others fixed. This process is repeated until the objective function converges. The algorithm effectively performs iterative power control, time scheduling, and offloading data allocation.

Datasets

Not discussed in the paper. The paper focuses on theoretical modeling and simulation-based evaluation.

Key Insights and Ideas

  • Synergy of FD, NOMA, and EH: The paper demonstrates that combining FD, NOMA, and EH in MEC systems can lead to significant energy savings. Each technology contributes to overcoming the limitations of the others.
  • User Pairing Matters: The choice of how users are grouped (user pairing) significantly impacts overall energy consumption. The "strong-middle" (SM) pairing strategy, where users with strong and moderate channel conditions are paired, was found to be most energy-efficient. This suggests that balancing channel gains within groups can optimize resource utilization.
  • Iterative Optimization is Effective: The proposed iterative algorithm, which leverages the problem's conditional convexity, is shown to converge quickly and effectively find near-optimal solutions. This is a practical approach for real-world deployment.
  • Trade-offs in Resource Allocation: The results highlight inherent trade-offs. For instance, NOMA improves spectral efficiency but adds complexity at the BS (due to SIC). Increasing cloud computation capacity reduces user offloading energy but might not always lead to proportional energy savings beyond a certain threshold.

Key Results

  • The proposed algorithm achieves better total energy consumption performance compared to conventional methods like NOMA HD (half-duplex) and OMA FD (orthogonal multiple access, full-duplex).
  • The SM user pairing strategy yields the lowest total energy consumption among the considered pairing methods (SW, SM, SS).
  • The proposed algorithm demonstrates rapid convergence, typically requiring only a few iterations to reach a stable solution, showcasing its efficiency.
  • Total energy consumption decreases with increasing time slot duration and decreases with increasing cloud computation capacity up to a certain point, after which it stabilizes.

Limitations and Future Work

  • User Pairing Strategy: While the paper investigates user pairing, the optimal pairing strategy might be more complex than the simple methods considered. Future work could explore dynamic or more sophisticated pairing mechanisms.
  • Channel State Information (CSI): The proposed algorithm assumes perfect CSI. In real-world scenarios, CSI is often imperfect. Investigating the algorithm's performance under imperfect CSI would be a valuable extension.
  • Mobility: The current model assumes users are relatively static. Incorporating user mobility and its impact on channel conditions and resource allocation would enhance the model's realism.
  • Complexity of SIC: The paper mentions the computational complexity of Successive Interference Cancellation (SIC) at the BS for NOMA. Further analysis or optimization of this aspect could be beneficial.

Summary updated

BibTeX
@inproceedings{yang2019resource,
  author = {Zhaohui Yang and Zhaohui Yang and Jiancao Hou and Mohammad Shikh-Bahaei},
  title = {{Resource Allocation in Full-Duplex Mobile-Edge Computing Systems with NOMA and Energy Harvesting}},
  booktitle = {ICC},
  year = {2019},
  doi = {10.1109/icc.2019.8761939},
}
Figure