Optimizing Air-borne Network-in-a-box Deployment for Efficient Remote Coverage

Sidrah Javed, Yunfei Chen, Mohamed‐Slim Alouini, Cheng‐Xiang WangIEEE Internet of Things Journal 2024

TL;DR

This paper proposes an optimized deployment strategy for "Network-in-a-Box" (NIB) systems mounted on low-altitude platforms (LAPS) to provide efficient remote wireless coverage. By jointly optimizing NIB deployment, user association, beamforming, and resource allocation, the system achieves significant improvements in sum-rate, fairness, and energy efficiency compared to conventional approaches.

What Problem Does This Paper Address?

The United Nations' Sustainable Development Goals aim to eliminate digital inequality, highlighting the need for on-demand and flexible wireless coverage in remote, sparsely populated, or disaster-stricken areas. Traditional terrestrial networks struggle to provide this coverage cost-effectively. Aerial platforms like drones (LAPS) and high-altitude platforms (HAPS) offer a promising solution due to their mobility and flexible deployment. However, efficiently coordinating these platforms, especially when using a multi-layer architecture (LAPS for access, HAPS for backhaul), presents significant challenges. These include determining the optimal number and placement of LAPS (NIBs), associating users with the most suitable NIB, optimizing beamforming for efficient coverage, and allocating resources (power, spectrum) to maximize system performance while ensuring quality of service (QoS). This paper tackles these interconnected optimization problems to create an efficient and adaptable aerial network.

Methods

The authors propose a sequential optimization algorithm to address the complex, non-convex mixed-integer programming problem. This algorithm breaks down the problem into several sub-problems that are solved iteratively:

  • UAVB-NIB Deployment: A Geometric Disk Cover (GDC) algorithm is used to determine the minimum number of UAVB-NIBs and their optimal ground locations to cover the desired area. This is contrasted with a conventional cellular approach, showing GDC requires fewer NIBs.
  • User Association (UA): A greedy algorithm is employed to associate each user with a single UAVB-NIB that offers the maximum received signal power, ensuring user fairness and interference mitigation.
  • Beam Optimization: The location parameters (beam centers, radii, altitudes, beamwidths) of the UAVB-NIBs are optimized to minimize beam radius and maximize channel gain, effectively creating directive beams. This involves solving a minimum enclosing circle problem.
  • Resource Allocation:
    • Backhaul: Non-Orthogonal Multiple Access (NOMA) is used for the HAPS-to-UAVB-NIB backhaul link. A closed-form solution is derived for power allocation factors to maximize the sum rate, considering channel strength ordering and successive interference cancellation (SIC) at the NIBs.
  • Access Link (AL): For the UAVB-NIB-to-Ground User (GU) link, NOMA is employed with Regularized Zero-Forcing (RZF) precoding. Successive Convex Approximation (SCA) is used to iteratively optimize power allocation for users, maximizing the achievable sum rate under QoS constraints.

The overall algorithm proceeds sequentially, with each sub-problem's solution feeding into the next, aiming for an optimal system configuration.

Datasets

Not discussed in the paper. The authors use simulated data based on realistic parameters for HAPS altitude, coverage radius, user densities, and RAT probabilities.

Key Insights and Ideas

  • NIBs as Building Blocks: The "Network-in-a-Box" (NIB) concept is highlighted as a crucial enabler for flexible and on-demand aerial networks, packing a full wireless network into a portable unit.
  • Synergy of LAPS and HAPS: Combining LAPS (for access) with HAPS (for backhaul) creates a powerful multi-layer architecture that can overcome limitations of individual platforms and extend coverage significantly.
  • NOMA for Efficiency: NOMA is effectively leveraged in both the backhaul (HAPS-NIB) and access link (NIB-GU) to improve spectral efficiency and sum-rate by allowing multiple users/NIBs to share the same resources.
  • Importance of Joint Optimization: The paper strongly emphasizes that optimizing deployment, association, beamforming, and resource allocation jointly is critical for maximizing system performance, rather than optimizing each in isolation.
  • GDC for Deployment Efficiency: The Geometric Disk Cover approach for NIB deployment is shown to be significantly more efficient, requiring fewer NIBs than conventional cellular deployment strategies.
  • User-Centric Beamforming: Optimizing beam parameters to align with user locations and demands leads to more concentrated power, higher SINR, and improved coverage.

Key Results

  • Reduced NIB Deployment: The proposed GDC deployment scheme requires up to 50% fewer UAVB-NIBs compared to conventional deployment for a given coverage area.
  • Improved Sum-Rate: The optimized resource allocation strategy leads to a significant increase in the average sum-rate of the AL, with up to 185.24% improvement observed in dense urban areas.
  • Enhanced Fairness: The proposed power allocation strategy significantly improves Jain's fairness index, showing up to 85% improvement over uniform power allocation.
  • Higher Energy and Spectral Efficiency: The NOMA-based approach outperforms OFDMA in terms of energy efficiency (AEE) and spectral efficiency (ASE) for the backhaul link. Similarly, the AL achieves higher AEE and SE with optimal power allocation.
  • SINR Improvement: The UA-SNR algorithm demonstrates a substantial improvement in average received SINR at ground users compared to random or distance-based association.

Limitations and Future Work

  • Computational Complexity: The paper acknowledges that the overall optimization problem is a non-convex mixed-integer program, making it computationally challenging. The sequential optimization approach provides a sub-optimal solution.
  • Perfect CSI Assumption: The analysis assumes perfect Channel State Information (CSI) for NIB ordering and error-free decoding in the backhaul link. Real-world scenarios with imperfect CSI would introduce performance degradation.
  • Simplified Propagation Models: While considering both small-scale and large-scale fading, the propagation models might not capture all the complexities of real-world aerial communication environments.
  • Future Work: The authors suggest exploring more advanced optimization techniques to achieve near-optimal solutions and investigating the impact of imperfect CSI and more realistic propagation models. They also mention the potential for further research into dynamic deployment and resource allocation in response to changing network conditions.

Summary updated

Abstract

Among many envisaged drivers for the sixth generation, one is from the United Nations Sustainability Development Goals 2030 to eliminate digital inequality. Remote coverage in sparsely populated areas, difficult terrains, or emergency scenarios requires on-demand access and flexible deployment with minimal capex and opex. In this context, network-in-a-box (NIB) is an exciting solution that packs the whole wireless network into a single portable and re-configurable box to support multiple access technologies such as WiFi/2G/3G/4G/5G etc. In this paper, we propose low-altitude platform stations (LAPS) based NIBs with stratospheric high-altitude platform station (HAPS) as backhaul. Specifically, backhaul employs non-orthogonal multiple access (NOMA) with superposition coding at the transmitting HAPS and successive interference cancellation (SIC) at the receiving NIBs, whereas the access link (AL) employs superposition coding along with the regularized zero-forcing (RZF) precoding at the NIB in order to elevate the computational overhead from the ground users. The required number of airborne NIBs to serve a desired coverage area, their optimal placement, user association, beam optimization, and resource allocation are optimized by maximizing the sum rate of the AL while maintaining the quality of service. Our findings reveal the significance of thorough system planning and communication parameters optimization for enhanced system performance and best coverage under limited resources.
BibTeX
@article{javed2024optimizing,
  author = {Sidrah Javed and Yunfei Chen and Mohamed‐Slim Alouini and Cheng‐Xiang Wang},
  title = {{Optimizing Air-borne Network-in-a-box Deployment for Efficient Remote Coverage}},
  journal = {IEEE Internet of Things Journal},
  year = {2024},
}
Figure