Novel Methods for Load Estimation in Cell Switching in HAPS-Assisted Sustainable 6G Networks

TL;DR

The paper tackles the energy efficiency challenge in vHetNets by addressing the practical gap of estimating traffic loads for sleeping SBSs, which is critical for feasible cell switching with a HAPS-assisted architecture. It introduces three spatial interpolation schemes—clustering-based, distance-based, and random neighboring selection—to estimate sleeping SBS loads using a Milan dataset, and analyzes how estimation errors influence the network's power consumption. The vHetNet power minimization is formulated as minimizing $P(\Delta)$ under load-dependent constraints, with MBS and HAPS kept always-on and SBSs switchable. The findings demonstrate that accurate load estimation is essential for applying power-optimization methods in vHetNets, and the proposed methods offer concrete, data-driven routes to close the gap between theory and practice.

Abstract

In the evolving landscape of vertical heterogeneous networks, the practice of cell switching particularly for small base stations faces a significant challenge due to the lack of accurate data on the traffic load of sleeping SBSs. This information gap is crucial as it hinders the feasibility and applicability of existing power consumption optimization methods; however, the studies in the literature predominantly assume perfect knowledge about the traffic load of sleeping SBSs. Addressing this critical issue, our study introduces innovative methodologies for estimating the traffic load of sleeping SBSs in a vHetNet including the integration of a high altitude platform as a super macro base station into the terrestrial network. We propose three distinct spatial interpolation-based estimation schemes: clustering-based, distance based, and random neighboring selection. Employing a real data set for empirical validations, we compare the estimation performance of the developed traffic load estimation schemes and assess the impact of estimation errors. Our findings demonstrate that accurate estimation of sleeping SBSs' traffic loads is essential for making network power consumption optimization methods both feasible and applicable in vHetNets.

Figures (5)

• Figure 1: A vHetNet model with an MBS, multiple SBSs, and a HAPS-SMBS.
• Figure 2: Estimation error for distance-based cell selection with weighting.
• Figure 3: The estimation error for different methods. Two $x$- and two $y$-axes are considered: blue one is for clustering based while black one is for the rest of the approaches.
• Figure 4: Decision change for different number of layers in MLC.
• Figure 5: Network power consumption for different number of layers in MLC.