HAPS-Enabled 6G TN-NTNs: Achieving Superior QoS-Energy Tradeoffs Compared with LEO
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: Electrical Engineering. The day in electrical engineering: signals, systems and the hardware that carries them.
Rosa: I'm Rosa.
Dev: And I'm Dev, here with Taro.
Rosa: We're looking at the paper: "HAPS-Enabled 6G TN-NTNs".
Dev: Here's the gist of the paper. A Traffic-Aware QoS-Energy Consumption Tradeoff Study for Integrated 6G TN & NTNs investigates how increasing HAPS density can significantly reduce terrestrial energy consumption while maintaining or improving quality of service compared to LEO satellite systems Core Investigation and Motivation The research extends the BLASTER framework to evaluate the trade-off between network quality of service (QoS) and terrestrial energy consumption in integrated terrestrial and non-terrestrial networks using different HAPS deployment densities.
Rosa: The study compares one low Earth orbit (LEO) satellite at 600 km with 1-, 4-, and 9-highaltitude platform station (HAPS) deployments at 20 km under a realistic diurnal traffic profile.
Dev: The motivation stems from the Hypercell concept, which introduces a new architectural perspective where a wide-area HAPS coverage layer assumes part of the responsibility traditionally assigned to terrestrial macro cells, allowing for dynamic deactivation of both terrestrial coverage and capacity cells based on traffic demand.
Rosa: System Modeling and Optimization Framework The system model considers a heterogeneous terrestrial cellular network partitioned into urban (Au) and rural (Ar) regions, with the total set of MBSs being the union of terrestrial (T) and non-terrestrial (S) MBSs.
Title and authors: Rosa: So we're looking at this paper now, "HAPS-Enabled 6G TN-NTNs: Achieving Superior QoS-Energy Tradeoffs Compared with LEO <ref:2609.22288#pg1>." The title tells us right away that they're not just comparing different networks; they are specifically looking at how adding High Altitude Platform Stations, or HAPS, helps manage the trade-off between network quality and energy use when mixing terrestrial and non-terrestrial networks.
Dev: Yeah, that's what it is. They are taking a look at integrating these three types of services—terrestrial cellular, satellite communication from Low Earth Orbit satellites like LEO, and HAPS—and seeing how the HAPS deployment density changes the outcome compared to just using one LEO satellite setup.
Rosa: It's interesting because they aren't just throwing more satellites in there; they are focusing on a traffic-aware approach, which is a big deal. They are investigating how increasing HAPS density can actually reduce terrestrial energy consumption while keeping the quality of service the same or even making it better.
Dev: Exactly. And the authors, Azim Akhtarshenas, Henri Alam, David Lopez-Pérez, and Matteo Bernabe from UPV Valencia, they are tackling this using a framework called BLASTER and extending it to handle these multi-HAPS scenarios.
Rosa: So what does that mean for us on the ground? It suggests that we can dynamically configure a wide-area HAPS layer to take on some of the work normally done by terrestrial macro cells, which is what they call the Hypercell concept.
Dev: Right. Because this lets them decide whether to keep terrestrial coverage or capacity cells active based directly on how much traffic there is, which sounds really dynamic for managing resources in real time.
Rosa: It moves the decision-making from a static setup to something that reacts to the actual traffic demand throughout the day. And they compare one LEO satellite at six hundred kilometers altitude against three different HAPS deployments—one with one station, one with four, and one with nine stations—all under a realistic twenty-four hour traffic profile.
Dev: That comparison is key because it shows how much performance you can get from scaling up the HAPS deployment before you hit diminishing returns or just more complexity.
Rosa: And they mention that increasing the number of HAPS improves traffic offloading and enables additional terrestrial base station shutdowns, which is a direct path to energy savings.
Dev: That's the mechanism here. The core idea is that by having more HAPS, you can offload some of the traffic onto those platforms, meaning you don't need to keep as many terrestrial base stations running at full power.
The paper's summary: Rosa: So, summarizing the core investigation of this paper, it’s about systematically looking at how increasing HAPS density—going from one station up to nine stations—impacts user association, bandwidth allocation, network throughput, and the potential for terrestrial base station shutdowns.
Dev: They are essentially setting up a system model that partitions an area into urban and rural parts of the terrestrial network, and they define the total set of active base stations as being both terrestrial and non-terrestrial MBSs.
Rosa: The way they look at the physics is by using a 3GPP terrestrial channel model, but they extend it for both satellite links and HAPS links, defining achievable downlink rate based on bandwidth available at the serving MBS <ref:2609.22288#pg2>.
Dev: And the optimization goal is quite specific: they aim to maximize the sum of logarithmic throughput, which we call SLT for sum logarithmic throughput, while simultaneously penalizing terrestrial power consumption through a reweighted sparsity-promoting relaxation.
Rosa: That means they are trying to find that sweet spot where you get high data rates for users without wasting too much power on the terrestrial side. They aren't just optimizing one thing; they are balancing two competing objectives.
Dev: And the experiments involve running these simulations across different HAPS configurations against a fixed LEO-based architecture under varying diurnal traffic patterns, specifically looking at how UE density changes throughout the day—like seeing that density go up to one point in the evening.
Rosa: The key findings they report are pretty striking, showing that SC4 with four HAPS achieves thirty-eight percent and two hundred sixty-eight percent higher average throughput per non-terrestrial served user compared to SC3 and SC2 configurations.
Dev: And they also found that this is about twenty times higher than the single LEO configuration, which points to a significant performance gain from this multi-HAPS approach.
The paper's improvements: Rosa: Now let's talk about what the authors suggest as their main contributions, because they lay out some clear improvements they found in their study of HAPS-Enabled 6G TN-NTNs: first, they extend BLASTER to support multi-HAPS deployments involving one, four, and nine HAPS stations and compare them against a LEO-based architecture under realistic diurnal traffic variations <ref:2609.22288#pg1>.
Dev: That extension into the multi-HAPS space is important because it allows for more flexible resource management across the network, moving away from simpler single configurations.
Rosa: They also investigate exactly how increasing the number of HAPS affects things like user association decisions, bandwidth allocation strategies, and whether terrestrial base stations get shut down based on demand.
Dev: And then they characterize this trade-off mathematically through system-level simulations to quantify the additional terrestrial base station shutdown opportunities that different non-terrestrial network configurations enable under comparable quality of service requirements.
Rosa: It’s about characterizing the QoS energy consumption trade-off by analyzing how increasing HAPS density affects terrestrial power reduction versus changes in per user throughput, allowing for a precise determination of optimal operating points based on a reference metric like SLT.
Dev: That sounds like they are providing a mathematical tool to find the exact point where saving energy doesn't hurt your service quality too much.
Rosa: And they also suggest using traffic-aware radio resource management frameworks, like BLASTER extended for multi-HAPS deployments, to jointly optimize user association, bandwidth allocation, and terrestrial base station activation while minimizing the total energy consumption of integrated terrestrial and non-terrestrial networks.
Dev: So it’s not just about the configuration; it’s about a whole framework that manages everything from how users connect to how much power those different network elements consume together.
Conclusion: Rosa: Wrapping this up, the paper on HAPS-Enabled 6G TN-NTNs shows that a nine-HAPS configuration reduces terrestrial energy consumption by thirty-four point one percent over the full day and sixty point six percent during the quietest hour when compared to LEO <ref:2609.22288#pg1,HAPS configuration reduces terrestrial energy consumption by>.
Dev: And even at the same sum logarithmic throughput, that nine-HAPS setup achieves twenty percent additional terrestrial energy savings relative to LEO while substantially improving cell-edge throughput for those users who are near the network edge.
Rosa: The main implication is that enhanced non-terrestrial offloading can actually reduce the energy consumption of your terrestrial infrastructure, and this happens while protecting those cell-edge users significantly.
Dev: It shows that multi-HAPS deployments enable deeper terrestrial energy savings while substantially protecting cell-edge users who are often the hardest to serve efficiently.
Taro: From my view, I'm interested in how this system handles the world misbehaving when you have those dynamic shutdown opportunities; what happens to service continuity when the network decides to power down a terrestrial base station because of traffic changes?
Rosa: That’s a good point, Taro. The study focuses on the optimization goal and how it balances quality, but I wonder if there are specific safety cases they outline for when that traffic shift happens really fast.
Dev: Yeah, the paper focuses on the mathematical characterization of the trade-off at an aggregate level, so it doesn't go into minute real-time failure modes or recovery protocols under extreme conditions.
Taro: Right, so it’s a very high level optimization view compared to what I’d need to test for graceful recovery when things go really wrong on the ground.
Azim Akhtarshenas, Henri Alam, David Lopez-Pérez, Matteo Bernabe
Universitat Politecnica de Val á Valencia · Nokia Standards France, Nokia, France
eess.SP
Submitted: 2026-09-13
Updated: 2026-10-30
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Importance score: 70/100
The gist: A Traffic-Aware QoS-Energy Consumption Tradeoff Study for Integrated 6G TN & NTNs investigates how increasing HAPS density can significantly reduce terrestrial energy consumption while maintaining or…
Key concepts
- Hypercell Concept
- This architectural idea suggests using a wide-area HAPS layer to take over the role of traditional terrestrial macro cells. This allows the network to dynamically turn off both terrestrial coverage and capacity cells based on current traffic needs, optimizing resource usage.
- SLT (Sum of Logarithmic Throughput)
- This is a metric used to quantify network quality, representing the sum of logarithmic throughput across all users. The optimization goal is to maximize this value while simultaneously minimizing the energy spent by terrestrial base stations and non-terrestrial links.
- Terrestrial Energy Savings ($ riangle PT_N(oldsymbol{ ho}^*)$)
- This measures the reduction in total terrestrial power consumption achieved by shutting down unnecessary base stations. The research shows that increasing HAPS density allows for greater energy savings, but this saving is balanced against potential drops in per-user throughput.
- BLASTER Framework Extension
- BLASTER is an AI radio resource management framework adapted here to handle multi-HAPS deployments. It jointly optimizes user association, bandwidth allocation, and terrestrial base station activation to find the best balance between network performance and energy efficiency.
Terminology
Summary
A Traffic-Aware QoS-Energy Consumption Tradeoff Study for Integrated 6G TN & NTNs investigates how increasing HAPS density can significantly reduce terrestrial energy consumption while maintaining or improving quality of service compared to LEO satellite systems
Core Investigation and Motivation
The research extends the BLASTER framework to evaluate the trade-off between network quality of service (QoS) and terrestrial energy consumption in integrated terrestrial and non-terrestrial networks using different HAPS deployment densities. The study compares one low Earth orbit (LEO) satellite at 600 km with 1-, 4-, and 9-highaltitude platform station (HAPS) deployments at 20 km under a realistic diurnal traffic profile. p. 0 The motivation stems from the Hypercell concept, which introduces a new architectural perspective where a wide-area HAPS coverage layer assumes part of the responsibility traditionally assigned to terrestrial macro cells, allowing for dynamic deactivation of both terrestrial coverage and capacity cells based on traffic demand. p. 0
System Modeling and Optimization Framework
The system model considers a heterogeneous terrestrial cellular network partitioned into urban (Au) and rural (Ar) regions, with the total set of MBSs being the union of terrestrial (T) and non-terrestrial (S) MBSs. The channel model adopts the 3rd Generation Partnership Project (3GPP) terrestrial channel model and its extension to satellite and HAPS links. p. 1 The achievable downlink rate is defined by the formula Rj = Wj/kj log2(1 + γi), where Wj is the bandwidth available at the serving MBS. p. 1 The joint optimization problem maximizes the sum of logarithmic throughput (SLT) while penalizing terrestrial power consumption through a reweighted sparsity-promoting relaxation. p. 2
Performance Gains from Multi-HAPS Deployment
The results demonstrate that increasing the number of HAPS significantly improves performance metrics across all configurations. Specifically, SC4 achieves 38% and 268% higher values in average throughput per NT-served UE compared to SC3 and SC2, and approximately 20× higher than LEO (SC1). p. 3 These gains are attributed to two multiplicative effects: the bandwidth available per NT-served UE in SC4 is approximately 11× that of SC1 due to a larger NT bandwidth fraction and reuse across nine HAPS, and the shorter propagation distance of HAPS links increases average spectral efficiency by approximately 1. p. 36×.
Trade-off Analysis and Energy Savings
The analysis characterizes the QoS–energy consumption trade-off by showing that additional terrestrial energy savings through MBS shutdown can come at the expense of lower per-UE throughput due to increased loading and less favorable serving links. The aggregate terrestrial power reduction is quantified as ∆PT N (ε∗) = T − Mon(ε∗) ∆Poff, and the trade-off is shown by dR¯T N/d∆PT N < 0. p. 3 For instance, at the same SLT as SC1, SC4 reduces terrestrial energy consumption by an additional 28. p. 54% relative to SC1. p. 4 Conversely, forcing SC1 to deactivate as many MBSs as SC4 reduces its mean UE throughput by 50. p. 57%.
Conclusion of Findings
The study concludes that the 9-HAPS configuration reduces terrestrial energy consumption by 34.1% over the full day and 60. p. 06% at the quietest hour compared to LEO. p. 5 At the same SLT, this multi-HAPS configuration achieves 20% additional terrestrial energy savings relative to LEO while substantially improving the throughput of celledge UEs. p. 5 This demonstrates that enhanced NT offloading can reduce terrestrial infrastructure energy consumption. p. 5 The findings show that multi-HAPS deployments enable deeper terrestrial energy savings while substantially protecting cell-edge UEs. p. 5
Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.
Improvements for AI systems
-
Network Architecture Optimization for Edge Computing Systems: The improved system can dynamically configure a wide-area HAPS coverage layer to assume part of terrestrial macro cell responsibility, enabling both terrestrial coverage and capacity cells to be deactivated according to traffic demand, as suggested by the Hypercell concept.
-
Energy Efficient Resource Allocation Framework: The AI system can utilize a traffic-aware radio resource management framework like BLASTER extended for multi-HAPS deployments (1, 4, or 9 HAPS) to jointly optimize user association, bandwidth allocation, and terrestrial base station activation while minimizing the total energy consumption of integrated terrestrial and non-terrestrial networks.
-
Dynamic Terrestrial Infrastructure Management: The improved system can exploit NTN-assisted traffic offloading to deactivate lightly loaded terrestrial base stations, achieving significant energy savings while maintaining service continuity under time-varying traffic conditions, as shown by the finding that
additional MBS shutdowns translate into greater terrestrial power savings.
-
Throughput and Energy Trade-off Quantification: The system can mathematically characterize the QoS energy consumption trade-off by analyzing how increasing HAPS density affects terrestrial power reduction versus changes in per-UE throughput, allowing for precise determination of optimal operating points based on a reference QoS metric like SLT.
-
Predictive Traffic-Aware Scheduling: By solving the optimization problem independently for each hour of the day using traffic profile data (Table I), the AI can implement a predictive scheduling mechanism that anticipates low-traffic periods to maximize terrestrial energy savings before they occur.
Sources
- Bridging Earth and Space: A Survey on HAPS for Non-Terrestrial Networks
- Optimizing Integrated Terrestrial and Non-Terrestrial Networks Performance with Traffic-Aware Resource Management
- HAPS as a Hypercell: Enabling Coverage and Capacity Carrier Shutdown in Cellular Networks
Related papers
- Open-H-RF: A Large-Scale Ultrasound Raw Channel Dataset Initiative
- Scalable Attention for 5G NR Channel Estimation
- Low-Altitude Wireless Networks: A Comprehensive Survey
- Sensing-Assisted OFDM with Index Modulation: A Waveform for Optical Wireless Integrated Communication and Localisation
- Iteration-Free Candidate Pruning and an Index-Domain VLSI Architecture for Sort-Free MIMO Box Decoding up to 1024-QAM
- A Foundation Model for Large-Scale Wireless Network Planning, Operation and Optimization