Introducing Energy Efficient Routing in UAV-Satellite NTNs for Dynamic 6G Interconnectivity

IEEE TRANSACTIONS ON COMMUNICATIONS 1 Introducing Energy Efficient Routing in UAV-Satellite NTNs for Dynamic 6G Interconnectivity George Amponis, Graduate Student Member, IEEE, Thomas Lagkas, Senior Member, IEEE, Pavlos Bouzinis, Panagiotis Radoglou-Grammatikis, Member, IEEE, Antonios Sarigiannidis, Panagiotis Sarigiannidis, Member, IEEE, and Vasileios Argyriou Abstract—The integration of Unmanned Aerial Vehicles (UAVs) and Low-Earth Orbit (LEO) satellites as aerial nodes in non-terrestrial networks (NTNs) presents both opportunities and challenges for on-demand 6G interconnectivity. This paper presents a new Composite Cost Metric (CCM) which improves energy-efficient routing performance in combined UAV-satellite constellations. We consider incorporating cumulative Free Space Path Loss (FSPL) and residual energy into the route selection process for both proactive and reactive protocols, our approach refines the routing decisions of classical protocols. The proposed CCM-driven modifications and protocol-specific integration typologies can improve overall route stability, reduce energy consumption per delivered packet, and optimize network reliability by dynamically selecting relays with lower attenuation and higher energy availability. We develop an NS-3-based simulation framework that integrates realistic satellite orbital mechanics, UAV mobility models, and a hybrid energy model that includes solar energy harvesting for satellites. Simulation results demonstrate that our enhancements can indeed outperform baseline implementations in packet delivery ratio, energy efficiency, and end-to-end delay which makes them viable for next-generation NTN-supported 6G networks, at the expense of some additional control overhead. With this set of developments we aim to pave the way for global-optimum and energy-aware emergency and disaster relief communications. Index Terms—LEO Constellations, Multi-Altitude Flying Adhoc Networks (FANETs), Aerial Base Stations, Satellite-Drone Relaying, Energy-Aware Routing, Non-Terrestrial Networks, 6G Connectivity. This project has received funding from the European Union’s Horizon Europe research and innovation programme under grant agreement No. 101097122 (ACROSS). Disclaimer: Funded by the European Union. Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or European Commission. Neither the European Union nor the European Commission can be held responsible for them. Corresponding author: Pavlos Bouzinis George Amponis is with the Dept. of R&D, K3Y Ltd., Sofia, Bulgaria, and the Dept. of Informatics, Democritus University of Thrace, Kavala Campus, Greece (email: [email protected], [email protected]). Thomas Lagkas is with the Dept. of Informatics, Democritus University of Thrace, Kavala Campus, Greece (e-mail: [email protected]). Pavlos Bouzinis is with MetaMind Innovations P.C., 50100 Kozani,Greece, Thessaloniki, Greece (e-mail: [email protected]). Panagiotis Radoglou-Grammatikis is with the Dept. of R&D, K3Y Ltd., Sofia, Bulgaria, and the Dept. of Electrical and Computer Engineering, University of Western Macedonia, Kozani, Greece (e-mail: [email protected], [email protected]). Antonios Sarigiannidis is with the Dept. of R&D, K3Y Ltd., Sofia, Bulgaria (e-mail: [email protected]). Panagiotis Sarigiannidis is with the Dept. of Electrical and Computer Engineering, University of Western Macedonia, Kozani, Greece (e-mail: [email protected]). Vasileios Argyriou is with the Dept. of Networks and Digital Media, Kingston University, London, UK (e-mail: v[email protected]). I. INTRODUCTION THE integration of Unmanned Aerial Vehicles (UAVs) into Low-Earth Orbit (LEO) satellite networks represents a critical threshold for next-generation wireless solutions. Strategic relay selection in these hybrid 6G environments creates a paradigm that enhances communication range on-demand while maximizing network capabilities. Our work studies how optimizing relay selection enhances performance—predominantly from an energy perspective—alongside overall communication network robustness in UAV-LEO Multi-Altitude Constellations. We delve into the potential trade-offs, networking implications, and solutions to relay selection optimization problems that the industrial and research landscape will face as on-demand and mission-critical communications become prevalent. UAVs operating as network nodes have transformed traditional mobile base stations into dynamic elements for communication networks. LEO satellites represented by Starlink with its specific orbital parameters demonstrate a solid answer to the increasing need for widespread high-speed connectivity inside 6G networks. High-altitude platforms (HAPs) together with FANETs form a complicated yet optimally efficient system for future communication requirements when integrated with satellite constellations. The successful implementation of this system depends on optimized relay selection that accounts for UAV battery performance together with signal propagation delay and the changing dynamics of aerial and spaceborne nodes. Mobile Ad-hoc Networks (MANETs) have brought significant changes to decentralized and dynamic deployments by revolutionizing their communication capabilities. The introduction of hybrid UAV and LEO satellite networks into advanced technological frameworks requires upgrading traditional MANET routing protocols, which tend to encounter significant challenges. a) Dynamic Topology, High Mobility, and Energy Constraints: Hybrid NTNs exhibit extreme node mobility and continuously shifting topologies, making routing a complex and energy-intensive task. Traditional MANET protocols struggle to maintain consistent route information due to frequent topology changes, leading to excessive control overhead, route instability, and increased energy consumption. In UAV-LEO constellations, the high velocity of satellites and intermittent connectivity introduce additional challenges, exacerbating link This article has been accepted for publication in IEEE Transactions on Communications. This is the author's version which has not been fully edited and content may change prior to final publication. Citation information: DOI 10.1109/TCOMM.2025.3634253 This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/ IEEE TRANSACTIONS ON COMMUNICATIONS 2 disruptions and requiring frequent route recalculations. Conventional routing strategies fail to account for the compounded impact of mobility and energy constraints, where UAVs must optimize transmission power and relay selection to sustain network longevity. Addressing these issues necessitates routing mechanisms that integrate predictive topological adaptation with energy-aware link selection, ensuring stability while minimizing energy expenditure across dynamic, multi-altitude architectures. b) Heterogeneous Link Characteristics and EnergyOptimized Routing: UAV-LEO networks operate over highly heterogeneous links, combining UAV-based ad hoc relaying with satellite-based long-range connectivity. These network segments exhibit distinct physical and energy-related constraints: UAVs are energy-constrained and require powerefficient routing strategies, while LEO satellites introduce high-latency, dynamically shifting link characteristics influenced by orbital mechanics. Conventional routing protocols fail to integrate these disparate constraints, leading to suboptimal path selection that disregards both energy sustainability and link stability. Effective routing solutions for UAV-LEO networks must incorporate composite cost metrics that jointly consider free-space path loss, energy availability, and link reliability, ensuring that relay selection maximizes transmission efficiency while extending the operational lifespan of energy-limited nodes. Energy efficiency is critical, particularly for UAVs implementing communication relaying in critical environments [1]. The majority of relevant routing protocols fail to prioritize energy optimization when making decisions which results in sub-optimal path selection energywise. Energy-aware routing decisions represent an essential requirement because they determine how long UAVs survive in the network and how effective they remain throughout their operation. The vast separation between satellites and ground nodes creates long delays and unpredictable bandwidth conditions and special topology limitations. Traditional protocols are not designed to handle discrepancies stemming from the ”marriage” of aerial ad hoc networks with space-born counterparts, with a distance spanning hundreds of thousands of meters, which often leads to inefficient routing decisions and poor utilization of the available resources [2]. Critical applications enabled by UAV-LEO network hybrids need smart protocols to achieve a ”global optimum”. Increased mobility, scale, and link parameter heterogeneity demand advanced solutions capable of adapting to changing network conditions while ensuring efficient, secure, and reliable communication. This sets the stage for the development of innovative routing protocols that leverage the CCM to intelligently determine the optimal path, effectively balancing the trade-offs between immediacy, reliability, and resource consumption. Controlmessage enhancements specific to UAV-LEO satellite communications were also considered by D. Shumeye Lakew et. al. [3] as a means of bringing about seamless coverage in 6G. For comparative baselines, we include Ad-hoc On-demand Multipath Distance Vector (AOMDV) [4], [5] and a CyberPhysical Routing protocol exploiting Trajectory Dynamics (CPR-TD) [6]. The main contributions of this paper consist of three parts which focus on improving energy-efficient routing mechanisms for UAV-assisted cellular networks that integrate LEO satellites. First, we introduce enhanced versions of the OLSR and AODV protocols, integrating our CCM to enable multi-criteria route selection based on residual energy and estimated link quality; these modifications do indeed go beyond traditional routing protocol implementations, embedding energy-awareness and link attenuation directly into routing decisions (calculated discreetly in each case) with the purpose of addressing the nature and challenges of heterogeneous NTNs. Second, we propose a comprehensive simulation framework built on the NS-3 platform, incorporating real-world satellite orbital dynamics (particularly regarding the Starlink constellation), UAV mobility patterns, and a new energy model, providing a robust testbed for evaluating the performance of energy-efficient routing protocols in diverse scenarios. Third, as hinted above, we extend the native NS-3 core energy source and model classes (discussed in detail in Subsection V), and introduce a composite model which integrates the standard energy source with a solar energy harvesting mechanism which is applicable to LEO-satellite constellations for greater realism. The described contributions form a platform to evaluate routing in NTN scenarios. We begin by developing our proposal from OLSR and AODV before implementing our new metric for enhancement. The modifications made to OLSR and AODV protocols use their fundamental structures to incorporate the CCM in different ways. The CCM metric in OLSR gets implemented through the willingness parameter which leads to MPR selection modifications based on link and energy parameters and replaces the ETX metric which we consider as a baseline). Similarly, AODV uses CCM as a replacement for ETX in RREQ and RREP messages to enable nodes to select routes based on their cumulative FSPL and residual energy during route discovery operations. Our performance evaluation uses AODV-ETX and OLSRETX as primary baselines, and includes AOMDV and CPRTD as additional state-of-the-art comparators, to demonstrate CCM metric advantages across diverse routing paradigms. The selection of AODV-ETX over vanilla AODV is deliberate because ETX provides an effective link reliability assessment through expected transmission count estimation which better suits dynamic UAV–satellite environments. Likewise, OLSRETX offers a fairer baseline than hop-count-only OLSR when assessing link-aware energy efficiency. II. RELATED WORK D. Huo, Q. Liu, Y. Sun, and H. Li in [7] focus on optimizing network performance in large-scale satellite swarm systems. The paper addresses challenges like fast node movement, long communication distances, and dynamic topologies. The paper proposes an inter-satellite routing protocol, IS-OLSR, based on the traditional MANET protocol OLSR. This protocol aims to improve network performance by implementing subnet division, abstraction, and gateway election. IS-OLSR has been shown through simulations to significantly reduce routing overhead as well as data transmission delay. The study is relevant in the context of satellite network communication technologies, especially in fields such as crossregion communication, disaster relief, and national defense. This article has been accepted for publication in IEEE Transactions on Communications. This is the author's version which has not been fully edited and content may change prior to final publication. Citation information: DOI 10.1109/TCOMM.2025.3634253 This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/ IEEE TRANSACTIONS ON COMMUNICATIONS 3 H. Liming et al in [8] propose a novel load-balancing strategy for routing in LEO satellite constellations. Their approach focuses on optimizing the traffic distribution across the network by considering historical traffic data of satellite links. This method significantly improves network congestion and packet loss issues in high-traffic areas, enhancing overall network performance. The paper emphasizes the relevance of managing traffic effectively in the dynamic topology of LEO satellite networks, particularly for space-based IoT networks. The authors employ simulation studies to demonstrate the efficacy of their proposed load-balancing method in reducing delay performance degradation and packet loss under high-load conditions. Z. Liu et al. in [9] composed a research which focuses on addressing the challenges of uneven satellite network load and unstable link connections. The authors propose a globally adaptive satellite network routing strategy, G-AODV, which enhances the existing Ad hoc On-Demand Distance Vector Routing (AODV) protocol by incorporating a traffic prediction mechanism during the route discovery phase. This startegy aims to avoid heavily loaded nodes from becoming intermediate nodes and introduces a path replacement strategy to replace paths before node congestion occurs, achieving load balancing. The paper also discusses the highly volatile (and occasionally completely unpredictable) nature of satellite link instability and its impact on maintaining a consistent communication environment. The study is significant in improving satellite network routing efficiency, particularly in dynamic and complex network environments. S. Suhaimi, K. Mamat, and K. Daniel Wong in [10] propose a method to improve the Optimized Link State Routing (OLSR) protocol in mobile ad-hoc networks (MANETs) by considering battery power status. The study emphasizes the relationship between power status and OLSR’s functionality, particularly in maximizing the use of battery power sources. It involves modifications to the OLSR source code to enhance node ’willingness’ based on battery power status, demonstrating the improved utilization of battery resources in MANETs through experiments. The paper also discusses the significance of energy awareness in wireless ad hoc networks and proposes methods for optimizing power usage in OLSR, especially when nodes rely on battery power in infrastructure-free scenarios. S. Jiao et. al. in [11] focus on the deployment of LEO satellite networks with intersatellite links. These networks have gained significant attention due to their high throughput, wide coverage, and cost-effective control and management. This paper particularly emphasizes the role of these networks in supporting 5G mobile services (and potentially 6G in the future), acting as a space bearer to connect distant users with the ground core network. The authors propose an architecture for the space bearer network and develop routing protocols and signaling methods to address challenges like large-scale flooding and prolonged routing convergence. Moreover, they introduce coordinated signaling techniques across different network segments to ensure Quality of Service (QoS) consistency. The experimental validation of their routing design demonstrates successful User Equipment (UE) to UE connections, highlighting the paper’s contribution to advancing satellite-based 5G networks. To position our study within broader NTN and SpaceAir-Ground Integrated Network (SAGIN) developments, we reference recent works on real-time large-scale satellite–UAV systems [12], maritime hybrid Satellite–UAV–terrestrial architectures [13], and cell-free Satellite–UAV networks for widearea IoT [14]. From a UAV communications viewpoint, the UAV-to-Everything (U2X) paradigm [15] generalizes beyond D2D and motivates our protocol-agnostic CCM integration across proactive and reactive stacks. Finally, energy-aware satellite–terrestrial computing [16] and LEO-focused SAGIN outlooks [17] further underline the importance of jointly optimizing link quality and energy sustainability in heterogeneous non-terrestrial networks. Recent advancements in NTN and Flying Ad-Hoc Networks (FANETs) have further highlighted the need for sophisticated routing strategies. Dong et al. [18], [19] provide stochasticgeometry models for analyzing uplink performance in heterogeneous non-terrestrial networks, offering critical insights under harsh conditions. For highly dynamic FANETs, Yang et al. [20] proposed a betweenness centrality-based DSR variant, while Wang et al. [21] introduced a hybrid proactive–reactive ant-colony routing protocol with link quality prediction. Furthermore, Hu et al. [6] developed a cyberphysical routing protocol that leverages trajectory dynamics for mission-oriented FANETs. These studies underscore the trend toward more intelligent, context-aware routing, yet they often focus on either the satellite or the aerial layer in isolation. Our CCM complements this body of work by providing a protocol-agnostic, pluggable metric that captures a global energy–attenuation optimum across heterogeneous links. A significant research gap persists in the integration of these advancements with the dynamic and energy-intensive characteristics of hybrid networks between UAV deployments and LEO satellite constellations. Current methodologies often overlook the compounded complexity brought by the integration of aerial and satellite components, especially in terms of energy efficiency and the adaptability of routing protocols in high-mobility environments. This work, introducing CCM integrated into routing protocols with different core route selection algorithms (namely AODV and OLSR in this study), addresses this gap because it uniquely combines and couples considerations for energy efficiency, link stability, and the distinct attributes of satellite and UAV communications in scenarios of communication range extension. III. BACKGROUND This section is dedicated to the analysis of two considered ad hoc routing protocols, both of which have constituted the basis of a significant amount of developments targeting similar environments. Moreover, this section targets the analysis of matters relevant to LEO satellite constellations’ orbital mechanics and the modelling of FSPL for such wireless links. Figure 1 visualizes the envisioned communications relaying paradigm leveraging dynamic satellite backhauls. The choice of AODV-ETX and OLSR is deliberate: AODV-ETX, an advanced version of AODV, focuses on link quality, similar to CCM’s goals while the default implementation of OLSR offers a proactive routing perspective, contrasting AODV’s reactive This article has been accepted for publication in IEEE Transactions on Communications. This is the author's version which has not been fully edited and content may change prior to final publication. Citation information: DOI 10.1109/TCOMM.2025.3634253 This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/ IEEE TRANSACTIONS ON COMMUNICATIONS 4 Wireless backhaul UAVs service area Sat. service area LEO satellites High altitude UAV Figure 1: Envisioned use-case of hybrid UAV-LEO satelliteenabled relaying nature. Header overheads comparability was also a factor we considered, in order to ensure a fair assessment of CCM’s impact on network performance. In the context of both OLSR and AODV (and their subsequent enhancements), we refer to networked entities as ”nodes”. This includes UEs, UAVs, and LEO satellite relays. The modeled system represents a hybrid network topology comprising the above-mentioned entities, interconnected through a single routing protocol across all link types. UEs are ground-based devices that communicate with UAVs, which act as aerial relays, while LEO satellites provide backhaul connectivity and facilitate inter-node communication across broader distances. The network operates with a hierarchical structure where UAVs connect UEs to satellites or other UAVs, depending on the route selected. Path selection is influenced by the composite cost metric. a) Expected Transmission Count (ETX): Traditional routing protocols often use hop count as the primary metric for route selection. However, hop count doesn’t account for the link quality. ETX provides a more nuanced metric that considers the loss rate of links, which can lead to the selection of more reliable and higher-throughput routes. ETX considers the transmission success rate, factoring in aspects like packet loss, which is vital when dealing with varying signal qualities in UAV networks. Implementing ETX in AODV for UAVs would ensure that routes are not just the shortest but the most reliable, leading to enhanced data transmission quality. We use ETX [22] as the link-quality baseline. For a link (u, v)with forward and reverse delivery ratios (df, dr), the per-link ETX is ETX(u, v) = 1 df·dr , df, dr∈(0,1].(1) For a path p, the cumulative cost is the sum of link ETX values. Integration follows protocol semantics: in OLSR-ETX, Dijkstra runs on ETX-weighted links; in AODV-ETX, RREQs accumulate per-link ETX and the RREP returns the minimumETX route. A. The OLSR Protocol OLSR is a proactive routing protocol for mobile ad hoc networks, standardized in RFC 3626 [23]. Unlike AODV (analyzed in the next subsection), which seeks routes on demand, OLSR continuously maintains routes to all destinations in the network. It achieves efficiency through the use of Multipoint Relays (MPRs) to minimize flooding of control traffic. MPRs are selected nodes that forward broadcast messages during the flooding process. This reduces the number of transmissions required and optimizes the control traffic overhead. Equation (3) formally describes the MPR selection process for two-hop neighbors. For multi-hop routing, this principle is extended by recursively selecting MPRs to cover all nodes in the network, forming a tree that minimizes control traffic overhead while maintaining full routing information, as detailed in the standard. Equation (3) formally describes the MPR selection process using set theory, where MPR(s)is the set of Multipoint Relays selected by node s. Firstly, let N(s)denote the one-hop neighbors of a node s. Following that, N2(s)denotes the two-hop neighbors of s, which are the neighbors of s’s neighbors but not directly connected to s. More formally, N2(s)can be defined as N2(s) =  [ m∈N (s) N(m) \ N(s),(2) indicating that N2(s)consists of the neighboring nodes of s’s neighbors, i.e., N(m), m ∈ N(s), while excluding nodes that can be reached from s, i.e., N(s). The optimal MPR∗(s) set is described as the smallest possible subset M ⊆ N(s), such that all two-hop neighbors of scan be reached through the subset of one-hop neighbors M. This gives rise to the following minimization problem MPR∗(s) = argmin M⊆N(s)(|M|,s.t., [ m∈M N(m)⊇ N2(s)), (3) where |M| denotes the cardinality of M. a) Efficiency in Dynamic Networks: The dynamic nature of UAV-enabled networks, characterized by frequent topological changes, demands an adaptive and efficient routing protocol. OLSR, with its proactive routing strategy, is wellsuited for such environments. However, the key to its efficiency lies in the strategic selection of Multipoint Relays (MPRs). In UAV-assisted networks, MPRs are dynamically chosen based on current network conditions and node mobility. This approach ensures that the control traffic is minimized, and the routes are quickly established and maintained despite potential changes in nodes’ locations. b) Control Messages in OLSR: OLSR disseminates local and global topology using periodic control messages: 1) HELLO: Link sensing and neighbor discovery/status. 2) TC (Topology Control): Advertises MPR selectors to build the topology graph. 3) MID: Declares multiple interfaces when present. This article has been accepted for publication in IEEE Transactions on Communications. This is the author's version which has not been fully edited and content may change prior to final publication. Citation information: DOI 10.1109/TCOMM.2025.3634253 This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/ IEEE TRANSACTIONS ON COMMUNICATIONS 5 c) Route Computation Model: The route calculation in OLSR can be mathematically modeled as finding the shortest path in a graph. In the hop-count formulation, this is described as Route(s, d) = min p∈Ps,d {Length(p)},(4) where Route(s, d)is the shortest path from sto d,Ps,d denotes the set of all possible paths from sto d, and Length(p)is the length of the path p, calculated using hop count. d) Scalability and Control Overhead: As networks grow in scale, managing control traffic becomes crucial to maintain network performance. OLSR’s MPR system inherently reduces the number of transmissions required for route discovery and maintenance. However, in large-scale UAV networks, even this optimized approach can lead to significant overhead. Techniques such as adaptive control message intervals, MPR selection based on node density, and region-based MPR clustering can be explored to enhance scalability and reduce control traffic further. B. The AODV Routing Protocol The AODV routing protocol, standardized in RFC 3561 [24], is designed for mobile ad hoc networks and provides an efficient routing mechanism that establishes routes between nodes only as desired by source nodes. It belongs to the family of Distance Vector routing protocols and uses a sequence number to ensure the freshness of routes and prevent routing loops. On a message-exchange level, the protocol works as follows. a) Efficiency in Dynamic Networks: AODV exhibits particular efficiency in dynamic networks, a characteristic paramount in the volatile topology of UAV-assisted 6G environments and LEO satellite networks. AODV’s on-demand nature allows it to adapt swiftly to frequent changes, as routes are established only when required, significantly reducing the overhead associated with maintaining a full map of the network at all times. By minimizing the route discovery frequency through the utilization of sequence numbers and maintaining up-to-date routes to active destinations, AODV can effectively manage the balance between the need for up-todate path information and the desire to minimize signaling traffic. Furthermore, its ability to quickly respond to link breaks with Route Error (RERR) messages and efficiently establish new routes through Route Requests (RREQs) and Route Replies (RREPs) makes it well-suited for environments where network topology is subject to frequent and unpredictable changes. As such, AODV stands out as a robust solution for ensuring communication reliability and maintaining consistent performance in highly dynamic and mobile network settings. b) Control Messages in AODV: AODV operation centers on three control messages: 1) RREQ: Route discovery broadcast from the source when no valid route is known. Includes source/destination addresses, broadcast ID, and sequence numbers. 2) RREP: Unicast reply from destination or an intermediate node with a fresh enough route; carries destination sequence number and hop count. 3) RERR: Error notification when a link break invalidates active routes, triggering local repair or new discovery. c) Route Computation Model: As seen in Equation (5), the path discovery process is a simple optimization problem, where AODV selects the path with the minimum cost. Route(S, D) = min p∈PSD {Cost(p)}(5) where: Route(S, D) = Optimal path from Sto D PSD =The set of all possible paths from Sto D Cost(p) = Cost of p, calculated based on hop count. AODV maintains routes as long as they are active. If a link break is detected, a RERR message is used to notify affected nodes. In LEO satellite constellations and similar high-mobility environments: 1) Adaptiveness: ETX adapts to changing link conditions in real-time, providing a more reliable and consistent network experience. 2) Throughput Optimization: By avoiding low-quality links, ETX can improve end-to-end throughput, which is crucial for bandwidth-sensitive applications. 3) Approximation of cross-layering: The ETX metric builds upon statistical metrics inherent to the routing layer and presents a statistically significant tool to compute the expected number of (re)transmissions at the Medium Access Control (MAC) layer. d) Scalability and Control Overhead: AODV scales well in dynamic settings by avoiding global state, but frequent discoveries and repairs increase signaling under high mobility. Using ETX as the baseline path metric reduces retransmissions and helps contain overhead; CCM integration further stabilizes routes at the cost of larger control payloads. Parameter tuning (e.g., ring search TTL, local repair) is important to balance responsiveness and overhead. C. Orbital Mechanics Overview For completeness, we briefly outline the orbital-mechanics concepts that underpin the LEO satellite positions used in our simulations. Each satellite follows a nearly circular low-Earth orbit (Starlink-like, altitude h= 550 km). The instantaneous radius is therefore r≈RE+hwhere RE= 6378 km is Earth’s radius. The orbital period follows from Kepler’s third law, T= 2πr(RE+h)3 GM , with GM = 3.986 ×1014 m3s−2, yielding T≈95 min. Satellite positions at simulation time tare obtained from the true anomaly ν(t) = 2πt/T and standard ECI rotations (see Figure 2). The large slant range between a UAV at altitude 300 m and a LEO satellite motivates the use of the Friis free-space pathloss model which, for carrier frequency f= 1.8 GHz and range d, is FSPLdB = 20 log10(d) + 32.44 + 20 log10(fGHz).(6) This article has been accepted for publication in IEEE Transactions on Communications. This is the author's version which has not been fully edited and content may change prior to final publication. Citation information: DOI 10.1109/TCOMM.2025.3634253 This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/ IEEE TRANSACTIONS ON COMMUNICATIONS 6 X (East) 1e7 1.00 0.75 0.50 0.25 0.00 0.25 0.50 0.75 1.00 Y (North) 1e7 1.00 0.75 0.50 0.25 0.00 0.25 0.50 0.75 1.00 Z (Up) 1e6 6 4 2 0 2 4 6 X (East) Y (North) Z (Up) Satellite UAV = 330° Satellite and UAV Positions in 3D Space Figure 2: Visualization of LEO satellite on a circular orbit with a 330 deg true anomaly (not true to scale) These FSPL values feed directly into the cost metric (Eq. 7) and thus influence route selection without requiring more intricate propagation models. Sensitivity to shadowing and fading is captured separately by the physical-layer error model in NS-3 as discussed in Section V. IV. PROPOSED METHOD This section outlines a proposed routing protocol designed to tackle the intrinsic challenges posed by such a complex system. Central to this approach is the implementation of the CCM, which dynamically calculates and selects the optimal path for data transmission. Unlike traditional methods, this advanced protocol doesn’t just consider a single metric but rather a combination of several crucial factors, each representing a different dimension of the communication link. The CCM incorporates elements such as energy efficiency, link stability, and the specific characteristics of satellite and UAV links. By considering end-to-end delay, the protocol ensures that time-sensitive information is prioritized, addressing the critical latency issues that often plague satellite communications. Energy efficiency is another vital component, particularly for UAVs where battery life is a premium. Optimizing routes based on energy consumption can significantly prolong operational time and enhance the overall sustainability of the network. Link stability is a dynamic factor, especially in a network characterized by high mobility and variable environmental conditions. The CCM takes into account the stability of each link, preferring routes that offer consistent performance over those that might lead to frequent disconnections or require constant rerouting. This aspect is particularly crucial in maintaining a reliable communication standard, essential for applications requiring a consistent data stream. In scenarios where direct UAV-to-UAV communication is feasible and efficient, the system might minimize the use of satellite links, reducing latency and preserving bandwidth for scenarios where such links are indispensable. Conversely, when satellite paths transiently offer higher stability or lower end-to-end delay, the stack biases decisions toward the LEO backhaul, exploiting wide-area coverage and predictable geometry. This adaptivity suits heterogeneous missions where requirements vary over time and a network-level optimum (not a myopic per-hop choice) is preferred. Our contribution is not a new protocol but a pluggable CCM with reference integrations in OLSR and AODV for hybrid NTN deployments. Implementation is via a software shim that hooks the metric interface without altering base state machines or timers. Required telemetry (residual energy normalized to [0,1], link/path-loss in dB, optional timestamp) is conveyed in a TLV (Type–Length–Value) piggybacked on existing control messages—HELLO/TC for OLSR and RREQ/RREP for AODV. The TLV uses a private type code and compact encoding; legacy nodes ignore unknown TLVs and continue with their baseline metric, preserving backward compatibility in mixed fleets. CCM-capable nodes parse the TLV, update per-neighbor state, and apply CCM at next-hop/route selection. No base headers are repurposed, checksums remain valid, and an optional capability bit can advertise CCM support. The CCM and its associated protocols are designed to adapt to rapid network topology and node state changes. In AODVCCM, the on-demand nature of route discovery ensures that the most current link and energy state information is used when establishing a new route. In OLSR-CCM, the periodic exchange of HELLO and TC messages, now augmented with CCM data, allows the network to continuously adapt to changes, although there is a risk of routing update delays if the update interval is not tuned to the mobility speed. Future work could explore dynamically adjusting the update interval based on network volatility. While we focus on residual energy, the CCM framework is extensible; other energy management strategies, such as incorporating energy consumption rate or prioritizing nodes with more stable energy sources, could be integrated by adding new terms to the metric. This would allow for even more nuanced and context-aware routing decisions. 1) The Composite Cost Metric: We define a Composite Cost Metric to balance link attenuation and node battery constraints in path selection. Although we primarily focus on the more significant losses between UAVs and LEO satellites, we also include inter-UAV link losses (which are typically smaller or practically negligible comparatively) for completeness. Let ndenote the total number of links (or hops) in a candidate endto-end route. For each link i∈ {1, . . . , n}, we first calculate its free space path loss FSPLiin decibels (dB) using (6) and we subsequently proceed with the summation of these dB values is a convenient way to represent total path attenuation: FSPLpath(dB) = n X i=1 FSPLi(dB).(7) To ensure dimensional consistency with energy-based metrics in our final cost function, we convert this total path loss from This article has been accepted for publication in IEEE Transactions on Communications. This is the author's version which has not been fully edited and content may change prior to final publication. Citation information: DOI 10.1109/TCOMM.2025.3634253 This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/ IEEE TRANSACTIONS ON COMMUNICATIONS 7 dB into the factor: Lpath = 10FSPLpath(dB)/10.(8) Hence, rather than converting each link’s loss to linear individually, we simply exponentiate the summed dB values once to obtain Lpath. Residual Energy: Next, we capture the average battery status of the nodes along this path (including ground relays or UAVs). Let Eibe the current residual energy of node i, and let Eimax denote its maximum energy capacity. To quantify how depleted or sustainable the route’s relays are, we define Er=1 n n X i=1Ei Eimax ×100%,(9) which represents the average residual energy fraction (as a percentage) across the nnodes participating in the path. Definition of CCM: Finally, we combine these two factors—path attenuation and residual energy fraction—into a cost metric: CCM = Lpath Er .(10) Here in Equation (10), Lpath is the aggregate linear pathloss factor, and Eris the percentage of residual energy. A high CCM indicates either large cumulative attenuation or low node energies (or both), making the path less desirable. Although FSPL does not inherently depend on energy, this formulation allows us to penalize routes that either have very poor signal quality or are likely to lose relaying nodes prematurely due to battery depletion. Since Lpath is dimensionless (obtained from converting dB to linear scale) and Eris also dimensionless (expressed as a percentage), CCM becomes a pure, dimensionless score for ranking candidate end-toend routing paths. While no direct physical law couples path loss and node energy, combining them in one metric ensures that routing decisions avoid both high-attenuation links and severely depleted nodes, thereby improving overall network reliability. A. Enhancing AODV with the CCM Metric a) Message Structure Enhancement: The RREQ and RREP messages in AODV have been modified to accommodate the components of the CCM. For the RREQ message, the original structure: [Type | Flags | Hop Count | RREQ ID | Destination IP | Destination Sequence | Origin IP | Origin Sequence] is expanded to include fields for FSPL and residual energy. This extension allows each node to compute and forward the FSPL and the residual energy of nodes along the proposed route. The enhanced RREQ structure is [Type | Flags | Hop Count | RREQ ID | Destination IP | Destination Sequence | Origin IP | Origin Sequence | FSPL | Avg Residual Energy]. The structure of the RREQ messages can be seen in Figure 3. For the RREP message, the key enhancement is the inclusion of a field to carry back the aggregated CCM value of the route. The CCM is a crucial element for the route selection process, as it combines the FSPL and residual energy metrics 0123 01234567890123456789012345678901 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Type |J|R|G|D|U| Reserved | Hop Count | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | RREQ ID | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Destination IP Address | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Destination Sequence Number | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Originator IP Address | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Originator Sequence Number | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | FSPL | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Residual Energy | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Figure 3: AODV-CCM Route Request (RREQ) Packet Structure 0123 01234567890123456789012345678901 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Type |R|A| Reserved |Prefix Sz| Hop Count | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Destination IP address | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Destination Sequence Number | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Originator IP address | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Lifetime | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | CCM | +-+-+-+--+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Figure 4: AODV-CCM Route Reply (RREP) Packet Structure for the entire route. The revised RREP structure will include a field for the CCM value, resulting in [Type | Flags | Prefix Size | Hop Count | Destination IP | Destination Sequence Number | Origin IP | Lifetime | CCM]. The RREQ conveys individual link metrics (FSPL and energy) while the RREP carries the aggregate route CCM, allowing for more informed and efficient routing decisions. The structure of the RREP messages can be seen in Figure 4. b) Route Calculation Mechanism and Packet Loss Model: In our enhanced AODV protocol, which now incorporates the CCM, route selection is systematically guided by both link attenuation and node energy constraints. Figure 5 outlines the step-by-step process: 1) Reception of RREQ: The process begins when a node receives a RREQ message carrying two key parameters: (i) the cumulative FSPL along the path from the origin node to the current node, and (ii) the average residual energy of all intermediate nodes so far. 2) Computation of CCM: Upon receiving a RREQ, each node computes the CCM for the partial route from the source to itself. This calculation aggregates the total FSPL (as a proxy for link quality) and the average residual energy of the path’s relays. Nodes with higher path loss or depleted batteries yield a higher CCM, indicating a less favorable route. 3) Forwarding the RREQ: If the node is not the final destination, it appends its own local FSPL and residual energy information and forwards the updated RREQ. This article has been accepted for publication in IEEE Transactions on Communications. This is the author's version which has not been fully edited and content may change prior to final publication. Citation information: DOI 10.1109/TCOMM.2025.3634253 This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/ IEEE TRANSACTIONS ON COMMUNICATIONS 8 Consequently, subsequent nodes can recalculate the CCM with more complete path information. 4) Generation of Route Reply (RREP): When the RREQ reaches the destination or an intermediate node with a valid route, a RREP is generated. This message includes the final, aggregated CCM value for the entire path. 5) Route Selection Based on CCM: As RREP messages travel back toward the source, each intermediate node updates its routing table and prioritizes paths with the lowest CCM. This ensures that final route decisions favor energy-efficient, lower-attenuation paths. Enhanced AODV with CCM: Detailed Route Discovery Source Source Intermediate Node 1 Intermediate Node 1 Intermediate Node n Intermediate Node n Destination Destination RREQ from Source to Node 1 RREQ [FSPL=0, E_r = S.Energy%] (1) The Source initializes FSPL=0 and sets its local residual energy fraction. Compute partial CCM partial_CCM = (FSPL + FSPL_N1link) / updated_Er Store in route table with (previous hop = S). alt [N1 has an active route to D?] Generate RREP immediately (optional case) If Node 1 already knows a valid route to D, it can respond with RREP right away. RREP [CCM from cached route] Forward RREQ [Accum FSPL, E_r] RREQ from Node 1 to Node n Compute partial CCM partial_CCM = (accumulated FSPL + FSPL_Nnlink) / updated_Er Update route table with partial CCM. alt [Nn has an active route to D?] Generate RREP immediately RREP [Cached route CCM] Same logic: if Node n can reach D, reply now to save another hop. Forward RREQ [Final FSPL, E_r] At Destination Compute final path FSPL & E_r CCM = FSPL_path / E_r Fill RREP with CCM RREP Back to Node n RREP [CCM, Destination= D] Node n updates route table (destination = D, CCM value). RREP [CCM] Similarly, Node 1 updates its table for destination D with new CCM. RREP [CCM] Final Route Choice at Source Compare CCM among all RREPs Select route with lowest CCM The Source installs the best path (lowest CCM) in its route table. Figure 5: Sequence Diagram of the CCM-enhanced AODV route establishment process While CCM employs FSPL as a surrogate for link quality, packet-loss events in our ns-3 simulations are determined by the underlying physical-layer error model, specifically the YansErrorRateModel. In this model, the receiver’s signal-to-noise ratio (SNR) is calculated by subtracting the cumulative FSPL from the transmit power, incorporating additional attenuation factors such as fading. The error rate model then applies modulation-specific BER equations to probabilistically determine whether each incoming packet is successfully decoded or lost. Higher FSPL naturally results in lower SNR, thereby increasing the likelihood of packet drops. Consequently, CCM indirectly influences the packet delivery ratio (PDR): routes with excessive path loss incur 0123 01234567890123456789012345678901 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Reserved | Htime | Willingness | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Link Code | Reserved | Link Message Size | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Neighbor Interface Address | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | FSPL | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Residual Energy | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Neighbor Interface Address | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | FSPL | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Residual Energy | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ : . . . : : : +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Figure 6: Enhanced OLSR HELLO Message Structure higher packet losses, leading to larger CCM values and making them less likely to be selected. This approach mirrors the intent of metrics like ETX, which account for the average number of transmissions needed by considering packet loss probabilities. By integrating the CCM into the AODV’s route calculation mechanism, the protocol significantly enhances its ability to select the most efficient and stable paths . This approach not only optimizes network performance but also contributes to the longevity and sustainability of the network by prioritizing more efficient routes. B. Enhancing OLSR with CCM-Mapped MPR Willingness a) Message Structure Enhancement in OLSR: The OLSR protocol has been adapted to incorporate the components of the CCM, specifically the FSPL and residual energy metrics. This adaptation enhances the protocol’s ability to make more informed decisions regarding route selection and Multi-Point Relay (MPR) selection. In the enhanced OLSR protocol, the Hello and Topology Control (TC) messages are modified to include FSPL and residual energy data. For the Hello message, the original structure is expanded to include fields for FSPL and the node’s residual energy. This extension enables nodes to broadcast their link quality (FSPL) and energy status to their immediate neighbors. The updated Hello message structure becomes [Type | VTime | Willingness | Link Code | Link Message Size | Neighbor Interface Address | FSPL | Residual Energy]. The structure of the new OLSR HELLO message can be seen in Figure 6. For the TC message, similarly, the TC message is augmented to carry the FSPL and residual energy information of the MPRs. This information is used by other nodes in the network to calculate the CCM for routes passing through these MPRs. The revised TC message structure is [Type | ANSN | Reserved | Advertised Neighbor Main Address | FSPL | Residual Energy]. The structure of the TC message can be seen in Figure 7. In this enhanced OLSR framework, each node calculates the CCM for its potential routes locally, based on the FSPL and residual energy information received from peers. This local This article has been accepted for publication in IEEE Transactions on Communications. This is the author's version which has not been fully edited and content may change prior to final publication. Citation information: DOI 10.1109/TCOMM.2025.3634253 This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/ IEEE TRANSACTIONS ON COMMUNICATIONS 9 0123 01234567890123456789012345678901 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | ANSN | Reserved | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Advertised Neighbor Main Address | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | FSPL | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Residual Energy | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Advertised Neighbor Main Address | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | FSPL | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Residual Energy | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | ... | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Figure 7: Enhanced OLSR TC Message Structure calculation of CCM informs the node’s willingness to serve as an MPR, as well as its decision-making process when selecting MPRs for efficient route establishment. This approach ensures that route and MPR selection are grounded in the latest, most relevant data. b) Route Calculation Mechanism: In the Enhanced OLSR protocol integrating the CCM, a sophisticated mechanism for dynamic willingness adjustment is employed. This mechanism is designed to respond to changes in the CCM, which reflects the cost-efficiency of potential routes. Figure 8 demonstrates the function of this mechanism, which is also described as: 1) Each OLSR node continuously monitors the CCM of potential routes. The CCM calculation is based on the cumulative FSPL for the entire path and the average residual energy of all nodes along that path. 2) If a node observes that the CCM of a potential route decreases in two (or more, depending on network policy) consecutive updates, this indicates an improvement in the route’s efficiency (either through reduced path loss or increased residual energy). In response, the node increases its willingness to become a Multi-Point Relay (MPR) by 1 point. This increment in willingness signifies a higher propensity to participate in routing due to the improved route quality.Conversely, if the CCM increases in one or two consecutive updates (again, based on network policy), this signals a decline in the route’s efficiency. The node will then decrease its willingness to serve as an MPR, reflecting the less favorable conditions for routing. 3) This adjustment mechanism is carefully balanced to ensure responsiveness to changes in network conditions while avoiding excessive fluctuations in willingness. A node only changes its willingness after observing consistent trends in the CCM, thereby ensuring that transient or minor variations in network conditions do not lead to abrupt changes in routing behavior. 4) Impact on MPR Selection and Routing: These dynamic adjustments in willingness directly influence MPR selection. Nodes with increased willingness are more likely to be chosen as MPRs, thus promoting the use of more efficient routes. Conversely, nodes with decreased willingness are less likely to be selected, steering the network away from less efficient paths. This approach to willingness adjustment for relay selection allows the network to continuously adapt to changing conditions, optimizing routing decisions for both energy efficiency and link stability. Dynamic willingness adjustment makes OLSR highly adaptive, ensuring that the network’s routing decisions are better aligned with the current optimal conditions, leading to improved overall performance - albeit some additional control overhead. Enhanced OLSR with CCM: Route and MPR Selection NodeA NodeA NodeB NodeB NodeC NodeC Periodic HELLO Messages Collect local FSPL, Residual Energy HELLO [FSPL_A->B, E_A] HELLO [FSPL_A->C, E_A] Collect local FSPL, Residual Energy HELLO [FSPL_B->A, E_B] HELLO [FSPL_B->C, E_B] Collect local FSPL, Residual Energy HELLO [FSPL_C->A, E_C] HELLO [FSPL_C->B, E_C] Each node learns neighbors' FSPL and residual energy, storing them for local CCM calculations. Periodic TC Messages TC [Neighbors, FSPL, E] TC [Neighbors, FSPL, E] TC [Neighbors, FSPL, E] TC [Neighbors, FSPL, E] TC [Neighbors, FSPL, E] TC [Neighbors, FSPL, E] Compute CCM (FSPL, E) from neighbors Compute CCM (FSPL, E) from neighbors Compute CCM (FSPL, E) from neighbors Dynamic Willingness Adjustment if (CCM improves) => willingness++ else if (CCM worsens) => willingness-- if (CCM improves) => willingness++ else if (CCM worsens) => willingness-- if (CCM improves) => willingness++ else if (CCM worsens) => willingness-- Each node adjusts MPR willingness based on observed CCM trends. MPR Selection & Routing Select MPRs (neighbors w/ higher willingness) Select MPRs (neighbors w/ higher willingness) Select MPRs (neighbors w/ higher willingness) By choosing MPRs that have better (lower) CCM, routes are more efficient. Update routing table (MPR + CCM) Update routing table (MPR + CCM) Update routing table (MPR + CCM) Figure 8: Sequence Diagram of the CCM-enhanced OLSR route establishment process The added complexity of calculating and mapping CCM to willingness in this almost congestion window-like manner, must be offset by significant performance improvements. The research must demonstrate that the benefits in network performance justify the increased complexity. The dynamic nature of CCM-mapped willingness may lead to more frequent changes in MPR structure. Algorithms must ensure that these changes don’t lead to instability or excessive control traffic. Considerations will be made regarding the practical implementation of such a system, including computational This article has been accepted for publication in IEEE Transactions on Communications. This is the author's version which has not been fully edited and content may change prior to final publication. Citation information: DOI 10.1109/TCOMM.2025.3634253 This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/ IEEE TRANSACTIONS ON COMMUNICATIONS 16 [7] D. Huo, Q. Liu, Y. Sun, and H. Li, “A Large Inter-satellite Nondependent Routing Technology – IS-OLSR,” in 2021 17th International Conference on Mobility, Sensing and Networking (MSN), 2021, pp. 25– 31. [8] H. Liming, K. Shaoli, S. Shaohui, M. Deshan, H. Bo, and Z. Meiting, “A load balancing routing method based on real time traffic in LEO satellite constellation space networks,” in 2022 IEEE 95th Vehicular Technology Conference: (VTC2022-Spring), 2022, pp. 1–5. [9] Z. Liu, Z. Liu, L. Wang, and W. Li, “Traffic-Predictive Routing Strategy for Satellite Networks,” Electronics, vol. 13, no. 1, 2024. [10] S. Suhaimi, K. Mamat, and K. D. Wong, “Enhancing the ”Willingness” on the OLSR Protocol: Methodology on Experimenting the Real Testbed,” in 2009 5th International Conference on Wireless Communications, Networking and Mobile Computing, 2009, pp. 1–4. [11] S. Jiao, C. Wang, R. Sun, Y. Wang, L. Jia, and J. Chen, “Routing Design and Experimental Validation on a Satellite 5G Bearer Network,” in 2022 IEEE 8th International Conference on Computer and Communications (ICCC), 2022, pp. 638–643. [12] M.-H. T. Nguyen, T. T. Bui, L. D. Nguyen, E. Garcia-Palacios, H.-J. Zepernick, and T. Q. Duong, “Real-Time Large-Scale 6G Satellite-UAV Networks,” in 2023 IEEE Statistical Signal Processing Workshop (SSP), Hanoi, Vietnam, 2023, pp. 100–104. [13] Y. Wang, W. Feng, J. Wang, and T. Q. S. Quek, “Hybrid SatelliteUAV-Terrestrial Networks for 6G Ubiquitous Coverage: A Maritime Communications Perspective,” IEEE Journal on Selected Areas in Communications, vol. 39, no. 11, pp. 3475–3490, Nov 2021. [14] C. Liu, W. Feng, Y. Chen, C.-X. Wang, and N. Ge, “Cell-Free SatelliteUAV Networks for 6G Wide-Area Internet of Things,” IEEE Journal on Selected Areas in Communications, vol. 39, no. 4, pp. 1116–1131, Apr 2021. [15] S. Zhang, H. Zhang, and L. Song, “Beyond D2D: Full Dimension UAVto-Everything Communications in 6G,” IEEE Transactions on Vehicular Technology, vol. 69, no. 6, pp. 6592–6602, Jun 2020. [16] Q. Wang, X. Chen, and Q. Qi, “Energy-Efficient Design of SatelliteTerrestrial Computing in 6G Wireless Networks,” IEEE Transactions on Communications, vol. 72, no. 3, pp. 1759–1772, Mar 2024. [17] Z. Jia, C. Dong, K. Guo, and Q. Wu, “The Potential of LEO Satellites in 6G Space-Air-Ground Enabled Access Networks,” 2023. [Online]. Available: [https://arxiv.org/abs/2307.00234] [18] W.-Y. Dong, S. Yang, and S. Chen, “Uplink performance analysis of heterogeneous non-terrestrial networks in harsh environments: A novel stochastic geometry model,” IEEE Transactions on Communications, 2025, early access, Jan. 2025. [19] W.-Y. Dong, S. Yang, P. Zhang, and S. Chen, “Stochastic Geometry Based Modeling and Analysis of Uplink Cooperative Satellite-AerialTerrestrial Networks for Nomadic Communications With Weak Satellite Coverage,” IEEE Journal on Selected Areas in Communications, vol. 42, no. 12, pp. 3428–3444, 2024. [20] S. Yang, W. Zhao, C.-M. Wang, W.-Y. Dong, and X. Ju, “Betweenness centrality based dynamic source routing for flying ad hoc networks in marching formation,” IEEE Transactions on Vehicular Technology, 2025, early access, Mar. 2025. [21] C.-M. Wang, S. Yang, W.-Y. Dong, W. Zhao, and W. Lin, “A Distributed Hybrid Proactive-Reactive Ant Colony Routing Protocol for Highly Dynamic FANETs With Link Quality Prediction,” IEEE Transactions on Vehicular Technology, vol. 74, no. 1, pp. 1817–1822, 2025. [22] D. S. J. De Couto, D. Aguayo, J. Bicket, and R. Morris, “A HighThroughput Path Metric for Multi-Hop Wireless Routing,” Wireless Networks, vol. 11, pp. 419–434, Jul 2005. [23] T. Clausen and P. Jacquet, “Optimized Link State Routing Protocol (OLSR),” RFC 3626, IETF, RFC, Oct. 2003. [Online]. Available: [https://www.rfc-editor.org/info/rfc3626] [24] C. Perkins, E. Belding-Royer, and S. Das, “Ad hoc On-Demand Distance Vector (AODV) Routing,” RFC 3561, IETF, RFC, Jul. 2003. [Online]. Available: [https://www.rfc-editor.org/info/rfc3561] [25] G. Amponis, T. Lagkas, V. Argyriou, I. Moscholios, M. Zevgara, S. Ouzounidis, and P. Sarigiannidis, “Anchored self-similar 3D GaussMarkov mobility model for ad hoc routing scenarios,” IET Networks, vol. 12, no. 5, pp. 250–259, 2023. [26] G. Amponis, “NS-3 Composite Energy Source Model,” Software, 2025, implements a Li-Ion energy source with solar harvesting (LEO cycle and fixed window) for ns-3. [Online]. Available: [https://github.com/g-ampo/NS3-composite-energy-source-model] George Amponis received his B.Sc. degree from the Department of Information and Electronic Engineering of the International Hellenic University, School of Engineering in 2020, and his PhD from the Department of Informatics of the Democritus University of Thrace, School of Natural Sciences in 2025. He has experience in ad hoc networking, embedded systems and firmware development for a wide range of applications. His research interests include wireless communications, routing protocols for drone swarms, security in 5G environments, secure real-time communications, linkand transportlayer protocols with an emphasis on congestion avoidance. He is a member of IEEE and has authored numerous publications in the field of routing, ad hoc networks and next generation cellular communications. Dr. Thomas Lagkas is Assistant Professor at the Department of Computer Science of the Democritus University of Thrace and Director of the Laboratory of Industrial and Educational Embedded Systems. He graduated with honours from the Department of Informatics, Aristotle University of Thessaloniki and awarded PhD on Wireless Networks. He also completed MBA studies at the Hellenic Open University and received a postgraduate certificate on Teaching and Learning from The University of Sheffield. He has been scholar of the Aristotle University Research Committee and postdoctoral scholar of the National Scholarships Institute of Greece. His research interests are in the areas of IoT communications with numerous highly cited publications. Dr. Lagkas is an IEEE Senior Member, Fellow of the Higher Education Academy in the UK, and member of the Editorial Board of reputable scientific journals. Moreover, he actively participates in several EU-funded research projects. Pavlos Bouzinis received the Diploma (five years) and Ph.D. degrees in electrical and computer engineering from the Aristotle University of Thessaloniki, Greece, in 2019 and 2023, respectively, where he was a member of the Wireless Communications and Information Processing Group. Currently, he works as a research engineer at MetaMind Innovations P.C. His main research interests include machine learning, optimization, and intrusion detection systems. He has served as a reviewer for several scientific journals and was an exemplary reviewer of IEEE WIRELESS COMMUNICATIONS LETTERS, in 2021 (top 3% of reviewers). Dr. Panagiotis Radoglou-Grammatikis received Diploma (five years) and PhD from the Dept. of Electrical and Computer Engineering, University of Western Macedonia, Greece, in 2016 and 2023, respectively. His main research interests focus on AI-driven cybersecurity, intrusion detection and security games. He has published more than 50 research papers in international scientific journals, conferences and book chapters, while he has received five best paper awards. He was included in Stanford University’s list (shared by Elsevier) of the Top 2% of Scientists in the World for 2021 and 2022. Currently, he is working as a research director at K3Y Ltd, while he is also a postdoc researcher at the ITHACA Lab of the University of Western Macedonia and co-founder of MetaMind Innovations P.C. He is involved in several national and international projects. Finally, he is a member of IEEE, ACM and the Technical Chamber of Greece. Antonios Sarigiannidis received the B.Sc. degree in Information Technology from the Aristotle University of Thessaloniki in 2007 and the M.Sc. degree in Communication Systems and Technologies, specialising in advanced optical and wireless technologies, from the Aristotle University of Thessaloniki in 2009. He obtained his Ph.D. in Information Technology from the Aristotle University of Thessaloniki in 2016. His Ph.D. thesis includes the development of bandwidth allocation algorithms in Communication Networks. His research interests include machine learning mechanism and optimisation techniques as well as visualisation techniques regarding analytics, big data and security analysis. Recently, he has been involved in IoT and M2M research towards in coverage analysis and security services. He actively participated in both national and EU funded projects. He is the author of more than 30 publications in leading international journals and conferences. This article has been accepted for publication in IEEE Transactions on Communications. This is the author's version which has not been fully edited and content may change prior to final publication. Citation information: DOI 10.1109/TCOMM.2025.3634253 This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/ IEEE TRANSACTIONS ON COMMUNICATIONS 17 Prof. Panagiotis Sarigiannidis is the Director of ITHACA Lab, Co-Founder of MetaMind Innovations P.C. and Full Professor at the Department of Electrical and Computer Engineering, University of Western Macedonia, Kozani, Greece. He received his B.Sc. and Ph.D. in computer science from the Aristotle University of Thessaloniki, Thessaloniki, Greece, in 2001 and 2007, respectively. His research interests include telecommunication networks, Internet of Things and cybersecurity. He has published over 270 papers in international journals, conferences and book chapters, while he has also received five best paper awards. He is involved in several national and international projects. He served as the project coordinator of three H2020 projects, namely SPEAR, EVIDENT and TERMINET. Moreover, he has coordinated national and Erasmus+ KA2 projects, while he served as a principal investigator in SDN-microSENSE and three Erasmus+ KA2: ARRANGE-ICT, JAUNTY and STRONG. Finally, he participates in several editorial boards of various journals. Prof. Vasileios Argyriou received the B.Sc. degree in computer science from the Aristotle University of Thessaloniki, Greece, in 2001, and the M.Sc. and Ph.D. degrees in electrical engineering working on registration from the University of Surrey, in 2003 and 2006, respectively. From 2001 to 2002, he held a research position with Aristotle University, with a focus on image and video watermarking. He joined the Communications and Signal Processing Department, Imperial College London, London, in 2007, where he was a Research Fellow working on 3D object reconstruction. He is currently a Professor with Kingston University, London, working on computer vision and AI for crowd and human behavior analysis, computer games, entertainment, and medical applications. Also, research is conducted on educational games and on HCI for augmented and virtual reality (AR/VR) systems. This article has been accepted for publication in IEEE Transactions on Communications. This is the author's version which has not been fully edited and content may change prior to final publication. Citation information: DOI 10.1109/TCOMM.2025.3634253 This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/