Paper deep dive
STR-Agent: An LLM-Driven Agent for QoS-Aware Routing in LEO Satellite Networks
Bowen Lu, Mugen Peng, Yaohua Sun, Hongyu Wang, Kerui Guo, Wenjia Xu
Intelligence
Status: succeeded | Model: Gemma-4-26B-A4B | Prompt: intel-v1 | Confidence: 91%
Last extracted: 9/19/2026, 1:34:11 AM
Summary
The paper introduces STR-Agent, an LLM-driven framework for Quality-of-Service (QoS)-aware routing in Low Earth Orbit (LEO) satellite networks. It addresses the challenge of translating unstructured natural language service requests into adaptive routing decisions by integrating a Perception Module for intent parsing, an Execution Module for tool-based routing, an Experience Buffer for historical data, and a Reflection Module for dynamic policy adaptation. The system uses a specialized fine-tuned model to interpret service types (delay-sensitive, bandwidth-sensitive, best-effort) and adjusts routing algorithms based on real-time congestion and past performance, significantly outperforming baselines like DQ-Dijkstra in delay reduction and intent understanding accuracy.
Entities (10)
Relation Signals (9)
STR-Agent → uses → Reflection Module
confidence 95% · the Reflection Module dynamically adjusts the service-to-routing-policy mapping
STR-Agent → uses → Execution Module
confidence 95% · the Execution Module first maps the source and destination coordinates... and then performs routing through external tool calls.
STR-Agent → uses → Perception Module
confidence 95% · The key innovation of STR-Agent lies in unifying intent perception... Specifically, the Perception Module converts natural-language requests
STR-Agent → outperforms → DQ-Dijkstra
confidence 92% · it reduces end-to-end delay by up to 60% compared with DQ-Dijkstra
Reflection Module → adjusts → routing_policy
confidence 90% · the Reflection Module dynamically adjusts the service-to-routing-policy mapping according to real-time congestion conditions
Qwen2.5-7B → isfinetunedwith → LoRA
confidence 90% · We fine-tune Qwen2.5-7B using LoRA
Perception Module → isimplementedby → Qwen2.5-7B
confidence 90% · We fine-tune Qwen2.5-7B... This enables the Perception Module to accurately interpret diverse service intents
STR-Agent → uses → Experience Buffer
confidence 90% · Experience Buffer for outcome statistics
Cypher Suggestions (0)
No Cypher suggestions yet.
Abstract
Abstract:LEO satellite networks feature dynamic topologies, time-varying links, and diverse service requirements, which make conventional routing schemes difficult to support fine-grained quality-of-service (QoS) provisioning. Existing studies mainly optimize routing over network states with predefined objectives, but rarely address the practical challenge of translating unstructured natural-language service requests into adaptive routing decisions. To bridge this gap, we propose STR-Agent, an LLM-driven framework for QoS-aware routing in LEO satellite networks. The key innovation of STR-Agent lies in unifying intent perception, tool-based execution, experience accumulation, and reflection-based policy adaptation within a single agent architecture. Specifically, the Perception Module converts natural-language requests into structured routing semantics, while the Reflection Module dynamically adjusts the service-to-routing-policy mapping according to real-time congestion conditions and historical routing outcomes, rather than relying on a fixed routing objective. In addition, we develop a specialized perception model, and construct a domain-specific supervised fine-tuning dataset for LEO service understanding. Simulation results in a Walker-Delta constellation show that STR-Agent significantly outperforms conventional baselines: it reduces end-to-end delay by up to 60% compared with DQ-Dijkstra, improves average intent-understanding accuracy from 45.4% to 92.45% after supervised fine-tuning, and the Reflection Module further reduces the delay by 120 ms at 600 Mbps. These results demonstrate the potential of LLM-driven agent architectures to enable service-aware and adaptive QoS routing in future LEO satellite networks.
Tags
Links
- Source: https://arxiv.org/abs/2609.20347v1
- Canonical: https://arxiv.org/abs/2609.20347v1
Trouble viewing inline? Open PDF directly →
Full Text
30,433 characters extracted from source content.
Expand or collapse full text
STR-Agent: An LLM-Driven Agent for QoS-Aware Routing in LEO Satellite Networks Bowen Lu, Mugen Peng, Yaohua Sun, Hongyu Wang, Kerui Guo, and Wenjia Xu * State Key Laboratory of Networking and Switching Technology Beijing University of Posts and Telecommunications Email: xuwenjia@bupt.edu.cn, bowenlu@bupt.edu.cn Abstract—LEO satellite networks feature dynamic topologies, time-varying links, and diverse service requirements, which make conventional routing schemes difficult to support fine-grained quality-of-service (QoS) provisioning. Existing studies mainly optimize routing over network states with predefined objec- tives, but rarely address the practical challenge of translating unstructured natural-language service requests into adaptive routing decisions. To bridge this gap, we propose STR-Agent, an LLM-driven framework for QoS-aware routing in LEO satellite networks. The key innovation of STR-Agent lies in unifying intent perception, tool-based execution, experience accumulation, and reflection-based policy adaptation within a single agent archi- tecture. Specifically, the Perception Module converts natural- language requests into structured routing semantics, while the Reflection Module dynamically adjusts the service-to-routing- policy mapping according to real-time congestion conditions and historical routing outcomes, rather than relying on a fixed routing objective. In addition, we develop a specialized perception model, and construct a domain-specific supervised fine-tuning dataset for LEO service understanding. Simulation results in a Walker–Delta constellation show that STR-Agent significantly outperforms conventional baselines: it reduces end-to-end delay by up to 60% compared with DQ-Dijkstra, improves average intent-understanding accuracy from 45.4% to 92.45% after supervised fine-tuning, and the Reflection Module further reduces the delay by 120 ms at 600 Mbps. These results demonstrate the potential of LLM-driven agent architectures to enable service- aware and adaptive QoS routing in future LEO satellite networks. Index Terms—LEO satellite network, STR-Agent, large lan- guage model I. INTRODUCTION With the large-scale deployment of Low Earth Orbit (LEO) satellite constellations, LEO systems are becoming critical infrastructure for global communications [1]. Compared with Medium Earth Orbit (MEO) and Geostationary Earth Orbit (GEO) satellites, they provide superior deployment flexibility, enhanced spectrum efficiency, and reduced link loss [2]. Many LEO initiatives therefore aim to deploy tens of thousands of satellites and integrate them with terrestrial cellular net- works for seamless global connectivity [3]. However, LEO networks are inherently dynamic, with high node mobility and rapidly varying inter-satellite and satellite-to-ground links [4], fundamentally differing from relatively static terrestrial networks. As a result, terrestrial routing mechanisms often cannot adapt to topology evolution and link variation, leading *Corresponding author. Code is available at: https://github.com/IntelliSensing/STR-Agent. to path mismatch, degraded forwarding efficiency, and unstable service quality [5]. Therefore, routing strategies with real-time network awareness and adaptive decision-making are essential. Recent advances in artificial intelligence have shifted LEO routing from conventional static approaches to data-driven, learning-based methods [6]. The focus has moved from fixed rule-based path selection to learning network state, predicting link dynamics, and optimizing routing decisions [7]. Rein- forcement learning-based routing algorithms, in particular, of- fer greater adaptability in dynamic environments [8]. However, existing learning-based routing methods in LEO networks still face two important limitations. First, their optimization objectives are typically predefined and fixed during training, making it difficult to flexibly balance latency, throughput, load balancing, and reliability across heterogeneous services [9]. As terrestrial and satellite networks become increasingly integrated, user applications exhibit growing diversity, and different services impose distinct QoS requirements [10]. Second, most methods decouple routing optimization from service understanding, assuming service type, priority, or QoS requirements are structured inputs, whereas user demands are often expressed in natural language. Existing approaches can optimize predefined objectives, but cannot interpret service intent, extract key constraints, or translate them into QoS- aware routing actions [11], [12]. This limitation is more pro- nounced in integrated terrestrial–satellite networks, where fine- grained QoS provisioning requires both path optimization and accurate service understanding. The key challenge is therefore to bridge natural-language service semantics and QoS-aware routing actions within a unified framework. To address this issue, we formulate QoS-aware routing in LEO networks as a closed-loop intent-to-routing problem, where natural-language requests are translated into structured service semantics and then mapped to adaptive routing policies based on network congestion and historical routing outcomes. Based on this, we propose the LLM-driven Satellite Task- aware Routing Agent (STR-Agent), integrating Perception, Execution, Experience Buffer, and Reflection into a unified perceive–execute–reflect framework for service understanding, routing, experience summarization, and policy refinement. To improve domain understanding in LEO routing scenarios, we further construct a domain-specific dataset for service-request understanding and develop a specialized perception model. The main contributions are summarized as follows: arXiv:2609.20347v1 [cs.NI] 17 Sep 2026 Please deliver emergency medical telemetry from London to Shanghai with low delay. Perception Module Okay, the uplink is from London (51.5°N, 0.13°W) to Shanghai (31.2°N, 121.5°E) for a delay-sensitive task. (51.5°N, 0.13°W) and (31.2°N, 121.5°E) connect to satellites SAT-102 and SAT-514, respectively. Tools: Satellite Access Tool Execution Module I should check current congestion and experience buffer, the congestion level is low. Tools: Network State Tool Execution Module Reflection Module Based on the current network congestion and scenario, the delay-sensitive routing algorithm is recommended. I need to use the delay-sensitive routing algorithm, the next hop is SAT-103. Tools: Delay-Sensitive Routing Tool Execution Module ... Reflection Module Congestion is high. Experience buffer shows a bandwidth-sensitive algorithm reduces latency. I need to use the bandwidth-sensitive routing algorithm, the next hop is SAT-490. Tools: Bandwidth-Sensitive Routing Tool Execution Module ... Destination reached. Access Satellite: SAT-102 and SAT-514. Path: SAT-102SAT-103 ... SAT-515 SAT-514. Fig. 1: Workflow of STR-Agent for QoS-aware routing in LEO satellite networks. • We propose STR-Agent, a closed-loop intent-to-routing framework for LEO satellite networks, which transforms natural-language service requests into structured routing semantics and further into QoS-aware routing actions through coordinated perception, execution, experience buffering, and reflection. • We propose an adaptive service-to-algorithm mapping mechanism that leverages both the current congestion profile and summarized routing experience to dynami- cally assign routing policies to different service types, rather than relying on a fixed routing objective. • We construct a domain-specific supervised fine-tuning dataset for LEO service-request understanding, and val- idate through intent-understanding and routing experi- ments the contribution of the perception and reflection modules to QoS-aware routing performance. Fig. 1 illustrates the STR-Agent workflow. Given a natural- language service request, STR-Agent identifies the service intent, and performs multi-hop routing through tool execution and reflection-driven strategy adjustment. I. SYSTEM MODEL We consider a Walker–Delta low Earth orbit (LEO) mega- constellation for global broadband services. The network is modeled as a time-varying graph with packet forwarding over inter-satellite links (ISLs). The constellation, link, delay, and QoS-oriented routing objective are defined as follows. A. Constellation Model The constellation is denoted by (N p ,N m ,F ), where N p is the number of orbital planes, N m is the number of satellites per plane, and F is the phasing factor. The total number of satellites is N = N p N m . Each satellite operates at altitude H with orbital radius r = R e +H , where R e is the Earth radius. At time t, the network is represented as G(t) = (V,E(t)), where V and E(t) denote the satellite set and active ISL set, respectively. Each satellite maintains four links, including two intra-orbit links within the same plane and two inter-orbit links to adjacent planes, forming a regular mesh topology. B. Link Model For any active link (i,j) ∈ E(t), letp i (t),p j (t) ∈R 3 denote the positions of satellites i and j. The inter-satellite distance is d ij (t) = ∥p i (t)−p j (t)∥ 2 . Assuming free-space propagation, the path loss is L ij (t) = 4πf c d ij (t) c 2 ,(1) where f c is the carrier frequency and c is the speed of light. The received power is P r,ij (t) = P t G t G r /L ij (t), where P t , G t , and G r denote the transmit power and the transmit and receive antenna gains, respectively. Since practical inter-satellite links employ fixed modulation and coding schemes, we use an E b /N 0 -based model to ap- proximate the achievable transmission rate [13]: R ij (t) = P r,ij (t) (E b /N 0 )k B T s ,(2) where E b /N 0 is the required energy-per-bit to noise spectral density ratio, k B is the Boltzmann constant, and T s is the system noise temperature. C. Delay Model For a packet of size L p transmitted over link (i,j) at time t, the propagation, transmission, and queuing de- lays are τ prop ij (t) = d ij (t)/c, τ tx ij (t) = L p /R ij (t), and τ que ij (t) = L p Q ij (t)/R ij (t), respectively, where Q ij (t) de- notes the queue length at satellite i for packets forwarded to satellite j. The total one-hop delay is τ ij (t) = τ prop ij (t) + τ tx ij (t) + τ que ij (t).(3) D. QoS-Oriented Routing Objective Based on the above model, QoS-aware routing is performed on the time-varying graph G(t). For a service request r k , the perception module extracts structured semantics and identifies the service type b k ∈B delay ,B bw ,B hop . The routing module then outputs an end-to-end path P k (t) =v 0 ,v 1 ,...,v m .(4) Instead of optimizing a fixed global metric, STR-Agent adopts a service-dependent objective: P ⋆ k (t) = arg min P∈Π k (t) J (P | b k ,G(t)),(5) where Π k (t) is the feasible path set for request r k , and J (P | b k ,G(t)) is the service-adaptive path cost. For delay-sensitive services, the objective is to minimize cumulative latency, i.e., J delay (P) = P (i,j)∈P τ ij (t). For User Request: “Send urgent medical information from London to Shanghai. Perception Module Execution Module Experience Buffer Reflection Module Service Scenarios emergency rescue real-time drone swarm remote sensing data cloud-based video data energy monitoring IoT periodic reporting intent understanding Source location destination location service scenario “src_lat”:“ 51.5°N” , “src_lon”:“ 0.13 °W”, “dst_lat”:“ 31.2 °N”, “dst_lon”:“ 121.5 °E”, “service_type”: “ delay_sensitive” Satellite Access Tool Network State Tool Bandwidth Routing Tool Best-Effort Routing Tool next hop Delay Routing Tool congestion profile ( ത 푄푡,푄 90 푡,푁 ℎ표푡 푡) per-hop experience recording (푏 ℎ ,푎 ℎ ,푙 ℎ ,푄 ℎ 푑푒푐 ,푄 ℎ 푎푟 ,휏 ℎ ) Summarized statistics 푆푏,푎,푙=(ҧ휏 푏,푎,푙 , ത 푄 푏,푎,푙 푑푒푐 , ത 푄 푏,푎,푙 푎푟 ,푁 푏,푎,푙 ) Strategy refinement network state agent thinking adaptive strategy 푀푡=푏→푎 푏 (푡) Fig. 2: Architecture of the proposed Satellite Task-aware Routing Agent (STR-Agent). bandwidth-sensitive services, the objective is to avoid con- gested or highly imbalanced regions, which can be written as J bw (P) = P (i,j)∈P σ(Q n (t) : n ∈ N (i)), where N (i) denotes the neighbor set of node i and σ(·) denotes the standard deviation of queue lengths. For best-effort services, the objective is to reduce routing complexity and forwarding overhead, approximated by the hop count J hop (P) =|P|− 1. This formulation provides the modeling basis for the routing tools in Section I: the routing policy is determined not by a fixed objective, but by a service-adaptive cost selected according to the inferred service type. I. SATELLITE TASK-AWARE ROUTING AGENT We propose the Satellite Task-aware Routing Agent (STR- Agent), an LLM-driven framework for QoS-aware routing in LEO satellite networks. As shown in Fig. 2, STR-Agent con- sists of four modules: Perception for intent parsing, Execution for tool-grounded route construction, Experience Buffer for outcome statistics, and Reflection for adaptive policy selection. Together, they form a closed loop from natural-language requests to adaptive multi-hop forwarding. A. Perception Module Given natural-language request r k , the Perception Module outputs P (r k ) = φ src k ,λ src k ,φ dst k ,λ dst k ,b k ,(6) where (φ,λ) denote latitude and longitude, and b k denotes the inferred service type. This task contains two coupled subtasks: geographic entity extraction and service-type classification. The service type b k is drawn fromB delay ,B bw ,B hop , cor- responding to delay-sensitive, bandwidth-sensitive, and best- effort services, respectively. A typical output is "src_lat":"51.5°N","src_lon":"0.13°W", "dst_lat":"31.2°N","dst_lon":"121.5°E", "service_type":"delay_sensitive". This structured representation provides the geographic end- points and service semantics required by downstream routing. Errors in service-type prediction may cause the Execution Module to invoke an unsuitable routing policy; therefore, in- tent understanding accuracy directly affects downstream QoS. B. Execution Module Given P (r k ), the Execution Module first maps the source and destination coordinates to an access-satellite pair, and then performs routing through external tool calls. STR-Agent adopts hop-by-hop forwarding: at each step, it queries the current congestion profile, obtains the service-to-algorithm mapping from the Reflection Module, selects the routing tool, and computes the next hop. This enables online adaptation to time-varying congestion. The Execution Module uses five tools: • Satellite Access Tool, which maps ground coordinates (φ,λ) to the visible access satellite with the highest elevation angle; • Network State Tool, which returns the current conges- tion profile C(t) = ̄ Q(t),Q 90 (t),N hot (t) ,(7) where ̄ Q(t) is the mean queue length, Q 90 (t) is the 90th- percentile queue length, and N hot (t) is the number of satellites whose queue length exceeds a preset threshold; • Delay-Sensitive Routing Tool B delsy ; • Bandwidth-Sensitive Routing Tool B bw ; • Best-Effort Routing Tool B hop . All three routing algorithms are based on the Dijkstra Short- est Path (DSP) method, with edge weights varying according to service type. The weights are defined as follows. For delay- sensitive traffic, the edge weight is w delay ij = d ij c + L p R ij ,(8) which favors low propagation and transmission delay. For bandwidth-sensitive traffic, the edge weight is w bw ij = σ(Q n : n∈N (i)),(9) where N (i) denotes the neighbor set of node i, Q n is the queue length of neighbor n, and σ(·) denotes the standard deviation. This metric encourages routing through regions with more balanced queue states and hence alleviates queue buildup for bandwidth-intensive traffic. For best-effort traffic, the edge weight is w hop ij = 1,(10) which reduces to minimum-hop routing. C. Experience Buffer The Experience Buffer stores and summarizes hop-level routing outcomes. After each hop, STR-Agent records e h = b h ,a h ,ℓ h ,Q dec h ,Q arr h ,τ h ,(11) where b h is the service type, a h ∈ B delay ,B bw ,B hop is the selected routing tool, ℓ h is the congestion level at decision time, Q dec h and Q arr h are the queue lengths observed at forwarding and arrival, and τ h is the realized one-hop delay. To preserve congestion dependence, these records are ag- gregated as S(b,a,ℓ) = ̄τ b,a,ℓ , ̄ Q dec b,a,ℓ , ̄ Q arr b,a,ℓ ,N b,a,ℓ ,(12) where ̄τ b,a,ℓ denotes the mean delay, ̄ Q dec b,a,ℓ and ̄ Q arr b,a,ℓ denote the average queue lengths, and N b,a,ℓ is the sample count. Each hop contributes one record, and the summary is updated in a rolling manner after each request or time window so that recent observations dominate policy adjustment. D. Reflection Module The Reflection Module maps the current network state and accumulated experience to a service-to-algorithm mapping M (t) : b7→ a b (t),(13) where a b (t) is the routing tool for service type b. Under normal conditions, each service type uses its default routing tool. The Reflection Module regards the network as severely congested when the average satellite queue length is at least 5 or at least 50 satellites have queues longer than 5. It then uses the historical hop-level records in the latest reflec- tion window: if both delay-sensitive and bandwidth-sensitive records exist and the bandwidth-sensitive records have a lower average per-hop delay, delay-sensitive traffic is temporarily switched to B bw ; otherwise, the default delay-sensitive tool is kept. Bandwidth-sensitive and best-effort traffic continue to use B bw and B hop , respectively. IV. MODEL FINE-TUNING To enhance the Perception Module of STR-Agent, we fine- tune a domain-specific model for mapping natural-language service requests to structured routing semantics, with emphasis on geographic entity extraction and service-type recognition. A. Dataset Construction We construct a supervised fine-tuning dataset tailored to LEO satellite networks. Satellite communication services are divided into three QoS-oriented categories, namely delay- sensitive, bandwidth-sensitive, and best-effort. For each cat- egory, 14 representative scenarios are selected with reference to 3GPP TR 22.822 [14], covering typical LEO applications such as emergency communications, data backhaul, and large- scale IoT services. The geographic space is instantiated by 49 global cities, from which source–destination pairs are sampled to form realistic service requests. All samples are generated by Qwen2.5-7B from scenario descriptions and city pairs in a structured instruction, input, output format. The input is a natural-language request containing the upstream and downstream locations and scenario context, while the output gives the corresponding source and destination coordinates and service type. The training set contains 30,000 samples, with 10,000 per class, and the test set contains 3,000 samples, with 1,000 per class. B. Supervised Fine-Tuning We fine-tune Qwen2.5-7B using LoRA to adapt the model for mapping natural-language service requests to structured routing semantics. This enables the Perception Module to accurately interpret diverse service intents and provide reliable inputs to the Execution and Reflection Modules, enabling precise, routing-oriented decision-making. V. EXPERIMENTS In this section, we evaluate STR-Agent through simulations in a Walker–Delta LEO constellation. A. Simulation Setup The constellation configuration is (N p ,N m ,F )= (72, 22, 39), corresponding to 1584 satellites at an altitude of 550 km. The carrier frequency is 23.28 GHz, the channel bandwidth is 2 GHz, the transmit EIRP is 35 dBW, the receiver quality factor is G r /T s = 6.8 dB/K, and the modulation scheme is 8-PSK with a target BER of 10 −7 . The LoRA rank is 8, the scaling factor is 16, the dropout rate is 0.05, the learning rate is 2× 10 −4 , the batch size is 64, and the number of training epochs is 3. Service requests follow the dataset construction in Sec- tion IV-A. Performance is evaluated separately for different service types to reflect QoS-differentiated routing behavior. During routing, STR-Agent performs hop-by-hop forwarding: at each step, it queries the current congestion profile and selects the routing tool accordingly. 90120150180210240270 Data Rate (Mbps) 100 200 300 400 500 600 700 800 Average Delay (ms) STR-Agent (SFT) STR-Agent (Prompt) STR-Agent (Base) QSMR DQ-Dijkstra (a) Average end-to-end delay 90120150180210240270 Data Rate (Mbps) 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 Average per-hop queue STR-Agent (SFT) STR-Agent (Prompt) STR-Agent (Base) QSMR DQ-Dijkstra (b) Average per-hop queue length 90120150180210240270 Data Rate (Mbps) 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 Average per-hop queuing delay (ms) STR-Agent (SFT) STR-Agent (Prompt) STR-Agent (Base) QSMR DQ-Dijkstra (c) Average per-hop queuing delay 90120150180210240270 Data Rate (Mbps) 25 30 35 40 45 50 55 60 65 70 75 80 85 90 Average Hop STR-Agent (SFT) STR-Agent (Prompt) STR-Agent (Base) QSMR DQ-Dijkstra (d) Average hop count Fig. 3: (a) Average end-to-end delay, (b) average per-hop queue length, (c) average per-hop queuing delay, (d) average hop count under different data rates. TABLE I: Intent Understanding Accuracy (%) ModelDelay-Sens. Bandwidth-Sens. Best-EffortAvg. Base Model78.0249.98.2745.4 Prompt-Based Model82.4186.4973.3280.74 SFT-Based Model97.795.4584.292.45 B. Intent Understanding Evaluation We first evaluate the perception module on natural-language service requests. The metric is intent-understanding accuracy, defined as the proportion of requests for which the source, destination, and service type are all correctly identified. The test set is generated following Section IV-A. We compare three models: the Base Model, namely the original Qwen2.5-7B; the Prompt-Based Model, which uses task-specific prompts; and the SFT-Based Model, which is fine- tuned as described in Section IV-B. As shown in Table I, the Base Model achieves only 45.4% average accuracy, indicating limited domain-specific parsing ability. Prompting improves the average accuracy to 80.74%, and supervised fine-tuning further raises it to 92.45%. Best-effort requests are consistently harder than the other two categories. In the Base Model, their accuracy is only 8.27%, far below the delay-sensitive and bandwidth-sensitive cases, mainly because their weaker semantic cues make them more easily confused with other scenarios. These results directly affect downstream routing, since misclassified service types lead to unsuitable routing-tool selection, explaining the routing gap among the Base, Prompt- Based, and SFT-Based models. C. Routing Performance Evaluation We evaluate routing performance under varying traffic con- ditions, with traffic arrivals modeled as a Poisson process. The data rate ranges from 90 to 270 Mbps, covering low- to high- load conditions, and each simulation lasts T = 10,000 ms. Performance is evaluated separately for different service types to reflect QoS-differentiated routing behavior. We compare the following methods: • DQ-Dijkstra: DSP with edge weights combining delay and queue length, balancing latency and congestion. • QSMR [15]: A multi-routing algorithm that jointly con- siders propagation delay and residual link capacity. • STR-Agent (Base): Base-model STR-Agent. • STR-Agent (Prompt): Prompt-based STR-Agent. • STR-Agent (SFT): SFT-based STR-Agent. We use four metrics corresponding to different service types: average end-to-end delay for delay-sensitive traffic, average per-hop queue length and per-hop queuing delay for bandwidth-intensive traffic, which reflect the load-balancing capability of the method, and average hop count for best-effort traffic, which evaluates routing simplicity and efficiency. Fig. 3a shows that STR-Agent (SFT) consistently achieves the lowest delay over the entire load range. At 270 Mbps, it reduces the delay from 696.01 ms with QSMR to 260.11 ms, while STR-Agent (Prompt) also surpasses DQ-Dijkstra and QSMR from 180 Mbps onward. Fig. 3b and Fig. 3c show that, although DQ-Dijkstra and QSMR perform better at low load, STR-Agent (SFT) outperforms the baselines as the load increases and, at 270 Mbps, reduces the average per-hop queue length from 1.06 to 0.71 and the per-hop queuing delay from 5.22 ms to 2.50 ms compared with DQ-Dijkstra, indicating stronger load balancing under congested conditions. Fig. 3d shows that STR-Agent (SFT) achieves the smallest hop count at medium and high load, reducing it from 53.03 with QSMR to 25.94 at 270 Mbps. Overall, among the three agent variants, STR-Agent (SFT) performs best, STR-Agent (Prompt) ranks second, and STR-Agent (Base) yields the weakest performance, indicating that more accurate service understanding leads to better routing-policy selection and reduced routing-policy mismatch, especially under moderate and high traffic load. D. Reflection Module Evaluation 350400450500550600 Data Rate (Mbps) 500 600 700 800 900 1000 1100 1200 1300 1400 Average Delay (ms) STR-Agent SFT (Full) STR-Agent SFT (No Reflect) Fig. 4: Ablation study of Reflection Module for average delay in Delay-Sensitive traffic. We further evaluate the Reflection Module under high-load conditions by comparing STR-Agent SFT (Full) and STR-Agent SFT (No Reflect) with the offered load, modeled as a Poisson process, ranging from 350 to 600 Mbps. In the latter setting, each service type always uses its default routing tool. Fig. 4 shows that the Reflection Module improves delay mainly under medium- to high-load conditions. At 350– 400 Mbps, the two configurations exhibit nearly identical delays, indicating limited benefit under light congestion. As the load increases, the gap becomes more pronounced, and at 600 Mbps the average delay decreases from 1320 ms to 1200 ms. Overall, STR-Agent SFT (Full) consistently achieves lower delay, with gains increasing as congestion intensifies. The Reflection Module enables STR-Agent to adapt rout- ing based on historical experience and real-time congestion, avoiding overloaded paths, which confirms its importance for delay-sensitive tasks under high-load conditions. VI. CONCLUSION This work shows that QoS-aware routing in LEO satel- lite networks can be effectively addressed by STR-Agent, a closed-loop Satellite Task-aware Routing Agent that unifies perception, execution, experience buffering, and reflection in a QoS-aware control framework. Rather than relying on a fixed routing objective, STR-Agent adapts routing decisions to heterogeneous service demands and changing congestion con- ditions. Experimental results show significant gains in delay and load-balancing performance over conventional baselines. Overall, STR-Agent provides a practical direction for adaptive and service-aware routing in future LEO satellite networks. Future work will examine human-labeled requests, learning- based baselines, and runtime overhead. VII. ACKNOWLEDGEMENT This work has been funded by the National Natural Science Foundation of China under Grant 62301063. REFERENCES [1] I. Shayea, A. A. El-Saleh, M. Ergen, B. Saoud, R. Hartani, et al., “Integration of 5G, 6G and IoT with Low Earth Orbit (LEO) networks: opportunity, challenges and future trends,” Results in Engineering, vol. 23, p. 102409, 2024. [2] H. Al-Hraishawi, H. Chougrani, S. Kisseleff, E. Lagunas, and S. Chatzinotas, “A survey on nongeostationary satellite systems: the com- munication perspective,” IEEE Communications Surveys & Tutorials, vol. 25, no. 1, p. 101–132, 2022. [3] S. Liu, Z. Gao, Y. Wu, D. W. K. Ng, X. Gao, et al., “LEO satellite constellations for 5G and beyond: how will they reshape vertical domains?” IEEE Communications Magazine, vol. 59, no. 7, p. 30–36, 2021. [4] Z. Han, C. Xu, G. Zhao, S. Wang, K. Cheng, et al., “Time-varying topology model for dynamic routing in LEO satellite constellation networks,” IEEE Transactions on Vehicular Technology, vol. 72, no. 3, p. 3440–3454, 2022. [5] X. Qi, J. Ma, D. Wu, L. Liu, and S. Hu, “A survey of routing techniques for satellite networks,” Journal of Communications and Information Networks, vol. 1, no. 4, p. 66–85, 2016. [6] G. Fontanesi, F. Ort ́ ız, E. Lagunas, L. M. Garc ́ es-Socarr ́ as, V. Monzon Baeza, et al., “Artificial intelligence for satellite communication: a survey,” IEEE Communications Surveys & Tutorials, vol. 28, p. 1381– 1435, 2025. [7] Y. Huang, S. Wu, Z. Kang, Z. Mu, H. Huang, et al., “Reinforcement learning based dynamic distributed routing scheme for mega LEO satellite networks,” Chinese Journal of Aeronautics, vol. 36, no. 2, p. 284–291, 2023. [8] S. Zhang, A. Liu, C. Han, X. Xu, X. Liang, K. An, et al., “GRLR: routing with graph neural network and reinforcement learning for mega LEO satellite constellations,” IEEE Transactions on Vehicular Technology, vol. 74, no. 2, p. 3225–3237, 2024. [9] A. Fuad, A. H. Ahmed, M. A. Riegler, and T. ˇ Ci ˇ ci ́ c, “An intent-based networks framework based on large language models,” in Proc. IEEE 10th International Conference on Network Softwarization (NetSoft), 2024, p. 7–12. [10] Y. Yin, C. Huang, N. N. Xiong, D.-F. Wu, and S. Huang, “Joint dynamic routing and resource allocation in satellite-terrestrial integrated networks,” Computer Networks, vol. 231, p. 109823, 2023. [11] Z. Xing, H. Qi, X. Di, J. Liu, R. Xu, et al., “A multipath routing algorithm for satellite networks based on service demand and traffic awareness,” Frontiers of Information Technology & Electronic Engineer- ing, vol. 24, no. 6, p. 844–858, 2023. [12] N. Tu, S. Nam, and J. W.-K. Hong, “Intent-based network configuration using large language models,” International Journal of Network Man- agement, vol. 35, no. 1, p. e2313, 2025. [13] Y. Ran, Y. Ding, S. Chen, J. Lei, and J. Luo, “Fully-distributed dynamic packet routing for LEO satellite networks: a GNN-enhanced multi-agent reinforcement learning approach,” IEEE Transactions on Vehicular Technology, vol. 74, no. 3, p. 5229–5234, Mar. 2025, doi: 10.1109/TVT.2024.3499933. [14] 3rd Generation Partnership Project (3GPP), Technical Specification Group Services and System Aspects, “Study on using satellite access in 5G (Release 16),” 3GPP TR 22.822, version 16.0.0, 2018. [15] T. Zhang, H. Li, S. Zhang, J. Li, and H. Shen, “STAG-based QoS support routing strategy for multiple missions over the satellite networks,” IEEE Transactions on Communications, vol. 67, no. 10, p. 6912–6924, Oct. 2019.