ETSI GR ENI 036 V4.1.1 (2025-08)

Experiential Networked Intelligence (ENI); Space-Ground Cooperative Network Slicing

DGR/ENI-0036V411_SGC_NetSlicin

General Information

• Status: Not Published
• Technical Committee: ENI - Experiential Networked Intelligence
• Current Stage: 12 - Citation in the OJ (auto-insert)
• Due Date: 13-Aug-2025
• Completion Date: 28-Aug-2025

Standards Content (Sample)

GROUP REPORT
Experiential Networked Intelligence (ENI);
Space-Ground Cooperative Network Slicing
Disclaimer
The present document has been produced and approved by the Experiential Networked Intelligence (ENI) ETSI Industry
Specification Group (ISG) and represents the views of those members who participated in this ISG.
It does not necessarily represent the views of the entire ETSI membership.

2 ETSI GR ENI 036 V4.1.1 (2025-08)

Reference
DGR/ENI-0036V411_SGC_NetSlicin
Keywords
network, slicing
ETSI
650 Route des Lucioles
F-06921 Sophia Antipolis Cedex - FRANCE

Tel.: [phone]  Fax: [phone]

Siret N° 348 623 562 00017 - APE 7112B
Association à but non lucratif enregistrée à la
Sous-Préfecture de Grasse (06) N° w061004871

Important notice
The present document can be downloaded from the
ETSI Search & Browse Standards application.
The present document may be made available in electronic versions and/or in print. The content of any electronic and/or
print versions of the present document shall not be modified without the prior written authorization of ETSI. In case of any
existing or perceived difference in contents between such versions and/or in print, the prevailing version of an ETSI
deliverable is the one made publicly available in PDF format on ETSI deliver repository.
Users should be aware that the present document may be revised or have its status changed,
this information is available in the Milestones listing.
If you find errors in the present document, please send your comments to
the relevant service listed under Committee Support Staff.
If you find a security vulnerability in the present document, please report it through our
Coordinated Vulnerability Disclosure (CVD) program.
Notice of disclaimer & limitation of liability
The information provided in the present deliverable is directed solely to professionals who have the appropriate degree of
experience to understand and interpret its content in accordance with generally accepted engineering or
other professional standard and applicable regulations.
No recommendation as to products and services or vendors is made or should be implied.
No representation or warranty is made that this deliverable is technically accurate or sufficient or conforms to any law
and/or governmental rule and/or regulation and further, no representation or warranty is made of merchantability or fitness
for any particular purpose or against infringement of intellectual property rights.
In no event shall ETSI be held liable for loss of profits or any other incidental or consequential damages.

Any software contained in this deliverable is provided "AS IS" with no warranties, express or implied, including but not
limited to, the warranties of merchantability, fitness for a particular purpose and non-infringement of intellectual property
rights and ETSI shall not be held liable in any event for any damages whatsoever (including, without limitation, damages
for loss of profits, business interruption, loss of information, or any other pecuniary loss) arising out of or related to the use
of or inability to use the software.
Copyright Notification
No part may be reproduced or utilized in any form or by any means, electronic or mechanical, including photocopying and
microfilm except as authorized by written permission of ETSI.
The content of the PDF version shall not be modified without the written authorization of ETSI.
The copyright and the foregoing restriction extend to reproduction in all media.

© ETSI 2025.
All rights reserved.
ETSI
3 ETSI GR ENI 036 V4.1.1 (2025-08)
Contents
Intellectual Property Rights . 4
Foreword . 4
Modal verbs terminology . 4
Introduction . 4
1 Scope . 5
2 References . 5
2.1 Normative references . 5
2.2 Informative references . 5
3 Definition of terms, symbols and abbreviations . 6
3.1 Terms . 6
3.2 Symbols . 6
3.3 Abbreviations . 6
4 Overview . 6
4.1 Introduction . 6
4.2 Architecture . 7
4.2.1 Space-Ground Cooperative Network Slicing Architecture . 7
4.2.2 Space-ground Slicing Session Collaboration . 7
4.2.3 Intelligent Slice Mapping . 11
4.2.4 Intelligent slicing technology for space-ground collaborative network resources on demand . 14
5 Network Structure . 17
5.1 Forward and Backhaul Link . 17
5.2 Identification and restriction of satellite access types . 17
5.3 UE location Identification . 18
6 Mathematical model . 20
6.1 Mathematical model introduction . 20
6.2 Mathematical model based on the slices of the optimal weighted graph matching . 22
6.3 Simulation and verification of network resource slicing generation technology . 22
6.4 Network resource slicing scheduling demand forecasting technology . 24
6.4.1 Introduction. 24
6.4.2 Mathematical model for slice prediction based on spatiotemporal correlation . 24
6.4.3 Design of network resource slice migration scheme . 25
7 Conclusion and recommendations . 26
History . 28

ETSI
4 ETSI GR ENI 036 V4.1.1 (2025-08)
Intellectual Property Rights
Essential patents
IPRs essential or potentially essential to normative deliverables may have been declared to ETSI. The declarations
pertaining to these essential IPRs, if any, are publicly available for ETSI members and non-members, and can be
found in ETSI SR 000 314: "Intellectual Property Rights (IPRs); Essential, or potentially Essential, IPRs notified to
ETSI in respect of ETSI standards", which is available from the ETSI Secretariat. Latest updates are available on the
ETSI IPR online database.
Pursuant to the ETSI Directives including the ETSI IPR Policy, no investigation regarding the essentiality of IPRs,
including IPR searches, has been carried out by ETSI. No guarantee can be given as to the existence of other IPRs not
referenced in ETSI SR 000 314 (or the updates on the ETSI Web server) which are, or may be, or may become,
essential to the present document.
Trademarks
The present document may include trademarks and/or tradenames which are asserted and/or registered by their owners.
ETSI claims no ownership of these except for any which are indicated as being the property of ETSI, and conveys no
right to use or reproduce any trademark and/or tradename. Mention of those trademarks in the present document does
not constitute an endorsement by ETSI of products, services or organizations associated with those trademarks.
DECT™, PLUGTESTS™, UMTS™ and the ETSI logo are trademarks of ETSI registered for the benefit of its
Members. 3GPP™, LTE™ and 5G™ logo are trademarks of ETSI registered for the benefit of its Members and of the
3GPP Organizational Partners. oneM2M™ logo is a trademark of ETSI registered for the benefit of its Members and of ®
the oneM2M Partners. GSM and the GSM logo are trademarks registered and owned by the GSM Association.
Foreword
This Group Report (GR) has been produced by ETSI Industry Specification Group (ISG) Experiential Networked
Intelligence (ENI).
Modal verbs terminology
In the present document "should", "should not", "may", "need not", "will", "will not", "can" and "cannot" are to be
interpreted as described in clause 3.2 of the ETSI Drafting Rules (Verbal forms for the expression of provisions).
"must" and "must not" are NOT allowed in ETSI deliverables except when used in direct citation.
Introduction
Space-ground cooperative network includes the mobile communication network on the ground and the satellite network
in the space, and the slicing configuration rules of the two networks are different. A slicing adaptation technology
connecting mobile communication network and satellite network can effectively support the requirement of the
end-to-end slicing service guarantee for space-ground cooperative network. Through the adaptation mapping of data
plane and the collaborative management of control plane for Network Slicing, it can improve the customized service
capability of space-ground cooperative network for differentiated services.

ETSI
5 ETSI GR ENI 036 V4.1.1 (2025-08)
1 Scope
The present document intends to describe a method of network architecture and slicing mapping for the interconnection
between the mobile communication network slicing and satellite network slicing. The detailed plan includes:
• Support identity resolution such as VLAN and IP address on the data plane, support precise identification and
control for user services, and realize the slicing adaptation between mobile communication network on the
ground and satellite network.
• Exchange the slicing control information, using the control plane of ground mobile communication network
and satellite network (5GC and Satellite Network Operation Control Center (SNOCC)), optimize the global
service quality of service for the network slicing, and ensure the consistency and continuity of slicing service
in space-ground cooperative network environment.
• Leverage Graph Convolutional Networks (GCN) and Gated Recurrent Unit (GRU) to predict traffic patterns
and optimize slice-resource mapping in real time.
The present document will deliver research and investigation activities and insights that will further explore the related
techniques that can be used to employ connection improvement for space-ground network slicing.
2 References
2.1 Normative references
Normative references are not applicable in the present document.
2.2 Informative references
References are either specific (identified by date of publication and/or edition number or version number) or
non-specific. For specific references, only the cited version applies. For non-specific references, the latest version of the
referenced document (including any amendments) applies.
NOTE: While any hyperlinks included in this clause were valid at the time of publication, ETSI cannot guarantee
their long-term validity.
The following referenced documents may be useful in implementing an ETSI deliverable or add to the reader's
understanding, but are not required for conformance to the present document.
[i.1] ETSI GR ENI 004: "Experiential Networked Intelligence (ENI); Terminology".
[i.2] ETSI GS ENI 005: "Experiential Networked Intelligence (ENI); System Architecture".
[i.3] ETSI GR ENI 008: "Experiential Networked Intelligence (ENI); InTent Aware Network
Autonomicity (ITANA)".
[i.4] NIST Special Publication 800-207: "Zero Trust Architecture".
[i.5] ETSI TS 138 413: "5G; NG-RAN; NG Application Protocol (NGAP) (3GPP TS 38.413)".
[i.6] ETSI TS 123 008: "Digital cellular telecommunications system (Phase 2+) (GSM); Universal
Mobile Telecommunications System (UMTS); LTE; 5G; Organization of subscriber data (3GPP
TS 23.008)".
[i.7] ETSI TS 123 501 (V17.8.0): "5G; System architecture for the 5G System (5GS) (3GPP TS 23.501
version 17.8.0 Release 17)".
[i.8] ETSI TS 123 273: "5G; 5G System (5GS) Location Services (LCS); Stage 2 (3GPP TS 23.273)".
ETSI
6 ETSI GR ENI 036 V4.1.1 (2025-08)
[i.9] ETSI TS 123 122: "Digital cellular telecommunications system (Phase 2+) (GSM); Universal
Mobile Telecommunications System (UMTS); LTE; 5G; Non-Access-Stratum (NAS) functions
related to Mobile Station (MS) in idle mode (3GPP TS 23.122)".
[i.10] ETSI TS 124 501: "5G; Non-Access-Stratum (NAS) protocol for 5G System (5GS); Stage 3
(3GPP TS 24.501)".
3 Definition of terms, symbols and abbreviations
3.1 Terms
For the purposes of the present document, the terms given in ETSI GR ENI 004 [i.1] and ETSI GS ENI 005 [i.2] apply.
3.2 Symbols
Void.
3.3 Abbreviations
For the purposes of the present document, the abbreviations given in ETSI GR ENI 004 [i.1], ETSI GS ENI 005 [i.2]
and ETSI GR ENI 008 [i.3] apply.
4 Overview
4.1 Introduction
With the evolution of network services, it has become increasingly challenging to dynamically match network resources
to the diverse and simultaneous demands of network capacity. Multiple Network Slicing (NS) technology brings an
excellent solution for the mismatch between supply and demand of network capacity. NS virtualizes multiple network
slices within a network to provide customized services tailored to diverse performance requirements. It can not only
meet the performance requirements of different services, but also maximize the network resource utilization, save the
cost of network construction, and improve the profitability of operators. Finally, it achieves the network service and
cost-benefit balance.
To meet diverse user requirements, NS integrates various network elements on a shared physical platform to create
independent, end-to-end service subnets. While the underlying architecture is modular and flexible, NS enables tightly
integrated service delivery for a wide range of business demands. In recent years, NS technology has advanced rapidly
across wireless access, mobile core, IP bearer, satellite and other network environments. However, the interconnection
of network slices across heterogeneous domains, such as terrestrial and satellite networks, remains a significant
challenge. The configuration and management rules for slicing differ between these environments. Therefore, adaptive
slicing technologies that bridge mobile and satellite networks are essential for delivering consistent, end-to-end service
guarantees in space-ground cooperative networks. By enabling adaptive mapping at the data plane and the collaborative
management at the control plane, these technologies enhance the ability of space-ground networks to deliver tailored,
high-quality services.
ETSI
7 ETSI GR ENI 036 V4.1.1 (2025-08)
4.2 Architecture
4.2.1 Space-Ground Cooperative Network Slicing Architecture
• The space-ground cooperative network slicing architecture is shown in Figure 1. The architecture includes the
deployment of a programmable slicing gateway and a space-ground cooperative slicing control system,
positioned between the terrestrial mobile communication network and the satellite network. The programmable
slicing gateway is the transit channel for the slicing service data flows. With definable message parsing (the
capability of a gateway to be programmed to identify, interpret, and extract specific information from data
packets based on a given configuration policy, enabling precise control and forwarding of different slicing
services), processing and forwarding capabilities, the gateway accurately identifies and controls slicing
services, and achieves the data mapping between slices according to the configuration policy provided by the
control system. It can ensure the service consistency and continuity of service data in space-ground
cooperative network slicing and realize the adaptation of heterogeneous network slices. Example of
heterogeneous network slices:
• QoS Parameters: A slice for Ultra-Reliable Low-Latency Communication (URLLC) like autonomous vehicle
control would prioritize stringent latency and reliability guarantees, while a slice for massive IoT would
prioritize connection density and energy efficiency over low latency.
• Session Management: A terrestrial mobile network slice might establish a session with a simple handover
between cell towers, while a space-ground slice requires complex, predictive session management to handle
handovers between satellites moving at high orbital speeds and ground stations.
The space-ground cooperative slicing control system interacts with the space-ground network slicing control planes to
open up the slicing session channel between the space and ground network. Taking into account the differences between
mobile communication network and satellite network in slicing service classification, slicing quantity and slicing
construction form, the control system can optimize the matching mode of service traffic and network resources, and
intelligently generate the configuration policy of the programmable slicing gateway, thus improving the end-to-end
quality of slicing service in space-ground cooperative network. Optimize the matching mode of service traffic and
network resources can be realised by the following methods:
• Dynamic Resource Allocation: The control system could direct high-bandwidth, non-delay-sensitive traffic
(e.g. software updates for ships) to a satellite slice with abundant bandwidth but higher latency, while reserving
scarce, low-latency terrestrial resources for real-time video calls.
• Predictive Traffic Steering: For a user on a high-speed train, the system could predict the route and pre-
emptively allocate resources and establish sessions on satellite network slices to maintain service continuity
before the terrestrial network connection is lost.
Session      Session
establish release
Slicing session
Session Cooperative
Cooperative unit
NRF NSSF AUSF UDM NWDAF S-NSSF S-AUSF S-UDM
Ground slicing view Intelligent Space-based slice view
Intelligent slicing
decision
S-AMF S-SMF S-PCF S-AF
AMF SMF PCF NEF AF
Mapping unit
Traffic prediction Network Resources Statistics
5GC  Space-ground cooperative slicing control system  s-NOCC

Ground Space Space- Ground-
message message ground space
5G Slicing     ble        able
Satellite
ina           ramm
Message Processing
Prog
Def
slicing
parsing
forwarding
Satellite network
Programmable slice gateway
5G user plane   user plane
Figure 1: Space-ground cooperative network slicing architecture
4.2.2 Space-ground Slicing Session Collaboration
The main function of slice-session collaboration is to coordinate the management of PDU sessions in mobile
communication network and satellite network, and establish PDU session channels from UE to ground-based 5G mobile
communication network, space-based satellite network and up to Data Network.
ETSI
8 ETSI GR ENI 036 V4.1.1 (2025-08)
Ground-based Space-based
core network Session co-processing core network
interface interface
module
module module
Ground-based Space-based
5G mobile
Ground Space
communication 切片Sess会话ion m协同app单in元g satellite network
-based -based
management module
network
Session session
Core network view view Core network
Space-ground session mapping library

Figure 2: Slicing session collaboration architecture
As shown in Figure 2, the functional modules of slicing session collaboration unit include slicing mapping management
module, session collaborative processing module, ground-based core network interface module, and space-based core
network interface module. The slice mapping management module is mainly responsible for maintaining the mapping
relationship between ground-based PDU sessions and space-based PDU sessions. The session cooperative processing
module can cooperate with the process of establishing, modifying and releasing sessions of ground-based and
space-based networks, according to the mapping relationship maintained by the slice mapping management module.
The interface module of ground-based core network is responsible for the interface with the core network of
ground-based 5G mobile communication network. The space-based core network interface module is responsible for the
interface with the space-based satellite network core network.
The establishment process of UE-initiated PDU sessions is used as an example to illustrate the slicing session
collaboration process. In the following example, assuming that the mapping rule is based on service type, UE1 and UE2
initiate PDU sessions of the same type to access Data Network.
UE1
Ground-based Ground-based Slicing session Space-based
Space-based
RAN
DataNetwork
core network UPF cooperative unit core network
UPF
1. PDU session
establishment
request
2. Ground-base core
network internal
operation, select SMF
and UPF, etc
3. Establish N4 session with UPF
4. Notifies currently established
ground-based session
5. Select the space-
Build RAN tunnel
based session ID
6. Notify and establish
space-based session
7. Perform
operations inside
the core
network, such as
SMF and UPF
8. Establish N4
session with UPF
Figure 3: Data channel establishment process of UE1
For the PDU session initiated by UE1, the data path establishment includes three stages, as shown in Figure 3.
The first stage is PDU session establishment process from UE to ground mobile communication network:
Step1: UE1 initiates a PDU session establishment request.
Step2: The request is processed by the ground-based core network, and is used to selects the
ground-based SMF and UPF for the session.
Step3: The ground-based core network establishes an N4 session with the selected ground-based UPF.
Step 4: The ground-based core network notifies the space-ground cooperative session management unit of
the currently established ground-based session information.
ETSI
9 ETSI GR ENI 036 V4.1.1 (2025-08)
The ground-based core network notifies the information about the current ground-based session to space-ground session
management unit. At the same time, the ground core network notifies RAN and users to build RAN tunnels.
slicing session collaborative unit for slicing mapping:
The second stage is the
Step 5: After receiving the notification from the ground-based core network, the slicing session
collaborative unit carries out the space-based session mapping.
Since the session of UE1 is a new service type, a new space-based session ID needs to be assigned to the session of
UE1.
The third stage is the PDU session establishment process of the satellite network:
Step 6: The slicing session collaboration unit notifies the establishment of a new space-based session to
the space-based core network.
Step 7: The space-based core network selects the space-based SMF and UPF for the session after
receiving a session establishment notification.
Step 8: The space-based core network establishes the N4 session with the selected space-based UPF.
At this point, for the PDU sessions of UE1, the channel from UE1 to ground-based 5G mobile communication network,
space-based satellite network and up to Data Network has been established and opened.
UE2 UE1
Ground-based Ground-based Slicing session Space-based
Space-based
RAN
DataNetwork
core network
UPF cooperative unit core network UPF
1. PDU session
establishment
request
2. Ground-base core
network internal
operation, select SMF
and UPF, etc
3. Establish N4 session with UPF
4. Notifies currently established
ground-based session
5. Space-ground session
mapping, and aggregate
Build RAN tunnel
ground-based sessions of
UE1 and UE2
6. Notify and establish
space-based session
7. Modify internal
session of space-
based core network
8. Notify UPF for
modification
Figure 4: Data channel establishment process of UE2
As shown in Figure 4, after UE1 has established the channel to the Data Network, and when UE2 intends to access the
Data Network, the establishment of the data channel also includes three stages, as shown in Figure 4.
The first stage is the process of PDU session establishment from UE to ground-based mobile communication network.
Step1~Step4: The process of ground-based network for the session establishment request of UE2 is the same as
that of UE1.
The second stage is the slicing session collaborative unit for slicing mapping:
Step 5: The slicing session collaborative unit carries out the space-ground session mapping after receiving
the notification of the ground-based core network. Based on resource allocation, the control unit
determines that UE2 and UE1 sessions can be aggregated into the same space-based session (e.g. if
they share the same service type and QoS requirements).
The third stage is the PDU session establishment process of the satellite network:
Step 6: The slicing session collaborative unit notifies the space-based core network to modify the
space-based session, and the modification can be for QoS parameters.
ETSI
10 ETSI GR ENI 036 V4.1.1 (2025-08)
Step 7: The space-based core network performs a modification operation for the session after receiving the
session modification notification.
Step 8: The space-based core network notifies the corresponding space-based UPF to perform session
modifications.
At this point, for the PDU sessions of UE2, the channel from UE2 to ground-based 5G mobile communication network,
space-based satellite network and up to Data Network has been established and opened. PDU sessions of the same
service type in UE1 and UE2 are allocated to the same slice, and the slicing sessions terminates (this refer to the end of
the process).
ETSI
11 ETSI GR ENI 036 V4.1.1 (2025-08)
4.2.3 Intelligent Slice Mapping
In the space-ground cooperative network, there are many types of service requirements. The performance requirements
of services such as real-time voice, data transmission, control signalling, and short message have different performance
requirements, and the service delay, bandwidth, and security requirements all change in real time. To meet the
differentiated application requirements of wide-area information networks, the space-ground cooperative network needs
to dynamically construct differentiated network slices involving different service characteristics, accurately match the
resource requirements of different service data, and realize multi-service converged application.
The mechanism achieves this through the intelligent, real-time decision-making capabilities of the Space-Ground
Cooperative Slicing Control System introduced in clause 4.2.1. This system acts as the central orchestrator:
• Reconciling Conflicting QoS: The control system can prioritize and reconcile conflicting requirements by
leveraging the heterogeneous resources of both terrestrial and satellite networks, as described in the "Optimize
the matching mode" examples (clause 4.2.1). For instance, a service with strict low-latency requirements (e.g.
real-time voice) would be matched to a terrestrial network slice or a Low-Earth Orbit (LEO) satellite link with
minimal delay. A service demanding high bandwidth but tolerating higher latency (e.g. data transmission for
software updates) would be directed to a geostationary (GEO) satellite slice with abundant bandwidth. The
control system intelligently makes this choice based on its global view of all network resources and service
requirements.
• Orchestration: The orchestration is performed by this control system. It is responsible for the end-to-end
lifecycle management of slices across both domains. It "intelligently generates the configuration policy of the
programmable slicing gateway" and, by interacting with both ground and space core network control planes, it
orchestrates the overall resource matching and slice mapping process.
The space-ground cooperative network proposes an intelligent slice mapping mechanism based on spatial-temporal
correlation.
The intelligent slice mapping mechanism is the brain, while the programmable gateway and session collaboration unit
are the executing limbs:
• Connection to the Programmable Slicing Gateway (clause 4.2.1): The control system (which houses the
intelligent mapping logic) generates the configuration policy for the gateway. Based on the mapping
decisions, it commands the gateway on how to perform "data mapping between slices" - for example, which
specific service flows to steer onto which network paths (terrestrial or satellite) to meet their QoS
requirements. The gateway executes these policies using its "definable message parsing" capability.
• Connection to the Session Collaboration Unit (clause 4.2.2): The session collaboration unit is a key
functional component that implements the mapping decisions for session management. The intelligent
mapping mechanism likely provides the mapping rules and logic (e.g. "based on service type" or "based on
resource allocation") that the slice mapping management module within the session collaboration unit follows.
This is shown in the UE1/UE2 example where the session unit decides to create a new space-based session for
a new service type (UE1) or aggregate sessions into an existing one (UE2) based on these rules.
Traffic prediction is used to establish the prediction model of resource demand of network services:
• Model Type: The spatial-temporal correlation depend on the use of advanced ML models capable of analyzing
patterns across both time (e.g. time of day, network usage cycles) and space (e.g. user location, satellite
coverage, ground network congestion). Models used including the following:
- Time Series Forecasting models (e.g. LSTMs - Long Short-Term Memory networks) were used to
predict traffic load fluctuations.
- Reinforcement Learning models, where the control system learns optimal mapping and resource
allocation policies through continuous interaction with the network environment.
- Graph Neural Networks were used to model the complex topology of the space-ground network and
optimize resource paths.
ETSI
12 ETSI GR ENI 036 V4.1.1 (2025-08)
This enables the space-ground cooperative network to respond to the service characteristics and the transformation of
access node in real time. Thus, the slices of network resources can be matched as needed with the fluctuating traffic in
the space-ground cooperative network.

Figure 5: Intelligent slice mapping based on spatial-temporal correlation
Figure 5 shows the smart slice mapping diagram based on spatial-temporal correlation. Graph Convolutional Network
(GCN) and Gated Recurrent Unit (GRU) are used to extract the temporal-spatial characteristics of the historical traffic
load of each node in the space-ground cooperative network slicing, which is to provide a decision basis for slice
mapping. Firstly, the network topological features are captured by GCN to obtain the spatial dependence. Secondly, the
dynamic changes of node attributes are captured by GRU to obtain the local time trend of traffic load. Finally, the
multi-output fully connected layer of artificial neural network is used to realize the transformation from traffic load to
resource demand, and output the predicted result. The system monitors the network resource status in real time, slices
are allocated network resources based on the predicted results of slicing service requirements to complete slicing
adaptation decisions.
The real-time monitoring of network resource status is achieved through a comprehensive telemetry data collection
framework. The Space-Ground Cooperative Slicing Control System acts as a central brain, continuously gathering and
analyzing real-time performance and status metrics from all key network elements across both terrestrial and satellite
domains. This is done through integrated agents and standardized interfaces:
• Terrestrial Network (Ground-based UPF & RAN): The system collects real-time telemetry on Key
Performance Indicators (KPIs) such as bandwidth utilization, port statistics, packet loss, and session
load from User Plane Functions (UPFs), and radio resource block usage, connected User Equipment (UE)
count, and handover events from the Radio Access Network (RAN).
• Satellite Network: The system interfaces with the satellite network's core and management systems to
monitor per-beam capacity utilization, Signal-to-Noise Ratio (SNR) metrics for different geographic
areas, satellite transponder load, and propagation delay characteristics.
• Unified View: This constant stream of spatial-temporal data provides the control system with a unified,
real-time view of the entire integrated network's health, capacity, and congestion points, forming the ground
truth against which AI predictions are validated and resource allocation decisions are executed.
ETSI
13 ETSI GR ENI 036 V4.1.1 (2025-08)
Figure 5 shows the smart slice mapping diagram based on spatial-temporal correlation. Graph Convolutional Network
(GCN) and Gated Recurrent Unit (GRU) form the core of a predictive AI model that extracts the temporal-spatial
characteristics from the historical traffic load of each node (e.g. base stations, satellite gateways, UPFs) in the
space-ground cooperative network. This model provides the intelligent decision basis for proactive slice mapping and
resource allocation. The process, also illustrated in Figure 5, works as follows:
1) Spatial Feature Extraction via GCN: The GCN layers process the physical and logical network topology.
Nodes represent network elements (e.g. gNBs, Satellites, UPFs), and edges represent the links between them.
The GCN effectively captures the spatial dependence and correlations between neighboring nodes,
understanding how traffic congestion in one cell might impact adjacent cells or a satellite beam.
2) Temporal Feature Extraction via GRU: The time-series data of historical traffic load for each node is fed
into the GRU layers. The GRU is adept at capturing dynamic changes and local time trends, learning
patterns such as daily usage cycles, periodic bursts of traffic, and long-term growth trends.
3) Prediction and Decision Output: The combined spatio-temporal features from the GCN and GRU are then
fed into a multi-output fully connected layer of an artificial neural network. This layer performs the
non-linear transformation from abstract features into a concrete predicted resource demand (e.g. required
Gbps of bandwidth, number of required network slices) for each segment of the network, outputting the final
forecast result.
4) Closed-Loop Resource Allocation: These predicted outputs are fed directly into the resource allocation
decisions of the control system. The system proactively instructs the programmable slicing gateway and
session collaboration unit to pre-emptively configure slices and allocate resources before the predicted demand
arrives.
5) The Feedback Loop for Real-Time Monitoring: Crucially, this is a closed-loop system. The
system monitors the network resource status in real-time via the telemetry framework described above.
This real-world data on actual resource utilization, Bit Error Rate (BER), and achieved data rates is
continuously fed back as new input into the AI model. This allows the model to compare its predictions with
reality, automatically learn from any discrepancies, and refine its future predictions, creating a self-optimizing
loop. Slices are thus dynamically adapted and network resources are allocated based on this continuous cycle
of prediction and observation, ensuring efficient resource utilization and meeting stringent end-to-end Quality
of Service (QoS) requirements.
In figure 5, the desired system performance can benefit from an adaptive mechanism of slicing. A well-designed slicing
algorithm takes into account the (unavoidable) tradeoffs between bandwidth and power efficiency). Haiyuan Li et al.
(arXiv:2310.17523) proposed a Multi-Agent Deep Reinforcement Learning (MADRL) Approach, highlighted a
state-of-the-art method that addresses the trade-offs between bandwidth and power efficiency through intelligent,
adaptive slicing algorithms that can be applied in the present document.
The slicing scheme here was initially intended to mitigate the data rate drop in a generic link, where interference is
present. The candidate switching schemes have been chosen based on the merit of combined power and bandwidth
efficiency. More realistic models will also be addressed in this. For model simplicity and discussion continuity, the
following assumptions are made:
a) The channel is a direct link channel.
b) Synchronization is maintained.
c) The system operates at 2 possible data rates (moderate and high).
The direct-link assumption simplifies initial validation; future iterations will incorporate 3D channel models to account
for Doppler, polarization and atmospheric effects.
-9 -7
In the test model two sets of BER test data will be used. The first set will range from 10 to 10 , to represent moderate
-8 -6
to 10 , to represent severe degradation. The data rate will be chosen as
degradation and second set will range from 10
a realistic 4 Mbps and a higher rate of 250 Mbps.
ETSI
14 ETSI GR ENI 036 V4.1.1 (2025-08)
Table 1: Data rate degradation in Space Ground network
Moderate Degradation (BER) Severe Degradation (BER)
-9 -7 -8 -6
10 → 10 10 → 10
Initial Rb (Mbps) Final Rb (Mbps) Initial Rb (Mbps) Final Rb (Mbps)
4 3 4 2,9
250 187,9 250 179,4
Assumptions (a-c) represent a controlled test case; future work will address multipath and mobility, 4 Mbps (narrowband
IoT) and 250 Mbps (broadband video) reflect 3GPP NTN use cases.
In Table 1, while in the moderate degradation, the data rates fell for 4 Mbps and 250 Mbps respectively to 3 Mbps and
187,9 Mbps, which is about 25 % of throughput loss; for severe degradation, the data rate fell for 4 Mbps and 250 Mbps
respectively to 2,9 Mbps and 179,4 Mbps, which is about 28 % of throughput loss.
Comparing data rate loss for systems operating at 4 Mbps, the difference between the moderate and severe model is not
significant. In contrast, for a system operating at 250 Mbps, the data rate loss between the moderate and severe model
indicates a large departure from the operating speed. This suggests that systems operating at a higher speed are more
susceptible to environmental change than their lower speed counterparts. The comparisons between these two data rate
bands can be further demonstrated.
4.2.4 Intelligent slicing technology for space-ground collaborative network
resources on demand
This clause explores intelligent slicing technology for on-demand space-ground collaborative network resource
allocation, focusing on efficient slice generation to meet diverse application requirements and a mathematical model
using optimal weighted graph matching to optimize the mapping process. The need to dynamically construct
differentiated network slices tailored to specific business characteristics, ensuring accurate alignment between resource
capabilities and service demands were also addressed. The importance of timeliness in slice mapping algorithms,
particularly when handling large volumes of real-time service requests, while also balancing node load and link
bandwidth to prevent network overload were emphasised. Additionally, a mathematical approach grounded in
adjacency matrix feature vector decomposition to formulate slice mapping as an optimal weighted graph matching
problem was introduced. This model translates resource slicing requirements into a virtual topology representation,
enabling efficient, simultaneous processing of multiple slice requests and enhancing the scalability and responsiveness
of space-ground collaborative networks:
1) Efficient generation technology of network resource slicing:
In order to meet the requirements of differentiated wide-area information network applications, the generation
and scheduling of intelligent slices of space-ground collaborative network. This generation needs to
dynamically construct differentiated network slices involving different business characteristics. It also,
accurately matches the resource requirements of different business data requirements, and realizes
multi-service integration applications. In the process of providing a wide variety of network services, the
space-ground synergy network needs to deploy a large number of network resource slices with different
functional sequences to meet the real-time needs of users, and the timeliness of the slice mapping algorithm is
very important in the face of a large number of real-time service requests. In addition, due to the limited
processing power and link bandwidth of nodes in the network, once some nodes and links are overloaded, the
carrying capacity of the network will be reduced. Therefore, the efficient generation technology of network
resource slicing needs to complete the rapid mapping of a large number of service requests while balancing the
node load and link bandwidth.
The "differentiated wide-area information network applications" refers to distinct types of data services that span large
geographical areas (wide-area) and have vastly different performance requirements (differentiated). These applications
cannot be served by a one-size-fits-all network and instead require dedicated, customi
...