Service Customized Space-Air-Ground Integrated Network for Immersive Media: Architecture,Key Technologies,and Prospects

Linhui Wei,Jiacheng Shuai,Yu Liu,2,*,Yumei Wang,2,Lin Zhang3,

1 School of Artificial Intelligence,Beijing University of Posts and Telecommunications,Beijing 100876,China

2 Research Center of Networks and Communications,Peng Cheng Laboratory,Shenzhen 518000,China

3 Beijing Big Data Center,Beijing Municipal Bureau of Economy and Information Technology,Beijing 100101,China

Abstract: The space-air-ground integrated network(SAGIN) is regarded as the key approach to realize global coverage in future network and it reaches broad access for various services.Being the new paradigm of service,immersive media(IM)has attracted users’attention for its virtualization, but it poses challenges to network performance, e.g.bandwidth, rate, latency.However, the SAGIN has limitations in supporting IM services, such as 4K/8K video, virtual reality,and interactive games.In this paper,a novel service customized SAGIN architecture for IM applications (SAG-IM) is proposed, which achieves content interactive and real-time communication among terminal users.State-of-the-art research is investigated in detail to facilitate the combination of SAGIN and service customized technology, which provides endto-end differentiated services for users.Besides, the functional components of SAG-IM contain the infrastructure layer,perception layer,intelligence layer,and application layer, reaching the capabilities of intelligent management of the network.Moreover, to provide IM content with ultra-high-definition and high frame rate for the optimal user experience,the promising key technologies on intelligent routing and delivery are discussed.The performance evaluation shows the superiority of SAG-IM in supporting IM service.Finally,the prospects in practical application are highlighted.

Keywords: space-air-ground integrated network; architecture; immersive media; service customized network;intelligent

I.INTRODUCTION

The fifth-generation(5G)network provides users with low latency services,preparing for the next generation of media service[1].The bandwidth demands on immersive media (IM) such as 4K/8K video and virtual reality(VR)increases rapidly[2].Due to the unstable network connection in urban areas and the high cost of broadband infrastructure construction in rural areas,it is difficult for people to enjoy real-time and low-cost IM services.

In order to solve the above problems, the concept of space-air-ground integration network(SAGIN)has been proposed [3].The SAGIN achieves great advances in realizing global network coverage.Satellite communication is an indispensable part to establish a global connection.Recently, deep space exploration and interstellar communication are hot missions over the world.In the space environment,lunar[4]or Mars[5]communication are realized by robots such as lunar detector and Mars rover.

The SAGIN combines the strength of heterogeneousnetworks to provide broad access.The detailed descriptions of different networks of SAGIN are shown in Table 1.The satellite networks mainly consist of the geostationary orbit (GEO) and low earth orbit(LEO) satellites.Different from terrestrial networks,the satellite networks have long propagation latency and reach a wide coverage.The aerial platforms are divided into high altitude platforms(HAPs)and low altitude platforms(LAPs).Flying objects with high mobility, such as unmanned aerial vehicles (UAVs) and balloons,enhance the network coverage for certain areas,especially in emergency scenarios.

Table 1. The detailed description of different networks.

However, providing high-quality service in SAGIN still face some problems.It is not a simple task to build an efficient and controllable SAGIN.In the integrated network, it needs cooperation among different segments to maintain service.Besides,the protocol is various in different segments.The transmission protocol is TCP/IP in the terrestrial network, which has designed different standards and architectures to meet the communication requirement.In satellite communication, the consultative committee for space data systems (CCSDS) protocol [6] contains advanced orbiting systems and IP over CCSDS.

Since the network status is complex and the node status is difficult to obtain in SAGIN,researchers turn their attention to software-defined networking (SDN)[7], which separates the control plane and the data plane.By introducing SDN into the mobile network[8], the SDN controller obtains the global view of the network topology to monitor the status information.Besides, network function virtualization (NFV)[9] enhances the reliability and agile of SAGIN.The hardware and software are decoupled to realize the dynamic resource deployment.

VR, interactive games, streaming media, and other new immersive services development greatly promote the demand for bandwidth and latency.It is urgent to meet users’individual requirements by customized network services.Service customized network(SCN)[10]is a new pattern of future network architecture,integrates the advantages of SDN and NFV technology,and tries to offer differentiated services with efficient resource scheduling.The concept of SCN is similar to the travel modes,such as aircraft,high-speed rail,and ordinary highway,people choose the affordable travel way under the maximum value of economic ability.This paper introduces the concept of SCN into SAGIN to realize the flexibility and efficiency architecture and provide users with a high-quality immersive experience.

In this paper, the service customized SAGIN architecture for IM services is proposed, called SAGIM.The difference of the proposed SAG-IM with other architecture is that it deeply integrates the multidimensional resources of SAGIN to support the realtime requirements of IM.The artificial intelligence(AI) mechanisms are considered to improve the resource management ability of the SAG-IM [11].The main contributions of this paper are summarized as follows:

•A detailed literature review of the SAGIN architecture is given.Based on the concept of SCN,SAG-IM and its functional components are proposed.The multi-layer SDN controller cluster and edge servers deployment strategy are designed.

•The intelligent mechanism is investigated to achieve high-quality IM services, such as 4K/8K video and interactive games.The key technologies on intelligent SAG-IM are analyzed.The sketches of routing strategy and edge content delivery are discussed.

•Performance evaluation is given to show the strength of the proposed SAG-IM.Configuration updating time is calculated to reflect the flexibility of the hierarchical architecture.The ability of SAG-IM to support representative IM service is verified.

The rest of this paper is organized as follows.Section II introduces the related work of SAGIN.Section III proposes the framework of SAG-IM architecture and its the functional components.Section IV discusses the promising key technologies.Section V gives the performance evaluation and analysis.Section VI presents the directions and challenges for future works and summarizes the paper.

II.RELATED RESEARCH ON SAGIN

SAGIN provides a collaborative information guarantee for various network applications.The terrestrial networks are the foundation of the SAGIN, the satellite networks,and aerial platforms are the extensions.In this section,the development of satellite communication system is reviewed and the literature overview of SAGIN in recent years is given.

2.1 Development of Satellite Communication System

SAGIN is the development trend of future network,and many companies have already used it for commercial purposes.TSAT system[12]consists of five GEO satellites that obtains data for the military.It makes use of laser communications and on-board packet routing to form the space high-speed backbone network,achieving high capacity information sharing.Project ISICOM[13]brought up by European Union is made up of three GEO satellites and HAPs to provide widecoverage services.It aims at the integration of future global communication networks.The company O3b also called the“other 3 billion”deploys 12-20 medium earth orbit(MEO)satellites at the height of 8,000 km.O3b is committed to providing ubiquitous Internet access for 3 billion people living in Africa, Asia, and South America.

Because of the low propagation latency, the LEO satellite constellation has gained widespread attention.Iridium [22] consists of 66 LEO satellites that orbits at the height of 780 km to provide global coverage.Globalstar [23] consists of 48 LEO satellites orbiting at the height of 1,414km.The propagation and process latency are less than 300 ms.Starlink mission[24]launched by SpaceX plans to realize low-cost access service using a network consisting of 12,000 LEO satellites.

Efforts on the standardization of satellite networks have been done by 3rd Generation Partnership Project(3GPP).The latest reports have described the importance of the combination of satellite networks with advanced technology.In TR.22.822 [25], the satellite networks can be implemented with transparent satellites(without on-board processing capabilities)and regenerative satellites (with on-board processing capabilities).In TR.23.737[26],the transparent satellite is equivalent to a radio frequency remote unit and fully transparent to the New Radio protocols.The regenerative satellite payload implements a full gNodeB to support a satellite-enabled NR-RAN.

Media Activities on Satellite and 5G.To meet the growing demands of users, satellite communication and 5G play an important role in the provision of truly ubiquitous geographic coverage.Media service is one of the most popular applications in 5G networks.The large capacity and high data rate of the satellite and 5G give the access chances for multimedia applications.

Some projects have been launched for satellite and 5G,which study the media activities.SaT5G[27]is a European Commission Horizon 2020 project, which brings satellite communication into 5G by defining optimal satellite based backhaul and traffic offloading solutions.SaT5G has demonstrated the effectiveness of edge delivery for multimedia content when the satellite is treated as the backhauling network.5GALLSTAR[28]funded by the European Commission Horizon 2020 EU-Korea programme is another project working on facilitating the integration of satellite networks into 5G.Besides,this project gives the demonstration environment for 8K and VR video streaming among three trial sites in France and Korea.

2.2 The Literature Review of SAGIN

Currently, state-of-the-art research on SAGIN architecture can be mainly classified into three categories.The first category is the prospect of potential SAGIN architectures, they analyze the resource allocation and communication protocol in the heterogeneous networks.The second category is the SDN based SAGIN architectures, which focuses on managing thenetwork infrastructures through intelligent management and orchestration systems.The third category is the SAGIN architecture in particular application scenarios.

For the first category,these proposed SAGIN architectures studied the ability in realizing seamless coverage[29].Shi et al.[30]discussed the performance of data delivery in cross-layer SAGIN architecture.Yao et al.[31] combined the concept of smart identifier with SAGIN to improve transmission efficiency and service effectiveness.Zhang et al.[15]presented the double-edge intelligent integrated architecture, which had strength in task offloading, content caching, and distribution.Xie et al.[16] proposed the edge computing enabled satellite-terrestrial networks,achieving resource computing in multi-layer architecture.

For the second category, researchers focus on the role of SDN controllers in SAGIN,including the location and function.There are usually two schemes of SDN controllers in the SAGIN,one is the centralized deployment and the other is hierarchical deployment.

Centralized controller deployment.Centralized controller deployment means to deploy controllers on the ground or satellites,realizing centralized management of the entire network.Bao et al.[32]presented an architecture called OpenSAN, which deployed the SDN controllers on GEO satellites.All routing strategies were made by the management plane and translated by GEO satellites.SD-STIN architecture proposed by Bi et al.[21] integrated space and terrestrial networks components and deployed controllers on the ground to achieve link discovery,topology control,and so on.In the SDSN architecture put forward by Yang et al.[33], controllers were deployed on the ground to improve resource management and achieve the goal of seamless handover.

Due to the high mobility of non-GEO satellites and the long-distance propagation latency of the intersatellite links (ISLs), this scheme cannot obtain network status information in time.But with the expansion of the network scale, it meets the bottleneck in this scheme.

Hierarchical controller deployment.Due to the limited resources in the centralized controller, it is hard to handle the requests originating from the data plane.Shi et al.[17] proposed the MLSTIN architecture and deployed the controllers in GEO satellites,HAPs, and GSs.The cross-domain SDN architecture reduced the time of configuration updating and routing decisions.Li et al.[18]proposed the SERvICE to maintain the global view of the network and forward routing decisions.The FRBSN presented by Sheng et al.[19]discussed the protocol translation and data forwarding.

The hierarchical deployment scheme deploys the SDN controllers in a multi-layer network, which exchanges information equally in the same layer.The results also outperform the centralized controller deployment scheme in network management.

For the third category, these architectures have strength in specific application scenarios.Internet of Vehicles (IoV) and Internet of things (IoT) have gained more attention with the development of 5G.The SSAGV [14] and SGIIN-IoV [34] based on IoV reached high reliability for vehicular service.Chien et al.[20] proposed an H-STIN architecture for IoT network and gave the potential technologies and challenges.Hong et al.[35]introduced the SAG IoT network paradigm characterized by high efficiency,flexible configuration,and high precision.Wang et al.[36]researched the hybrid SAGIN in emergency scenarios.

The development of relevant researches promotes the birth of a new architecture.Table 2 compares the difference of SAG-IM with other architectures in detail.

Table 2. The Comparison of SAG-IM and other architectures.

III.ARCHITECTURE AND COMPONENTS OF SAG-IM

In this section,the proposed SAG-IM architecture and its functional components are presented.The designed SAG-IM architecture makes it possible to provide endto-end IM services with low latency and high reliability.

3.1 Proposed SAG-IM Architecture

The emergence of new services brings massive data.For example, the explosive growth of video streaming poses challenges to the capacity of the network.Therefore,it is essential to provide a reliable network condition to meet the requirement of IM services.The architecture of SAG-IM is shown in Figure 1, which comprises the integrated network, controller clusters,and edge servers.

Figure 1. Illustration of the considered of SAG-IM architecture.

The integrated network is the core component of SAG-IM.It consists of space, air, and ground segment.The space segment includes GEO and LEO satellites.They communicate with each other through microwave or laser.LEO satellites support broadband Internet access for IM services.The air segment contains balloons, air planes, and UAVs, which achieves real-time network coverage for remote areas, especially in emergency scenarios.The ground segment realizes the inter-connected among various application scenarios.In this way, passengers in the urban area,or employees in the rural area,or adventurers in the remote area,can enjoy ubiquitous connections and global-area coverage.

In order to realize the efficient management of the SAG-IM, it is necessary to introduce the controller cluster into the integrated network.The hierarchical controller deployment strategy is designed.The SDN controllers are deployed on the server or in the cloud.Due to the heterogeneity of the SAG-IM, the communication protocols or standards are inapplicable in each segment.The SDN controllers in each segment achieve self-organization by collecting network state information,routing,and forwarding.To help the unified management of SAG-IM, the controllers of each segment are connected to the higher-level SDN controller,such as the central controller on the ground.

The edge servers with powerful computing have strength in providing low latency and high computational load services.Remote assistance is built on the efficient network connection, promoting the progress of interaction among patients, doctors, and devices.The entertainment services such as 3D games need to respond to user needs in a timely manner.Besides, these multi-party interactive applications require lower latency.The interactive data is processed on LEO satellites to save the downlink bandwidth.In addition,the edge servers provide the lower latency to all users in its coverage and respond to users’requests immediately.The computing platform is extended to the network edge,even the user terminal,meeting the requirements of multi-level network services.

3.2 Functional Components

According to the requirement in different IM scenarios,the service processing procedure based on SCN is considered.The proposed functional components architecture of SAG-IM is shown in Figure 2.It includes four parts: infrastructure layer,perception layer,intelligence layer,and application layer.

Figure 2. The functional components of SAG-IM.

As the core foundation of the SAG-IM architecture, the infrastructure layer provides the basic guarantee for the operation of the integrated network.The network infrastructure includes network entities and links, which guarantees the robustness of the network.In order to separate the services for different users,hardware resources are abstracted as communication,computing,and storage resource.These virtual resources make a guarantee for data processing and transmission of different services in the network.

The perspective layer collects the network state of the infrastructure layer,including nodes,links,and so on.The multi-dimensional traffic perception is used for network traffic prediction.Besides, the adaptive traffic scheduling is realized by the data calculation.In order to ensure the security of data transmission,it carries out real-time monitoring so that data is delivered safely in heterogeneous networks.

The intelligence layer is the brain of the architecture, which is responsible for scheduling the network resources effectively according to the current service requirements.It makes routing and forwarding policies to provide different levels of quality of service(QoS) guarantee for different IM requests.The hierarchical SDN controller deployment strategy achieves intelligent routing,throughput control,load balancing,and so on.The controllers in SAG-IM are deployed on data centers on the ground and satellites, in which the update and configuration performance have been greatly improved than centralized management.

The application layer guarantees the efficient operation of the SAG-IM, which contains the network services and application scenarios.It provides customized program interfaces for various technologies,supporting the dynamic deployment and management of applications by calling standardized interfaces like northbound interfaces.The network services are the technical support of the upper IM application,including cloud rendering,real-time interaction,and content delivery.

Through the scheduling of different modules in the network, it realizes the efficient utilization and flexible configuration of network resources,which lays the foundation for the promotion of IM.The SAG-IM also meets the requirement of massive information processing and achieves the demand of end-to-end service.

IV.PROMISING KEY TECHNOLOGIES ON INTELLIGENT SAG-IM

Intelligent information processing is considered a key technique to provide IM services with high quality in SAG-IM architecture.In this section, the promising key technologies about intelligent routing and content delivery are discussed.

4.1 Intelligent Routing Plan

The end-to-end multimedia routing plan is a potential key technology in SAG-IM.To achieve intelligent routing, the learning-based strategy for video streaming is emphasized.Through learning a large amount of data,the dynamic network traffic are predicted in time according to the current network state.In the SDNbased architecture,the learning-based routing strategy solves the problem in traffic engineering and achieves the optimum throughput and latency[37].

It is a tendency to optimize the performance of SAGIN using AI in recent years [11].AI provides various services with low latency and achieves the communication network intelligence [38].In the case of the vehicle in SAGIN,AI facilitates the decision of resource management in real-time[39].Since the timevarying topology of satellite networks has its shortcomings such as frequent handoff,it is still a challenge to implement AI-based routing in SAG-IM.Besides,the delay-sensitive service such as IM requires stable routing path.The accuracy routing policy for IM service should be considered in a dynamic environment.

Multi-path TCP (MPTCP) is regarded as an effective way to solve the efficient transmission of multimedia services in SAG-IM.The performance of MPTCP in the LEO satellite network has been studied to show the ability in achieving high throughput [40].The combination of deep reinforcement learning (DRL)and MPTCP makes routing more easier in the dynamic network.Figure 3 gives the scenario of DRL-MPTCP in SAG-IM.There are multiple flows from IM content providers to user terminals[41].The DRL-based neural network uses the state of the SDN controllers as input to help decision making[42].The monitor at terminal side monitors the user’s high quality of experience(QoE)in real time, which is the feedback of the current route as the improvement for the next route selection.

Figure 3. The DRL-based MPTCP routing strategy of SAG-IM.

4.2 Service-oriented Content Delivery

The SAG-IM is a large-scale network with dynamic and complex node states.There exist diverse QoS requirements when users request IM services.But it is difficult to guarantee the QoS of IM services.To improve the utilization of network resources, it is essential to design a suitable content delivery method in SAG-IM.

Edge computing is a potential way to solve computing task in large-scale network.To ensure users enjoy various services, network slicing is an indispensable technology in resource management [43].Each slice supports specific services and provides a diversified QoS capability for different requirements.In this way,communication,computing,and storage resources are abstracted to establish multi-dimensional SAG-IM resources.Mobile edge computing(MEC)improves the QoS and achieves low latency in the integrated network [44], which is gradually being integrated with AI to realize edge intelligence for IM content.

Multimedia providers and cloud platforms are the main source of traffic.The edge servers are deployed on each segment of SAG-IM to provide users with various IM applications(e.g.VR,gaming,video).Since there are various requirements from cloud servers directly, such as high bandwidth, intensive computing,and reliability, the DL-based edge computing is designed to execute computing tasks nearly the end devices, which promotes real-time content delivery in SAG-IM.The sketch of edge intelligence based on deep learning (DL) for SAG-IM is shown in Figure 4.Edge intelligence means the high resource utilization and rapid resource allocating[45].In this way,the computing tasks in the edge reduces the transmission latency obviously.

Figure 4. The sketch of edge intelligence based on deep learning in SAG-IM.

V.PERFORMANCE EVALUATION AND ANALYSES

To show the superiority of SAG-IM,the performance evaluation is given.In the evaluation,the topology of satellite networks is the two-layer constellation with GEO and LEO satellites.There are three GEO satellites evenly distributed on the earth,covering the lower layer satellites and ground stations.The LEO satel-lite network is a large-scale Walker delta constellation(80/4/1) with a height of 800 km and the inclination of 53°.The constellation is simulated in Satellite Tool Kit(STK)[46]at 1 Jul 2021 UTCG.The prototype of the integrated network is built on a Linux-based simulator named for Mininet [47], in which the satellite parameters are exported from STK.The RYU [48] is chosen as the SDN controller.The bandwidth of satellite and terrestrial links are set as 500 KBps and 800 KBps,respectively.

5.1 Configuration Updating

In the satellite network,three GEO satellites are used as controllers,and LEO satellites are only responsible for forwarding.The GEO satellites manage the satellite networks and regularly forward information to the main controller on the ground.When the main controller sends the network configuration information,the status of each satellite should be updated.Once the last LEO satellite is updated, the configuration command completes successfully.Table 3 shows the configuration updating time with different control deployment methods.It can be seen that the updating time of the centralized controller deployment takes more than 8 hours, while the updating time of the hierarchical controller deployment is only about 235 ms.

Table 3. Configuration updating time with different control deployment methods.

In the centralized controller deployment mode,only when the LEO satellites pass the ground station,the status information of the satellite can be updated.Therefore, the updating time is greatly increased.SAG-IM adopts the hierarchical controller mode,which achieves self-management in the satellite networks.Due to the wide coverage of GEO satellites,information are exchanged frequently while passing through the ground station.The configuration updating time mainly depends on the distance from ground stations to GEO satellites and GEO satellites to LEO satellites.In this way,SAG-IM effectively reduces the updating time and improves the flexibility of the network.

5.2 Multimedia Interaction

Once the satellite is deployed with the edge server,the information can be processed in orbit.In this way,the bandwidth of links between satellites and ground stations is saved.Besides, data is stored and computed in the satellites,which reduces the latency to some extend.This is a potential solution for solving the latency in upcoming applications such as VR and interactive games.

The edge servers in SAG-IM meet the need for lowlatency IM content delivery.The designed constellation does not achieve coverage in the polar area.The maximum and minimum RTT from the ground locations to satellites at different latitudes is between 6 ms and 22 ms.It should be noted that the latency is over 100 ms when the edge servers are set in the ground[49].

Interactivity is the most significant indicator for SAG-IM.When the users want to enjoy the IM service like integrative games with friends, the satellite network is a suitable choice.These interactive computing businesses should be deployed to the same server,so it needs to select the candidate edge server in the satellite network.Since the trajectory and direction of the satellite are predictable,the candidate edge servers can be selected adaptively.

For interactive applications, the stable handoff should be guaranteed to reduce latency and jitter.Figure 5 shows the duration time until the next handoff in different server selection approaches.The easiest way is to select the edge server with the shortest latency.It can be seen that the median of the duration time is about 250 s, but it results in frequent handoff.The longest connected approach reduces the handoff frequency by the cost of latency.

Figure 5. The duration time until next handoff in different server selection approaches.

5.3 Video Streaming Transmission

As one of the mainstream services of IM, the quality of video mainly depend on the network condition.The satellite network has the advantages of low latency and wide bandwidth.In the SAG-IM,providing users with high-quality video service is the core goal of the architecture.In this section,the performance to transmit video streaming is verified in an integrated network.

The video is transmitted adaptively according to the network conditions by using the DASH protocol[50].Figure 6 shows the transmission rate under different package loss rates (PLR) (0.15%, 0.5%, 1.5%, 3%,and 5%) in the satellite networks.It can be seen that the PLR leads to packet retransmissions and decreases the transmission rate.The transmission rate has a direct impact on the quality of video and users’QoE.

Figure 6. The transmission rate under different package loss rates.

Figure 7 shows the fluctuation of throughput in SAGIN scenario.The PLR is set as 0.15% to imitate the real satellite network.As can be seen from Figure 7, the terrestrial link is used for transmission at first.The gateway switch breaks down which causes the first handoff.The throughput falls to about 480 KBps at the second handoff.The traffic is transferred from terrestrial link to satellite link.But in the case of the terrestrial scenario,the throughput falls to 0 KBps at the second handoff.The flexible handoff in the integrated network keeps the relatively steady throughput and decreases the jitter for video streaming.

Figure 7. The throughput of the integrated network.

6.1 Secure Guarantee

VI.PROSPECTS AND CONCLUSIONS

In this section,the prospects and challenges to realize high-efficiency network management in SAG-IM are discussed.Besides, the conclusions of this paper is given.

IM services are widely used in the field of entertainment, which makes the immersive experience easier for everyone.The integrated network supports all kinds of IM terminals to be accessed at anytime and anywhere, but it also challenges information security.From the severity of video piracy, protect the copyright of IM applications in SAG-IM should not be neglected.

In order to ensure reliable IM service transmission,an efficient secure guarantee mechanism need to be established.In the infrastructure layer of SAG-IM,access authentication is the first step to ensure the security of the transmission content.When users request resources from resource storage nodes(such as MEC),the link is subject to various security threats such as eavesdropping, attacks, and so on.Effective safety guarantee is a significant approach to protect users’privacy and improves their awareness of copyright.

6.2 Service Reliability

Deploying edge servers brings opportunities and challenges in SAG-IM.The life cycle of satellites is limited.The onboard processing capability of satellites ensures the agility of the system and makes information transmission more convenient.But the cost of deploying satellites increases accordingly.

The edge servers need to support multi-party interaction,real-time video streaming,and so on.The dynamic environment of SAG-IM poses a challenge to the stability transmission of IM service.The fast speed of satellite movement and limited connection time lead to frequent handoff of servers.Fortunately, the satellite trajectory is predictable,and the corresponding algorithm can be proposed to solve these problems.

6.3 Conclusions

In this paper, the service customized architecture for SAGIN to support various IM services is studied.In particular, this paper reviews the literature of SAGIN architecture in detail and discusses their architectures,strength, and key indicators.On this basis, to support IM services more suitably, the service-oriented integrated architecture for IM(SAG-IM)is proposed.The hierarchical SDN controller deployment and edge server placement methods improve the network management ability.In the design of the functional components, a service-oriented processing procedure is put forward.Besides, the key technologies on intelligent SAG-IM are discussed,including routing and delivery.Furthermore,the performance evaluation on SAG-IM is given,analyzing the superiority in realizing flexible configuration updating and supporting IM service.At last, the potential directions and challenges of SAGIM in practical application are discussed.

ACKNOWLEDGEMENT

This work was supported by the National Key Research and Development Program of China(No.2019YFB1803103),and in part by the BUPT Excellent Ph.D.Students Foundation (No.CX2021113).The authors would like to thank the anonymous reviewers for their valuable comments and helpful suggestions.