Civil Aircraft Assisted Space-Air-Ground Integrated Networks: Architecture Design and Coverage Analysis

Shuxun Li,Qian Chen,Zhe Li,Weixiao Meng,*,Cheng Li

1 School of Electronics and Information Engineering,Harbin Institute of Technology,Harbin 150001,China

2 Research and Development Department,China Academy of Launch Vehicle Technology,Beijing 100076,China

3 Faculty of Engineering and Applied Science,Memorial University of Newfoundland,St.Johns A1C5S7,Canada

Abstract: In this paper, we propose a novel AIenabled space-air-ground integrated networks (SAGIN).This new integrated networks architecture consists of LEO satellites and civil aircrafts carrying aerial base stations, called “civil aircraft assisted SAGIN(CAA-SAGIN)”.The assistance of civil aircrafts can reduce the stress of satellite networks, improve the performance of SAGIN,decrease the construction cost and save space resources.Taking the Chinese mainland as an example, this paper has analyzed the distribution of civil aircrafts, and obtained the coverage characteristics of civil aircraft assisted networks(CAAN).Taking Starlink as the benchmark, this paper has calculated the service gap of CAAN, and designed the joint coverage constellation.The simulation results prove that the number of satellites in CAASAGIN can be greatly reduced with the assistance of civil aircrafts at the same data rate.

Keywords: AI-enabled space-air-ground integrated networks; civil aircraft assisted networks; satellite constellation design;joint coverage analysis

I.INTRODUCTION

With the continuous development of mobile communication technology,mobile users go for faster,higher and stronger communication requirements,and 5G era has come.Although people have a lot of imagination,the coverage problem of mobile communication networks has not been completely solved due to the geographical environment and infrastructure conditions.According to the Internet world state report, Internet users account for about 51%of the global population[1].In other words, 49%of the world’s population is still unable to access the Internet.In addition, more than 90%of the earth’s surface areas,such as oceans,deserts,mountainous and so on,still have no access to the Internet.

Satellite networks make the seamless coverage become possible.Compared with terrestrial networks,satellites have wider coverage range.Moreover, the construction cost and applicable service of satellite networks are relatively fixed.Besides, the on-board equipment improves and the costs of launching satellites decreases a lot.Therefore, the telecommunications industry has gradually expanded from ground to the space these years,giving birth to the idea of building a SAGIN.

Authors introduced the development history of SAGIN in detail [2], from which we can see that satellite networks have always maintained the core position, while other aerial communication platforms have gradually attracted attention.Satellites have always been very important in SAGIN because they have large coverage area,wide channel bandwidth and large communication capacity [3].In recent years,with the miniaturization of satellites,the costs of manufacturing and launching satellites have largely reduced.The low earth orbit (LEO) satellite constellation has become the mainstream, and many LEO mega-constellation schemes have been proposed.The most typical cases are OneWeb constellation and Starlink constellation.OneWeb constellation contains 720 satellites [4], providing seamless broadband access services,not only for terrestrial users but also for airborne users[5].Starlink constellation plans to launch 11,943 LEO satellites in order to achieve global seamless coverage [6].The average sum rate of a single satellite downlink can reach 20 Gbps [7].Theoretically, the delay of the network can be reduced to 25 ms,which is much lower than that of the current mainstream satellites,and even faster than some terrestrial networks.

Although the cost has decreased, with the increasing scale of constellation,the construction cost of SAGIN is still very high.And a large number of satellites will increase the risk of space exploration.In order to reduce the risks and achieve global seamless coverage, airborne vehicles have received more attention[8], such as high-altitude unmanned aerial vehicles(UAVs),balloons and so on.A typical case is Google’s Project Loon.Project Loon uses a large number of balloons as aerial base stations,which are evenly distributed at an altitude of about 20,000 meters[9].Balloons are equipped with communication instruments,solar cells, and so on.Each single balloon is equivalent to a base station, which can realize the communication between balloons, balloons and the ground,balloons and satellites.Balloons and satellites form an integrated sky ground networks.

What’s more, it is also novel to use civil aircrafts or other airborne platforms as aerial base stations to form an access network [10, 11].In the joint communication of SAGIN, the stratospheric communication is more popular.Stratospheric communication is a relatively novel communication system.Compared with satellite networks, high altitude platform station(HAPS)system has the advantages of low cost,quick construction, low signal fading and high communication capacity.Compared with the traditional terrestrial cellular networks, HAPS has a larger coverage area and is not limited by geographical conditions [12].Therefore, in recent years, HAPS has gained worldwide attention.The high-altitude communication platform can be equipped with multi-beam antenna to cover different directions of the coverage area, supplemented by code division multiple access(CDMA)and space division multiple access(SDMA)technology, which improves the spectrum utilization and system capacity [13].The key technologies include multi-beam antenna,coverage scheme,link budget characteristics, and cell division, etc.Domestic scholars have done a lot of works on the area division and coverage[14,15].However, the current research mainly focuses on the theoretical and experimental aspects[16].

However,the dedicated airborne vehicles have little contribution to the reduction of costs.Besides, taking off,landing,breakdown and other conditions will threaten the conventional aviation safety.Therefore,we propose a new idea: utilizing the existing airborne vehicles,such as civil aircrafts,to assist satellites,and form a novel SAGIN.Compared with previous SAGIN,this new architecture has the following 3 advantages:

•Lower costs.The airborne vehicles will undertake most of the data transmission tasks,which greatly reduces the number of satellites in SAGIN.Moreover,there is no need to launch dedicated airborne vehicles.

•Safer space/air environment.The reduction of the number of satellites is conducive to the sustainable development of the space environment, and is very important for future space exploration.In addition, the reduction of dedicated airborne vehicles is more conducive to aviation safety.

•Better communication performance.The flight altitude of existing airborne vehicles is lower than LEO satellites and HAPs.This can bring lower transmission delay and higher spectral efficiency.

As shown in Figure 1, the new SAGIN is composed of satellite networks and civil aircraft assisted networks (CAAN).In CAAN, civil aircrafts carrying aerial base station provide access service for terrestrial users and reduce the stress of satellite networks.In this case,the terrestrial users prefer the aircrafts to access the networks due to shorter transmission delay.When there is no civil aircraft in the surrounding airspace,users will access the networks through satellites.Nowadays, there are more than 11,000 aircrafts in the air.Such a considerable number of airborne vehicles can cover a large area, greatly reduce the pressures of satellite networks.Thus, we can reduce the dependence on satellites.Besides, as an undedicated airborne platform,carrying aerial base station is an additional service for civil aircrafts and does not affect their flight tasks.Since the coverage of CAAN is not uniform, the joint coverage design with satellite networks can realize more stable coverage of CAAN.

Figure 1. The architecture of civil aircraft assisted spaceair-ground integrated networks.

The addition of civil aircrafts will also bring many new problems and challenges to SAGIN, such as access selection, interference management, resource allocation and management, and so on.At this time,artificial intelligence (AI) will play a significant role.The application of AI in communication networks can be divided into spectrum resource allocation,baseband processing and network intelligent optimization [17].With AI, many problems that cannot be modeled and solved by traditional methods will be solved, such as interference, cell selection and handover, resource allocation and multi-user clustering [18].In particular,CAA-SAGIN has high-dynamic characteristics,which requires rapid response and automatic processing.

The main contributions of this paper are as follows:

• An AI enabled CAA-SAGIN is proposed.The addition of civil aircrafts greatly reduces the costs of SAGIN,while AI can improve the performance of SAGIN.

• The coverage capability of CAAN is analyzed,and the coverage gap of CAAN is analyzed with Chinese mainland as the target.Compared with Starlink,CAAN can undertake 90%of communication tasks.

• A joint coverage satellite constellation is designed.In order to achieve stable and seamless coverage,satellites are essential.

The structure of this paper is as follows: In section II,AI-enabled CAA-SAGIN is introduced.In section III,we analyze the coverage capability of CAAN and compare the changes in the number of satellites required before and after the introduction of CAAN with Starlink.In section IV,in order to ensure the seamless coverage of CAA-SAGIN,the joint coverage constellation is designed.Finally,in section V,a conclusion is given.

II.ARCHITECTURE OF CAA-SAGIN

CAA-SAGIN is consists of CAAN and a joint coverage constellation, which provides global seamless communication access services for users.Tens of thousands of aerial base stations improve the performance of CAA-SAGIN, but also make the architecture of the networks more complex.Moreover,the dynamic characteristics of CAA-SAGIN increase the difficulty of network management and resource scheduling.In this case,AI will play an important role.

2.1 The Architecture of AI-enabled CAASAGIN

The AI plane can implement learning, training and decision-making,which are the foundations for realizing service intelligence.Through the deep reinforcement learning and the interaction between agent and the environment, the networks can learn the action strategy (i.e., access, the content of caching and distribution) and improve the resource utilization.Software defined network (SDN), network functions virtualization(NFV)and other AI techniques can play an important role/function in access network(AN),transmission network(TN)and core network(CN).

Figure 2 shows an example of AI-enabled CAASAGIN.For simplicity,the satellites and civil aircrafts are collectively called “sky access platforms” (SAPs)in this paper.The control plane is separated from the data plane of the network, and AI techniques run through the entire network plane from user terminal to CN.

Figure 2. The architecture of AI-enabled CAA-SAGIN.

Figure 3. The architecture of CAA-SAGIN access network.

Figure 4. Main access modes of CAA-SAGIN.

There are some problems in SAP communication networks, such as limited payload capacity, longer transmission delay.Therefore, it is necessary to introduce dynamic and hierarchical management mechanism.LEO satellites in the same orbit and civil aircrafts can cluster respectively.These SAPs have the ability of communication,computing and caching,which are the edge nodes performing AI techniques.GEO satellites with large communication coverage and capacity are the master host of SAPs.

With the cooperation of AI,SAPs and BSs can utilize their characteristics to develop their advantages,and the AN,TN and CN can achieve effective integration.In this way,the system can realize the expansion of capacity,service-oriented dynamic resource allocation and flexible reconfiguration of heterogeneous networks.

Therefore,AI-enabled CAA-SAGIN becomes an intelligent architecture towards 6G.

2.2 Access Network Architecture

User equipments can freely choose to access aerial base station, satellite or terrestrial base station.And it can switch among them smoothly to ensure the continuity of service.When the terrestrial base station is available,the user equipment will preferentially access the CN through terrestrial base station.Otherwise,users will choose to access the CN through CAASAGIN.

The AN of CAA-SAGIN includes civil aircrafts and satellites,as shown in figure 3.Satellite is not a simple bent pipe,and it has ability of calculation and data processing.Satellites can communicate with other satellites, aerial base stations and gateways directly.The aerial base station is carried on civil aircraft, and is composed of active antenna array,computing processing unit (CPU) and cache unit.For the service with insensitive delay, the aerial base station can cache to the local first, and then transmit the data to the CN while passing through the terrestrial gateway.

CAA-SAGIN supports multiple access modes.The main access modes are shown in the figure 4.First,user equipment can directly access the CN through civil aircraft.For delay insensitive services, civil aircraft can cache them locally, carry these information and transmit them at the appropriate time.However,when the service is sensitive to the delay, the user equipment can take mode 2 to access the CN.In mode 2,the civil aircraft can not communicate with the gateway, and can not directly complete the access task of user equipment.At this time,civil aircraft can choose other nearby aerial base stations or the satellites for relay.When there is no civil aircraft passing by, mode 3 can be used to access the CN.At this time, users will directly access the CN through satellite.If a single satellite cannot transmit data to the CN gateway in time, other satellites can be selected for relay, as shown in mode 4.

III.COVERAGE PERFORMANCE ANALYSIS OF CAAN

Similar to satellites, civil aircrafts have strict flight schedules.However, their routes are not evenly distributed, which means CAAN can not realize seamless coverage alone.Therefore, we need to analyze the coverage performance of CAAN to compare the improvement of CAA-SAGIN.At the same time,it is convenient to design joint coverage constellation to realize seamless coverage.

3.1 Coverage of A Single Aircraft

Coverage performance analysis starts with a single aircraft.The radius of the area covered by the aircraft,R,is given in(1).

In (1),ris the radius of the earth,his the altitude of the aircrafts,andθis the communication elevation.According to the actual situation, we assume that the average flight altitude of the aircraft is 10 km,and the lowest communication elevation is 5°.Then,the coverage radius of a single aircraft is 104 km accordingly.

It is necessary to explain the rationality of the lowest communication elevation.The head and tail of the aircraft usually have corresponding radar and communication equipment,and their wings are equipped with fuel tank.Therefore, the best position to install the aerial base station is the cabin abdomen.Note that the aerial base station cannot affect the aerodynamic structure of the aircraft.Considering the arc structure of the cabin itself,the antenna of the base station should consist of several plane antenna arrays, which should be close to the arc surface of the cabin abdomen and installed inside the cabin.This arc structure makes it possible for the antenna of base station to point to a wider area.Therefore,the assumption that the lowest communication elevation is 5 degrees is reasonable.

Since the civil aircrafts and satellites need to cooperate,their frequency spectrum should match and use a specified frequency in Ka/Ku band.Due to shorter transmission range, the free space loss of aircrafts is 30 dB lower than that of satellites,and the power consumption is smaller accordingly.As the altitude of the aircraft is in the stratosphere and above the clouds,the influence of weather should be the same.Considering the worst weather conditions, the fading of aircrafts will not exceed 6 dB.

3.2 Coverage of CAAN

To analyze the coverage performance of CAAN, we need to obtain the geographic information and deter-mine the target area first.

Table 1. The coverage ratio of CAAN in Chinese mainland area.

For further analysis,it is necessary to project the latitude and longitude under the geographic coordinate system to the plane coordinate system.Since the target area is the Chinese mainland, we choose Albers projection to process the geographic information.In order to facilitate the statistics and analysis of characteristics, we divide the map into several square areas according to the projected coordinates, and each area is identified by matrix, which is called the map matrix.The “several square areas” mentioned here are different from the cells in the terrestrial cellular networks,or the spot beams in the multi-beam satellites.Square is the smallest unit used to process map graphics, which is projected and divided according to the longitude and latitude coordinates of each point on the map.Then, we need to confirm the coverage of each point on the map through the identification information on these minimum units, count the proportion of civil aircraft coverage area,and analyze the proportion of multiple coverage.The number and size of square areas are limited by the resolution and processing ability of graphic information.Although in map information processing, graphics can be divided into hexagonal structure, it is relatively complicated.It is necessary to recalculate and identify the abscissa and ordinate of each area after projection according to the geometric relationship.Moreover,limited by the resolution of the image,the division of the region cannot be more precisely.Although hexagon partition can fit the coverage area better,it has little influence on the analysis of coverage proportion.The division of square is based on the existing longitude and latitude coordinate information for projection, which can simplify the process of data processing and analysis.Although there are gaps in the fitting coverage of the square area,it has little influence on the analysis.According to the projection map, the East and West length of Chinese mainland is about 5200 km,and the north and the south are about 4000 km.So the map is divided into 300×250 squares with side length of 20 km are used to establish the map matrix.

Next,the collected real-time position information of civil aircrafts is projected and imported into the map matrix.Each aircraft can cover an area with a radius of 104 km,and identify the coverage of the unit of the map matrix to obtain the geometric coverage characteristics of CAAN.Figure 5 shows the coverage ratio of CAAN in Chinese mainland in 24 hours,and table 1 indicates the multiple coverage ratio of CAAN at a certain time.As can be seen from Figure 5,the coverage ratio of CAAN is not continuous.When the number of aircraft is less at night,the coverage of CAAN is the lowest.When the number of aircraft is more in the daytime, coverage ratio can reach nearly 55%.It can be concluded from table 1 that the distribution of civil aircrafts is relatively concentrated,because the proportion of multiple coverage is higher than that of single coverage.In other words,for economically developed and densely populated areas, the aircraft distribution is more concentrated, and the coverage performance of CAAN is better.

Figure 5.The coverage ratio of CAAN in Chinese mainland area in 24 hours.

Figure 6. The coverage ratio of CAAN in Beijing area in 24 hours.

Figure 7.The coverage ratio of CAAN in Heilongjiang area in 24 hours.

Figure 8. The coverage ratio of CAAN in Tibet area in 24 hours.

Figure 9. Temporal distribution of active network users.

Figure 10. Temporal distribution of service gap in Chinese mainland area.

Figure 12. Coverage ratio of joint coverage constellation in Chinese mainland area.

Figure 13.Satellites available across the Chinese mainland area.

Spatially,the distribution of CAAN coverage is also different.We chose Beijing, Heilongjiang and Tibet as representatives to compare the coverage of CAAN in areas with different population densities.Among them, Beijing is a high-population-density area, Heilongjiang is a medium-population-density area, and Tibet is a low-population-density area.The results are shown in figure 6-8.It can be seen that the coverage of CAAN is higher and more stable in areas with high population density.In areas with low population density,however,coverage of CAAN is very low.

3.3 Improvement Brought by CAAN

Here, we take Starlink as the benchmark and compare CAAN with it.Starlink constellation contains about 12,000 satellites.We assume that Starlink has uniform global coverage and the earth is considered as a uniform sphere.China’s land area is about 9.6 million square kilometers, accounting for 1.88% of the earth’s surface area.Since the spare star is not considered, the satellites are assumed to work at the same time, and that about 225 satellites are located above the Chinese mainland area.For fairness consideration,only 225 Starlink satellites are assumed to serve the Chinese mainland.In this paper, we use the data from SpaceX’s application to Federal Communications Commission (FCC), that is, the average downlink sum capacity of a single satellite is 20 Gbps,and the 225 satellites can provide 4500 Gbps capacity for Chinese mainland area.According to Baidu’s statistics, there are about 829 million Internet users in China.The temporal distribution of active Internet users is shown in figure 9,with the highest simultaneous online rate of 6.05%in a single day.In other words,the maximum number of people online is about 50 million at the same time in one day.Assuming that all users access the networks through Starlink satellite,the average capacity allocated to users by Starlink satellite is 90 kbps.Although this assumption is too harsh,not all users access the Internet through SAGIN.However, this assumption does not affect the fairness of the comparison.For CAAN, the same number of users will access the networks through CAA-SAGIN.

Although the distribution of civil aircrafts is uneven,it is relatively regular.Therefore, we can divide the target area into many small areas.A time-series prediction model is established for each small area, and then the coverage ratio of each small area can be predicted.According to the population distribution and coverage ratio of different small areas,the service gap of CAAN could be calculated.

First,we need to sample a large number of real-time position information of civil aircrafts.For any small area, the time-series prediction model can be established.

Here, the disturbance factor,ϵ, is set to 0.According to the time series prediction model, the following equation can be obtained.

It can be seen that the time coefficientbis satisfied

The time series prediction expression can be obtained by substituting(5)into(2).

whereτis the next prediction period.Two estimated parameters can be obtained from(6).

Then,we get the time-series prediction model,

For each small area, service gap is defined as the number of users who cannot access CAAN in the area.Thus, for any small area, the service gap can be expressed as

In (10), indexiis the number of small areas.cis total number of Internet users.ρiis the proportion of population in a small area.etis the proportion of active Internet users.Assume that all Internet users use CAAN to access the networks.By summing the service gaps of all small areas,the overall service gap of the target area can be obtained.The results are shown in figure 10.It can be seen that the biggest service gap of CAAN is 5 million.In other words, since the maximum number of online users is 50 million, even in the worst case, CAAN can meet the needs of 90%of network users.If the capacity allocated to each user is still 90kbps,the service gap of CAAN is 450 Gbps.Assuming the same performance of satellites,only 23 satellites are required.In other words,the introduction of CAAN has reduced the number of satellites in SAGIN by 89.89%,and these satellites can be set to sleep or even free from launching.

IV.JOINT COVERAGE CONSTELLATION

In order to achieve global seamless coverage and quantify the performance improvement of SAGIN by CAAN,it is necessary to design a joint coverage constellation.Walker constellation is the most widely used constellation in the world [19].In order to ensure the uniform distribution of joint coverage constellation, we select Walker-δconstellation configuration[20],and the orbit attitude is 1,248 km.To ensure that the number of satellites above the Chinese mainland area is no less than 25,the number of satellites in the constellation is not less than 1,328, and the coverage angle is 3.14°.According to the Walker-δconstellation design method, the number of orbits is 58, the inclination angle is 42°, the phase factor is 30, and the total number of satellites is 1,334.Walker-δConstellation parameter is(1,334/58/30/42°)[21,22].The constellation simulation is shown in figure 11.

Joint coverage constellation needs to ensure seamless coverage all the day.The simulation results are shown in figure 12.The x-axis represents time, and the y-axis represents the coverage ratio of the constellation in Chinese mainland.The coverage ratio of the joint coverage constellation is 100%for Chinese mainland area.

In addition,we simulated the visibility of joint coverage constellations.The simulation results are shown in the figure 13.The x-axis represents time and the yaxis represents the number of satellites available.The simulation results show that the number of satellites available is always greater than 93, which can fully meet the access requirements.

According to the assumption of this paper, combined with the route distribution of civil aircrafts and the population distribution characteristics of the service area,the CAA-SAGIN only needs 1,334 satellites to reach the capacity of 11,943 satellites in Starlink constellation.That is, CAAN can reduce the number of satellites by 88.9%under ideal conditions with the assistance of civil aircrafts.

The proposed satellite constellation scheme is to solve the uneven distribution problem of service gap,and the satellite constellation is ideal.However, we only give analysis from theoretical perspective based on some assumptions.In reality,the design of satellite constellation will be more complex,and we must consider the problem of backup satellite.Therefore, the actual satellite constellation should be larger in scale,and it is reasonable to leave 20%redundancy.

V.CONCLUSION

CAA-SAGIN is a new type of SAGIN,which consists of civil aircrafts carrying aerial base stations and LEO satellite constellation.Aerial base stations cooperate with LEO satellites to provide wireless access services for terrestrial users.The purpose of this paper is to analyze the coverage performance of CAAN,so as to demonstrate the feasibility of CAA-SAGIN and its potential advantages in performance.By analyzing the routes of civil aircrafts,we obtain the temporal-spatial characteristics of CAAN geometric coverage,and the peak of service gap in the target area.According to the prediction of the service gap and the average sum rate of Starlink,we design a joint coverage constellation.Compared with traditional Starlink,our proposed constellation scheme decreases 88.9%satellites at the same data rate.Through the research and analysis of this paper,AI-enabled CAA-SAGIN is a promising architecture towards 6G.Although some assumptions in this paper are ideal, the conclusion of this paper can provide theoretical guidance for the future work.

ACKNOWLEDGEMENT

This work was supported by National Nature Science Foundation of China(No.61871155).