Research on Selection Scheme for Integrated

Satellite-terrestrial Networks

A Thesis Submitted to the Department of Computer Science and Communications Engineering

the Graduate School of Fundamental Science and Engineering of Waseda University

in Partial Fulfillment of the Requirements for the Degree of Master of Engineering

Submission Date: January 25, 2021

Liu Xuanning

(5119FG12-1)

Advisor: Prof. Shigeru Shimamoto

Research guidance: Research on Wireless Access

2

Acknowledgments

First of all, I would like to sincerely gratitude to my advisor, Shigeru

SHIMAMOTO, for his patient guidance and tolerance during my two years of

master's studies.

Secondly, I would like to express my heartful gratitude to Ms. Liu Jiang,

Ms. Jenny Pan, and Ms. Megumi Saito for their help during my master's studies.

And I appreciate a lot of the help from Kenta Matsufuji, for his excellent

guidance on my topic. And I would like to thank Yang Chen and Dr. Huan Wang

for their advice in my life.

Finally, I would like to thank my family for their support and

encouragement for my study experience in Japan.

Thank you all in the Shimamoto lab for spending two happy and meaningful

years together, and I will always be grateful for the experience of learning and

progressing with you all at Waseda.

3

Abstract

An integrated satellite-terrestrial network is the integration of satellite network and

the terrestrial network, providing anytime, global seamless, ubiquitous services of

accessing to Internet. The integration and complementation can achieve high speed,

wide access, and multi-application commercial effect. That is a major technological

opportunity and challenge for the development of future networks. This paper considers

both QoS requirements of applications and network conditions. And discuss selection

algorithm using classic multiple attributes decision methods with special attention on

the characteristics of the integrated satellite-terrestrial network. Finally, simulation

results are obtained according to several application scenarios.

4

Table of contents

Acknowledgments.................................................................................................. 2

Abstract .................................................................................................................. 3

Chapter 1 ................................................................................................................ 7

Introduction ............................................................................................................ 7

1.1 Background ............................................................................................. 7

1.2 Motivation ............................................................................................. 10

Chapter 2 .............................................................................................................. 11

Overview of integrated satellite-terrestrial network ............................................ 11

2.1 LEO communication ............................................................................. 11

2.1.1 System composition .................................................................... 12

2.2 5G .......................................................................................................... 14

2.3 Integrate ................................................................................................. 16

Chapter 3 .............................................................................................................. 18

Radio Propagation for LEO ................................................................................. 18

3.1 Study on coverage of LEO .................................................................... 18

3.2 LOS and shadowing between LEO and mobile terminal in urban areas

............................................................................................................................. 20

3.2.1 LOS/NLOS model ...................................................................... 20

3.2.2 Probability of LOS and shadow situation ................................... 21

3.2.3 Influencing factors in the experimental area ............................... 22

5

3.3 LOS/NLOS judgment experiment ......................................................... 23

3.3.1 Experimental subjects ................................................................. 23

3.3.2 Model .......................................................................................... 24

Chapter 4 .............................................................................................................. 27

Network Selection ................................................................................................ 27

4.1 Overview ............................................................................................... 27

4.1.1 Horizontal handover and vertical handover ................................ 27

4.1.2 Handover process ........................................................................ 28

4.1.3 Network selection. ...................................................................... 30

4.2 QoS study of application scenarios ....................................................... 31

4.2.1 Analysis of QoS parameter weights based on service category .. 32

4.3 TODWA Network selection algorithm .................................................. 39

4.3.1 Network Selection Process ......................................................... 39

4.3.2 Candidate network evaluation based on TOPSIS ....................... 40

4.4 Simulation ............................................................................................. 42

4.4.1 Simulate scenario： .................................................................... 42

4.4.2 Simulation result ......................................................................... 43

Chapter 5 .............................................................................................................. 52

Conclusion and Future Work ............................................................................... 52

5.1 Conclusion ............................................................................................. 52

5.2 Future work ........................................................................................... 52

6

Reference ............................................................................................................. 54

Research Achievements ....................................................................................... 58

7

Chapter 1

Introduction

1.1 Background

As mobile data traffic explodes with the popularity of mobile smart

terminals and the diversification of their applications, MSS (Mobile Satellite

System), an important component of next-generation networks, are also facing

the challenge of providing high-bandwidth connectivity to users. Our

requirements for wireless networks require continuity in crisis on the one hand,

and comprehensive coverage on the other. Currently, it is mainly cellular

networks that are providing wireless coverage, but they are far from meeting the

requirements[1].

Satellite communications rely on line-of-sight (LOS) connections, and

signal strength will be significantly degraded in an obscured environment, so the

coverage characteristics of terrestrial networks are used to complement satellite

coverage, and the two are fused to build an integrated satellite-ground network.

It can provide the most effective coverage for sparsely populated areas as

well as high-capacity and economical services for densely populated areas, and

it is not affected by natural disasters and other unexpected events, which is an

effective way to achieve "any time" and "any place" communication goals. It is

an effective means to achieve the goal of "any time" and "any place"

communication and is a hot issue for academic research. Terrestrial networks can

maintain high-speed services at a low cost, while satellite networks can help

8

achieve 100% coverage. The two can work together to meet the needs of the

network.

Fig 1.1 Integrated satellite-terrestrial network

From 1G to 4G terrestrial mobile communication base station construction

density is proportional to the density of population distribution and economic

development in geography, base station construction priority in densely

populated and economically developed cities and industrial parks, and then

expand to relatively sparsely populated suburbs and rural areas.

At present, terrestrial mobile communication covers only about 20% of the

land area, which is less than 6% of the earth's surface area. As the goal of 5G is

to serve thousands of industries and interconnections, it is reasonable to assume

that the future distribution of 5G base stations will depend on population

distribution and industry application demand. In the industrial applications,

especially in the air, sea, forests, desert areas, and other sparsely populated areas,

it is difficult to achieve 5G for airplanes and drones, offshore oil wells and ships,

forest fire prevention and wildlife video monitoring, inspection along power

lines and railroads, border control and other application scenarios. Even for such

applications on land if 5G coverage is used, the pre-business model faces great

9

challenges, with the scale of revenue not matching the cost of 5G station

construction and operation and maintenance[2]. This brings business

opportunities to low-orbit satellite communications, low-orbit satellite

communications global coverage, and its cost sensitivity and industry

applications of geographic location and communications access points regional

density are not directly related, especially for low-density user access scenarios

of broadband interconnection and communications more advantageous.

At present, the global LEO field is undergoing a profound industrial, system,

and technological change, which has triggered a disruptive revolution in the core

technologies of satellite manufacturing and launching, and the expansion of

satellite application fields and modes of innovation, as shown in Table 1.

While competing with terrestrial networks, LEO communication networks

are also constantly learning from the advanced technologies developed by

terrestrial networks and moving toward complementarity and integration with

terrestrial networks, and the development of sky-ground-sea integration is

gradually becoming a trend.

Tab1 Low Orbit Satellite Constellation

10

1.2 Motivation

In essence, the satellite-terrestrial integrated network is a heterogeneous

wireless network, which, in terms of topology, consists of space-based network

and terrestrial-based network: the space-based network is a mobile satellite or

constellation located in different orbits, and the ground-based network is a

cellular network deployed based on terrestrial wireless standards, both of which

are unified and managed by the mobile satellite network to form an integrated

network[3]. The characteristics of the integrated network are that the mobile

satellite network can improve the coverage of the blocked area of traditional

satellite signals by deploying terrestrial cellular network based on MSS band on

the ground, which can lead to an increase of the number of system users and the

better utilization of network resources.

In this paper, a network selection judgment algorithm based on users’

application QoS demand is proposed and simulated and analyzed based on the

communication characteristics of integrated satellite-terrestrial network and

selection technology mechanism.

Keywords: LEO satellite communications, LOS, 5G, Network selection,

QoS

11

Chapter 2

Overview of integrated satellite-terrestrial

network

2.1 LEO communication

Low earth orbit (LEO) satellite communication is an essential part of

building 6G (The sixth-generation mobile communication system). Compared

with ground communication, LEO has wider coverage and is more suitable for

global communication in desert Gobi, deep forest, sea and other uninhabited

areas. The LEO is more suitable for global communication in desert, Gobi, deep

forest, sea and other unmanned areas. Compared with high-orbit satellite

communication, LEO has the advantages of low transmission loss, low time

delay and low cost[4]. Therefore, the research and design of LEO satellite mobile

communication is of far-reaching significance for the construction of integrated

air-space-sea communication.

12

Fig 2.1.&Table 2 Satellite orbits and major orbital characteristics.

Characteristics of LEO Systems:

1. “Anytime, Anywhere”

2. Blends cellular and satellite technologies Satellite Terrestrial Gateways

3. 500 – 2000 km orbit Below Van Allen Belts

4. Fiber-like propagation delay

5. Voice and data capabilities

6. Hand-held multi-mode phones

However, LEO communication also has some disadvantages, such as higher

cost of LEO satellite launch and onboard payload, high link transmission

attenuation, and high transmission delay, etc. LEO has gradually broken through

some unified satellite communication technologies, such as Space X's multi-

arrow, rocket recovery and reuse technology, etc.

2.1.1 System composition

LEO communication is to achieve communication between two or more

points by relaying radio waves used for communication between mobile users or

between mobile users and fixed users through LEO satellites[5]. The LEO

system architecture is shown in figure 2.2.

13

Fig 2.2. Satellite communication system composition and structure

The space segment, i.e., the constellation, consists of multiple low-orbit

satellites, which are interconnected by inter-satellite links and have the capability

of global networking and data exchange routing.

The ground segment is the control center, data exchange center and

operation center of the system, and consists of signal gateway stations,

measurement and control stations, mobile communication networks, operation

and control systems, integrated network management systems and business

support systems. The signal gateway stations and measurement and control

stations are located according to the coverage requirements and deployed

globally in a distributed manner, while the rest of the network equipment is

14

deployed in a centralized manner to meet the reliability requirements of telecom

operations.

The user segment consists of various types of user terminals, and the user

segment consists of various user terminals, including handheld terminals,

vehicle-mounted stations, shipboard stations, airborne terminals, etc.

2.2 5G

Key technologies for 5G include large-scale multiple-input multiple-output,

dense heterogeneous networks, millimeter-wave communications, D2D

communications, energy-aware communications energy harvesting, cloud-based

radio access networks (Cloud-Radio Access Network, C-RAN)[6], and

virtualization of network resources, among others. Compared to 4G cellular

networks, 5G:

1. increases throughput by at least 1000 times;

2. supports higher network density;

3. significantly reduces wait times;

4. improves energy efficiency;

5. supports high-density mobile broadband users, D2D communications,

ultra-reliable and large-scale machine-type communications.

The application of 5G is manifested in three application scenarios.

Enhanced Mobile Broadband (eMBB) scenarios, massive Machine Type of

Communication (mMTC) scenarios, and Ultra-reliable and Low Latency

Communications (URLLC) scenarios[7].

15

Fig2.3 5G three typical application scenarios and features

However, in order to meet the three typical application scenarios, 5G's

(Radio Resource Management, RRM) also faces many challenges, such as user

access, spectrum allocation, and power management. The main causes of these

problems include

1. heterogeneity of HWNs and dense distribution of UDs;

2. heterogeneous radio resources;

3. unbalanced coverage and traffic load of base stations;

4. high switching frequency;

5. capacity limitation of forward and backhaul links;

6. a large number of users and stakeholders with different objectives such

as operators.

Among them, our research focuses on the access selection problem in 5G,

which focuses on designing a network selection scheme based on the

characteristics of different types of service requests, the scenarios where users

5G

URLLC mMTC

eMBB

High accuracy positioning

Ultra-high reliability

High availability

Ultra-low latency

Ultra-low energy

Ultra-low density

Deep coverage

Ultra-low interruption time

Ultra-high spectral efficiency

Extreme data rates

16

are located (e.g., in-vehicle users)[8], and the state of the network to select the

appropriate network for users requesting specific types of services to meet their

needs, improve the utilization of network resources in 5G, and enable network

operators and users to reach This will improve the utilization of 5G network

resources and balance the interests of network operators and users.

2.3 Integrate

According to the definition of the International Telecommunication Union

(ITU), the satellite-terrestrial communication system can be divided into two

categories: integrated satellite-terrestrial system and hybrid satellite-terrestrial

system.

To address the issue of satellite and terrestrial 5G convergence, the ITU has

proposed four application scenarios for satellite-terrestrial 5G convergence,

including relay-to-station, cell backhaul, dynamic transit and hybrid multicast

scenarios, and the key factors that must be considered to support these scenarios,

including multicast support[9], Intelligent routing support, dynamic cache

management and adaptive flow support, latency, consistent quality of service,

NFV (Network Function Virtualization)/SDN (Software Defined Network)

compatibility, and flexibility of business models.

3GPP has been working on star-ground convergence since R14. In

TS22.261, the role and advantages of satellites in 5G systems are discussed. As

one of the multiple access technologies for 5G, satellites have significant

advantages in some industrial application scenarios that require wide-area

coverage. Satellite networks can provide low-cost coverage solutions in areas

with weak terrestrial 5G coverage. For M2M/IoT in 5G networks, and for

17

providing ubiquitous network services to passengers on high-speed mobile

carriers, broadcast/multicast information services can be provided to network

edge network elements[10] and user terminals with the superior

broadcast/multicast capability of satellites.

Fig2.4. Network architecture of integrated satellite-terrestrial network1

Determine the service classification of each communication scenario of the

convergence between heaven and earth, such as the service types of pipeline,

mobile and IoT/data collection; and determine the performance requirements

under each service scenario, including satellite coverage, channel bandwidth,

communication delay, reliability and maximum number of connections.

1 https://www.researchgate.net/figure/Integrated-satellite-terrestrial-architecture-of-prevalent-standards-to-

connect-with_fig1_319901529

18

Chapter 3

Radio Propagation for LEO

The analysis of the bit error rate (BER) performance of direct sequence code

division multiple access (DS-CDMA) communication systems based on low

earth orbit (LEO) satellites under various propagation channel models and

communication scenarios has been well studied by many researchers in almost

all cases. However, most of these works consider a hypothetical snapshot of the

communication scenario[11], such as the worst-case scenario, to evaluate the

BER performance under that scenario, without considering the probability of

occurrence of that scenario.

3.1 Study on coverage of LEO

From the perspective of the ground station, the position of the satellite

within its orbit is defined by the azimuth (Az) and elevation (°) angles. The

azimuth angle is the angle of the satellite's orientation, measured clockwise from

geographic north in the plane of the horizon. The azimuth angle ranges from 0º

to 360º. The elevation angle is the angle between the satellite and the observer's

(ground station) horizon plane. The range of the elevation angle is 0º to 90º. The

coverage area of a single satellite is a circular area of the Earth's surface within

which the satellite can be seen at an elevation angle equal to or greater than the

minimum elevation angle determined by the system link budget requirements.

The coverage area of a satellite on the Earth depends on the orbital parameters

and is maximum at an elevation angle of 0.

19

The ground station (GS) can communicate with LEO satellites only when

the ground station is below the coverage area (satellite footprint). Because LEO

satellites move too fast over the Earth, the duration of visibility and the duration

of communication will vary for each LEO satellite as it passes the ground

station[12]. As the satellite moves, the footprint also moves, taking the ground

station out of the footprint and thus losing communication with the ground

station.

Fig3.1. LEO coverage geometry

𝜀0 + 𝛼0 + 𝛽0 = 90°

sin 𝛼0 =𝑅𝑒

𝑅𝑒 + 𝐻cos 𝜀0

𝑆𝐶𝑜𝑣𝑒𝑟𝑎𝑔𝑒 = 2𝜋𝑅𝑒2(1 − cos 𝛽0)

As in Fig 3.1, H is the elevation of the satellite, 𝑅𝑒 is the radius of the Earth,

20

and 𝜀0 is the elevation angle from the observation point to the satellite, from

which the relationship between the elevation angle of the LEO satellite

(assuming the altitude is 1200km) and the signal coverage area can be studied as

Fig 3.2.

Fig 3.2. The LEO Coverage of the Earth’s surface

3.2 LOS and shadowing between LEO and

mobile terminal in urban areas

3.2.1 LOS/NLOS model

Satellite signals are always largely reflected and be blocked in cities[13].

Due to the influence of the building, the signal shows two manifestations of line-

of-sight (LOS) and non-LOS (NLOS) as shown in Fig 3.3.

21

Fig 3.3 LOS and NLOS of satellite signal

3.2.2 Probability of LOS and shadow situation

According to the model in Figure 3.3, it is possible to know the variation of

the probability of LOS and NLOS by the elevation angle for the satellite orbit

altitude at 1200 km and assuming the average of buildings’ height of 20m.

Fig 3.4 Probability of LOS influenced by the change in elevation angle

NLOS

NLOS

LOS

22

Fig 3.5 The probability of NLOS influenced by the change in elevation angle

3.2.3 Influencing factors in the experimental area

Here I examine the relationship between average building height and

building density in several major areas of Tokyo Shinjuku is my main study site2.

2 Land Use in Tokyo https://www.toshiseibi.metro.tokyo.lg.jp/seisaku/tochi_c/tochi_5list.html

23

Fig 3.6 the relationship between average building height and building density

The study result shows the average height of buildings and building density.

The circled marks represent the average height of buildings in meter. The

waterfront has the highest average height of building followed by Shinjuku,

Shibuya, Ueno/Asakusa.

3.3 LOS/NLOS judgment experiment

3.3.1 Experimental subjects

Satellite ephemeris can accurately calculate, predict, depict and track the

time, position, speed and other operational status of satellites and flying bodies.

According to these data, do a simulation of 110 OneWeb satellites’ operation[14].

And to study the positional relationship and visibility between the observation

point and the LEO satellites.

24

3.3.2 Model

Fig 3.7 Experiment model

As shown in figure 3.7. If the coordinates of the building and the UE are

known, and the height of the building is known, an angle θ can be derived, and

this angle can be compared with the maximum elevation angle at the observation

point learned from the satellite tle to know whether the observation point is LOS

at this time.

Due to the shape of the Earth, the latitude and longitude elevation

coordinate vectors are converted to actual coordinate vectors[15] as shown in

Figure 3.8.

Fig 3.8 Representation of latitude and longitude elevation coordinates

25

The conversion between latitude and longitude coordinates and xyz

coordinates is performed by the following formula.

x = (R + h)cosφcosλ

y = (R + h)cosφsinλ

𝑧 = [(1 − 𝑒2)𝑅 + ℎ] sinφ

R = 63710004m

𝑒 = 0.0167 (Eccentricity of earth)

From a point outside the laboratory building (35°42', 139°42', 27m), the

number of OneWeb satellites that can be accessed in real time, as well as

information about them.

The minimum elevation angle of the observation point UE and the building

can be calculated by the previous equation as 44.47deg. Comparing this angle

with the results calculated by 110 satellites respectively, it can be determined

whether it is LOS.

Fig 3.9 Satellite visible judgment test result

Since the changes are in real time, this data collection was done on January

6, 2021 at 17:17 PM. The figure3.9 shows a part of the test about 1 satellite, and

26

the data of another 109 satellites are not shown here.

By studying the relationship between elevation angle and distance, derive

the number of the visible satellites.

Fig.3.9 Visible number of satellites

27

Chapter 4

Network Selection

4.1 Overview

In order to meet the diversification of communication service requirements

and provide users with the best quality of communication services, the

interoperability and convergence of wireless networks with different

architectures and characteristics will be an inevitable trend. In the environment

of heterogeneous network convergence, the complexity and heterogeneity of

network conditions make it necessary for users to frequently change network

affiliation and management domains, so how to seamlessly provide users with

high-quality data transmission services, regardless of the user's location and

conditions, the best access method according to the current network state[16],

adaptive to provide the best experience for users and according to the user's

location and service situation in the Smooth handover between different access

methods is a key issue that must be addressed.

4.1.1 Horizontal handover and vertical handover

Mobility management technology is the key technology to solve these

problems. Handover management is one of the important components of

mobility management. Network handover can be divided into different

categories from different perspectives, including horizontal handover and

vertical handover according to the type of network[17]. When handover occurs

in the same network technology between different access points is called

28

horizontal handover. When the handover occurs between different network

technology access points is called vertical handover, as shown in Figure 1.

Because heterogeneous networks mainly study the communication between

different types of network access methods, so the handover management

problem of heterogeneous networks is mainly the problem of vertical handover.

Fig4.1. Horizontal handover and vertical handover

4.1.2 Handover process

There are mainly two kinds of handover brought by satellite or terminal

movement, one is the handover within the satellite system, for the low-orbit

satellite system, its position relative to the ground changes rapidly, the terminal

is continuously covered by the same satellite for only ten minutes, for the low-

orbit satellite using multiple beams, the same beam continuously covers the

29

terminal for only a few minutes, so the inter-satellite or inter-beam handover

must be performed quickly and prevent data loss during the handover process.

The second is the handover of terminals between the terrestrial 5G network and

the satellite network, and the inter-network handover process needs to consider

various factors.

- Simultaneous support for on-satellite processing and bend-transparent

forwarding architecture

- Handover preparation and handover failure handling

- Time synchronization

- Coordination of measurement objects

- Support for lossless handover

It should be noted that the handover direction is different and the triggering

conditions are not the same. For example, when the signal of the terrestrial

cellular network is sufficient, the terminal handover from the satellite network to

the terrestrial network; however, the terminal leaves the cellular network only

when the signal of the cellular network is very weak.

The process of handover can be divided into three phases: handover

initialization, handover decision and handover execution. The handover

initialization phase is mainly for collecting some parameters, such as received

signal strength, network load, etc. The handover decision phase is to use the

collected information to select the target cell. The handover execution phase

sends a series of signaling to complete the handover according to the selected

target cell.

30

4.1.3 Network selection.

In the heterogeneous wireless network environment, how to choose the

most effective and suitable access network for mobile terminals to meet the

quality of service (Quality of Service, QoS) of the service has been a hot research

topic in recent years. And the terminal in the heterogeneous network

environment mobile necessarily involves the problem of handover between

networks. The existing vertical handover algorithms mainly include the

handover algorithm based on Received Signal Strength (RSS) and a series of

improved algorithms.

Since the QoS of a network includes parameters such as bit rate, packet loss

rate, delay and jitter, the selection problem of a heterogeneous network with the

goal of meeting the QoS requirements of the service is a multi-parameter

decision problem. The literature proposes a network selection framework

structure with user QoS as the core, but no specific selection algorithm is given.

The literature applies both fuzzy logic and multiparametric decision making to

the network selection process. The literature uses a combination of hierarchical

analysis and entropy value method to determine the weight values of network

parameters when performing multi-parameter decision making. However, when

the above algorithms are applied to network selection with the goal of satisfying

QoS, the weight values of each QoS parameter are often assigned equally. Since

the differences in QoS requirements between different types of services are not

considered, the network with the best QoS access is always selected regardless

of the type of service, and thus the load of each candidate network cannot be

reasonably and effectively balanced.

31

4.2 QoS study of application scenarios

The traditional communication services specified by the 3rd Generation

Partnership Project (3GPP) are classified into conversational services (CS),

background services, (BS) streaming services (SS) and interactive services (IS)

according to the requirements of multimedia on network latency3.

Conversational services are more sensitive to latency, tolerate certain packet

loss and BER, and generally require little bandwidth. The maximum tolerable

delay and packet loss rate for a video session are 150ms and 0.001%, respectively.

The streaming service does not require high latency, tolerates certain packet loss

and BER, and has higher bandwidth requirements. Live online video streaming

requires at most 300ms of network latency and 0.000001% of packet loss.

Interactive services require less delay, unlike the above two types of services,

interactive services have higher requirements for packet loss and BER, and

require less bandwidth resources. Contextual services are not sensitive to

network attributes, but they must ensure the security and reliability of data

transmission, so they have higher requirements for packet loss and bit error rate

and require very few bandwidth resources.

To address the above problems, this paper analyzes the relative importance

of QoS parameters for the four types of services defined by 3GPP, sets the

weights of QoS parameters for each type of services using Analytic Hierarchy

Process (AHP), and then uses Technique for Order Preference by Similarity to

3 https://spectrum.ieee.org/telecom/wireless/3gpp-release-15-overview

32

Ideal Solution (TOS). Preference by Similarity to Ideal Solution (TOPSIS) is

used as the basis to design a business class-oriented algorithm for differentiating

weights for heterogeneous network selection, and finally, the algorithm is

verified by simulation experiments.

4.2.1 Analysis of QoS parameter weights based on

service category

The types of business considered, which mainly include session-based

business, interaction-based business, stream-based business and backend-based

business, are shown in the table.

Session-based services are usually real-time services, and latency is critical

to the user's service experience. For background services, the latency has almost

no impact. The large data volume transmission of streaming media services is

very dependent on high network bandwidth. The session class services tend to

be less sensitive to bandwidth. Based on the above analysis, this paper selects 5

parameters that best reflect the performance of wireless networks, namely,

Received Signal Strength (RSS), Bandwidth (B), Delay (D), Jitter (J), and Packet

Loss Rate (PLR), and uses fuzzy hierarchical analysis to find the weights of each

network parameter based on the corresponding network services, and analyzes

their impact on the QoS performance of different mobile network services one

by one.

The algorithm AHP is a method to derive the relative weight relationship

between all parameters by a two-by-two comparison between parameters[19].

The relevant parameters of the candidate object are analyzed and the

33

parameter hierarchy is represented by the parameter hierarchy diagram shown in

Fig. A nine-level scale is usually used in AHP to express the relative importance

of each parameter based on subjective perceptions, feelings and knowledge.

Fig.4.2 Parameter hierarchy diagram

In the fuzzy logic criterion layer, the relative importance of each factor to

the target layer is composed in the form of a matrix. The quantitative scaling

value is obtained by a two-by-two comparison between factors. In this paper, the

fuzzy complementary judgment matrix 𝑅 = (𝑟𝑖𝑗)𝑛×𝑛 is obtained by using the

0~0.9 scaling method.

To ensure that the weights are obtained more reasonably and accurately, the

fuzzy consistency matrix is introduced to improve the fuzzy mutual judgment

matrix. The specific weight acquisition process is as follows.

1. Establishing a hierarchy of top-down dominance relationships

As shown in the figure, the target problem is first hierarchized. In general,

it can be divided into an objective layer, a criterion layer, and a solution layer. In

the network selection algorithm, the objective layer is the optimal network to be

selected. The criterion layer is the various judgment factors that affect the

network selection. In this paper, five factors, namely, received signal strength,

34

jitter, bandwidth, transmission delay, and packet loss rate, are chosen to measure

the performance of different networks. The scheme layer is the candidate

networks in the ground-satellite converged network environment, i.e., terrestrial

5G and low-orbit satellite communication networks.

Fig.4.3 Hierarchical structure model

2. Construction of fuzzy complementary judgment matrix

Fuzzy judgment matrix calculation method, the relative importance of each

factor in the criterion layer to the target layer is composed in the form of a matrix.

The quantitative scale values are obtained by two-by-two comparison between

factors and analysis. Thus, the fuzzy complementary judgment matrix 𝑅 =

(𝑟𝑖𝑗)𝑛×𝑛 is obtained. The method of taking the value of 𝑟𝑖𝑗 is 0~0.9 scaling

method, as shown in the table.

Scale Definition Description

0.5 Equal importance Two elements contribute

equally to the objective

0.6 Moderate importance Experience and judgment

moderately favor one element

35

over another

0.7 Strongly importance Experience and judgment

strongly favor one element

over another

0.8 Very strong importance One element is favored

very strongly over another, its

dominance is demonstrated in

practice

0.9 Extreme importance The evidence favoring

one element over another is of

the highest possible order of

affirmation

0.1 0.2 0.3 0.4 inverse comparison Inverse comparison of

the above comparison

Table.2 0~0.9 Scale Method

3. Weight vector calculation

The weight vector 𝑊 = (𝑊1, 𝑊2, … , 𝑊𝑛)−1 is obtained by the weight

value calculation formula, where ∑ 𝑊𝑖𝑛𝑖=1 = 1，𝑊𝑖 ≥ 0, i=1, 2, ..., n.

4. Consistency test of the fuzzy matrix 𝑹 = (𝒓𝒊𝒋)𝒏×𝒏

36

Firstly, construct the characteristic matrix 𝑊∗ = (𝑊𝑖𝑗)𝑛×𝑛 of matrix R,

where,

Matrix compatibility calculations are then performed,

If the compatibility value 𝐼(𝑅, 𝑊∗) ≤ 𝑇 , then the consistency is found to

meet the criteria, otherwise the matrix 𝑅 = (𝑟𝑖𝑗)𝑛×𝑛 needs to be adjusted and

reweighted. The critical value of compatibility index T=0.1 is usually set.

Whether the fuzzy judgment matrix is consistent or not is a crucial issue in

constructing the matrix, often in the obvious case because of the intricacies of

the problem to be solved and the limitations of human thinking leading to the

lack of consistency. The method based on the fuzzy complementary matrix to

find the weights better overcomes some of the shortcomings of the AHP

algorithm, although there are still shortcomings for the guarantee of consistency.

Therefore, the fuzzy consistency matrix is introduced to improve it so that the

weights are more reasonable and accurate.

5. Establishing fuzzy consistent judgment matrix

Firstly, the fuzzy complementary judgment matrix 𝑅 = (𝑟𝑖𝑗)𝑛×𝑛 is

summed up by rows and by columns, which are denoted as,

37

The following transformations are then implemented.

Derive the fuzzy consistent judgment matrix F.

𝐹 = (𝑓𝑖𝑗)𝑛×𝑛

6. Final weight calculation (hierarchical single ranking)

This layer is used to calculate the priority weights of the importance of the

factors in this layer for the previous layer. This step is used to calculate the weight

vector 𝑊𝑠 = (𝑤1, 𝑤2, … , 𝑤𝑛) of each judgment factor for the business S.

Which,

According to the above process of fuzzy hierarchical analysis (FAHP)

algorithm, in order to obtain the relative weights of each network parameter

index for different businesses.

A fuzzy complementary judgment matrix about four types of services is

38

established. The fuzzy complementary judgment matrices are obtained for

session class, background class, streaming class and interactive class services.

The final weight vector of each network parameter indicator with

concerning different services is obtained through a series of operations processed

39

by the BJTE algorithm.

In the judgment process, to ensure fair and effective network utility

evaluation, the judgment factors are normalized to eliminate the differences

between them.

4.3 TODWA Network selection algorithm

4.3.1 Network Selection Process

1. Network discovery and static information collection

Assume that the entire network is configured with Media Independent

Handover Function (MIHF) of IEEE802.21, and adopts the handover mode of

terminal decision. When mobile terminals communicate in the current Radio

Access Network (RAN), they can find adjacent RAN through MIHF and collect

the static information of adjacent RAN, which can be stored in the local database.

2. Determine the candidate network and QoS parameter value

collection

When the mobile terminal detects that the Radio Signal Strength (RSS) of

the current RAN drops to the handover threshold, the RSS of the adjacent RAN

40

is detected successively according to the static information of the adjacent RAN

in the local database, the adjacent RAN above the communication threshold is

included into the candidate network, and then the current QoS parameter value

of the candidate network is collected through MIHF.

3. Select target network

Candidate QoS parameter values into judgment matrix structure, the main

business categories, based on the current mobile terminal using chapter 2 QoS

parameters of the corresponding weight vector, using section 3.2 evaluation

algorithm of computing candidate network the ideal approximation value,

choose the ideal approximation candidate networks as the target of maximum

value, a handover request.

4.3.2 Candidate network evaluation based on TOPSIS

TOPSIS (Technique for Order Preference by Similarity to Ideal Solution)

algorithm is proposed to solve the multi-objective problem of a single decision

maker[20], which is close to the linear weighting method. Its basic idea is: the

selected satisfactory scheme should be as close to the positive ideal scheme as

possible, and as far away from the negative ideal scheme as possible.

Assuming that the number of candidate networks determined by mobile

terminals is N, the decision matrix is constructed by using the obtained candidate

network QoS parameter values.

41

Each element is a related network QoS parameters of the original values,

all the QoS parameters of the 𝑖th a network 𝛮𝑖

can show as matrix row vector

{𝑑𝑖′, 𝑗𝑖

′, 𝑝𝑟𝑖′, 𝑟𝑠𝑠𝑖

′, 𝑏𝑖′} .

Then, the elements of the judgment matrix are normalized using the following

equation, and to obtain the matrix R.

𝑟𝑖𝑗 =𝑥𝑖𝑗

√∑ 𝑥𝑖𝑗2𝑚

𝑖=1

; 𝑗 = 𝐷, 𝐽, 𝑃𝑅, 𝑅𝑆𝑆, 𝐵

𝑋’ =

𝛮1

⋮

𝛮𝑖

⋮

𝛮𝑚

𝑑1’

⋮

𝑑𝑖’

⋮

𝑑𝑚’

𝑗1’

⋮

𝑗𝑖’

⋮

𝑗𝑚’

𝑝𝑟1’

⋮

𝑝𝑟𝑖’

⋮

𝑝𝑟𝑚’

𝑟𝑠𝑠1’

⋮

𝑟𝑠𝑠𝑖’

⋮

𝑟𝑠𝑠𝑚’

𝑏1’

⋮

𝑏𝑖’

⋮

𝑏𝑚’

𝐷 𝐽 𝑃𝑅 𝑅𝑆𝑆 𝐵

𝑋 =

𝛮1

⋮

𝛮𝑖

⋮

𝛮𝑚

𝑑1

⋮

𝑑𝑖

⋮

𝑑𝑚

𝑗1

⋮

𝑗𝑖

⋮

𝑗𝑚

𝑝𝑟1

⋮

𝑝𝑟𝑖

⋮

𝑝𝑟𝑚

𝑟𝑠𝑠1

⋮

𝑟𝑠𝑠𝑖

⋮

𝑟𝑠𝑠𝑚

𝑏1

⋮

𝑏𝑖

⋮

𝑏𝑚

𝐷 𝐽 𝑃𝑅 𝑅𝑆𝑆 𝐵

42

According to the idea of TOPSIS, the result of judging duan should be as close as

possible to the positive ideal solution and the principle negative ideal solution. The

distance between each candidate network and the positive and negative solutions is

calculated and compared to get the judgment result.

4.4 Simulation

4.4.1 Simulate scenario：

Performing network selection scheme based on AHP and TOPSIS which

considering differentiated weight of criterion according to specific access

network.

Fig.4.4 Simulation Scenario

The simulation scenario is shown in Figure4.4 It is assumed that the 5G

network has full coverage and the satellite footprint is dynamic. The observation

points start moving from static in an urban environment.

UE’s position:

Latitude: 35°42’25.60” Longitude: 139°42’32.19”; Elevation: 27m

43

Building’s position:

Latitude:35°42’25.61”; Longitude:139°42’31.28”; Elevation:61m

The initial environment is a single building impact. And then UE get in

movement at a uniform speed of 1.5m/s along a direction 9 degrees west of south.

4.4.2 Simulation result

In this simulation, the services analyzed are backend services. The

importance weights of a set of AHP criteria and the weights of the network that

provides this service. The weight does not indicate that the network is more

dominant in terms of QoS, but only that it is sufficient to provide a weight of

network capacity sufficient for that service.

Background class

When the end-user sends and receives data-files in the background, this

scheme applies. Background traffic is one of the classical data communication

schemes that on an overall level is characterized by that the destination is not

expecting the data within a certain time. The scheme is thus more or less delivery

time insensitive. E.g.: background downloading of mails, sending of SMSs over

the internet.

44

Fig 4.5 Weights associated with the criteria for background

Comparison of possibilities when performing network selection algorithms

at fixed points.

Fig 4.6 Network selection probability of background (Fixed)

45

Comparison of network selection possibilities when UE is moving at a

uniform speed of 1.5m/s along direction 9 degrees west of south.

Fig 4.7 Network selection probability of background (Mobile)

The same approach was used to study conversational, streaming and

interactive services.

Conversational class

It is the service class with the highest QoS requirements including least

transfer delay and time relation between information elements. This is the only

scheme where the required characteristics are strictly given by human perception.

E.g.: telephony speech, VoIP, video conferencing and other real time

activities.

46

Fig 4.8 Weights associated with the criteria for conversational

Comparison of possibilities when performing network selection algorithms

at fixed points.

Fig 4.9 Network selection probability of conversational (Fixed)

47

Comparison of network selection possibilities when UE is moving at a

uniform speed of 1.5m/s along direction 9 degrees west of south.

Fig 4.10 Network selection probability of conversational (Mobile)

Streaming Class

This scheme is one of the newcomers in data communication, raising a

number of new requirements in both telecommunication and data

communication systems. It is characterized by that the time relations (variation)

between information entities (i.e. samples, packets) within a flow shall be

preserved, although it does not have any requirements on low transfer delay.

Acceptable delay variation is thus much greater than the delay variation given

by the limits of human perception. E.g.: streaming of audio and video.

48

Fig 4.11 Weights associated with the criteria for streaming

Comparison of possibilities when performing network selection algorithms

at fixed points.

Fig 4.12 Network selection probability of streaming (Fixed)

49

Comparison of network selection possibilities when UE is moving at a

uniform speed of 1.5m/s along direction 9 degrees west of south.

Fig 4.13 Network selection probability of streaming (Mobile)

Interactive class:

When the end-user, is on line requesting data from remote equipment (e.g.

a server), this scheme applies. Interactive traffic is the other classical data

communication scheme that on an overall level is characterized by the request

response pattern of the end-user. At the message destination there is an entity

expecting the message (response) within a certain time. Round trip delay time is

therefore one of the key attributes. Another characteristic is that the content of

the packets shall be transparently transferred.

50

Fig 4.14 Weights associated with the criteria for interactive

Comparison of possibilities when performing network selection algorithms

at fixed points.

Fig 4.15 Network selection probability of interactive (Fixed)

51

Comparison of network selection possibilities when UE is moving at a

uniform speed of 1.5m/s along direction 9 degrees west of south.

Fig 4.16 Network selection probability of interactive (Mobile)

52

Chapter 5

Conclusion and Future Work

5.1 Conclusion

In this paper, the research on wireless network selection mechanism is mainly

focused on the requirement of communication terminals needing to access multiple

wireless communication networks in integrated satellite-terrestrial network. Using

the traditional multi-attribute decision theory method, a network selection

judgment algorithm based on user experience and mobility is proposed, and on this

basis, the network selection mechanism under the star-ground integrated network

architecture is modeled, and finally the network selection status of the proposed method

under different application scenarios is studied through simulation.

5.2 Future work

This research focuses on the network selection algorithm for integrated

satellite-terrestrial network, which achieves seamless service QoS-based study

in these two wireless networks, and proposes a network selection algorithm

based on it. further work lies in.

1. this research is only for LEO and terrestrial 5G communication networks,

there are many more types of wireless networks that have been included or will

appear in today's heterogeneous wireless network systems, so further derivation

is needed to study the expression of other wireless networks so that the service

QoS requirements are seamlessly delivered throughout the heterogeneous

53

wireless network system, making the design of the network selection algorithm

more convenient.

2. In this research, only the design of the network selection algorithm is

discussed, and other processes of the access control and vertical switching

process are not studied, such as the triggering conditions of switching, etc.

Meanwhile, the algorithm considers the mobility of users, which may differ from

the actual scenario of the system and needs further study.

3. From the existing network selection algorithms, we can see that the

network selection algorithms are diverse and pluralistic, and there is no unified

standard to measure the performance of the algorithms. Therefore, designing a

unified standard to measure the performance of each network selection algorithm

is the next research direction.

54

Reference

[1] V. R. Anpilogov, "Satellite communication and broadcasting systems in the

21st century," 3rd International Conference on Satellite Communications

(IEEE Cat. No.98TH8392), Moscow, Russia, 1998, pp. 144-145 vol.1, doi:

10.1109/ICSC.1998.741397.

[2] Mamane, M. E. Ghazi, G. Barb and M. Oteșteanu, "5G Heterogeneous

Networks: An Overview on Radio Resource Management Scheduling

Schemes," 2019 7th Mediterranean Congress of Telecommunications (CMT),

Fès, Morocco, 2019, pp. 1-5, doi: 10.1109/CMT.2019.8931369.

[3] Luglio, Michele & Zampognaro, Francesco. (2008). Performance Analysis

of Wireless Systems in Telemedicine - Hybrid Network for Telemedicine

with Satellite and Terrestrial Wireless Links.. 1. 200-204.

[4] Y. Meng et al., "A Global Navigation Augmentation System Based on LEO

Communication Constellation," 2018 European Navigation Conference

(ENC), Gothenburg, 2018, pp. 65-71, doi:

10.1109/EURONAV.2018.8433242.

[5] Y. Su, Y. Liu, Y. Zhou, J. Yuan, H. Cao and J. Shi, "Broadband LEO Satellite

Communications: Architectures and Key Technologies," in IEEE Wireless

Communications, vol. 26, no. 2, pp. 55-61, April 2019, doi:

10.1109/MWC.2019.1800299.

[6] Q. Zhang, F. Liu and C. Zeng, "Adaptive Interference-Aware VNF

Placement for Service-Customized 5G Network Slices," IEEE INFOCOM

2019 - IEEE Conference on Computer Communications, Paris, France, 2019,

55

pp. 2449-2457, doi: 10.1109/INFOCOM.2019.8737660.

[7] Bolhasani, Hamidreza. (2020). 5G New Services - Opportunities and

Challenges. 10.13140/RG.2.2.20570.36805.

[8] J. Butler, "5G Spectrum challenges," 5G Radio Technology Seminar.

Exploring Technical Challenges in the Emerging 5G Ecosystem, London,

2015, pp. 1-9, doi: 10.1049/ic.2015.0028.

[9] V. Deslandes, J. Tronc and A. Beylot, "Analysis of interference issues in

Integrated Satellite and Terrestrial Mobile Systems," 2010 5th Advanced

Satellite Multimedia Systems Conference and the 11th Signal Processing for

Space Communications Workshop, Cagliari, 2010, pp. 256-261, doi:

10.1109/ASMS-SPSC.2010.5586925.

[10] Kota, Sastri & Giambene, Giovanni & Kim, Sooyoung. (2011). Satellite

component of NGN: Integrated and hybrid networks. Int. J. Satellite

Communications Networking. 29. 191-208. 10.1002/sat.972.

[11] W. P. Osborne and Yongjun Xie, "Propagation characterization of

LEO/MEO satellite systems at 900-2100 MHz," 1999 IEEE Emerging

Technologies Symposium. Wireless Communications and Systems (IEEE

Cat. No.99EX297), Richardson, TX, USA, 1999, pp. 21.1-21.8, doi:

10.1109/ETWCS.1999.897339.

[12] S. Xia, Q. Jiang, C. Zou and G. Li, "Beam Coverage Comparison of LEO

Satellite Systems Based on User Diversification," in IEEE Access, vol. 7, pp.

181656-181667, 2019, doi: 10.1109/ACCESS.2019.2959824.

[13] M. Kong, O. Yorkinov and S. Shimamoto, "Evaluations of Urban

Shadowing Characteristics for HAPS Communications," 2008 5th IEEE

56

Consumer Communications and Networking Conference, Las Vegas, NV,

2008, pp. 555-559, doi: 10.1109/ccnc08.2007.129.

[14] Hornillo-Mellado, Susana; Martín-Clemente, Rubén; Baena-Lecuyer,

Vicente. 2020. "Prediction of Satellite Shadowing in Smart Cities with

Application to IoT" Sensors 20, no. 2: 475.

[15] Gerdan, George & Deakin, Rod. (1999). Transforming Cartesian

Coordinates X, Y, Z to Geographical Coordinates φ, λ, h. Australian

Surveyor. 44. 55-63. 10.1080/00050351.1999.10558773.

[16] Min Chen, Lei Xie and Huifang Chen, "A network selection approach

based on network coding in heterogeneous wireless networks," 2011 IEEE

13th International Conference on Communication Technology, Jinan, 2011,

pp. 989-993, doi: 10.1109/ICCT.2011.6158027.

[17] P. Tinkhede and P. Ingole, "Survey of handover decision for next

generation," International Conference on Information Communication and

Embedded Systems (ICICES2014), Chennai, 2014, pp. 1-5, doi:

10.1109/ICICES.2014.7033849.

[18] S. Kang and J. Kim, "QoS-aware network selection for seamless

multimedia service," 2012 IEEE International Symposium on Circuits and

Systems (ISCAS), Seoul, 2012, pp. 544-547, doi:

10.1109/ISCAS.2012.6272087.

[19] V. T. Lokare and P. M. Jadhav, "Using the AHP and TOPSIS methods for

decision making in best course selection after HSC," 2016 International

Conference on Computer Communication and Informatics (ICCCI),

Coimbatore, 2016, pp. 1-6, doi: 10.1109/ICCCI.2016.7479937.

57

[20] B. Han, X. Zhang and Y. Yi, "Multi-criteria project selection using fuzzy

preference relations based AHP and TOPSIS," 2019 Chinese Control And

Decision Conference (CCDC), Nanchang, China, 2019, pp. 3809-3816, doi:

10.1109/CCDC.2019.8833175.

58

Research Achievements

Kenta MATSUFUJI, Xuanning LIU, Kazutoshi YOSHII, Pan Zhenni,

Megumi SAITO, Jiang LIU, Shigeru SHIMAMOTO, Takuya OKURA, Syuu

MIURA, Hiroyuki TSUJI, Naoko YOSHIMURA, Morio TOYOSHIMA,

“Expansion of Satellite Communication to the ground using Engineering Test

Satellite-9”, RCS IEICE Technical Conference, Japan, March 9-12, 2021.