A Dual-Link Soft Handover Scheme for C/U Plane Split ... - IEEE Xplore

0 downloads 0 Views 455KB Size Report
train relay station (TRS) and two antennas in the train, the handover outage ...... 80. Location of the train (km). Signal quality (dB). Macro eNB i. Macro eNB j.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2018.2794770, IEEE Access 1

A Dual-Link Soft Handover Scheme for C/U Plane Split Network in High-Speed Railway Junhui Zhao, Senior Member, IEEE, Yunyi Liu, Yi Gong, Senior Member, IEEE, Chuanyun Wang, and Lisheng Fan

Abstract—The heterogeneous network architecture based on control/user (C/U) plane split is a research hot spot in the fifth generation (5G) communication system. This new architecture for the high-speed railway (HSR) communication system can provide high quality of service (QoS) for the passengers, such as higher system transmission capacity, better transmission reliability, and lower co-channel interference. The relatively critical C plane is expanded and maintained in a reliable low-frequency band to guarantee transmission reliability, and the U plane is supported by the available high-frequency band to meet the increasing system capacity demands. However, there are still many problems to be solved in the C/U plane split network to ensure reliable transmission. In the HSR communication system, the C plane and the U plane are supported by the macro evolved NodeBs (eNBs) and the small eNBs respectively. The handover between the different macro eNBs involves two types of handovers, which directly reduces its applicability and reliability in HSR. Therefore, a dual-link soft handover scheme for C/U plane split network in HSR is proposed in this paper. By deploying a train relay station (TRS) and two antennas in the train, the handover outage probability will be reduced. Moreover, the bicasting is adopted to decrease the communication interruption time, and the signaling flows of the intra-macro eNB handover and inter-macro eNB handover are designed in detail. Simulation results show that the proposed handover scheme can significantly reduce the outage probability and improve the handover success probability in the inter-macro eNB handover. Keywords—High-speed railway (HSR), C/U plane split, handover, dual-link, bi-casting.

I. I NTRODUCTION

I

N order to meet the high comfort experience for passengers and promote the sustainable and rapid expansion of

This work was supported in part by the National Natural Science Foundation of China (61661021, 61471031), in part by the Open Research Fund of National Mobile Communications Research Laboratory, Southeast University (No. 2017D14), in part by the State Key Laboratory of Rail Traffic Control and Safety (Contract No.RCS2017K009), in part by Science and Technology Project of Jiangxi Provincial Transport Bureau (No. 2016D0037), in part by Science and Technology Program of Jiangxi Province (20172BCB22016, 20171BBE50057). J. Zhao (corresponding author) is with the School of Electronic and Information Engineering, Beijing Jiaotong University, Beijing, 100044, China and also with the School of Information Engineering, East China Jiaotong University, Nanchang, 330013, China (e-mail: [email protected]). Y. Liu is with the School of Electronic and Information Engineering, Beijing Jiaotong University, Beijing, 100044, China (e-mail: [email protected]). Y. Gong is with the Department of Electrical and Electronic Engineering, Southern University of Science and Technology of China, Shenzhen, 518055, China (e-mail: [email protected]). C. Wang is with School of Information Engineering, East China Jiaotong University, Nanchang, 330013, China (e-mail:[email protected]). L. Fan is with school of Computer Science and Educational Software, Guangzhou University, Guangzhou, 510006, China (E-mail: [email protected]).

China’s high-speed railway (HSR), it is significant to meet the broadband mobile communication requirements for passengers while ensuring the safe, efficient and reliable operation of the train, such as video calls, video on demand, online games and so on [1]. However, the existing HSR communication system is far from being able to meet the needs of various broadband services, which promotes various countries and research institutions to study the next generation of railway wireless communication system [2]. Due to the interference level and coverage area limit, the operators usually choose low-frequency band to deploy networks for ensuring reliable quality of service (QoS) and mobility support [3]. With the rapid development of wireless communication, the low-frequency band has been fully utilized [4]–[6]. In the public mobile network, in order to satisfy the demand of communication service, the high-frequency band that more than 5GHz (the maximum is up to 300GHz) will be used in the fifth generation (5G) communication system to get a wider spectrum [7]–[9]. However, the transmission of wireless signal in high-frequency band will experience severe propagation loss, which will limit the coverage of the signal and cause frequent handovers and access failure [10]. The resulting signaling exchange is a huge burden on the network and the user equipments (UEs) [11]. Currently, the cellular networks are becoming more and more dense, and many problems have been exposed in the traditional network [12]. In the future, HSR wireless communication system should consider how to improve the network capacity and rational use of different frequency bands [13]. Although the idea of C/U plane split has been implemented in the long term evolution/system architecture evolution (LTE/SAE) architecture, the C plane information and the U plane information are still transmitted through the same frequency band, and the decoupling network is not really achieved [14]. In order to improve the system transmission capacity and reduce the network access signaling overhead, the C/U plane split network architecture is a current research hot spot, which can provide high transmission capacity and reliability for HSR communication system [15]. With the global system for mobile communication for railway (GSM-R) being replaced by LTE for railway (LTER) in the future, the 4 MHz low-frequency band (930-934 MHz for downlink and 885-889 MHz for uplink in China) previously allocated to GSM-R makes it possible to apply C/U plane split network in HSR [16]. The C plane information and some U plane information which requires higher transmission reliability such as train control and train scheduling can be

2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2018.2794770, IEEE Access 2

supported by the macro evolved NodeBs (eNBs) with the low-frequency band. The U plane information requires high transmission capacity, which can be supported by the small eNBs with the 5GHz high-frequency band to enhance the system capacity, handle serious propagation loss, and guarantee the transmission reliability [17]. Since macro eNBs and small eNBs use different frequency bands , the co-channel interference between the macro eNBs and the small eNBs does not exist [18], [19]. However, the HSR communication system with this architecture will inevitably lead to many new problems [20]. The handover is more serious than that in the traditional cellular network. The handover between the different macro eNBs needs to be switched in respective C plane and U plane, which will result in lower system performance [21]. The reasons are as follows: (1) The more handover frequency. As the speed of the train increases, the handover is more frequent. Assuming that a macro eNB covers three small eNBs in the coverage area, the UEs will perform a inter-macro eNB handover and two intra-macro eNB handovers when crossing a macro cell. As the more handovers are executed, the outage possibility will be higher. (2) The higher inter-macro eNB handover outage probability. In the C/U plane split HSR network, the inter-macro eNB handover consists of two parts. First, the UEs will perform a macro-macro eNB handover to access the target macro eNB. Then, the target macro eNB will control the target small eNB to perform a small-small eNB handover and establish a connection with the UEs. Therefore, the inter-macro eNB handover will be successfully executed after the macro-macro eNB handover and the small-small eNB handover are completed. (3) The more handover steps. If the handover is not completed before the high-speed train leaves the overlapping area between the two small eNBs, the inter-macro eNB handover will fail. However, the small-small eNB handover, as the second step of the inter-macro eNB handover, must be performed after the macro-macro eNB handover. Therefore, the inter-macro eNB handover must be completed in the overlapping area. (4) The handover trigger hysteresis. The inter-macro eNB handover is only triggered when the signal quality of the target macro and small eNBs simultaneously exceeds a certain hysteresis of the source macro and small eNBs, respectively. As a result, it causes that the inter-macro eNB handover is relatively lagged. Therefore, HSR communication network based on the C/U plane split requires very high handover performance, while frequent handovers will also cause a lot of system overhead. In order to ensure the smooth progress of handover and guarantee the QoS, a double-link soft handover scheme based on C/U plane split in HSR communication network is proposed in this paper. We set up a train relay station (TRS) and two antennas in high-speed train. The two antennas cooperate to execute the soft handover and reduce the handover outage probability. Meanwhile, we make use of the bi-casting technology to decrease the communication interruption time and ensure QoS. The signaling flows of the intra-macro eNB handover and the

MME/S-GW S1-MME

S1-U X3 X2

Macro eNB j

Macro eNB i

v

Small eNB

Small eNB a

Small eNB b

Rear antenna

Small eNB

Front antenna UE

AP

TRS

Fig. 1. High-speed railway communication system model based on C/U plane split

inter-macro eNB handover are designed in detail. The rest of this paper is organized as follows. The system model, handover scheme and handover signaling flows are presented in Section II. In Section III, we describe and analysis the theoretical performance of the proposed handover scheme. Simulation results and analysis are given in Section IV. Finally, Section V concludes the entire work. II. S YSTEM M ODEL , AND H ANDOVER S CHEME A. System Model Fig. 1 illustrates the high-speed railway communication system model based on C/U plane split. In the proposed scheme, we set up a TRS and two antennas in the high-speed train, which are located at the front and rear of the train. When the train is traveling in a single cell, the TRS will choose a better signal quality received from the two antennas and establish a communication link with the eNB. When the train is approaching an eNB, the signal quality received from the rear antenna is usually worse than that of the front antenna. Therefore, the front antenna will be chosen to establish the communication link. Conversely, if the train is moving far from an eNB, the rear antenna will be selected to establish a communication link and get a better signal quality. When the train moves across the cell edge, the handover is performed by two antennas. When the front antenna performs a handover and establishes a communication link with the target eNB, the rear antenna maintains a connection with the source eNB. Then, the rear antenna will synchronize its operating frequency to the target eNB after the handover is successfully executed. As a result, the communication will not be interrupted in the process of handover. The rear antenna can perform an another handover again in case of the handover with the front antenna is failed to execute. Mobility management entity (MME) mainly has mobility management, non-access stratum (NAS) signaling tracking, serving gateway (S-GW) selection, air interface security control, authentication and other functions. The S-GW has local mobility support, downlink packet caching, data transmission

2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2018.2794770, IEEE Access 3

TRS

UE Data Packets

Source Macro eNB

Target Small eNB

Source Small eNB

MME

S-GW

Data Packets

Data Packets 1. Bi-casting Decision

2. Bi-casting Request 3. Bi-casting Request ACK Data Bi-casting

Data Packets

Data Packets

4. Group Measurement Control 5. Group Measurement Reports

6. Handover Decision 7. Handover Request 8. Admission Control 9. Handover Request ACK

10. Handover Command Data Packets to Rear Antenna 11. Front Antenna Synchronization 12. Front Antenna Synchronization ACK

Data Packets to Front Antenna

Legend

13. Rear Antenna Synchronization 14. Rear Antenna Synchronization ACK

L3 Signaling

L1/L2 Signaling U-plane Packets

15. Uplink Allocation 16. Handover Confirm 17. Bi-casting Complete Request 18. Bi-casting Complete Request ACK

19. Release Resource Data Packets

Data Packets

Data Packets

Fig. 2. Intra-macro eNB handover signaling

from the packet data network (PDN) to eNBs and other functions. After the handover is successfully performed, the S-GW should select the downlink data packets forwarding path and send them to the target eNB. Since the TRS is connecting to the source and target eNBs during the handover, the MME needs to maintain two different connections for the TRS simultaneously. The S-GW would keep two routing paths in its routing table to support bidirectional forwarding. The train with well shield carriages can cause higher penetration loss of the radio signals (20dB-35dB) to/from UEs inside the train [22]. It is difficult for UEs in the train to maintain a reliable connection directly with the eNBs. Therefore, the UEs in the train are directly connected to the access points (APs) in every carriages. The TRS controls all APs, which could support multiple wireless access technologies, such as WiFi, 2G and 3G. APs transmit the data packets to the terrestrial eNBs via TRS. TRS connects to the antennas outside the train through the fiber to overcome the penetration loss and achieve reliable communication with eNBs [23]. Compared with the non-relay scheme, the use of TRS can reduce the penetration loss of the train about 10-30dB [24], [25]. In addition, by using the group mobility of TRS instead of each user to achieve the handover, it could greatly reduce the handover signaling overhead. Another problem in HSR scenarios is the hard handover, which may cause communication interruption. In recent years, the handover scheme based on bi-casting is also a hot spot

[26]. In the traditional LTE handover, the data forwarding mechanism is used to reduce the data packets loss rate. When the handover process is enabled, the received data packets are transmitted to the target eNB by the the source eNB through the X2 interface, and the target eNB sends the data packets to the UE or the S-GW. There is a delay in data forwarding between eNBs, and the source and target eNBs cannot send data packets to the UE simultaneously. Therefore, the communication interruption time consists of handover processing delay and forwarding delay. In bi-casting mechanism, the SGW sends data packets to the source and target eNBs after the handover is triggered simultaneously. The source eNB processes the received data packets and sends them to the UE, and the target eNB discards the received data packets until the UE successfully connects to the target eNB. Because the target eNB receives data packets immediately from S-GW after the handover is triggered, the communication interruption time is approximately equal to the handover processing delay. Thus, the communication interruption time in bi-casting is shorter than the forwarding, which is more suitable for some realtime communication services. In addition, the S1 interface between the eNB and S-GW has a higher throughput for data transmission than the X2 interface. B. Intra-macro eNB handover The intra-macro eNB handover refers to the handover between different small eNBs within the same macro eNB.

2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2018.2794770, IEEE Access 4

The intra-macro eNB handover signaling is shown in Fig. 2. When the train enters the overlapping area of two adjacent small eNBs in the same macro eNB, the source macro eNB performs a bi-casting decision and sends a bi-casting request to the S-GW. The S-GW sends the data packets to the source small eNB and the target small eNB, the source small eNB receives and forwards the data packets to the TRS, and the target small eNB discards the data packets until it becomes the source eNB. In the handover preparation phase, according to the group measurement control message, the group measurement report is transmitted to the source macro eNB by the TRS. When the source macro eNB determines to trigger an intra-macro eNB handover based on the handover decision algorithm, the source macro eNB will transmit the handover request message to the target small eNB. The target small eNB prepares and retains the resources for the TRS based on admission control message, and then the target small eNB transmits the handover request acknowledgement (ACK) to the source macro eNB. During the handover execution phase, when the handover command message is received by the TRS from the source macro eNB, the front antenna is separated from the source small eNB and synchronized to the target small eNB to establish a connection. After the rear antenna enters the overlapping area, the TRS changes the working frequency of the rear antenna to be the same as the target small eNB. With the help of bi-casting, TRS can receive data packets from the source small eNB through the rear antenna during handover, thereby avoiding communication interruption. However, once the front antenna fails to execute a handover , the TRS will send the group measurement report of the rear antenna to the source macro eNB. When the handover trigger condition is satisfied, the rear antenna will execute a handover again. After the synchronization is completed, the TRS and the source macro eNB perform uplink resource allocation and handover confirmation signaling interaction to complete the handover. After the handover is completed, the source macro eNB requests the S-GW to end the bi-casting and the S-GW is configured accordingly. The source small eNB releases the resources associated with the TRS. The downlink data packets are only sent to the target small eNB and the intra-macro eNB handover is completed. C. Inter-macro eNB handover The inter-macro eNB handover signaling is shown in Fig. 3. For this type of handover, the TRS must execute the macromacro eNB handover and the small-small eNB handover in succession. When the train enters the overlapping area of two adjacent macro eNBs, the target macro eNB performs a bi-casting decision and sends a bi-casting request to the S-GW. The SGW sends the data packets to the source small eNB and the target small eNB, the source small eNB receives and forwards the data packets to the TRS, and the target small eNB discards the data packets until it becomes the source eNB. In the handover preparation phase, the source macro eNB needs to determine whether the target small eNB and the

target macro eNB satisfy the respective handover trigger conditions depending on the measurement report from TRS. If all conditions are met at the same time, the source macro eNB transmits the handover request message to the target macro eNB. The target macro eNB reserves the corresponding resources in the admission control, including the dedicated random access resource, necessary parameters and the RRC connection reconfiguration information. Then, the handover request ACK is transmitted by the target macro eNB to the source macro eNB. During the handover execution phase, the TRS performs a handover according to the handover command, which includes RRC signaling, with the necessary integrity protection and encryption. The front antenna is then separated from the source macro eNB and synchronized to the target macro eNB to establish a connection. With the help of the bi-casting, the rear antenna can receive data packets from the source small eNB. After the rear antenna in the train travels into the overlapping area, it will adjust its working frequency to be the same as the target macro eNB. If the handover executed with front antenna fails, the group measurement report of the rear antenna will be send to the source macro eNB by the TRS . When the handover trigger condition is satisfied, the handover will be executed again by rear antenna. After the handover is completed, the TRS transmits the RRC connection configuration complete message to the target macro eNB, and the macro-macro eNB handover is completed. The next small-small eNB handover is substantially the same as the intra-macro eNB handover. After the inter-macro eNB handover is completed, the target macro eNB requests the S-GW to end the bi-casting and the S-GW is configured accordingly. Finally, the source macro and small eNBs release the resource. It is remarkable that the signaling transmitted between the target small eNB and the MME must be transmitted via the target macro eNB. III. P ERFORMANCE A NALYSIS A. Distance Calculation The system parameters are shown in Table I at the next page. Assuming that the vertical point of the macro eNB i to the rail is the coordinate origin, the vertical line from the macro eNB i to the rail is the Y-axis, and the rail is the X-axis. x represents the position of the front antenna of the train at the rail, and the abscissa of each eNBs is xi = 0km, xj = 4.8km, xa = 1.6km, xb = 3.2km. Therefore, the distance calculation formula for the front antenna at the position of x to each eNBs are as follows: q 2 di (x) = (hm − ht ) + dm 2 + x2 , q 2 2 dj (x) = (hm − ht ) + dm 2 + (Dm − x) , q (1) 2 2 da (x) = (hs − ht ) + dp 2 + (x − xa ) , q 2 2 db (x) = (hs − ht ) + dp 2 + (xb − x) . Similarly, the distance from the rear antenna to each eNBs at the position of x is d (x − L).

2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2018.2794770, IEEE Access 5

UE

Source Small eNB

Source Macro eNB

TRS Data Packets

Target Macro eNB

Target Small eNB

MME

S-GW

Data Packets

Data Packets 1. Bi-casting Decision

2. Bi-casting Request 3. Bi-casting Request ACK

Data Bi-casting

Data Packets

Data Packets

4. Group Measurement Control

5. Group Measurement Reports 6. Handover Decision

7. Handover Request 8. Admission Control 9. Handover Request ACK 10. Handover Command 11. RRC Connection Reconfiguration Request 12. Front and Rear Antenna Synchronization 13. Synchronization ACK 14. Uplink Allocation 15. RRC Connection Reconfiguration Complete

16. Handover Request

Legend

17. Admission Control

18. Handover Request ACK

L3 Signaling 19. Handover Command

L1/L2 Signaling U-plane Packets

20. Front and Rear Antenna Synchronization 21. Synchronization ACK 22. Handover Confirm 23. Bi-casting Complete Request 24. Bi-casting Complete Request ACK 25.Release Resource

Data Packets

Data Packets

Data Packets

Fig. 3. Inter-macro eNB handover signaling

For the sake of convenience, we only consider the interference caused by two co-channel eNBs. Taking the macro eNB i as an example, the distance from the front antenna at the position of x to its co-channel eNBs are: q 2 2 di1 (x) = (hm − ht ) + dm 2 + (x + Im ) , q (2) 2 2 di2 (x) = (hm − ht ) + dm 2 + (Im − x) . B. The C/U plane split handover scheme in HSR When the train arrives at the position of x, the signal strength received by the source macro eNB i and the source small eNB a can be respectively calculated as: P ri (x) = P tm − P Lm [di (x)] − ε (i, x) , P ra (x) = P ts − P Ls [da (x)] − ε (a, x) .

(3)

P Lm and P Ls denote the pass propagation loss of the macro eNB and small eNB, respectively. ε (i, x) and ε (a, x) represent the shadow fading of the macro eNB and small eNB and obey

the Gaussian distribution with the zero mean and standard deviation σ. Taking the macro eNB i as an example, the interference signals strength of the co-channel macro eNBs are: P ri1 (x) = P tm − P Lm [di1 (x)] − ε (i1 , x) , P ri2 (x) = P tm − P Lm [di2 (x)] − ε (i2 , x) . So, the power of the co-channel interference signal is:   Ii (x) = 10log10 10P ri1 (x)/10 + 10P ri2 (x)/10 .

(4)

(5)

The signal qualities received by the macro eNB i and the small eNB a can be given by SIRi (x) = P ri (x) − Ii (x) , SIRa (x) = P ra (x) − Ia (x) .

(6)

After the train enters the overlapping area, the TRS measures the signal qualities received from the source eNB and target eNB at the position of x. The handover will be triggered when the signal quality of the target eNB exceeds the source

2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2018.2794770, IEEE Access 6

TABLE I S YSTEM PARAMETERS

Parameters

Values

Parameters

Values

Frequency of macro eNB fc1

900MHz

Frequency of small eNB fc2

5GHz

Transmit power of macro eNB P tm

43dBm

Transmit power of small eNB P ts

33dBm

Path loss model of macro eNB

COST231-Hata

Path loss model of small eNB

WINNER II D2a

Radius of macro cell Rm

3km

Radius of small cell Rs

1.4km

Overlapping area of macro cell Am

1.2km

Overlapping area of small cell As

1.2km

Marco-to-macro eNB distance Dm

4.8km

Small-to-small eNB distance Ds

1.6km

Distance between macro eNB and rail dm

0.03km

Distance between small eNB and rail ds

0.002km

Macro eNB antenna height hm

0.03km

Small eNB antenna height hs

0.005km

Co-channel neighboring macro eNBs

Co-channel neighboring small eNBs

14.4km

distance Im Train length L

Train antenna height ht

0.4km

eNB Γ dB. Therefore, the macro-macro eNB handover trigger probability and small-small eNB handover trigger probability can be expressed as (7) at the bottom of the this page.

Obviously, these two type of handovers are independent of each other. When the both conditions are satisfied, the intermacro eNB handover will be triggered. Therefore, when the train arrives at the position of x, the inter-macro eNB handover trigger probability can be expressed as: P (x)ho = Pi,j (x)ho · Pa,b (x)ho .

9.6km

distance Is

(8)

Another important performance parameter of handover is the outage probability. If the signal quality received by TRS from an eNB is below Υ , the communication is interrupted. Thus, the outage probability of the macro eNB i and the small

0.0025km

eNB a are calculated as follows: Pi (x)out = P [SIRi (x) < Υm ] = P [P tm − P Lm (di ) − Ii (x) − ε (i, x) < Υm ] = P [ε (i, x) > P tm − P Lm (di ) − Ii (x) − Υm ]   P tm − P Lm (di ) − Ii (x) − Υm =Q . σi   P ts − P Ls (da ) − Ia (x) − Υs Pa (x)out = Q . σa (9) In this network, the communication interruptions from the macro eNB and small eNB can both result in outage, so the outage probability can be expressed as: P (x)out = Pi (x)out · [1 − Pa (x)out ] + [1 − Pi (x)out ] · Pa (x)out + Pi (x)out · Pa (x)out = Pi (x)out + Pa (x)out − Pi (x)out · Pa (x)out . (10)

[SIRj (x) − SIRi (x) ≥ Γ] = P [P rj (x) − Ij (x) − P ri (x) + Ii (x) ≥ Γ] [−P Lm (dj ) + P Lm (di ) − Ij (x) + Ii (x) − ε (j, x) + ε (i, x) ≥ Γ] [ε (i, x) ≥ Γ + P Lm (dj ) − P Lm (di ) + Ij (x) − Ii (x) + ε (j, x)] [ε (i, x) ≥ Γ + P Lm (dj ) − P Lm (di ) + Ij (x) − Ii (x) + ε0 |ε (j, x) = ε0 ] · P [ε (j, x) = ε0 ] !   ∞ 1 Γ + P Lm (dj ) − P Lm (di ) + Ij (x) − Ii (x) + ε0 ε20 q Q exp − 2 dε0 σi 2σj −∞ 2πσj2

Pi,j (x)ho = P =P =P =P Z =

(7)

Pa,b (x)ho = P [SIRa (x) − SIRb (x) ≥ Γ]     Z ∞ ε2 1 Γ + P Ls (db ) − P Ls (da ) + Ib (x) − Ia (x) + ε0 p = Q exp − 02 dε0 σa 2σb 2πσa2 −∞

2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2018.2794770, IEEE Access 7

The handover success probability at a certain position is defined as the eNB successfully triggers a handover at this position, meanwhile there are no interruptions occurred before and after the handover is completed. Thus, the macromacro eNB handover success probability and small-small eNB handover success probability are written as: Pi,j (x)suc = [1 − Pi (x)out ] · Pi,j (x)ho · [1 − Pj (x)out ] , Pa,b (x)suc = [1 − Pa (x)out ] · Pa,b (x)ho · [1 − Pb (x)out ] . (11)

TABLE II S IMULATION PARAMETERS

Parameters

Values

Noise power spectral density N0

-174dBm/Hz

Bandwidth B

10MHz

Lognormal shadow fading σ

4dB

Shadow fading related distance dp

50m

Measurement period t

500ms

Train speed v

360km/h

Handover trigger threshold Γ

3dB

Signal quality threshold of macro eNB Υm

15dBm

Signal quality threshold of small eNB Υs

22dBm

Then, the inter-macro eNB handover success probability is: P (x)suc = Pi,j (x)suc · Pa,b (x)suc .

(12)

C. Our Proposed Scheme In the proposed handover scheme in this paper, the trigger of handover includes two cases. The first one is the front antenna successfully triggers the handover, the second case is the front antenna unsuccessfully triggers a handover and the rear antenna successfully triggers the handover. Therefore, the macro-macro eNB handover trigger probability and small-small eNB handover trigger probability can be expressed R Rmas (13) at the bottom of this page. Where 1 · P (x − L)ho dx indicates the average hanRm +L−x x−L i,j dover trigger probability of the rear antenna from the position of x − L to the Rm after the front antenna failed to trigger a handover at the position of x. Pi,j (x − L)ho indicates the handover trigger probability of the rear antenna if the train arrives at the position of x. In the same way, the computational formula of the small-small eNB handover trigger probability Pa,b (x)ho s is similar to Pi,j (x)ho s . Thus, when the train is located at the position of x, the intermacro eNB handover trigger probability can be expressed as: P (x)ho

s

= Pi,j (x)ho s · Pa,b (x)ho s .

(14)

Similarly, the probabilities that an interruption occurs when the train is connected to the macro eNB i or the small eNB a can be expressed as: Z Rm 1 Pi (x)out s = Pi (x)out · · Pi (x − L)out dx. Rm + L − x x−L Z Rs 1 Pa (x)out s = Pa (x)out · · Pa (x − L)out dx. Rs + L − x x−L (15) R Rm 1 Where Rm +L−x · x−L Pi (x − L)out dx indicates the average handover outage probability of the rear antenna from the position of x − L to the Rm after the connection of front antenna has interrupted at the position of x. Meanwhile, the computational formula of Pa (x)out s is similar to Pi (x)out s .

Pi,j (x)ho Pa,b (x)ho

s

s

The interruption of the connection between a train and a macro eNB or a small eNB both can cause an interruption. Thus, the total handover outage probability is expressed as: P (x)out

s

= Pi (x)out s · [1 − Pa (x)out s ] + [1 − Pi (x)out s ] · Pa (x)out s + Pi (x)out s · Pa (x)out s =Pi (x)out s + Pa (x)out s − Pi (x)out s · Pa (x)out s . (16)

The definition of handover success probability is as the same as the one in the previous section. So, the success probabilities of macro-macro eNB handover and small-small eNB handover are: Pi,j (x)suc Pa,b (x)suc

s s

= [1 − Pi (x)out s ] · Pi,j (x)ho s · [1 − Pj (x)out s ] , = [1 − Pa (x)out s ] · Pa,b (x)ho s · [1 − Pb (x)out s ] . (17)

Then, the success probability of inter-macro eNB handover can be expressed as: P (x)suc

s

= Pi,j (x)suc s · Pa,b (x)suc s .

(18)

IV. S IMULATION R ESULTS AND A NALYSIS In section III, we analyze the performance of the dual-link soft handover scheme for C/U plane split network in HSR. The strengthes and qualities of the received signals, handover trigger probability, outage probability and handover success probability will be simulated in this section. The simulation parameters in this scheme are shown in Table II.

1 = Pi,j (x)ho + [1 − Pi,j (x)ho ] · · Rm + L − x

Z

1 = Pa,b (x)ho + [1 − Pa,b (x)ho ] · · Rs + L − x

Z

Rm

x−L Rs

x−L

Pi,j (x − L)ho dx (13) Pa,b (x − L)ho dx.

2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2018.2794770, IEEE Access 8

1

−20 Macro eNB i Macro eNB j Small eNB a Small eNB b

−30

0.9 0.8 Handover trigger probability

Signal strength (dBm)

−40

−50

−60

−70

LTE−R C/U plane split network C/U plane split network with grey system theory C/U plane split network with proposed scheme

0.7 0.6 0.5 0.4 0.3 0.2

−80 0.1 −90

0

1

2 3 Location of the train (km)

4

0 1.5

5

2 2.5 Location of the train (km)

3

Fig. 6. Handover trigger probability comparisons

Fig. 4. The received signal strength form each eNBs

0.2

80 Macro eNB i Macro eNB j Small eNB a Small eNB b

70 60

0.18 0.16

LTE−R C/U plane split network C/U plane split network with grey system theory C/U plane split network with proposed scheme

Outage probability

Signal quality (dB)

0.14 50 40 30

0.12 0.1 0.08 0.06

20 0.04 10 0

0.02

0

1

2 3 Location of the train (km)

4

5

0 1.5

2 2.5 Location of the train (km)

3

Fig. 5. The received signal quality form each eNBs

Fig. 7. Outage probability comparisons

Fig. 4 illustrates the relationship between the received signal strength of different eNBs and the location of the train. Fig. 5 shows the simulation results of the received signal quality from different eNBs at different locations. According to the system parameters set in Table I, the abscissa of the macro eNB i and j are 0 km and 4.8 km, respectively, while the small eNB a and b are located at 1.6 km and 3.2 km. The overlapping area of the macro eNBs and the small eNBs is (1.8, 3) on the X-axis. It can be seen that when the high-speed train moves away from the target eNBs, the received signal strength and signal quality begin to attenuate. In particular, since the small eNBs use a high-frequency band of 5 GHz, the signal from the small eNBs will suffer more severe path loss than the macro eNBs. Fig. 6 shows the simulation results of the handover trigger probability with different schemes. Because the inter-macro eNB handover is composed of macro-macro eNB handover and small-small eNB handover, the trigger probability in the

C/U plane split network is lower than the LTE-R network. In [21], the authors proposed a gray prediction handover scheme in C/U plane split network. Compared with the traditional network, its trigger probability increases slightly after 2.7 km. In this paper, our proposed handover scheme can further improve the trigger probability and ensure excellent performance. Fig. 7 shows the simulation results of the outage probability with different schemes. The outage probability in C/U plane split network is higher than that of LTE-R network when approaching the edge of cell. The gray prediction handover scheme can greatly reduce the outage probability. In the proposed handover scheme, when the handover triggered by the front antenna was failed, the TRS can use the rear antenna to connect with the target eNB, thus the outage probability can be greatly reduced. Fig. 8 shows the simulation results of the handover success probability with different schemes. The proposed handover scheme can greatly improve the handover success probability.

2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2018.2794770, IEEE Access 9

1 0.9

Handover success probability

0.8

LTE−R C/U plane split network C/U plane split network with grey system theory C/U plane split network with proposed scheme

0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 1.5

2 2.5 Location of the train (km)

3

Fig. 8. Handover success probability comparisons

Because the handover can be triggered again by the rear antenna if the front antenna executes handover failed. Compared with other handover schemes, the handover success probability in the proposed scheme is increased by about 35.7% at the position of 3km. V. C ONCLUSIONS In this paper, we proposed a dual-link soft handover scheme for C/U plane split network in HSR. Because of the characteristic of the C/U plane split network, the performance of inter-macro eNB handover in HSR scenarios is worse than the traditional LTE-R network. It severely restricts the applicability and availability of the C/U plane split network in HSR. The primary cause is that the high-speed train has to trigger the macro-macro eNB handover and small-small eNB handover respectively before leaving the overlapping area, otherwise the handover will fail. In order to solve this problem, the proposed handover scheme set up two antennas to execute handover cooperatively. Meanwhile, the bi-casting is adopted to decrease the communication interruption time. Simulation results show that the proposed handover scheme can effectively improve the handover success probability, reduce the outage probability and ensure the reliability of communication for C/U plane split network in HSR communication system. R EFERENCES [1] B. Ai, X. Cheng, T. K¨urner, Z.-D. Zhong, K. Guan, R.-S. He, L. Xiong, D. W. Matolak, D. G. Michelson, and C. Briso-Rodriguez, “Challenges toward wireless communications for high-speed railway,” IEEE Trans. Intell. Transp. Syst., vol. 15, no. 5, pp. 2143–2158, 2014. [2] J. Calle-S´anchez, M. Molina-Garc´ıa, J. I. Alonso, and A. Fern´andezDur´an, “Long term evolution in high speed railway environments: Feasibility and challenges,” Bell Labs Technical Journal, vol. 18, no. 2, pp. 237–253, 2013. [3] S. Hong, J. Brand, J. Choi, M. Jain, J. Mehlman, S. Katti, and P. Levis, “Applications of self-interference cancellation in 5G and beyond,” IEEE Commun. Mag., vol. 52, no. 2, pp. 114–121, 2014. [4] F. Boccardi, J. Andrews, H. Elshaer, M. Dohler, S. Parkvall, P. Popovski, and S. Singh, “Why to decouple the uplink and downlink in cellular networks and how to do it,” IEEE Commun. Mag., vol. 54, no. 3, pp. 110–117, 2016.

[5] H. Xie, F. Gao, and S. Jin, “An overview of low-rank channel estimation for massive MIMO systems,” IEEE Access, vol. 4, pp. 7313–7321, 2016. [6] H. Xie, B. Wang, F. Gao, and S. Jin, “A full-space spectrum-sharing strategy for massive MIMO cognitive radio systems,” IEEE J. Sel. Areas Commun., vol. 34, no. 10, pp. 2537–2549, 2016. [7] T. Nakamura, S. Nagata, A. Benjebbour, Y. Kishiyama, T. Hai, S. Xiaodong, Y. Ning, and L. Nan, “Trends in small cell enhancements in LTE advanced,” IEEE Commun. Mag., vol. 51, no. 2, pp. 98–105, 2013. [8] C.-X. Wang, F. Haider, X. Gao, X.-H. You, Y. Yang, D. Yuan, H. Aggoune, H. Haas, S. Fletcher, and E. Hepsaydir, “Cellular architecture and key technologies for 5G wireless communication networks,” IEEE Commun. Mag., vol. 52, no. 2, pp. 122–130, 2014. [9] T. S. Rappaport, S. Sun, R. Mayzus, H. Zhao, Y. Azar, K. Wang, G. N. Wong, J. K. Schulz, M. Samimi, and F. Gutierrez, “Millimeter wave mobile communications for 5G cellular: It will work!” IEEE Access, vol. 1, pp. 335–349, 2013. [10] L. Yan, X. Fang, and Y. Fang, “A novel network architecture for C/U-plane staggered handover in 5G decoupled heterogeneous railway wireless systems,” IEEE Trans. Intell. Transp. Syst, vol. 18, no. 12, pp. 3350–3362, 2017. [11] X. Xu, G. He, S. Zhang, Y. Chen, and S. Xu, “On functionality separation for green mobile networks: concept study over LTE,” IEEE Commun. Mag., vol. 51, no. 5, pp. 82–90, 2013. [12] S. Ni, J. Zhao, and Y. Gong, “Optimal pilot design in massive MIMO systems based on channel estimation,” IET Communications, vol. 11, no. 7, pp. 975–984, 2016. [13] L. Yan and X. Fang, “Reliability evaluation of 5G C/U-plane decoupled architecture for high-speed railway,” EURASIP Journal on Wireless Communications and Networking, vol. 2014, no. 1, p. 127, 2014. [14] A. N. Bikos and N. Sklavos, “Architecture design of an area efficient high speed crypto processor for 4G LTE,” IEEE Trans. Dependable and Secure Computing, 2016. [15] H. Ishii, Y. Kishiyama, and H. Takahashi, “A novel architecture for LTE-B: C-plane/U-plane split and phantom cell concept,” in Proc. IEEE Globecom Workshops, Anaheim, CA, USA, 2012, pp. 624–630. [16] A. Sniady and J. Soler, “An overview of GSM-R technology and its shortcomings,” in Proc. IEEE 12th ITST, Taipei, Taiwan, 2012, pp. 626– 629. [17] J. Wu and P. Fan, “A survey on high mobility wireless communications: Challenges, opportunities and solutions,” IEEE Access, vol. 4, pp. 450– 476, 2016. [18] L. Fan, X. Lei, N. Yang, T. Q. Duong, and G. K. Karagiannidis, “Secure multiple amplify-and-forward relaying with cochannel interference,” IEEE J. Sel. Topics Signal Process., vol. 10, no. 8, pp. 1494–1505, 2016. [19] N. Docomo, “Requirements, candidate solutions & technology roadmap for lte rel-12 onward,” 3GPP RWS-120010, June, 2012. [20] L. Tian, J. Li, Y. Huang, J. Shi, and J. Zhou, “Seamless dual-link handover scheme in broadband wireless communication systems for high-speed rail,” IEEE J. Sel. Areas Commun., vol. 30, no. 4, pp. 708– 718, 2012. [21] H. Song, X. Fang, and L. Yan, “Handover scheme for 5G C/U plane split heterogeneous network in high-speed railway,” IEEE Trans. Veh. Technol., vol. 63, no. 9, pp. 4633–4646, 2014. [22] Z. Jiang, C. Xu, J. Guan, H. Zhang, and S. Yu, “Loss-aware adaptive scalable transmission in wireless high-speed railway networks,” in Proc. IEEE ICC, Paris, France, 2017, pp. 1–6. [23] X. Zhu, S. Chen, H. Hu, X. Su, and Y. Shi, “TDD-based mobile communication solutions for high-speed railway scenarios,” IEEE Wireless Commun., vol. 20, no. 6, pp. 22–29, 2013. [24] A. Khan and A. Jamalipour, “Moving relays in heterogeneous cellular networksła coverage performance analysis,” IEEE Trans. Veh. Technol, vol. 65, no. 8, pp. 6128–6135, 2016. [25] L. Fan, R. Zhao, F.-K. Gong, N. Yang, and G. Karagiannidis, “Secure multiple amplify-and-forward relaying over correlated fading channels,” IEEE Trans. Commun., vol. 65, no. 7, pp. 2811–2820, 2017. [26] Y. Xia, X. Fang, W. Luo, M. Liu, S. Li, and Y. Zhao, “Coordinated of multi-point and bi-casting joint soft handover scheme for high-speed rail,” IET communications, vol. 8, no. 14, pp. 2509–2515, 2014.

2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.