An On-vehicle Dual-antenna Handover Scheme for High-Speed Railway Distributed Antenna System

Transcription

1 An On-vehicle Dual-antenna Handover Scheme for High-Speed Railway Distributed Antenna System ChongzheYang, Linghui Lu, Cheng Di, Xuming Fang Key Lab of Information Coding & Transmission State Key Laboratory of Rail Traffic Control and Safety Southwest Jiaotong University Beijing Jiaotong University Chengdu 63, China Beijing 44, China {linghuis, Abstract Distributed antenna technology, as one of the important next-generation wireless communication technologies, has aroused extensive attention. The technology has been applied in high-speed movement environment. Due to the high density coverage of distributed antennas, almost anywhere in the area has line-of-sight (LOS) to reach at least one fixed antenna. However it may correspondingly result in smaller overlap between adjacent cells and higher probability of handover failure in high-speed movement scenario. In order to solve these problems, this paper proposes a novel handover scheme based on on-vehicle dual-antenna for high-speed railway distributed antenna system (). On-vehicle antennas which collaborate each other, are mounted on the top of high-speed train (the one is in the front-end and the other is in the rear-end). The proposed scheme utilizes distributed transceivers and centralized processing technology. The numerical analysis results show that the novel scheme can pre-trigger handover appropriately, guarantee the higher handover success rate, and increase the system throughput by around 5%. In addition, the scheme is feasible and easy to be implemented. Keywords-Distributed Antenna System (), High-Speed Railway, Collaboration, Handover I. INTRODUCTION In cellular wireless communication system, some problems are caused by users high-speed movement, such as overfrequent handover, Doppler frequency shift, and channel fast variation. For the wireless communication of high-speed railway, handover success rate critically impacts the traffic safety. In order to ensure the system robustness and effectiveness, it requires shorter handover time and higher handover success rate, which cannot be achieved by the most current wireless communication systems. The future wireless communication network structure tends to be flat as the Internet network structure, rather than the traditional hierarchical tree structure. The mesh, relay and other existing distributed wireless communication technologies have gradually become the important support for future distributed wireless communication systems. With the distributed wireless communication technologies continuing to evolve, the concepts of mesh and relay are no longer confined to the traditional multi-hop communication networks. They are either fully distributed, or combined with the centralized control mode. The next generation wireless communication standards, such as LTE-Advanced and IEEE 8.6m, are being attracted by the distributed wireless communication technologies. The existing distributed antenna technology is mainly adopted by radio over fiber (RoF) system for WiMAX Bullet Train Field Trial in Taiwan []. Such system is characterized by using a large number of distributed antenna units along the railway to provide high-quality, seamless coverage of radio signals. On the one hand, different trains choose different antenna units to get access. The spatial diversity improves system capacity and Quality of Service (QoS). On the other hand, mobile trains take flexible handover manner to ensure reliable high-quality communications link when they move across two adjacent cells. To resolve problems for high-speed handover in Distributed Antenna System () for railway, the key is to design an elaborate handover scheme to guarantee the handover success rate. Currently, the related literatures are mainly to solve the problem of fast handover in the traditional cellular network. There are few researches about handover for high-speed railway. In [], Distributed Wireless Communication System (DWCS) is introduced to improve the system performance. DWCS is a new architecture for a wireless access system, which has the following features: distributed antennas, distributed processors, and distributed controlling. In DWCS, with the decrease of cellular radius, the power can be much saved for the base stations (BSs) and the mobile stations (MSs). However it may result in smaller overlap, so there is no enough time available for handover when the MSs quickly move through the overlap region. In [3], in order to solve the problem, it extends overlap region by disposing multiple antennas, thus ensures the MSs have enough time for the handover operation and improves the handover performance. In [4], a handover scheme for the is proposed which is implemented by updating the Active Antenna Set (AAS). When an MS moves through different cell, it should update the AAS, which can provide access service. The proposal changes the hard handover between the two adjacent cells to soft handover. In [5], to reduce the time of information process and transmission in the handover operation, a mechanism with a number of the antennas used by an MS is brought forward. In [6], a relay-assisted handover technique exploiting network coding for multi-hop cellular networks is described. It can enhance the signal quality in the overlap region. The MSs This work is supported by NSFC under the Grant 67785, Key Research Plan of MoR of China under the Grant 9X9-E and the State Key Laboratory of Rail Traffic Control and Safety under the Grant RCS8Κ of Beijing Jiaotong University //$5. IEEE

2 could obtain the neighboring BS s information earlier by network coding when the MSs pass through the overlap region. This paper proposes a novel handover scheme for highspeed railway. It is based on on-vehicle dual-antenna technology in which full-function antennas are mounted in the front and rear of the train respectively. Two antennas can either work independently, or collaborate whenever handover. The remainder of this paper is organized as follows: Section II provides the details of the scheme. Section III builds an analytical model, while Section IV presents the performance analysis. Finally, we conclude the paper in Section V. Figure 3. The train in the only logic cell II. ON-VEHICLE DUAL-ANTENNA BASED HANDOVER SCHEME Normally, in high-speed railway scenario, is configured by linear arrangement as shown in Fig., in which a logical cell is made up of one or multiple antennas. In a logic cell, all the antennas send same signals to the train, and only part of the antennas receive signals from the train. A central processing unit of the logical cell deals with the data transmission. As the future wireless communication systems will be allocated with higher frequency spectrum, for example,.5ghz, the coverage radius of the antenna unit is about.5- km. In this paper we select the typical coverage radius of km, and the width of the overlap region is.6km, namely the () antenna spacing is.4km. In Fig., the represents ()- remote antenna unit (RAU) in the logic cell i, and constitute a logical cell i which are centralized controlled by -(6) the central station i (CSi). Similarly, constitute a logical cell j which are centralized controlled by the CSj. According to the feature of, handover between the RAUs in same logic cell can be considered as the internal handover. It has lower handover delay, which weakly impacts the overall performance of the system. In this paper, we concern about the handover between and RAU j, namely the handover between two adjacent logic cells. As the Doppler frequency shift and channel estimation beyond the scope of this paper, and these problems can be solved through a certain Figure. High-speed railway scenario Figure. On-vehicle dual-antenna solution (a) (c) technologies in the physical layer, we ignore the relevant issues. Fig. shows the schematic diagram of on-vehicle dualantenna. Both the front and rear of the train are mounted a fullfunction antenna which is controlled by Central Control Station (CCS). On-vehicle dual-antenna either work independently, or collaborate when handover. As shown in Fig. 3, the train is completely in a logical cell, two antennas work independently at different frequency bands F and F, but the two antennas receive and transmit the same data. Therefore two antennas are mutual hot backup to form a robust communication link. Fig. 4(a) shows the location of a train in the overlapped region of two logic cells, or the handover region. In our scheme, firstly, the front antenna detects the received signal tends to deteriorate. If it drops to a certain value (a threshold), the handover process is triggered. Then the serving cell i will coordinate with the target logical cell j. The handover process starts. At the same time, the rear antenna is still in the original logic cell i within the normal communications. When the front antenna has successfully accessed to the target logical cell j (Fig. 4(b)), it will request resources in the new logical cell for both antennas. The target CSj communicates with the original CSi through the backbone. (b) Figure 4. Schematic diagram of train handover

3 Through the backbone, CSj obtains the data which CSi sends to the train, then CSj forwards it to the train at F3 band. At the moment, the front antenna receives the data from cell j, the rear antenna still receives the data from cell i. The original logical cell i and target cell j form a macro diversity system. When the rear antenna locates in the overlapped region (Fig. 4(c)), it will turn itself to use the resources pre-requested by the front antenna, and quickly adjust itself to the new environment and complete the handover in the cell j. The scheme transforms the hard-handover into macro diversity soft-handover, thereby improves the handover reliability. III. ANALYTICAL MODEL Let a train be located at x in the i-th cell (original logic cell). The received signal from original logical cell i is: L i / Six (, ) = P ix () ε (, ) / where P is the transmission power of original logical cell; L i is the propagation path loss with the COST3-Hata model [7]; ε (i,x) is a Gaussian distributed random variable with a zero mean and a standard deviation representing shadowing [8]. For simplicity, we do not consider the fast fading (including Doppler effect and multipath effect) in this paper. Similarly the signal from objective logical cell j is: L j / ε( j, x )/ S( j, x) = P () The co-channel interference is given by: I BS ix, N BS = Snx (, ) (3) n= where N BS is the number of co-channel interfering cells. S(n,x) represents interference signal power of co-channel cells.. Therefore we express the Signal-to-Interference-plus- Noise- Ratio (SINR) at x as: SINR = x N BS n= Six (, ) Snx (, ) + N In the on-vehicle dual-antenna system, the handover of front antenna was triggered at x at first, and then the handover of rear antenna was triggered at x actively. At x the handover of both antennas was triggered one after another. The Ant i represents the handover decision result of antenna i:, The i-th antenna needs handover Anti =, The i-th antenna does not need handover Thus the probability of the handover is: P = Pr[ Ant = or Ant = ] (6) handover If the signal strength of the target s better than the original RAU, and the difference is greater than a hysteresis margin Γ, handover is initiated. When Pr[ ε j, x = ε] ~ N(, σ j, x) and σ j,x is the standard deviation of ε j,x, we can get: + Li Lj Γ+ ε Pr[ Anti = ] = Φ σ i ε σ i Li Lmax Q e dε σi πσ where L i and L j are the propagation path loss. L max is the propagation path loss of the cell edge. After handover, the MS will be located at y in the j-th cell. If the SINR of the MS received is less than a threshold γ th, it will be outage. It can be defined as: outage y y th i (7) P = Pr( SINR < γ ) (8) where γ th is the minimum signal strength required for the normal communication. Communication outage may occur at x, where MS triggers handover to the edge of coverage area. The probability of outage: R Poutage = Pr( SINRy < γ th ) Pr[ y = z] (9) z= x where R is the radius of the cell. Then the probability of successful handover is: PHOsuccess = Phandover [ Poutage ] () = Pr[ Ant = ] ( Pr[ Ant = ]) Pr( S > S ) i i ( j, x) min The on-vehicle dual-antenna can also improve the throughput of the system. The distance of the two antennas is about 4m, and their SINRs will be different. Thus, we should always use the better received signal of the antenna to keep communication. We choose antenna which received good signal quality as the master antenna, another antenna is the slave antenna. Then we can use the adaptive modulation and coding (AMC) technique to obtain the system throughput.

4 IV. PERFORMANCE ANALYSIS In order to verify the proposal effect for handover operation, we present the MATLAB simulation results and analysis. In the simulation, suppose the radius of RAU coverage is km, the overlap is.6km. Fig. 5 shows the relationship between the handover probability and train location. The ordinate is handover probability, and the abscissa is the distance between the train and the serving RAU. The curve with blue star indicates the cumulative distribution curves of traditional handover probability. For example, when the abscissa is.7 and the ordinate is.3, it indicates at a distance of ~.7km, the MS has 3% handover probability within the region. The curve with the red triangle indicates the cumulative distribution curves of proposal handover probability. We can see when the train locates at.8km. The handover probability of proposal is larger than the traditional scheme with single antenna by %. Fig. 6 shows the probability density curve. In the with single antenna, the handover probability is 98.7% till the edge of the cell..in our proposal, the handover could occur earlier than the traditional scheme and the handover probability is.9.8 about 99.98% till the edge of the cell. Due to the appropriate earlier trigger handover, the handover success probability gets much improvement. Fig. 7 shows the simulation results of outage probability.. The abscissa in this figure represents the distance between the antenna on train and target RAU. The abscissa of.4 corresponds to the edge of the coverage of the target cell, where has the maximum outage probability. With the train moving towards the target RAU, its handover outage probability is gradually reduced. As shown at location of about.8 it is about. From the Fig. 7, it is clear that the premature handover will result in higher outage probability. Fig. 8 is the probability of successful handover. It can be clearly seen that the dual-antenna can significantly improve the handover performance. The successful handover probability increases from 98.4% to 99.8%. When the abscissa is.8km, the successful handover rate increases from 63% to 8%, which greatly enhances the handover performance. Fig. 9 shows the simple simulation results of the system capacity using the Shannon theorem, in which the train moves from the original RAU (distance = km) to the target RAU.5. outage of neighbor cell (99.5%coverage) handover probability(cdf) outage probability Figure 5. Comparison of handover performance (CDF) Figure 7. The outage probability handover probability (PDF) success probability Figure 8. The probability of a successful handover Figure 6. Comparison of handover performance (PDF)

5 capacity(mbps) throughput(mbps) Dual-antenna one-antenna Figure 9. The system capacity one-antenna dual-antenna Figure. Simulation of the system throughput rate (AMC) (distance =.4km). If the train has only one antenna, we can obtain the capacity in case of 4MHz bandwidth in which the blue star curve shows in Fig. 9. Assume the implementation of handover process time is 4ms, the train is running at 45km/h, and then the abscissa from.675km to.75km is the distance for handover to be implemented. As for single antenna it is a hard handover, when carry out handover it is unable to support the seamless communications. Therefore, the capacity of the blue curve at that time is. As both the front and rear of the train are mounted a fullfunction antenna, centralized controlled by CCS, the two antennas can collaborate each other to keep one of the antennas receive better signal quality. The red curve of capacity is shown in Fig. 9 under 4MHz bandwidth. For the red curve (two antennas), the abscissa is from to.km, the train selects front antenna as a data transmission antenna. From.Km to.4km, the throughput increase is mainly contributed by the rear antenna moving closer to the RAU. The quality of received signal gradually changes better and better. Since then the proposal still takes the advantage of this antenna as the data sending and receiving until the.7 position (handover position). After the handover completes, the train gradually closes to the target cell RAU. The quality of received signal in the front antenna gradually changes better. It becomes a data transmitting and receiving antenna, and makes the system throughput increase. For the blue curve (single antenna), because the train has only single antenna, the hard handover will cause communication disruption. When the train moves away from the RAU, the system throughput gradually reduces. It can be seen that the dual-antenna proposal greatly improves the throughput, and the gain reaches about 5%. If considering AMC, and taking the method and parameters proposed in [9], we can get the throughput performance in Fig.. V. CONCLUSION In the for railway, all the antennas in the logical cell transmit the same information to a train. When the train passes through the cell, it is unnecessary to handover between the antennas in the same logical cell. With the on-vehicle dualantenna system, a train could selectively receive stronger signal to get better SINR. In this study we have presented an onvehicle dual-antenna based handover scheme in for railway scenario, and analyzed the handover performance. The results show that the proposal is much better than the traditional scheme. Besides, collaboration of the dual antennas mounted on the train is expected to be widely used to resolve Doppler frequency shift and channel estimation in the high speed communication environment. Certainly we will study the technology in depth in the future. REFERENCES [] Bruce Chow, Ming-Li Yee, Michael Sauer, Anthony Ng'Oma, Ming- Chien Tseng, Chien-Hung Yeh, "Radio-over-Fiber Distributed Antenna System for WiMAX Bullet Train Field Trial," mws, pp.98-, 9 IEEE Mobile WiMAX Symposium, 9 [] S. Zhou et al., Distributed Wireless Communication System: A New Architecture for Future Public Wireless Access, IEEE Commun. Mag., Mar. 3, vol. 4, no. 3, pp [3] FENG GUO, JIANLI WANG, WEI WANG, Apparatus realizing distributed wireless cell and communication method CN (A), MAR 5, 9 [4] Zhan-jun Jiang et al., A Handoff Model in Distributed Radio Mobile Communication System, JOURNAL OF APPLIED SCIENCES, 7. 5(3), pp. -6 [5] Hunjoo Lee et al., Novel Handover Decision Method in Wireless Communication Systems with Multiple Antennas, 6 IEEE 64 th Vehicular Technology Conference, VTC-6 Fall. 5-8 Sept. 6 [6] S. Jeon and S. Lee, A relay-assisted handover technique with network coding over multihop cellular networks, IEEE Commun. Lett., vol., issue. 3, pp. 5-54, Mar. 7. [7] Andrea Goldsmith, Wireless Communication, Stanford University. [8] A. Molisch, WirelessCommunications.Wiley-IEEE Press, 5 [9] M.H.Ahmed and H.Yanikomeroglu, Throughput fairness and efciency of link adaptation techniques in wireless networks, IET Commun, vol.3, issue.7, pp7-38, 9