您的瀏覽器不支援JavaScript語法,但是並不影響您獲取本網站的內容。

技術探索

Moving Networks for 5G Communication Systems

  Abstract—The design of 5G communications must take moving networks into account. Among scenarios considered in 5G, a moving network refers to a moving node with advanced network capabilities or such nodes gathered together to form a movable network that can communicate with its environment. In this paper, we review the key performance indicators introduced by pioneer 5G projects, such as METIS and NGMN, and the technical challenges faced in bridging the gap between current and expected 5G moving network performance. The architectures used in moving networks are also introduced. The promising technologies are grouped into clusters including high mobility, resource orchestration, and network sharing to shed the light in tackling the technical challenges.

  Index Terms—moving network, mobility, 5G communication

I. SCENARIOS AND REQUIREMENTS

Figure 1: 5G challenges and scenarios [ 1 ]

Figure 1: 5G challenges and scenarios [ 1 ]

  WITH the deployment of 4G communication systems around the world, more and more people have started to enjoy the convenience it brings to work and daily life. However,eventually, the demand for better communication experience will exceed the capability of current 4G technologies.

  Therefore, the focus of the industry has been moved to the next generation communication technologies, also known as 5th generation of communication systems, or 5G.

  For now, the scenarios of 5G are still vague and diverse, yet there are some commonalities between the vison of different scenarios, which can be summarized as [ 1 ] :

  • Faster (in terms of data rate, latency, and mobility);
  • Denser (people as well as machine); and
  • Cheaper (in terms of manufacturing cost and energy consumption).

  Among the versatile scenarios considered by the industry,mobility is one of the topics gaining much attention. Enabling mobility has always been one of the major driving forces since the development of cellular communications. The ability to support mobility also evolves from nomadic to vehicular speed. For instance, the design of LTE-Advanced can support mobility up to 350 km/hr. However, it becomes more and more evident that communications under such high speed is very challenging,since the channel conditional varies very quickly and is difficultto predict or estimate [2] [3].

 On the other hand, high mobility transportations, e.g., Japan Tohoku Shinkansen, German Intercity-Express (ICE), AGV Italo, Taiwan High Speed Rail (THSR) and Shanghai Maglev,will become more popular in the future, and the speed may be increased to higher than 500 km/hr [4]. Therefore, the design of 5G communications must consider moving networks. 

  By moving network, we mean a moving entity, possibly with high or average speed and carrying a few or hundreds of passengers, and many on board passengers like to access Internet and other services using their mobile phones and other mobile devices. Thus, it will be required in 5G that movingnetwork users under high mobility can communicate subject to specific quality of experience (QoE) constraints, i.e., the communication experience should be similar to non-movingcases. Consequently, more innovative services for moving network users can be realized. The concept can also be stated as “best experience follows you” as shown in Figure 2. In thispaper, we review the key performance indicators anticipated and the technical challenges in achieving the performance indicators. In addition, we review the candidate architecturesand provide a novel way to cluster the promising technologies in tackling the technical challenges.

II. KEY PERFORMANCE INDICATORS

Figure 2: Best experience follows you [5]

Figure 2: Best experience follows you [5]

  At first, some definitions are given for well describing the performance indicators:

  • E2E (end-to-end) latency, also known as one-trip time (OTT) latency, refers to the time it takes from the moment a data packet sent by the transmitter to the moment the packet received by the receiver.
  • Reliability: the probability that data of an end user device is successfully transmitted to another peer within a predefined time frame.
  • Availability: the percentage of places inside a coverage area where a service is provided according to the users’ requested QoE level.
  • Experienced user data rate: the data throughput an end-user device achieves on the media access control(MAC) layer (user plane only) averaged during a predefined time span; a possible measure for the QoE level a user experiences

Table 1: KPI for moving networks


III. TECHNICAL CHALLENGES FOR MOVING NETWORKS

  In this section, the technical challenges a user equipment (UE) may encounter in a moving network are introduced.

A. Channel modeling

  The channel model for moving networks varies a lot depending on different factors including:

  • Terrestrials: open space, viaduct, hilly terrain, tunnel, etc.
  • Deployment: Macro base station only, leaky cable or remote radio head (RRH)/remote antenna unit (RAU) intunnel
  • Architecture: direct link, with repeater in carriage, or two-hop architecture 

Each UE needs to cope with large Doppler shift and rapid Doppler transition except in two-hop architecture since two-hop architecture can handle the impact of Doppler Effect such that UEs can communicate with network regardless of Doppler Effect.

  Currently there are several related channel models, such as WINNER (Wireless World Initiative New Radio) II D2a for channel between fixed base stations and moving relay nodes mounted to train roof [8], and 3GPP high-speed train profile [9].However, new model may be required in the case of radio-over-fiber (RoF) deployment, where more than two paths(from different RAUs transmitting the same signal) are observed with independent time-power-frequency profile.

  On the other hand, for two-hop architecture, it is widely believed that the UEs in a carriage face static channel. In fact, in addition to waves travelling inside the carriage, there may be waves re-entering from outside the window and it becomes a hyper-Rayleigh fading [10] which can be modeled by two-wave diffuse power (TWDP) channel model [11]. It poses new challenges for UEs.

B. Mobility management

   A UE is either in connected state to actively transmit and receive data, or in idle state. If a UE is moving fast while connected, the handoff operation must be seamless for the UE, especially for real time communications, such as voice or conference call. The faster the UE is, the more frequent the handoff happens, and context-based handoff optimization is more necessary.

 On the other hand, even if the UE is idle, the system stillneeds some mechanism to keep tracking of the UE’s location to some extent. Similarly, how to handle the fast moving idle UE while assuring energy conservation for the UE is important.

C. Group handoff

  When many UEs move together, they will face the similar channel condition and trigger handoff operation simultaneously,which is known as group handoff. The signaling burst caused by group handoff will overload the serving and target base stations. One of the solutions to prevent group handoff is adopting two-hop architecture [12].

D. Wireless TCP performance

  The design of TCP (transmit control protocol) was for wired networks, and packet drops will be considered as the consequence of network congestion, which will be followed by TCP window reduction and throughput degradation. It has been long recognized that TCP in wireless network has performance issue since packet loss could be caused by wireless error rather than network congestion [13], and it could be worse for high mobility moving networks. Innovative services over moving networks are possible only if the TCP performance they rely on can be stable in the presence of wireless error.


IV. ARCHITECTURES AND TECHNOLOGIES

  To address the technical challenges that users face in moving networks, many technologies are studied. Each technical component must be applied to one or all architectures introduced in this section. Technologies described in METIS and NGMN for moving networks enhancement are groupedinto three main themes, namely high mobility, resource orchestration, and network sharing, as shown in Table 2, Table 3 and Table 4Table 4, respectively.

A. Architectures

  The first architecture is direct access, i.e. each user connects the roadside radio access networks (RAN) the way it does in non-moving networks. Since there is no new infrastructure introduced, it will be the most popular architecture in the beginning. However, there are some drawbacks related to direct access architecture, such as large energy consumption due to high transmission power to overcome signal penetration loss and high computing power to overcome high-speed channel impairments.

  The other architecture is mobile-relay assisted. Relays are installed in the vehicles and work as base stations from the users’ point of view. There are L1, L2 and L3 relays, with better and better performance in terms of achieved user signal-to-interference-and-noise-ratio (SINR), at the cost of increased latency and complexity [14].

  Another similar architecture is mobile-router assisted. In this architecture, small cells or Wi-Fi access points are installed for user access, and another mobile router is installed to provide the
wireless backhaul for small cells or access points. Compared to mobile-relay assisted architecture, mobile-relay assisted architecture can provide another degrees of freedom, such as
network sharing and high security provisioning.

B. High mobility technologies

  Air interface methods include waveforms, coding/decoding,and multi-antenna technologies. The objective of new airinterface is to improve the robustness of mobile communication links, which is important for services with strict reliability requirements such as road safety applications. The highly time variant channels also requires researches in novel channel estimation and channel prediction techniques. FBMC (Filter bank Multicarrier) and UFMC (Universal Filtered Multicarrier) schemes, use digital filtering methods to shape the multicarrier waveform for offering better PHY layer performance.Advanced multiple-input multiple-output (MIMO)technologies (such as MISO predictor antenna array and massive-MIMO) exploit channel features to increase reliability or create multiplexing gain.

  The interference identification methods aim at providing better interference awareness in heterogeneous networks. Adaptive Projected Subgradient Method (APSM) uses an adaptive projection algorithm to estimate/identify long-term interference couplings between Base Stations (BSs) and users,while Minimum Mean Square Error (MMSE) Estimation uses a statistical estimation approach. Interference Identification using multilayer inputs combines available information from several network points using the Interference Identification Entity (IIE), which undertakes the identification of the potential aggressors in the interferer.

  For handoff optimization, street-specific context is worth exploiting in achieving optimized handoff parameters, while Fuzzy Q-Learning-based approach provides a generic basis for enabling self-optimizing and self-healing network operations in moving networks.

  The UE-centric network means that a set of network nodes provide connectivity to a given device with functions tailored to that specific device. It is a consequence of several trends in cellular communication evolution, such as cell densification and RAN virtualization. On the other hand, dual-connectivity design enable a user terminal associated to two different base stations for uplink and downlink, respectively, with an aim at optimizing the quality of the uplink connection.

Table 2: High mobility technologies
                                                      

C. Resource orchestration technologies

  Radio resource management methods include smart mobility and resource allocation using context information, such as QoS indicators achieved by the UE on the passed cells and used for scheduling decisions in the current cell. Non-coherent MIMO communication enables data detection to be carried out without any knowledge of the MIMO channel coefficients at the
receiver side.

  Context information, such as user position, velocity, radio propagation map, profile of the current and upcoming services, can be exploited to assist decision-making process. It is important to design an efficient management and exchange mechanism for aggregating the context information. The mobility of users is usually direction oriented. Therefore, it is feasible to predict next cell for user transition, and load balancing or other radio resource management (RRM) schemes are triggered in the predicted next cell. Besides, the users tend to have similar behaviors in specific locations during specific time-periods. Exploiting user habits for better resource mapping will increase the user capacity and experienced throughput. Big data and data mining technologies will facilitate the exploitation.

D. Network sharing technologies

  While the technologies mentioned so far focus on UE-BS interactions to enhance user experience, network sharing technologies address networking issue for a group of users associated to different operators. The need for network sharing in moving networks comes from the space limit of moving entity, such as trains or buses. The radio access infrastructure needs to be shared among users associated to different operators. Shared elements can include infrastructure, spectrum,BTS (small cells and macro base stations), backhaul, fronthaul, etc. Technologies should be developed to optimize network sharing agreements and enable flexible business models/commercial relationships.

  To better implement network sharing functionalities, virtualized mobile core network and flexible split of RAN functions among network will be the next evolution. By abstracting software based functionality from common pool of hardware, mobile core network elements become virtualized functions decoupled from specialized hardware. Flexible split of RAN functions enables to perform centralization or decentralization of L1/L2 RAN protocols based on specific needs.

V. CONCLUSIONS

  With the increased popularity of high mobility transportations, moving networks need to be addressed in 5G communication systems so that people can have similar experience no matter they are moving or not. 

  Key performance indicators (KPI), including mobility, E2E latency and experienced user data rate are provided for 5G moving networks. The technical challenges for current UEs to reach those KPIs are stated, including accurate channel modeling, mobility management, group handoff and TCP performance issue in wireless communications. Architectures, classified into direct access, mobile-relay-assisted, and mobile–router-assisted, and technologies, clustered into main themes including high mobility, resource orchestration, and network sharing, can be the guideline in looking for candidate solutions to the technical challenges.

Table 3: Resource orchestration technologies
                                                       

Table 4: Network sharing technologies
                                                         

VI. REFERENCES

  1. A. Osseiran, "Challenges and Scenarios of the fifth Generation (5G) Wireless Communications System,"Nov 2013. [Online]. Available:https://www.metis2020.com/wp-content/uploads/present ations/W@kth_METIS_overview_scenarios_20131115_web.pdf.
  2. ITRI, "R4-150018: Performance Evaluation for High Speed Train Scenario," in 3GPP TSG-RAN WG4 Meeting#74, Athens, Greece, 2015.
  3. D. Inc., "R4-151490: Study on High Speed Scenarios," in 3GPP TSG-RAN WG4 Meeting #74bis, Rio de Janeiro, Brasil, 2015.
  4. "Japan maglev train breaks world speed record again,"BBC, 21 4 2015. [Online]. Available:http://www.bbc.com/news/world-asia-32391020.
  5. "Deliverable D1.1: Scenarios, requirements and KPIs for 5G mobile and wireless system," METIS, 2013.
  6. "The METIS 2020 Project – Laying the foundation of 5G," [Online]. Available: https://www.metis2020.com/.

  7. "NGMN," [Online]. Available:https://www.ngmn.org/home.html.
  8. “D1.1.2: WINNER II channel models,” WINNER,2008.
  9. “3GPP TS 36.101 v12.7.0,” 3GPP, 2015.
  10. J. Frolik, “A case for considering Hyper-Rayleigh fading channels,” IEEE Trans. Wireless Comm., vol. 6, no. 4, pp. 1235–1239, Apr. 2007.
  11. G. D. Durgin, T. S. Rappaport and D. A. D. Wolf, "New analytical models and probability density functions for fading in wireless communications," IEEE Trans. Comm.,vol. 50, pp. 1005-1015, 2002.
  12. Chang, Hsien-Wen, "Field trial results for integrated WiMAX and radio-over-fiber systems on high speed rail," in 2011 IEEE 22nd International Symposium on Personal Indoor and Mobile Radio Communications (PIMRC), Toronto, Canada, 2011.
  13. J. F. Kurose and K. W. Ross, Computer networking: a top-down approach, 6e, Pearson, 2012.
  14. Iwamura, Mikio, Hideaki Takahashi, and Satoshi Nagata."Relay technology in LTE-Advanced." NTT DoCoMo Technical Journal 12.2 (2010): pp. 29-36.

Hsien-Wen Chang

Hsien-Wen Chang

Hsien-Wen Chang received the B.S.and Master’s degree in electrical
engineering from National Tsing-Hua University, Taiwan, in 1999 and 2001,respectively. He is currently with Information and Communications Research Laboratories, Industrial Technology Research Institute, Taiwan.
His research interests include digital broadcasting and MIMO-OFDM communications. He is currently working on the broadband access and networking technologies for moving networks.
E-mail: seanchang@itri.org.tw

Chia-Lin Lai received her M.S. and Ph.D. degrees inNational Cheng-Kung University, Tainan, Taiwan in 2008and in 2014. She is currently an engineer in Industrial Technology Research Institute, Taiwan. Her researchinterests include QoS of high speed networks, optical
networks, and wireless networks.
E-mail: Chia-LinLai@itri.org.tw

Kun-Yi Lin received the B.S. degree and Ph.D. degree in electrical engineering from National Taipei University of Technology, Taiwan, in 2005 and 2013, respectively. He is currently with information and Communications Research Laboratories, Industrial Technology Research Institute,
Taiwan. His research interests include wireless and digital communications. He is currently working on networking technologies for moving networks and radio-over-fiber system design.
E-mail: LinKY@itri.org.tw

Hsu-Tung Chien received the B.S. degree in Computer Science from Tung Hai University, Taiwan, in 2014. He is currently studying for the Ph.D. degree in the Institute of Computer Science and Engineering, National Chiao-Tung University, Taiwan science 2014. His research interests
include computer networks and wireless networks.
E-mail: brachymsg@gmail.com