Mobile Backhaul Solutions

1 Introduction
    Recent advances in cellular technology have necessitated significant improvement in the mobile backhaul network. With increased use of smartphones and laptops, mobile communication has been moving from voice-centered infrastructure to data-centered infrastructure.

    New mobile technologies, such as high-speed packet access (HSPA+) and LTE, have significantly increased downlink and uplink speeds over the radio link. Improvements over the radio access link demand similar improvement in bandwidth on the backhaul network.

    However, current backhaul networks are based on legacy technologies that are optimized for voice transport. To benefit from advances in radio access link technologies, the backhaul network needs to be transformed from a slow, TDM-centered network into a high-speed, data-centered network.

    Driven by bandwidth demand of the Internet, engineers have been building high-speed, data-centered networks for many years. However, building a mobile backhaul network is very different from building a commercial data network. A mobile backhaul network requires

• QoS-based traffic with strict requirements on delay and jitter
• high availability
• timing synchronization
• frequency and phase synchronization for optimal handover
• coordinated multipoint transmission (CoMP) and interference cancellation in LTE-Advanced
• enhanced OAM
• multiple transport modes that support data traffic and TDM traffic for legacy base stations
• symmetric bandwidth requirements for the up-stream and down-stream between the central office and base stations.

In traditional wired broadband access, bandwidth is usually asymmetrical. As the use of cloud services increases, symmetrical bandwidth in both directions becomes more important when designing the network.

    In this paper, an Ethernet/MPLS-based solution is proposed to satisfy mobile backhaul requirements. The solution addresses the following issues:

• Depending on the situation, the migration path from the backhaul network (based on legacy technology) to the new infrastructure (capable of supporting the new cellular technologies such as LTE) may be different.
• High availability necessitates protection infrastructure. A means of efficiently reserving bandwidth on the protection path for an MPLS-based network needs to be articulated.
• OAM techniques of data-centered networks need to be incorporated into  traditional transport networks.
• Traffic aggregation needs to be performed closer to the base stations.
  In this paper, a new trend in mobile radio access network architecture is briefly discussed. In this architecture, called cloud radio access network (C-RAN), baseband processing of the radio signal is centralized. Although there are many advantages to a centralized approach in terms of power consumption and project engineering, among other things, C-RAN also requires a very different transport network between the radio amplifiers and antenna, and the central location where baseband signals are processed.

2 Basic Requirements
    During mobile communication, connectivity from the base station to the central office occurs over the mobile backhaul network.

    In this backhaul network, traffic for different generations of the radio access network is transported. Different generations of cellular technologies have different architectural names to indicate cell-site equipment and the corresponding equipment in the central office (CO). In this paper, the cell-site equipment is called base station unless otherwise noted (e.g. in LTE, the base station is eNodeB). The central office equipment is called message switch center-server (MSC-S) for voice, and gateway for data (e.g. SAE-GW for EPC, SGSN for UTRAN). The backhaul network is shown in Fig. 1.

    Connectivity between the base station and the CO site is determined by the technology used in the RAN, the location of the cell site, the bandwidth requirements, and local regulations.
In a GSM (voice and limited data traffic) and UMTS/CDMA 2000 (voice-centered with increasing data traffic) mobile network, many cell sites use microwave as backhaul technology. However, in the LTE era of wide spectrum and high data rate, the microwave-based backhaul network does not have adequate bandwidth. In this paper, we discuss a fiber-based mobile backhaul network.
The current practice for building mobile backhaul networks is a flat IP-centered architecture. To support legacy mobile technologies, one of the key requirements of a mobile backhaul network is support for a native TDM interface. Support for such an interface usually comes in the form of circuit emulation, for example, PWE3 emulation, in the flat IP network.

    Different generations of mobile network also have different delay and jitter requirements. For LTE, Next Generation Mobile Network Alliance (NGMN) specifies a maximum delay of 5 ms (one way), and there are no specific jitter requirements.

    The common requirement for LTE bandwidth, which also depends on spectrum width and the number of carrier-sections, is around 300-400 Mbit/s. In LTE, the physical layer can be gigabit Ethernet, 10G Ethernet (in sharing mode), GPON, or 10G EPON.

    Normal EPON is not included in the physical layer because the sharing capability of the link is a maximum of 1G. 10G GPON is also omitted because current 10G GPON has asymmetric bandwidth.

    Fig. 2 shows the basic requirements for a mobile backhaul for advanced mobile technologies such as LTE and LTE-Advanced.

3 MPLS-Based Solutions
    In this section, the specifics of using multiprotocol label switching (MPLS) technology for mobile backhaul application are articulated.

    Apart from the final access link (which can be xPON-based), inside the mobile backhaul network, the layer-2 technology is Ethernet. With Ethernet, technologies such as
Layer-2/Layer-3 virtual private network (VPN), pseudo-wire for circuit emulation in supporting TDM traffic, and MPLS can be easily supported.

    For mobile backhaul applications, a pure IP-based network with no MPLS can also provide adequate service with increases in delay and jitter. These increases are due to an increase in processing at each hop. With an IP transport network supporting multiple services, and considering IP router port cost and processing complexity, mobile backhaul can best be implemented via an MPLS-based network.

    Another contender for mobile backhaul could be optical transport network (OTN) with ODU-0 to transport gigabit Ethernet signals. But an OTN network is TDM in nature and cannot be as flexibly deployed as an MPLS network.

3.1 QoS and CoS
    To make the network scalable so that it can provide many different types of services, CoS is usually used with queuing and discard techniques. A mobile backhaul network needs to queue and discard priorities from the radio access network.

    In almost all radio access technologies, the relative queuing and/or discard priorities are marked. After baseband processing of the radio signals at the entrance of the mobile backhaul network, the markings on the radio signals need to be remapped onto the technologies used in the backhaul network.

    The core technology used in the mobile backhaul network is MPLS over Ethernet. For layer-2 (Ethernet), the 802.3p bits can indicate the priority of the packet. At the starting point of the MPLS label switched path (LSP), the EXP bits can also indicate the priority of the packet.

3.2 Timing Synchronization
    In a mobile backhaul network with Ethernet and MPLS, timing needs to be synchronized for better handoff support, better voice/video quality, lower interference, and better bandwidth (e.g. CoMP).

    There are a number of ways of supporting timing synchronization, each of which has different advantages and disadvantages:

• GPS based timing. This is accurate for frequency and can also be used for phase synchronization. The drawback is that GPS is often unavailable because of a poor environment for acquiring GPS signals or because the U.S. government has encrypted the signals.
• packet-based timing synchronization (IEEE1588v2). This synchronization mechanism has been widely evaluated over the past a few years. It is adequate for synchronization in universal mobile telecommunications system (UMTS) networks. A CDMA network uses GPS timing. The performance of this mechanism in synchronizing timing for more time-stringent requirements, such as CoMP, still requires further study.
• synchronous Ethernet. This scheme only works if Ethernet is provided end-to-end. In this case, the last access link also needs to be Ethernet, and an xPON-based solution is excluded. Synchronous Ethernet can be deployed if the carrier is not using xPON-based technologies for the access link between the backhaul network and the base station.

3.3 OAM for MPLS-Based Network
    OAM is an important and fundamental in transport networks[1]. It contributes to

• reducing operational complexity and costs because it allows for efficient and automatic detection, localization, handling, and diagnosis of defects. It also minimizes service interruptions and operational repair time.
• enhancing network availability by ensuring that defects, such as those resulting in misdirected customer traffic, are detected, diagnosed, and dealt with before a customer reports the problem.
• meeting service and performance objectives. OAM allows service-level agreements (SLAs) to be verified in a multimaintenance environment and allows service degradation caused by packet delay or packet loss to be determined.

    A backhaul solution should support end-to-end OAM in a multivendor environment and simplify network operations with OAM tools.

    Currently, the telecommunication standardization sector (ITU-T) SG15 and Internet engineering task force (IETF) are cooperating on MPLS transport profile (MPLS-TP). Compared with MPLS, OAM is a very important characteristic of MPLS-TP. Fault management (FM) OAM functions, such as continuity check (CC), continuity verification (CV), and remote defect indication (RDI), can automatically detect and localize defects that occur in network. Performance management (PM) OAM functions, such as packet loss measurement (LM), packet delay measurement (DM), and throughput measurement, can diagnose service degradation. OAM functionality is also the key for network survivability and triggering network protection.

4 Protection Bandwidth Reservation
    MPLS-based mobile backhaul is a mesh network with bandwidth assurance. When high availability is required, bandwidth also needs to be assured when the network experiences a limited fault, such as a single fiber cut or a single MPLS transport nodal failure.

    The current method of supporting high-availability is to provide a working path and a protection path. The required bandwidth is allocated on both the working path and protection path. If the working path experiences a failure, the protection path is used. Because the bandwidth is also allocated on the protection path, bandwidth is assured for the backhaul network. This arrangement is shown in Fig. 3.

    The protection bandwidth reservation scheme assures bandwidth when there is limited network failure, but shared protection between many different cell sites is not taken into account. This critical flaw is shown by the simple ring topology in Fig. 4.

    In the arrangement in Fig. 4, the protection bandwidth reserved on the link between nodes 5 and 6 (same as other links) is the sum of the bandwidth for the working paths A and B. If we assume a single failure on any node and take into consideration the ring structure, the maximum bandwidth required for protection is the maximum bandwidth for working paths A and B.
Bandwidth reservation for any link is performed in the egress direction. The adjacent node is concerned with the ingress direction, which is its egress direction on the link concerned.
For bandwidth reservation in a packet network, equivalent bandwidth is usually used. For any packet stream, the equivalent bandwidth is usually derived from the peak bandwidth, average bandwidth, maximum burst size, buffer size, and packet loss probability. The equivalent bandwidth is somewhere between the average bandwidth and the peak bandwidth. The theory of equivalent bandwidth, sometimes also called effective bandwidth, is discussed in [2] and [3].
In the following, for simplicity, the bandwidth required for reservation implies the equivalent bandwidth.

    A link k on node Y  is given by Y (k), k = 1,..., K y . There are K y links for Y. The egress bandwidth on a particular link has four main features:

    1) Normal working bandwidth. This is the bandwidth required if there is no failure inside the network. For a nodal link Y (k), 1 ≤ k ≤ Ky, the bandwidth,
BN (Y (k)) is the sum of the equivalent bandwidths of all the working paths over Y (k).

    2) Bandwidth may be absent on Y (k) because of the failure of link i  of node J (i.e. J (i ) failure). For any working path over Y (k) and J (i ), when there is J (i )failure, the working-path traffic is absent if the working path goes over
J (i ) before Y (k). There is a difference between 1+1 protection [4] and MPLS fast reroute protection [5]. The working path traffic is absent for 1+1 protection and may be absent for MPLS fast reroute, depending on topology and the detour path. Depending on the protection scheme, the working-path traffic may be absent if the working path goes over before Y (k). In this paper, this bandwidth is Ba (Y (k)),J (i )),where 1≤ k ≤K y  and 1≤ i ≤KJ.

    3) Bandwidth may be absent on Y (k) due to failure of J. For a complete failure of J, the bandwidth absent on link Y (k) is Ba (Y (k)),J (0)). In this case, the notation is the same as that for bandwidth absent because of link failure, and J (0) denotes the nodal failure.
4) Protection bandwidth on Y (k), 1≤ k ≤K y  because of failure on J (i ), 0≤ i  ≤KJ.. This failure includes both link and nodal failures. In such failures, the equivalent bandwidth for the protection path is BP (Y (k),J (i )), where 1≤ k ≤K y and 0≤ i ≤KJ.
All the required qualities, BN (Y (k)), Ba (Y (k),J (i )), and BP (Y (k),J (i )), where 0≤ i ≤KJ for any 1≤k≤K y , are available once the working path and the protection scheme is determined.

    With this information, we are ready to define a quantity

    where P (Y (k),J (i )) = BP (Y (k),J (i ))- Ba (Y (k),J (i )),and 0 ≤ i ≤ KJ.

    If there are N  nodes  in the network, for all values of J, where 1≤J≤N,J ≠Y, and for all values of i, where 0≤ i ≤KJ , a list  in descending order can be constructed based on P(Y(k),*) for a particular Y (k), 1≤k≤K y. This list is given by

    where                                     The notation J x is defined as J x ∈(1, ...,N ),
J x≠Y and 1≤ x ≤M.
In the ordered list, we have

    For a single link or node failure in the network, the protection bandwidth to be reserved for Y (k) is P (Y (k),J 1(i 1)).

    If the topology constraint is ignored, the upper bound for bandwidth reservation on Y (k) that is necessary to handle m  faults is the sum of the first m  members of the ordered list. That is,

    By using this method for reserving protection bandwidth, the mobile backhaul network is efficient and competitive. A tighter upper bound based on topology and traffic flow is also possible for multiple failures, and this will be left for future discussion.

5 New RAN Architecture:C-RAN
    To this point, there has been no change to the traditional mobile network architecture, where base stations consist of the radio unit (RRU) and baseband processing unit (BBU).
The RRU and BBU are connected with fiber optics. The RRU is located with the antenna on top of the cell tower, and the BBU sits on the ground.

    The BBUs for many different cell sites can be co-located to provide centralized baseband processing. The resulting architecture is called cloud radio access network (C-RAN) and is shown in Fig. 5.

    Because the modulated radio signals are transported between RRU and BBU, the requirements in terms of bandwidth, delay, and jitter are much more stringent than those in mobile backhaul networks. The preferred method of transport between the BBU and RRU are direct fiber connection or wavelength division multiplexing (WDM) if conservation of fiber links is required.

    Because of fiber connectivity in advanced mobile technology, variations, such as C-RAN, on traditional RAN architecture become feasible.

    C-RAN architecture is still in its infancy. Further division of the workload between BBU and RRU according to transport requirements warrants further study.

6 Conclusion
    In this paper, we have reviewed mobile backhaul requirement for current and future mobile access technologies. An MPLS-based mobile backhaul solution is proposed, and timing synchronization, MPLS OAM, and protection have been discussed. A new mechanism for reserving protection bandwidth is described. This mechanism ensures that protection bandwidth is efficiently reserved for a single fault and that an upper-bound protection bandwidth estimation mechanism is provided for multiple faults.

References
[1] Requirements for Operations, Administration, and Maintenance (OAM) in MPLS Transport Networks, RFC5860, 2010.
[2]ChengShang Chang, Performance guarantees in communication networks, Springer, 2000.
[3] Jean-Yves Le Boudec and Patrick Thiran, Network Calculus, Springer, 2001.
[4] Linear Protection Switching for Transport MPLS Networks, ITU-T G.8131, 2006.
[5] Fast Reroute Extensions to RSVP-TE for LSP Tunnels, RFC 4090, 2005.

Li Mo (mo.li@zte.com.cn)received his B.Eng. degree from the Department of Electrical Engineering, Queen’s University, in 1989. After graduation, he worked at IBM, Nortel, and Fujitsu before joining ZTE in 2001. Currently, Dr. Mo is the chief architect for ZTE USA and works as marketing specialist for ZTE headquarters in Shenzhen. Dr. Mo has worked in the networking industry for more than 20 years, and has numerous publications and U.S. patents. His research interests include fixed and mobile core networks, session control, QoS, and routing. Dr. Mo is also an active member of IEEE, ITU, ETSI, and IETF.

Fei Yuan (yuan.fei@zte.com.cn) is the acting vice general manager of the Department of Standard Development and Industry Relations at ZTE Corporation. He has been working in telecommunications for eighteen years and has a strong technical background in a range of fields. Fei Yuan guides the standardization and advanced research teams with innovative strategies and ideas.

Jian Yang (yang.jian90@zte.com.cn) joined ZTE after receiving her bachelor's degree from XiDian University. She is now the vice general standard engineer of bearer networks and has made many contibutions to MPLS-TP standard development, both in the ITU-T and IETF.

[Abstract] In this paper, we give an overview of mobile backhaul solutions and propose an MPLS-centered solution that takes into account timing synchronization, OAM, and protection. We also propose an evolved protection bandwidth allocation mechanism that makes the transport network as efficient as possible.

[Keywords] mobile backhaul; multiprotocol label switching (MPLS); optical networking; protection