5.3 Wire-Line Backhaul Technology

5.3.1 DSL Technologies

Digital Subscriber Line (DSL) technologies utilize the existing copper lines originally installed for Plain Old Telephone Service (POTS) for data transmission. DSL is used for access transmission to provide IP services for home computer users, but it is very common backhauling technology, too. The DSL subscriber line is called loop and it is typically a cable pair, but most DSL-variants can also combine (bundle) several pairs. The reach and maximum capacity of different xDSL technologies varies according to the loop rate, noise conditions and complexity of modulation and coding, see Table 5.5.

The most common technology is Asymmetric Digital Subscriber Line (ADSL). It has different standards for Europe and North America, while the newest ADSL+ is more harmonized. It uses the same line as POTS but a DSL filter allows the same copper line to be used for both voice and data transmission. Capacity is asymmetric; downlink direction to the BTS or customer premises is greater than uplink capacity. The configurations are star-like. The customer premises equipment (CPE) is small and cheap while the network end, called Digital Subscriber Line Access Multiplexer (DSLAM), usually terminates several tens or hundreds of access lines and connects to the data network by ATM/IP/SDH interface. With G.bond extension capacity can be doubled to 48/6 Mbit/s.

Single-Pair High-speed Digital Subscriber Line (SHDSL) is a symmetric DSL technology. SHDSL occupies the whole bandwidth for data transmission and POTS speech service is not possible in the same copper line. SHDSL has been popular in backhaul installations when copper pairs have been available and capacity requirement reasonable, because it provides directly capability for E1/T1, ATM and Ethernet transmission. SHDSL standard is ITU-T G.991.2 and it is also known as G.SHDSL. The updated version is known as G.SHDSL.bis or SHDSL.bis. SHDSL features symmetrical data rates up to 2,304 kbit/s for one-pair installations. The two pair feature may alternatively be used for increased reach by keeping the data rate low. Halving the data rate per pair will provide similar speeds to single pair lines while increasing the error/noise tolerance. Higher data rates may be achieved using two or up to four copper pairs.

Very high bit rate digital subscriber line (VDSL) is the fastest technology available. It enables configurable up/downlink speeds over 100 Mbit/s some hundreds of meters loop. VDSL is commonly used as ‘fiber extension’ to deliver FTTB further to the home or to the BTS site. It is enough for LTE eNB rates in the beginning when site rates are not very high, or for small cell sites with only a few users per eNB.

Copper line is neither stabile nor interference-free signal path. Usually the line rate is negotiated in the beginning of the connection, but electro-magnetic interferences can disturb the signal later causing errors. It depends on DSL modem implementation how well it can adapt to the situation and changing traffic patterns. Due to heavy interleaving there may also be challenges to keep packet time delay variation in control for synchronization purposes.

Figure 5.12 presents measured uplink and downlink rates from an IP-DSLAM node using VDSL2 link to CPE. Measurements are performed with and without cross-talk. The degradation caused the interference is worse at shorter distances.

Figure 5.12 Impact of cable length on VDSL2 data rates.

There are several advanced methods to further improve the performance of DSL. Dynamic Spectrum Management includes methods like rate adaptation and near/far-crosstalk cancellation. Simple methods balance the signal spectrum in one cable pair. More complex methods do the processing for all parallel signals in a node and bunch of cables. They require heavy signal processing in central office end and typically some processing also in terminal end. VDSL vectoring is a good example of how bit-rates have increased from 100 Mbit/s level to over 200 Mbit/s level. DSL rates up to 1 Gbit/s over a copper-pair have been demonstrated in laboratories.

Pair bonding is the technique of combining two or more parallel copper lines to provide greater bandwidth. In typical copper cable installations there are usually several spare pairs. More and more household and office building users are moving to wireless leaving pairs for MBH use. These pairs can be utilized as long as they are valid. With current MBH capacity requirements it is a viable alternative to newly laid optical fiber. Factors such as the accessibility of the pairs and the requirements of central-office equipment are also important when evaluating feasibility.

Table 5.5 Properties of some DSL techniques [ITU G.991, G.992, G.993].

5.3.2 Optical Technology

Optical fiber is a thin highly transparent coated class-tube inside which modulated laser light transfers. Modern fibers have attenuation about 0.2 dB/km meaning 100 km run attenuates signal only 20 dB which sits in the link budget of many optical systems. Fibers constitute a cable.

Multi-mode optical fiber has a larger core (typical 50 μm) that allows use of inexpensive connectors, optical transmitters and receivers. Multi-mode fiber material is more expensive and causes dispersion that limits signal bandwidth and causes attenuation. Single-mode fiber is thinner (typically < 10 μm). It allows wider bandwidth and lower attenuation, but requires more expensive components and interconnection methods.

Connecting plain optical fibers requires special equipment and skills. Once optical connectors are installed in the fiber ends connecting can be performed as with any other cable. Optical systems strength is the long reach that enables concentrating processing intensive equipment in mobile networks into a few central offices.

Fiber optic transmission utilizes certain wavelength ranges with low attenuation, called windows and avoids those wavelengths which have naturally high attenuation. The first window is at 800-900 nm. At the second window 1300 nm fiber attenuation and dispersion are much lower, enabling long distances. The third window at 1500 nm is most used nowadays due to widely available amplifiers and low fiber attenuation. Hydroxide causes high attenuation peak at about 1400 nm if OH molecules are present in the medium.

5.3.2.1 Wavelength Division Multiplexing

Wavelength Division Multiplexing (WDM) is a technique that utilizes several optical wavelengths in the same fiber. Basically, it is FDM technique, but in the optical spectrum usually word ‘wavelength’ or lambda is used instead of frequency. Each wavelength is a channel dedicated to a certain user or service (Figure 5.13).

Without wavelength (optical) multiplexing the whole spectrum must be converted to electrical signal and multiplexed in a conventional way (for example SDH). Optical-Electrical-Optical–conversion (OEO) decreases the length between active nodes and increases cost and complexity. Optical multiplexing is a passive operation. The only active components needed through the whole transmission path that can be even thousands of kilometers are optical amplifiers.

Figure 5.13 WDM de/multiplexing.

Coarse WDM (CWDM) uses wide wavelength range 1270...1610 nm with a coarse grid. ITU-T G.694.2 defines 18 channels. Due to relaxed accuracy requirements the reach of CWDM is limited to about 60 km and it is suitable for 2.5 Gbit/s rates. SFP TRX-modules are available for CWDM to allow upgrading existing systems to this technology. CWDM is also used to transmit upstream and downstream optical signals in a single fiber using bi-directional transceiver (BiDi SFP). Typically transmitters (lasers) are quite frequency coherent but receivers are more wide band. That is why frequency selective filters are needed in demultiplexers to pick up just the wanted channel.

Dense WDM (DWDM) uses C-band and L-band (1530...1620 nm) wavelengths that enable a much longer reach. DWDM systems range from 40 channels or ‘colors’ with 100 GHz grid to 160 channel with 25 GHz channel spacing. ITU-T G.694.1 defines frequency grid for DWDM. Systems use a fiber pair with transponders and multiplexers capable of handling TX and RX directions. Transceiver is a combination of optical transmitter and receiver. It can convert the electrical signal (OEO) directly to the wanted wavelength or to the ‘grey light’ containing all wavelengths. Transponder is a wavelength converter tuned to the wanted channel in the fiber network side and receiving either grey optical or electrical signal from the client system. WDM is used in very long haul high capacity core and international connections. WDM is used in MBH aggregation and usage also in access is increasing.

Reconfigurable optical add-drop multiplexer (ROADM) is an optical add-drop multiplexer that can be remotely configured to switch the individual of multiple wavelengths' traffic from optical signal. This allows data channels to be added or dropped from a WDM system without the need to OEO.

5.3.2.2 Passive Optical Network (PON)

Passive optical network (PON) is a point-to-multipoint fiber access technology that uses unpowered optical splitters to serve several end points. Hub-point or central office for PON is called optical line terminal (OLT) and end points are optical network units (ONU). In mobile backhaul use ONU may also be called Cellular Backhaul Unit (CBU). The point-to-multipoint fiber tree and branch options are called optical distribution network (ODN).

With PON less fiber and less central equipment is needed than with point-to-point optical links (Table 5.6). Typical splits are 16...128 ONUs per one OLT. Downstream signals are broadcast, i.e. every ONU can ‘see’ the signal. This may require coding/encryption. Upstream signals are combined using a multiple access protocol, usually time division multiple access (TDMA) that causes higher upstream delay. The OLTs measure the range to the ONUs in order to provide time slot assignments for upstream communication.

GPON (ITU-T G.984) networks have now been deployed in numerous networks across the globe. ITU G.987 defines 10G-PON (or XG-PON) with 10 Gbit/s downstream and 2.5 Gbit/s upstream. Framing is close to GPON and designed to coexist in the same network. Due to attenuation of splitters the range is typically limited to couple of tens of kilometers and split to 32.

Ethernet PON (EPON or GEPON) and 10G-EPON is included as part of the Ethernet in the first mile (EFM) definition. EPON uses standard Ethernet frames with symmetric 1 Gbit/s upstream and downstream rates. 10Gbit/s EPON supports simultaneous operation of 10 Gbit/s on one wavelength and 1 Gbit/s on a separate wavelength. The 10G-PON wavelengths differ from GPON and EPON, allowing it to coexist on the same fiber with either of the Gigabit PONs. EPON is the most widely deployed PON technology globally. EPON is also part of the DOCSIS Provisioning of EPON (DPoE) specifications. DPoE makes the EPON OLT act like a DOCSIS cable modem. In addition DPoE supports MEF 9 and 14 Ethernet services.

Wavelength Division Multiplexing PON (WDM-PON) utilizes several wavelengths. It allocates one or more dedicated wavelength(s) for each ONU or concatenates wavelengths to vary transport capacity. There is no WDM-PON standard available but some vendors are working on it.

Table 5.6 Comparison of PON technologies [Standards mentioned].

5.3.3 Ethernet Interfaces

Most common data access connection is 10/100/1000 Mbit/s Ethernet. Usually all speeds are supported by the same interface card. In contrary to DSL cabling Ethernet uses special cables categorized based on the maximum speed they can carry. The key for cables is to minimize cross-talk between pairs. Summary of Ethernet standards are in tables 5.7 and 5.8.

Cat 5 is the Fast Ethernet (100 Mbps) cable up to 100 m lengths. The cable, termination and verification are specified in ANSI/TIA/EIA-568 A or B. Standard connector is called RJ-45 or 8P8C modular connectors. Category 5e is the enhanced version. Basic cable type in today's installation is Cat 6 or higher. Cat 6 cable is for Gigabit Ethernet speeds and cable length up to 50 m. Length depends on the rate and total quality of the installation, and the bandwidth must be verified if targeting the maximum speeds. Using Cat-6a cable 10GBASET connections up to 100 m are possible. Even wider bandwidth Cat 7 cable standard have been defined but not widely used today.

Table 5.7 Fast Ethernet (FE, 100 Mbit/s) standards. MM=Multi-mode, SM=Single-mode.

10 and 100BASE-T use two pairs and with a splitter two 100BASE-T lines can be transferred over one cable. 1000BASET uses four pairs. 100BASE-FX use SC, ST, LC, MTRJ or MIC connectors. LC and SC connectors are the most commonly used ones. Telecom field installations and outside plants (OSP), however, require hardened fiber optic connectors like BX5. Telecommunication installations tend to use low-attenuating high-quality cables while the short range datacom typically use lower cost cables.

Table 5.8 Giga Ethernet (GE, 1 Gbit/s) standards. MM=Multi-mode, SM=Single-mode.

It is possible that Ethernet cards have separate detachable Gigabit Interface Converters (GBIC) that performs the interfacing functionality to copper or fiber. It makes it easier to change transport media, swap broken interface and manage spare parts. De-facto interface module nowadays is a small form-factor pluggable (SFP) module. Modules are not fully compatible and also some vendors may have vendor a lock-in feature that forces only proprietary modules to be used.

5.3.4 Ethernet in the First Mile

Ethernet in the First Mile (EFM) is the name for a set of Ethernet standards in amendment IEEE Std 802.3ah-2004. It is a set of additional specifications, allowing users to run the Ethernet protocol over various media, such as a pair of telephone wiring and single strands of fiber (Table 5.9).

Table 5.9 Ethernet in the First Mile (EFM) additions to 802.3.

Other extensions are 100BASE-LX10, 100BASE-BX10 and 1000BASE-BX10. These extensions make the EFM port types better suited for use in access networks and mobile backhauling.

5.3.5 DOCSIS

Data Over Cable Service Interface Specification (DOCSIS) is a method of transporting data over cable-TV network using modulated RF-signal. It was first standardized for the USA and there is also a version for Europe called EuroDOCSIS. First versions of DOCSIS were specified mainly for cable-only TV signal transmission, but later included improvements for internet-traffic. The service allows bi-directional transfer of data between the cable system headend (central-office or hub) and customer locations over an all-coaxial or hybrid fiber-coax (HFC) cable network. The headend interface is called as Cable Modem Termination System (CMTS) and customer premises equipment as Cable Modem (CM). Cross-version compatibility has been maintained across all versions of DOCSIS. [19]

A maximum optical/electrical distance between the CMTS and CM is 160 km (100 miles) in each direction. It uses 6 MHz RF-channels and speed is 38 Mbit/s down and 27 Mbit/s uplink. Channels can be bonded so that next generation 8 carriers bonding yield up to 304 Mbit/s.

DOCSIS is a potential last mile mobile backhaul access method because of large penetration of cable-TV lines in populated areas and high speed. However, it is not widely used.