PON (Passive Optical Networks)

In recent years, the Information Society developed rapidly, due in large part to the increased competitiveness driven by the deregulation of the Telecommunications market and the appearance of new broadband services.

These two factors have resulted in a need for better communications networks capable of offering greater bandwidth at lower cost. ADSL, a technology that continues to make use of the copper subscriber loop, is currently the undisputed star in the European panorama.

On the other hand, the ever-increasing bandwidth demands by users has forced both established and emerging operators to revise their strategies, starting with a race to double the speed of their lines, which in the eyes of the uninitiated, appears to be limitless. However, ADSL does have a significant technical limitation: the maximum bandwidth that it can offer can under no circumstances exceed 8Mbps downstream and 4 on the upstream channel. Also, these values drop drastically the further the user is from the central office.

And although new technologies like ADSL2 and ADSL+ provide a slight increase in the bandwidth offered to subscribers, the distance limitations, which are inversely proportional to bandwidth, represent a significant bottleneck that blocks the possibility of progressively increasing the quality of service. This has been called the “first-mile problem”.

In this sense, fiber-optic technology is presented as a solid solution to the problem thanks to its robustness, its unlimited potential bandwidth, and the continuous drop in the costs associated with lasers. If we combine these aspects with the new construction (new housing developments, new housing blocks, shopping centers) that are already integrated into the structured single-mode fiber-optic wiring, because of its low marginal cost in the project, we are talking about a scenario that is completely paid for to allow the deployment of fiber-optic connectivity solutions that reach all the way to the homes.

And in terms of architectures for the future, PON networks are positioned as a reliable wager: their contained costs in terms of electrical-optical equipment and the efficiency of branch-tree topologies provide an additional incentive versus traditional deployments based on point-to-point connectivity.  

Common characteristics of PON systems

Since 1995, telecom operators have been working on a network of integrated subscriber access services that give users fiber-optic broadband access and contain the costs of traditional point-to-point deployment (like ADSL does with the copper subscriber loop, or on optical fiber).

Passive Optical networks are modeled after recycled CATV networks to offer broadband services by enabling the return channel. A CATV network is made up of various optical nodes connected to the head end with optical fiber, and from these nodes, a shared architecture of coaxial cable branches out with the subscriber accesses. Normally in CATV, each optical node attaches a particular number of users (depending on the bandwidth to be assigned to users, using coaxial cable and electrical splitters. Passive optical networks replace the coaxial section with single-mode optical fiber and the electrical splitters with optical ones. This way, fiber’s increased capacity makes it possible to offer improved bandwidth on the downstream channel and above all, on the upstream channel, surpassing the typical limitation of 36Mbps per optical node in DOCSIS and EURODOCSIS cablemodem systems. 

This new architecture is a lower-cost evolution in comparison with traditional alternatives such as point-to-point networks or switched networks to the block, since it requires less equipment for the electrical-optical conversion and does away with the high-density network equipment needed for switching.

PON architectures are attracting the attention of the telecommunications industry as a way to attack the problem of the first mile, since it has obvious advantages:

• PON networks allow connection to users located up to 20 Km from the central office (or optical node). This is far greater than the maximum coverage distance of DSL technologies (maximum 5 Km from the central office).
• PON networks minimize fiber deployment on the local loop and make it possible to use tree-branch topologies that are much more efficient than point-to-point topologies. In addition, this type of architecture reduces the density of the equipment in the central office, and consequently, consumption.
• Passive optical networks offer higher bandwidth density per user because of fiber’s greater data transport capacity in comparison with copper alternatives (xDSL and CATV).
• As point-multipoint architecture, passive optical networks allow the possibility of superimposing an optical television signal from a CATV head end on another wavelength without making modifications to the data transport equipment (see section: VPON technology)
• PON networks increase quality of service and simplify network maintenance because they are immune to electro-magnetic noise, do not propagate electrical discharges from lightning, etc.
• PONs make it possible to grow to higher transfer rates by superimposing additional wavelengths.

Although PON networks as a concept have existed since the 90s, it has only been in the last two or three years that they’ve reached technological maturity to allow a large number of operators to begin to use them extensively. At this point, they appear to be the preferred option for building the future subscriber access network, once the growth potential of xDSL technologies has been exhausted.

 

Brief description of PON topologies

PON is a point-multipoint technology. All of the transmissions on a PON network are carried between the Optical Line Terminal (OLT), located at the optical node or central office, and the user’s Optical Network Unit(ONU). Normally, the OLT is interconnected with a transport network that collects the throughput from several OLTs and routes them to the network head end. The ONU unit is located on the user premises, configuring a FTTH (Fiber-To-The-Home) layout.

There are several suitable topologies for network access, including ring topologies (not common), tree, tree-branch, and linear optical bus. Each of the splits is made by chaining 1x2 or 1xN optical splitters together.

In some cases, depending on how critical the deployment is, network access may require protection. 

All PON topologies use single-fiber for deployment. On the downstream channel, a PON is a point-multipoint network. The OLT unit handles all of the bandwidth that is distributed to users in time intervals. On the upstream channel, the PON is a point-to-point network where multiple ONUs transmit to a single OLT. Working over single fiber, the way to optimize upstream and downstream transmissions without mixing them consists of working on different wavelengths using WDM (Wavelength Division Multiplexing) techniques. Most implementations superimpose two wavelengths, one for downstream transmission (1290nm) and the other for transmission to the head end (1310nm) upstream. The evolution of optical technology has made it possible to miniaturize the optical filters needed for this separation until they are integrated into the optical transceivers of user equipment. These optical carriers are used in a second window (instead of working in a third window) to minimize costs of optical-electrical equipment.

At the same time, PON architectures use time-based multiplexing techniques (TDMA) so that at different instants in time, determined by the OLT head-end controller, the ONUs can send their frames on the upstream channel. In the same way, the OLT head-end unit has to use a TDMA technique to send the information on the downstream channel in different time slots that will be selectively received by the user equipment (ONUs).

PON architectures have also had to resolve another important aspect: the transmission power of the OLT unit depends on how far away the ONU is located, as mentioned before, which may vary up to a maximum of 20 Km. Obviously, an ONU that is very close to the OLT will require lower burst power to avoid saturating its photodiode. Equipment that is very far away will need the burst to be transmitted at a higher power. This function has also been introduced recently in PON optical transceivers, which have significantly simplified the electronics that were previously required to act on gain control outside the transceiver. The new miniaturized optics integrate and simplify working with bursts with different power levels. 

 

APON, BPON and GPON

In 1995, seven telecommunications operators, seeing the possibilities of PON networks, founded the Full Service Access Network (FSAN), with the goal of unifying specifications for broadband access to homes. The FSAN also includes more than 30 equipment manufacturers.

The members of the FSAN developed a specification for a fully passive optical network that connected to a defined number of users from an optical node using ATM technology and its level-2 protocol.

Transmission on the downstream channel uses bursts of standard 53-byte ATM cells, adding a three-byte identifier that identifies the ONU that generated the burst. The maximum rate supported on the upstream channel, assuming a single ONU, is 155Mbps. This bandwidth is distributed according to the number of users assigned to the optical node (number of ONUs).

On the upstream channel, the frame is constructed using 54 ATM cells with two PLOAM cells inserted with the information of the recipient of each cell and the network maintenance and operation.

Although the system functions internally in ATM mode, which is more efficient than using Ethernet protocols, to the exterior, on both the user and “Central Office” sides, in addition to the native ATM interface, it also has TDM interfaces (i.e. 2 Mbit/s.) or Ethernet through emulation of both types of signals. The different manufacturers also normally have both user terminals (ONT/ONU) as well as the network core side (OLT) with different user interfaces adapted to conventional telephony or any data, video, or telemetry application.

APON may possibly provide the richest and most exhaustive set of operating and maintenance (OAM) characteristics  of all the PON technologies.

One of the drawbacks is that APON OLT head end equipment interconnects with transport networks at the SDH/ATM level and requires this type of transport infrastructure. Also, APON equipment bandwidth is limited to 155Mbps distributed among the users that make up an optical node. This limit was then increased to 622Mbps.

The APON term initially coined by the FSAN was replaced by BPON (Broadband PON – Broadband Passive Optical Networks) making reference to the possibility of supporting other broadband standards, including Ethernet, video distribution, VPL (virtual private lines), etc.

In 1997, FSAN sent the specifications to the ITU committee. After seven years, the ITU-T approved the following recommendations related to broadband passive optical networks. G.983.1 (general description), G983.2 (management and maintenance layer), G983.3 (quality of service in BPON), G983.4 (Dynamic bandwidth assignment), G983.5 (Protection mechanisms), G983.6 (OTN network control layer), G983.7 (Dynamic bandwidth network management layer), G983.8 (Support for IP, Video, VALN, and VC protocols).

The original recommendation specified in recommendation G.983.1 in the BPON architecture defines a symmetrical network with total bandwidth of 155Mbps, on both the upstream and downstream channels. This specification was modified in 2001 to allow asymmetrical configurations (622 downstream and 155 upstream) and increased capacity of symmetrical configurations (622Mbps).

BPON is not the FSAN’s latest contribution to passive optical networks. The increase in the bandwidth demanded by users, along with the balancing of the type of traffic exclusively to IP traffic, had a direct affect on the development of a new specification that was based on the BPON standard, which is very inefficient for IP traffic transport, which is improved by using an encapsulation procedure called GFP (General Framing Procedure, which increased architecture efficiency, allowing the mixture of variable size ATM frames.

This new recommendation standardized by the ITU-T and called Gigabit-capable PON (GPON) was approved in 2003-2004 by the ITU-T in recommendations G.984.1, G984.2, and G.984.3.

• Recommendation G.984.1 describes the general characteristics of a PON system that can transmit on ATM: its architecture, bit rates, range, signal transfer lag, protection, independent security and protection speeds.
• Recommendation G.984.2 describes a flexible fiber-optic access network capable of supporting the bandwidth requirements of services to companies and residential users.
• The GPON techniques make it possible to maintain the optical distribution network, the wavelength plan, and the design principles for the integrated service network specified in the G.983 Recommendations. Also, apart from increasing network capacity, the new standards allow more efficient IP and Ethernet handling.

GPON is a very powerful standard, but implementing what it has to offer is also very complex:

• Global multi-service support: including (TDM, SONET, SDH), Ethernet 10/100 Base T, ATM, Frame Relay, and many others
  Physical range up to 60km
• Support for several transfer rates, including symmetrical 622Mbps traffic, symmetrical 1.25Gbps traffic, and asymmetrical traffic with 2.5Gbps downstream and 1.25Gbps upstream.
• Important management, operation, and maintenance functions from the OLT head end to the ONU user equipment.
• Security at the protocol level (encryption) as a result of the protocol’s multicast nature.

The network organization and the terminology used is the same as for BPON networks. However, the GPON standard is expected to increase interoperability between different manufacturers even more, allowing the same system to use ONUs and OLTs from different manufacturers.

 

Ethernet PON, EPON

In January 2001, the IEEE (Institute of Electrical and Electronics Engineers) formed the Ethernet in the First Mile (EFM) task force. The purpose of this group was to extrapolate Ethernet technology to residential and business areas, carrying it to the home, taking advantage of the growth that this technology had been experiencing in recent years, because of its simplicity, performance, and ease of deployment.

This task force generated a new passive optical network specification called Ethernet PON (EPON). This new architecture is different from the previous ones in that it carries native Ethernet traffic directly instead of carrying ATM cells. It uses the 8b/10b (line encoding) standard, and whenever possible, remains true to Recommendation 802.3, including full-duplex use of access to the medium.

One of the possible advantages of this technology is its obvious optimization for IP traffic in comparison with the classic inefficiency of the ATM-based alternatives. Also, the interconnection of EPON islands is much simpler than the interconnection of APON/BPON, and GPON because no SDH architecture is needed for WAN transport. 

The entire EPON network architecture works at Gigabit Ethernet speed. The maximum bandwidth offer to users therefore depends on the number of ONUs connected to each OLT. If an optical node services 10 users, the maximum service capacity per user would b 1Gbps/10 = 100Mbps. Obviously, with 100 users per optical node, the bandwidth per user would drop to 10Mbps.

However, there are optical techniques – which can be extended to all PON architectures – such as the use of multiple optical carriers with different colors, WDM, to increase the bandwidth per optical node without modifying infrastructure.

In a medium-sized architecture, several head-end controllers coexist, depending on the maximum bandwidth that has to be guaranteed for users. A recommended value for this type of network could be 10 subscribers per optical node, but values of 64, 100, and 256 are also possible. Ranges of up to 20 kilometers can be obtained over fiber from the network head end to the subscriber.

The variations of the available interfaces for user equipment (ONU or VoIP gateway) include 10/100 ports (aimed at the home market), or Gigabit Ethernet ports (aimed at the business market), which require bandwidth granularity greater than 100Mbps)

EPON allows quality of service to be assigned on the downstream channel and on the upstream channel at the same time that it encodes all communications with the DES algorithm.

The use of EPON allows transport operators to eliminate complex and expensive ATM and SDH elements, simplifying networks and lowering implementation costs for subscribers. Currently, EPON costs per user units are approximately 10% lower than the equivalent GPON equipment.

Finally, IEEE has announced a new revision of the standard, which will use 10GbE technology to multiply bandwidth of first-generation EPON architecture by a factor of 10. This development effort will be covered by the future GEPON specification, a new IEEE standard that will move towards convergence with the ITU’s GPON standard.

 

 

Summary of the different standards

The table below summarizes the main characteristics of the dominant standards.

 

VPON

Thanks to a new type of optical transceiver, a video signal can be superimposed on traditional data throughput of A/B/GPON and EPON passive optical networks. This signal, transmitted at 1550nm and frequency-modulated from an ultra-linear CATV-type laser located at the network head end, can transport the UHF and VHF spectrum to all of the ONUs in the PON architecture. Using simple circuitry, this signal is extracted on user equipment by the optical transceiver, amplified by a bandwidth amplifier for the V/UHF range, and can be directly introduced into the antenna connector of analog televisions or digital decoders.

This system of operation, called Video RF, Video PON, or VPON, doesn’t use the bandwidth of the data signal to encapsulate video signals, but rather uses a much simpler system that can be implemented using a traditional CATV analog head end, which reduces the costs of digital IP encoders at the head end (and user decoders) for carrying MPEG2 frames.

 

Comparison between EPON and GPON

There are clearly significant differences between EPON and GPON technology, especially in layer 2. However, network architecture designers will also find differences in terms of bandwidth, range, efficiency, cost per user, and management. This section will cover these differences in greater detail:

1. Usable bandwidth

Bandwidth will vary between the two protocols. GPON promises 1.25Gbps or 2.5Gbps on the downstream channel and scalable bandwidth from 155Mbps to 2.5Gbps. EPON, on its part, offers a symmetrical bandwidth of 1Gbps which automatically wastes 250Mbps in 8b/10b encoding (until the 1.25Gbps lines speed is completed).

GPON doesn’t decrease bandwidth for encoding, since it uses an NRZ system and interlacing of typical SDH-network data. This gives GPON 25% more bandwidth on the upstream channel than EPON.

However, when aggregating traffic from several head-end controllers, GPON’s apparent bandwidth advantage is lost when the conversion to Gigabit Ethernet throughputs that the head-end switches require is done. In other words, in general terms, GPON adds bandwidth that will not be taken advantage of by operators when the GPON signal is transported on WAN Gigabit Ethernet networks.

2. Scope

As with any other protocol, the range over fiber is defined by the optical link’s dynamic range. Currently, the range of both protocols is defined at approximately 20Km, and it is limited by the number of ONUs defined for the node.

GPON offers support to up to 128 ONUs. With EPON, there is no limit on the number of nodes, although 256 is a maximum suitable value. In these conditions with the maximum node equipment, the maximum EPON range is obviously less than in GPON since there is more insertion loss resulting from the use of a greater number of optical splitters.

3. Cost per subscriber

The use of EPON completely eliminates the costly and complex ATM/SDH transport equipment of transport operators, simplifying their networks, and therefore their user costs. It has been estimated that EPON generates 10% less head-end equipment costs per user than GPON, and is on the same level as other access technologies like VDSL.

4. Efficiency of each standard

Both PON protocols add overhead to the frames of the protocol that they encapsulate (IP). EPON is an optimized standard for variable packet length (Ethernet frames up to 1518 bytes) according to the 802.3 Ethernet standard. In ATM PON systems (including GPON), data is transmitted in 53-byte fixed frames (cells) (48 bytes of payload and 5 bytes overhead). This format is extremely inefficient for carrying IP traffic, whose segments can vary in size up to 64KB.

GPON systems that carry IP traffic have to segment it into 48-byte sizes, adding the segmentation information in the 5-byte headers. This process, in addition to being complicated, adds lag.

It has been calculated that Ethernet encapsulation like what is done by EPON with IP traffic adds 7.42% inefficiency, while IP encapsulation over ATM increases this value to 13.22%.

Also, the 8B/10B encoding done by EPON and that wastes bandwidth becomes an advantage when carrying out electrical-optical conversion, since it requires much simpler and less precise synchronization electronics than GPON. 

5. Management systems

EPON bases its experience on Ethernet management systems over SNMP, which are much simpler than the ATM layer-2 management and maintenance models. This means that EPON management system can normally be integrated with existing operator solutions, such as HPOpenView or similar solutions.

7. Encryption

GPON uses the encryption defined in the ITU standard. However, GPON only limits encryption to the downstream channel.

EPON uses DES mechanisms for upstream and downstream channels. 

8.  Network protection

Both protocols have specific network protection mechanisms for each implementation by the manufacturer. These mechanisms include protection of the network frame and the transport operator interconnection frame.