Wavelength Division Multiplexing Passive Optical Networks Communications Essay

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

Passive optical networks (PONs) were originally developed in the 1980s as a cost effective method of sharing fibre infrastructure for narrowband telephony to business premises. Since those early days the application of PONs has moved on to interactive broadband networks implemented as either BPON (Broadband PON) or EPON (Ethernet PON) and now GPON (Gigabit PON).(Borghesani 2007)

All previous PON systems are based on the same idea of time sharing the optical medium by TDMA (time division multiplexed access). However, it has long been realized that using wavelength division multiplexing (WDM) offers an alternative method of sharing the capacity between multiple users and would offer advantages in terms of capacity, low latency and service transparency. Consequently, Wavelength Division Multiplexing Passive Optical Networks (WDM-PONs) have received a lot of attention as one of the best contenders for next generation optical access networks.(Borghesani 2007)

WDM-PON provides its users with protocol transparency, high flexibility in service convergence, sharable link infrastructure and large scalability in user bandwidth (W.Kim 2008) by employing WDM to support multiple wavelengths in both upstream and downstream directions. Each optical network unit (ONU) uses a unique wavelength in each direction to communicate with the optical line terminal (OLT). (Kani 2009) As a result, WDM-PONs are considered the next step from today's TDMA-based PONs to accommodate further traffic growth and facilitate new applications. Moreover, they are more adaptable to future technologies than conventional PONs due to their high transmission capacity and optical transparency.(Schussmann and Schirl n.d.)

The proposal of WDM-PON is more than two decades old. (Banerjee, et al. 2005) However, in spite of the superior features as compared to TDMA-PON, WDM-PON has not attracted yet much attention to practical application to commercial networks.(W.Kim 2008) In the past, the reason for its non-utilization included the lack of technology, low bandwidth requirements, high system cost and stability problem of optical system.  (and its high cost, as compared to TDMA-PON). However, these problems have been mitigated as technology has significantly improved and the demand for higher bandwidth is ever-increasing with every passing day. Even though, the equipment cost per user for WDM-PON is much more, as much as three times, than that of TDMA-PON, the effective cost per user is much lower than that of TDMA-PON.(W.Kim 2008). Moreover, WDM-PON has low power splitting losses and provides its users with scalability, as the same fiber infrastructure supports multiple wavelengths.(Banerjee, et al. 2005) Some of the other advantages of WDM-PON include;

  • Transparency to protocol and data rate
  • QoS guarantee via point-to-point connection
  • High security
  • Easy expansion for future network services
  • Easy upgrade for future traffic demand (Kim, Hwang and Yoo 2007)

WDM-PON Architecture.

Like TDMA-PONs, WDM-PONs also consist three basic parts; optical line terminal (OLT), remote node (RN) and multiple optical network units (ONUs). OLT and ONU have a set of transceivers to communicate with multiple wavelengths, and RN has equipments for splitting and combining wavelengths. Since each ONU equips with multiple transceivers to communicate with OLT, it is logically point-to-point network topology.(Banerjee, et al. 2005) For WDM-PONs, the remote node is also known as wavelength router (WR). The WR is a passive component and is also called arrayed waveguide grating (AWG) or optical phased array (PHASAR).(Schussmann and Schirl n.d.)

A simple WDM-PON employs a separate wavelength channel for each of the downstream (CO to end users) and upstream (end users to CO) directions. The usefulness of this approach is that point to point link is established between the central office (CO) and the user. For downstream traffic, the wavelength channels are routed from the OLT to the ONUs by means of an AWG router. Different wavelengths windows are employed for the upstream and downstream transmissions, which are separated using coarse WDM (CWDM). Dense WDM (DWDM) is then employed to separate wavelengths at intra-window level. For demultiplexing the received upstream signals, a demultiplexer is employed at the CO. (Banerjee, et al. 2005)

The advantage of this scheme is that each user can operate at a data rate up to the full bit rate of a wavelength channel, and different PON sub networks may utilize different set of wavelengths over same infrastructure. The WDM-PON architecture needs to be scalable in bandwidth as well as the number of users. For these requirements, economic optical devices should be utilized. (Banerjee, et al. 2005)

There are two options for implementing WDM namely CWDM and DWDM. A CWDM implementation employs wavelengths which are more than 20nm apart. A total of 18 CWDM channels are available if complete wavelength range i.e. 1271 nm to 1611 nm is utilized with 20 nm spacing. Furthermore, the wavelength multiplexer with low channel crosstalk can be implemented easily for CWDM. A CWDM implemented PON is cheaper as compared to a DWDM PON because a thermoelectric cooler TEC is not required. Cost decrease of up to 40% has been achieved for the CWDM-PON systems. (Banerjee, et al. 2005)However, when normal single-mode fiber with water-peak attenuation range is used, then, CWDM lacks scalability and the available channels are reduced. Another disadvantage is the limiting of the transmitting distance or splitting ratio due to higher losses at shorter wavelength channels. (Banerjee, et al. 2005)

The other implementation, namely DWDM, employs a significantly smaller wavelength spacing of less than 3.2 nm.DWDM has been developed to transmit many wavelengths in a limited spectrum region where an erbium-doped fiber amplifier (EDFA) can be used.DWDM PON is regarded as the ultimate PON system which can provide significant bandwidth to many users. Crosstalk between adjacent channels is controlled by monitoring carefully the center wavelength of WDM filter. Wavelength tuned devices and temperature control makes implementation of DWDM costly than CWDM. (Banerjee, et al. 2005)

An optical source emits a fixed wavelength from each component. Required wavelength is achieved using a wavelength monitoring circuit and controller for every component. A receiver part usually consists of a photodetector and electronic components like amplifier preamplifier clock and data recovery circuits. As every wavelength is working independently in WDM-PON, each receiver has different configurations. (Banerjee, et al. 2005)

PON Topologies.Sir plz reduce this section around 450 words (now it is of 600 words length)

There are several topologies suitable for the access network: tree, ring or bus.

Tree Topology.

It is the most commonly used in access networks and uses a single fibre from the OLT to an intermediate splitting point. From this splitting point, there is a fibre for each ONU connected to the network. In principle, the tree topology consists of cascaded splitting points and topologies with a single splitting point are in general termed as star topology. However, due to the special relation between OLT and ONU there is directivity and therefore this topology, if applied to access networks is commonly termed as tree topology. (Prat 2008)

The main advantage of this topology is that the splitting is concentrated on a single point thus it is simple to detect a network problem. Another advantage is that all ONUs has the same power budget which means that they all will receive roughly the same optical signal quality. (Prat 2008)

Among several PON topologies, the tree topology has been widely accepted as the standard for deployment and possible performance evaluation purposes. (Souza, Dini and Lorenz 2004)Figure 2 shows a tree based PON consisting of Optical Line Terminal (OLT), splitters, Optical Network Units (ONUs) and users. All transfers occur via the OLT and no direct traffic exists between the ONUs. The preference for the tree topology is due to its flexibility in adapting to a growing subscriber base and increasing bandwidth demands. (Freire, et al. n.d.)

Bus Topology.

The bus topology uses also a single fibre front the OLT, thus the same worst case failure and capacity/utilization issues arise as for the tree topology. Each final subscriber is connected to it by means of a tap coupler that extracts a small part of the power that is being transmitted front the OLT. The two advantages of this topology are that it is the one that uses minimal amount of optical fibre (if ONUs are directly connected to the tap coupler) and allows flexible deployments as a new ONUs can he connected to the network very easily by adding one more tap. The main problems are: on the one hand, that the signal is degraded when passing through each tap coupler, and therefore the ONUs located far from the OLT are receiving weak and degraded signals; on the other hand, that the required total fibre length is high for covering a two-dimensional area. (Prat 2008)

Ring Topology.

The ring topology is mainly used in metropolitan networks because it offers resilience capability with a minimal number links. As there are two possible ways to reach the OLT, it is still possible to establish and maintain a data link in case of a fibre cut. However, it requires two fibres to he used at the OLT and more complicated equipment at each ONU with switching capabilities to be able to send and receive the signals being transmitted from the two directions of the fibre ring. It also shows the same problem as the bus topology in terms of power budget. When the optical signal passes through each ONU, the signal is degraded and attenuated. This factor is the most restrictive one in terms of transmission capabilities and restricts the number of ONUs that can he connected to the ring. Capacity is also shared among all ONUs it resilience is used, thus the second fibre from the OLT does not increase the network capacity, i.e. the total number of customers is limited to the same number as for the tree and bus topology.(Prat 2008)

Linear Add Drop (LAD) Architecture.

In the simple WDM-PON architecture, each ONU has fixed number of wavelengths for both upstream and downstream traffic. It has advantage of simplicity in hardware and in resource management. However, when the system has been deployed, it is hard to reconfigure the system for resource allocation. Thus, it suffers from under or overutilization problem when the traffic demand is unbalanced. (Kim, Hwang and Yoo 2007)

Linear Add Drop (LAD) provides a solution by offering a higher capacity PON architecture which is not only scalable and flexible but also provides bidirectional and symmetric WDM data services. LAD is basically a WDM-based, low cost, high data rate tree-like architecture, which provides easily accessible triple play services (TSP) at the user end at a much lower cost. (Khan, Chang and Yu)

PON is the technology of the future and in the near future, it will be serving a large number of users simultaneously. Obviously all these users can't be served from a single split point. Therefore, the LAD architecture proposes multiple split points in its architecture, thus, leading to a multistage design. The multistage nature of this scalable architecture allows serving a large number of users in a much efficient manner. Moreover, it can be deployed in any terrain or any other demographic requirements. (Khan, Chang, et al. 2005)

LAD WDM-PON is novel and unique in the sense that it incorporates both WDM and Power Split (PS) techniques for traffic in both upstream and downstream directions to boost up the network capacity. LAD provides a flexible architecture as shifting a wavelength from one stage to some other requires no specific modifications. The architecture allows adding a new stage anywhere easily if wavelengths are left or power budget allows for it. LAD also alleviates the bottlenecks due to its symmetric and broadband nature. (Khan, Chang and Yu)

Initially pure WDM and pure PS were suggested but the use of both in a single fiber was not researched to exploit the capacity of fibre. Different architectures for WDM PON used AWG router which had drawbacks like cost, availability, coarseness, lack of flexibility, and scalability.

The complete LAD WDM-PON architecture revolves around the notion of stages, trunk, split points and other fibers. The trunk functions as the backbone of the architecture and serves an area known as the fiber serving area (FSA). FSA is further divided into multiple smaller areas, called stages, which operate on a group of specific upstream and downstream wavelengths. Stage Split Point (SSP) functions as the multiplexing point for upstream traffic and the demultiplexing point for downstream traffic as far as a stage is concerned. Wavelength split point (WSP), on the other hand, uses a power splitter to split the wavelengths, thus, allowing increased capacity by enabling multiple users to share the same set of wavelengths. The backbone fibre connects the OLT with the stage fibre at the Trunk Split Point (TSP). Similarly the Feeder Fibre and the Drop fibre serve to connect the SSP with the WSP and WSP with the ONU respectively. [no re-wording reqd, just check 4 grammar & continuity]

Three types of wavelengths are associated with the network namely DSD (downstream data), USD (upstream data) and DSV (downstream video). Each of them can be either a multiplexed group of wavelengths or a single wavelength. Each ONU is provided with a separate DSD and USD wavelength; however, multiple ONUs can use these wavelengths to share large capacity of optical fibre. DSV on the other hand provides broadcast services and, therefore, must be received by all users. [no re-wording reqd, just check for grammar & continuity]

The LAD architecture is designed in a multi-stage manner with every stage serving an area falling in about 1 km radius. In these networks, a new stage is formed by putting a TSP, which is basically a broadband coupler, on the trunk fibre. SSP is a set of passive complex devices, which are used to demultiplex DSD wavelengths, to multiplex the USD wavelengths, and to ensure the reception of DSV to each ONU. (Khan, Chang and Yu) If DSD and USD is shared by several users, then WSP is used, which is a simple broadband power coupler.

Stage Split Point (SSP).

The SSP design is very important part of the network as it acts as the main distribution and collection point. It serves as a remote node and also ensures the broadcast portion of the design. The physical layout of the SSP depends on the number of wavelengths and the wavelength bands used for DSD, USD, and DSV.

In the following we explain two general situations; other situations for specific requirements can be derived from these.

Case-1: According to G.984 recommendations, 1310 nm and 1510 nm ranges are used for upstream and downstream traffic respectively. However, AWG multiplexers (MUXs) are not available for these frequency ranges. In order to solve the problem while multiplexing USD and DSD wavelengths, Coarse AWG (CAWG) along with Power Splitters are employed. The CAWG typically operate in the 1310 nm range and provides support for six channels which are 20 nm apart. The multiplexing losses incurred are around 3dB. These systems can also be morphed to provide 24 channels (with 10dB loss) or 48 channels (with 13 dB loss). C and L bands and the 1.3-1.5 μm band are utilized at different places of the SSP to combine or separate these three different types of wavelengths.

Case-2: Recommended G.984 wavelength assignments do not meet the specifications of DWDM, therefore, a new simpler design is proposed. Here, the DSD and USD wavelengths both lie in the C band and the wavelengths are only separated by 0.2 nm. In this case, only a C and L band WDM coupler is used because only two bands are used.

Optical Line Terminator (OLT).

OLT is the high power device, located at the CO, which interfaces with the metropolitan network and serves to adapt the incoming traffic from the metropolitan rings into the PON transport layer. In case of WDM-PON, the OLT consists of multiple devices:

  • A transmitter and an amplifier for DSV transmission;
  • Several transmitters, a post amplifier and a multiplexing device for DSD transmission;
  • Several receivers, a preamplifier and a demultiplexing device for USD transmission;
  • Wavelengths combining and separating devices e.g. WDM couplers or circulators.

Case-1: The setup employs an amplifier before the USD demultiplexing device, which serves as a preamplifier for USD receivers, compensates for the multiplexing loss at the SSP and also boosts up the signal so that it is easily detectable at the receivers. As each port carries multiple USD wavelengths, therefore, tunable filters are used to filter out the specific wavelength on the port.

Case-2: The setup employs simple C-band AWG MUXs as the USD demultiplexer and DSD multiplexer. Depending on their position, the amplifiers used can either serve as preamplifiers or post-amplifiers.

Optical Network Unit (ONU).

ONU is the device, located at the customer premise, and serves as an interface between the customer equipment and the PON.  The ONU has two separate receivers, one each for DSD and DSV traffic and an optical transmitter for USD transmission. The simple design ensures low cost of Set Top Box (STB). WDM couplers combine or separate the three types of signals depending on their wavelength.

Salient Features of the Multistage Linear Add-Drop Design.

The most challenging characteristics desired for practical deployment of an access network include the technical requirements, flexibility, scalability, speed, capacity, coverage area and services. Multistage LAD PON accommodates all these functionalities in a much effective manner.


The flexibility of an access network can be defined as the ease of shifting a connection orservice from one user or geographic area to another user or geographic area. (Khan, Chang and Yu) Shifting a connection in LAD PON means the shifting of wavelength form one stage to another. Let us suppose that there are three wavelengths in stage 1 and two wavelengths in stage 2. For shifting wavelength, we just need to disconnect at the SSP in stage 1 and connect it the appropriate SSP in stage 2. This is achieved by hooking them up at the appropriate port of the AWG.

This shift is completely transparent, except at the SSP, and OLT or any other point is not affected by this change. The flexibility in the network is achieved by the use of wavelength nonselective devices at the TSP, such as use of a power splitter instead of wavelength selective devices.

A multistage LAD WDM PON is also flexible in the sense that the number of USD and DSD wavelength pairs do not have to be the same in all the stages, i.e., each stage can have different data wavelength pairs.

Scalability. Lot of Attention Reqd

Scalability is an important feature of an access network so that it can keep pace with the growing demand. Scalability of the WDM PON can be measured by its capability to add more wavelengths, more users, and more stages. WDM-PON uses the same fiber infrastructure to support multiple wavelengths, hence, providing its users with scalability and low power splitting losses. (Banerjee, et al. 2005)

The LAD WDM-PON architecture is highly flexible in the sense that adding new wavelengths is possible by simply putting more lasers on the OLT if there are some ports vacant at the AWG MUX or changing the size of the AWG MUX; for example, the AWG MUX can be changed from 1:32 to 1:40 or 1:64 or similarly, a 100 GHz AWG MUX can be replaced by a 50 GHz AWG MUX. Similarly, for added performance an interleaver in combination with an AWG MUX can be introduced. Adding more users to share the same wavelength is a straightforward process. It can be done by changing the split ratio at the WSP; for example, if eight users shared a wavelength initially, an increase in the split ratio to 1:16 or 1:32 thus allowing as many as 16 or 32 users/wavelength, respectively.

To add a stage, extra wavelengths available at the OLT can be utilized to add an appropriate TSP and generate the stage at the desired location. However, if the network is already using all the available wavelengths, then, we have to free some wavelengths at the existing stages by shifting their users to other wavelengths and shift the freed wavelengths to the new stage. The disconnected users are reconnected to some other wavelength by using a 1:2 power coupler at the SSP. This decreases the data rate of these users but this has to be done because it is never possible to go beyond the maximum network capacity. A new stage is created by adding a TSP at the appropriate place, connecting it to the appropriate stage of the SSP through fiber, and making the suitable connections at the new SSP for new users of the new stage.

High Speed and High Capacity.

High capacity plays an important role in the reducing the cost of services provided to the users as it remarkably decreases the effective cost per user. The effective cost includes the installation, equipment, and maintenance cost. A major driving force for switching to PONs is the high speed it provides to its users. Typical data rates of the order of 100Mbps for residential and 1 Gbps for corporate clients are common nowadays. (Khan, Chang and Yu) The LAD WDM PON architecture has a high capacity as it uses both WDM and PS technology to maximize the capacity of the network. It is also high speed and capable of operating at symmetric rates of 2.5 and 10 Gbps. (Khan, Chang and Yu)

The minimum guaranteed traffic, which is the available data rate even if all the users are in demand for traffic, provided by LAD WDM-PON architecture is promising. To provide a user with 200 Mbps of data traffic (100 Mbps in either direction) with the wavelength operating at 2.5 Gbps, the maximum number of users can be served by one wavelength by use of a 1:32 split ratio at the WSP and connecting only 25 users to the available 32 ports so that 200 Mbps per user can be guaranteed at a wavelength operating at 2.5 Gbps. Seldom is the case when all the users are utilizing their minimum guaranteed traffic. In practice, a dynamic allocation bandwidth scheme can further push this data rate to several folds.

Coverage Area.

PONs are usually recommended to operate at the range of 20 km from the OLT. For a high capacity PON it is not feasible to have one split point or remote node from which all the wavelengths are distributed to the ONUs. Therefore, multistage is a practical approach to cater to large coverage areas and to reduce the cost of fiber installation by maximization of the multiplexed signal fiber length.

Services. Lot of Attention Reqd

A PON must at least provide TPS to force a shift from HFC and DSL networks. The LAD architecture can provide these services as it is completely capable of broadcasting a complete 20 nm band that can have video and other future broadcast services. Previously data bandwidth has been proposed to transmit the video signals. This is not a good solution because usually a limited number of channels are being watched by most of the users and transmitting identical signals more than once over the same and different wavelengths results in bandwidth wastage. Second, use of data bandwidth for video restricts the number of channels simultaneously delivered to the user thus restricting the number of television (TV) sets that can be simultaneously used at the user premises. Table 1 gives a brief summary of the TV channels supported by our system in the simplest configuration. Third, use of a data stream for video (TV) channels reduces the practical data bandwidth to approximately 10 Mbps, which is not a drastic improvement from existing DSL or HFC data rates. With regard to voice, Voice over Internet Protocol (VoIP) or time-division multiplexing (TDM) channels in a data bandwidth are recommended since the minimum data rate per user in our case is 75 Mbps each way. If we dedicate only 2 Mbps out of that, we can easily achieve 30 uncompressed TDM voice channels (64 kbps each) or 240 VoIP channels with a compression ratio of 1:8.

Table 1 - Number of Video (TV) Channels for Multistage LAD WDM-PON

2.5 Gbps


Wavelengths Used



Total Available Rate

2.5 Gbps


HDTV (20 Mbps) Channels



SDTV (6 Mbps) Channels



Analysis of the Multistage Linear Add-Drop Design.

The analysis of LAD WDM-PON Architecture is done by means of eye diagrams. The eye diagram of different DSD and USD channels are almost identical so only a single eye diagram and BER for each DSD and USD operating at 1557.8nm and 1540.8nm respectively has been shown. (Khan, Chang and Yu) It is worth mentioning that by observing the eye diagrams of all the types of channels, one can see that there is not significant degradation in the signal even after the transmission. As observed in BER curves, the power penalty due to transmission is approximately 1 dB for DSD, USD, and DSV wavelengths.

The proposed LAD WDM-PON architecture has the following advantages compared with the GPON architecture:

  • Higher Capacity. The LAD WDM -ON has a huge capacity in terms of number of users it can serve compared with the GPON, because the GPON uses WDM to separate USD, DSD and DSV bands only, whereas the LAD WDM-PON uses CWDM for that purpose and further uses DWDM to increase the number of channels of each type. In this way, a LAD WDM-PON that uses 40 wavelengths can have as many as 40 times more users than a GPON that operates at the same data rates.
  • Higher Data Rates. The LAD WDM-PON can provide higher rates to its users because it operates at symmetric 2.5 Gbps as well as at10 Gbps, whereas the GPON is recommended for the maximum data rate of symmetric 2.5 Gbps.

The LAD WDM-PON has the ability to support larger and more scalable network throughput. The LAD WDM-PON that operates with 40 wavelengths can deliver as much as 100 Gbps or 400 Gbps each way when operating at 2.5 Gbps or 10 Gbps per wavelength, respectively. In addition, the LAD WDM-PON can also broadcast 2.5 or 10 Gbps of video per wavelength in a 20 nm wavelength band. For comparison, a 10 Gbit Ethernet provides only 10 Gbps of bandwidth that is shared by all users' data and video traffic.


Passive Star architecture is designed to alleviate some of the flexibility challenges of a traditional PON topology. Instead of pushing the splitters all the way out to the customer premises location, they are pulled back and aggregated in a more centralized location, typically housed in a cabinet. This design helps drive more efficiency and lightens the burden of troubleshooting since the splitters are now more centralized. But Passive Star is still subject to the inherent drawbacks of a PON network. One of these is the lack of diverse paths through the network. PONs, by their nature, subscribe to tree-based topologies. Even if the splitters are pulled back to an aggregation cabinet, there is still only one physical path upstream, and that introduces a dangerous dependency on that link. Because these splitters have no intelligence, there is no ability to provide emergency fail over to a diverse path in the event of a link failure. Another drawback is high first subscriber costs. As mentioned earlier, each PON port carries a high price because it is expected to be divided by 32 subscribers. Therefore, to activate that first subscriber, a significant CapEx investment must be made to provide them service. (Allied Telesyn 2004)

Sir, I want a comparison between the LAD and Tree Architecture, which I am leaving for you. If you want comparison between any other architecture, you are welcome to do so. But let me mail the sources and quote their sources in the brackets. Sir also see the Tree1 & Tree2 for additional information

We have proposed and demonstrated a high capacity, high speed, fully symmetric, flexible, and scalable WDM PON based on a multistage LAD test bed architecture for broadband access networks that can serve as many as 1280 users over a 20 km area with a single fiber for upstream as well as downstream communication. We operated the test bed at both 2.5 and 10 Gbps to show the ability of the architecture to serve a diverse customer base as well as provide three key services, i.e., video, voice, and data (TPS). The prototype test bed and proposed multistage LAD WDMPON offers many features that are desirable for a scalable access network. With the recent technological advancements in laser sources, amplifiers, and DWDM devices, this architecture design promises future PON customers the most in terms of high speed and high capacity along with TPS. This WDM PON architecture test bed demonstrates a practical design and approach in terms of ease of deployment, maintenance, management, scalability, and flexibility for broadband service providers.


  • Allied Telesyn, Inc. Active vs. PON - FTTx Technology Choices. Technical Brief, www.alliedtelesyn.com, 2004.
  • Banerjee, Amitabha, et al. "Wavelength-division-multiplexed passive optical network (WDM-PON) technologies for broadband access: a review." Journal of Optical Networking (Optical Society of America) IV, no. 11 (November 2005): 737-758.
  • Borghesani, Anna. "Optoelectronic Components for WDM-PON." 9th International Conference on Transparent Optical Networks. Rome, 2007. 305-308.
  • Infocellar. PON - Passive Optical Network. http://www.infocellar.com/networks/new-tech/PON/PON-real.htm (accessed May 28, 2010).
  • Kani, Jun-ichi. "Enabling Technologies for Future Scalable and Flexible WDM-PON and WDM/TDM-PON Systems." IEEE Journal of Selected Topics in Quantum Electronics, 2009.
  • Khan, Muhammad Faisal, Gee Kung Chang, and Jianjun Yu. "Design, implementation, and demonstration of a scalable three-stage, linear add-drop wavelength-division multiplexing passive optical network experimental test bed." Journal of Optical Networking (Optical Society of America) IV, no. 11 (November 2005): 696-713.
  • Khan, Muhammad Faisal, Gee Kung Chang, Jianjun Yu, Benny Bing, and Oladeji Akanbi. "Multistage Linear Add/Drop (LAD) Architecture for High Capacity WDM-PON Network Testbed." 2005 Conference on Lasers & Electro-Optics (CLEO). 2005. 1046-1048.
  • Kim, Bongtae, and Byoung Whi Kim. "WDM-PON Development and Deployment as a Present Optical Access Solution." OSA/OFC/NFOEC 2009. 2009.
  • Kim, Hyowon, Junho Hwang, and Myungsik Yoo. "A Cost-Efficient WDM-PON Architecture Supporting Dynamic Wavelength and Time Slot Allocation." ICACT2007. 2007. 1564-1568.
  • Schussmann, Jürgen, and Thomas Schirl. "Design and Cost Performance of WDM PONs for Multi-Wavelength Users."
  • W.Kim, B. "R&D and First Commercial Service ofWDM-PON." GSA / AGE 2008. 2008.
  • Yu, Jea-Hoon, Byoung-Whi Kim, and Nam Kim. "WDM/TDMA Hybrid-PON : Wx-PON system." ICACT 2009. 2009. 1707-1711.