Networks can be classified in three basic categories which include opaque, transparent, and semitransparent networks. Aracil & Callegati (2009) says that “opaque optical networks convert the signal from optical to electrical at each node port meaning that each link is optically isolated by transponders doing O/E/O conversions” (p. 24). The signal is normally switched electrically in each node and it is only maintained optically between adjacent nodes (Aracil & Callegati, 2009). In transparent optical networks they do not convert the optical signal to electrical in the intermediate nodes of the end to end path. Aracil & Callegati (2009) says that the absence of wavelength converters, a light-path must occupy the same wavelength on all the fiber links through which it traverses.
They further said that in both these types of networks the RWA algorithms must take into consideration that the transmission of the signal on the fiber has to follow certain restrictions as a result of technological limitations. Aracil & Callegati (2009) also indicated that wavelength continuity constraint can easily result in rejection of a connection although the required capacity is available on all links of the path but not on the same wavelength. Aracil & Callegati (2009) argued that “since the optical signal is subject to impairments, a limit exists in the maximum distance reachable by an optical signal on a transparent network” (p. 24).
Kitayama, Masetti-Placci & Prati (2005) further says that in wavelength-division multiplexing optical networks, the bandwidth request of a traffic stream can be much lower than the capacity of a lightpath. In an optical that is transparent wavelength routed network entails the establishment of point to point lightpaths between every edge node pair. Kitayama, Masetti-Placci & Prati (2005) also indicated that “these lightpaths may span multiple fiber links and thus virtual adjacency is created between the ingress and egress nodes via the established lightpath, even when the two nodes are located to far apart” (p. 238).
In all-optical wavelength routing approach presents two advantages in which the optical bypass eliminates the need for Optical-Electrical-Optical (OEO) conversion at intermediate nodes (Kitayama, Masetti-Placci & Prati, 2005). In this context the node cost decreases significantly because in such case the number of required expensive high speed electronics, laser transmitters, and receivers is reduced. Kitayama, Masetti-Placci & Prati (2005) further says that the second advantage is that because in all-optical routing which is transparent with regard to the bit rate and the format of the optical signal.
On the other hand wavelength routing presents two major disadvantages. Kitayama, Masetti-Placci & Prati (2005) says that the first drawback is that routing at a wavelength granularity puts a serious strain on the number of wavelengths required in a large network. The second disadvantage of transparent networks is that the rigid routing granularity entailed this type of approach. Kitayama, Masetti-Placci & Prati (2005) also indicated that this granularity is large and can lead to bandwidth waste especially when only a portion of wavelength capacity is used. In wavelength routed networks, this efficiency is probable only when there is adequate traffic among pair nodes to fill the entire capacity of wavelengths.
According to Kitayama, Masetti-Placci & Prati (2005) wavelength routing is performed in a similar way as in an all-optical network which implies that signals remain in the optical domain from end to end and are optically switched by intermediate OXVs (239). Because the lightpath remains transparent at intermediate nodes a MAC (Media Access Control) protocol is required to avoid collision between transient optical packets and local ones injected into the lightpath (Kitayama, Masetti-Placci & Prati, 2005). This type of mechanism guarantees collision free packet insertion on the transient wavelength at the add port of an intermediate node.
With the introduction of distributed aggregation there are multiple connections with fractional demands that can be multiplexed into shared lightpaths. Kitayama, Masetti-Placci & Prati (2005) also indicated that this helps to eliminate the wasted bandwidth problem found mostly in pure wavelength routed networks. Also Kitayama, Masetti-Placci & Prati (2005) says that “because of the sharing of wavelength channels, the number of admissible connections in the network will be increased” (p. 240). Connections from several nodes to the same destination are aggregated on the same lightpath, the destination node will receive less lightpaths. This implies that less physical components can be used hence resulting in the save of a great deal of equipment costs. This also means that it is possible to eliminate the scalability issues encountered in all optical wavelength routed networks (Kitayama, Masetti-Placci & Prati, 2005).
What are the limitations in transparent networks without wavelength conversion?
Bouillet & Ellinas (2007) says that the intricacy and volume of telecommunication networks and the speed of information exchange have increased at an unprecedented rate over the last few decades. This is because nowadays people are employing several devices that facilitate easier exchange of information. Bouillet & Ellinas (2007) says that based on the fact that the current trend in multimedia communication there is a high demand for flexible networks with extremely high capacities that can actually accommodate the expected vast growth in the network traffic volume.
The big potential of optical fiber to satisfy the demand for these networks has been well established in the last few years. According to Bouillet & Ellinas (2007) this is because optical fiber is highly reliable especially in commercially deployed systems and also its capable of accommodating longer repeater spacing’s besides having unlimited growth potential.
The use of transparent networks is considered as an attractive vision. Bouillet & Ellinas (2007) says that in transparent networks a signal wavelength passing an office space does not undergo optic-electronic conversion. Bouillet & Ellinas (2007) established that since a signal from a client NE connected via a specific wavelength must remain on the same wavelength when there is no wavelength conversion only a small size switch fabric is needed to interconnect the WDMs and NEs in a node. In a transparent network, it may include a single large fabric in place of having multiple switch matrices of small port counts.
Transparent networks provide significant footprint and power savings and on the surface suggest cost savings (Bouillet & Ellinas, 2007). They on the other hand indicated that although transparent network may be a viable option for small scale networks with pre-determined routes and limited numbers of nodes, it may not be a practical solution for a core mesh optical network (Bouillet & Ellinas, 2007).In a completely transparent optical network it is unworkable in wide-area national network. This is because most light paths will transparently surpass intermediary switch nodes, thus it is not viable for these nodes to glance into the aspect of the carried frames, thus not possible for this sort of network to perform an advanced light path performance monitoring.
Limitations of transparent networks
There are several limitations of transparent networks. The first is that this network does not allow wavelength conversion and thus it essentially creates a network of many WDM channels of disjoint layers. Bouillet & Ellinas (2007) says that “inflexible usage of wavelengths in this network can lead to increased bandwidth and network operational cost thus negating all savings that may result from the elimination of O/E conversion” (p. 5). Besides that it is important that for this technology to be effective and in order to construct or come up with a flexible network for unrestricted routing and redundant capacity sharing, an all optical 3R-regeneration function must be available (Bouillet & Ellinas, 2007).
The second limitation of transparent network is that the absence of wavelength conversion in transparent networks is that only client based dedicated backup path protection can easily be provided (Bouillet & Ellinas, 2007). The limitation is that the wavelength continuity constraint on backup paths makes resource sharing in the network difficult in transparent networks and consequently no shared backup path protection. This implies that the capacity requirement for protected services is significantly higher that is 80% to 100% for transparent compared to opaque networks (Bouillet & Ellinas, 2007).
The third limitation as indicated by Bouillet & Ellinas (2007) is that “transparent networks may experience physical impairments such as chromatic dispersion, polarization mode dispersion, fiber nonlinearities, polarization degradations, WDM filter pass band narrowing, component crosstalk, amplifier noise among others” (p. 6). All this may accumulate over the physical path of the signal because of lack of O/E conversion. In addition it is important to understand that the accumulation of these impairments requires engineering of end to end systems in fixed configurations and therefore it is not possible to build a large network with an acceptable degree of flexibility (Bouillet & Ellinas, 2007).
Bouillet & Ellinas (2007) says that the fourth limitation of transparent networks is that the design of high capacity DWDM systems is based on intricate proprietary techniques, excluding any hope of interoperability among multiple vendors in the foreseeable future. Because a signal is launched at the client NE through the all optical switch directly into the WDM system without O/E conversion and it is not possible to develop a standard for the interface for a high capacity WDM system, the operators will not have the flexibility to select the client NE vendor and the WDM vendor independently. In this context transparent networks by necessity are single vendor solutions (Bouillet & Ellinas, 2007).
With all the above limitation, Bouillet & Ellinas (2007) says that the challenge of performance engineering continental scale transparent reconfigurable wavelength-routed networks remains severe and in networks that push limits, remains unsolved despite some attempts at formalizing the routing problem. This implies that a wide range of carrier requirement that include dynamic configuration, wavelength conversion, multi vendor interoperability of transport equipment (WDM), low network level cost would be very hard to meet in transparent network architecture. Bouillet & Ellinas (2007) argued that the implication to the above issues is that an opaque network solution will remain for now the only practical and cost effective way of building a dynamic, scalable, and manageable core backbone network.
Opaque network solution are considered to be more expensive in terms of equipment costs when the core network capacity increases significantly, the opaque network offers the following fundamental ingredients for a large scale management network (Bouillet & Ellinas, 2007). The advantages of opaque networks is that there no cascading of physical impairments which eliminates the need to engineer end to end systems and thus allows full flexibility in signal routing (Bouillet & Ellinas, 2007).
In opaque networks wavelength conversion is enabled. Bouillet & Ellinas (2007) says that “network capacity can be utilized for service without any restrictions and additional significant cost savings can be offered by sharing redundant capacity in mesh architecture” (p. 6). The use of an all optical switch fabric without any compromise of the control and management functions. Also overhead visibility provides support for the management and control functions that are taken for granted in today’s networks.
In opaque networks Bouillet & Ellinas (2007) indicated that the network size and the length of the light paths can be large since regeneration and retiming are present along the physical path of the signal. Bouillet & Ellinas (2007) also say that “besides this link-by-link network evolution permits link-by link incorporation of new technology as the network is partitioned into point to point optical links” (p. 6).
Because transparent network architectures are likely to remain unrealistic for quite some time, opaque network architectures in which WDM systems utilize transponders. Bouillet & Ellinas (2007) says that today’s opaque networks have opaque switches with an electronic switch fabric in an opaque network. The interfaces to the fabric are opaque interfaces which imply that transceivers are present at all interfaces to the switch and these transceivers provide an OE (input) and EO (output) conversion of the signal.
In their research, Bouillet & Ellinas (2007) continues to say that the opaque transceivers provide support for fault detection and isolation, performance monitoring, connection verification, neighbor/topology discovery and signaling as well as support for implementation of the network routing and recovery protocols. Bouillet & Ellinas (2007) thus says that “wavelength-selective cross connects can also be used in ultra long haul applications in the core network in a completely transparent manner” (p. 11). These network elements tolerate for end to end bit rate and data format precision, they face a number of challenges (Bouillet & Ellinas, 2007).
Opaque switches will always remain for the embedded service base even after the transparent switches are introduced in the network. Bouillet & Ellinas (2007) further says “that opaque switches as used in opaque networks provide the grooming and multiplexing functions as well as some of the necessary control and management functions and will scale and decrease in cost with rapid progress in electronics” (p. 11). Transparent networks have not yet materialized on a large scale even transparent switch in opaque networks still face technological as well as control and management challenges.
Grover (2004) noted that all logical connections between node pairs are realized by assigning a specific DWDM wavelength path (of suitably low loss, noise and distraction) end to end. In this context the resulting path is said to be transparent not ironically because it is optical all the way but because any payload can be modulated on the respective optical carrier wavelength and then received at the far end (Grover, 2004). Grover (2004) indicated that “transparency in this case is used to refer to the contempt that there is no dependency on the payload being in a specific format in terms of framing, bit-rate, line-coding, power level, jitter and so on as is usually the case when electrical circuits have to handle the signal en route” (p. 69).
Transparency nowadays requires a wavelength assignment that must be uniquely reserved for the path on each fiber en route. Grover (2004) says that “in principle a transparent network will allow a light path to occupy different wavelengths on its links when all optical wavelength conversion technology is available” (p. 69). The major advantage of wavelength path networking is that wavelength convertors are not needed and the path is transparent to payload (Grover, 2004).
The disadvantages associated with transparent networking are that wavelength assignment is much more complex because light paths must use the same wavelength end to end Grover (2004). Grover (2004) also says that to assign a wavelength then requires that it is free on all spans en route. In transparent network, logically is like operating a SONET network in which payloads cannot change their timeslot assignments in successive spans. According to Grover (2004), to avoid wavelength blocking, where a path cannot be routed because a single wavelength is not available on every span of any route, more total capacity is needed on each span and the DWDM technology employed must also support more wavelengths per fiber than a corresponding network where wavelengths can be re-assigned at each node.
Unlike in opaque networks, in transparent networks, the benefit of regeneration is not obtained to reset the accumulation of optical transmission impairments at each node. Grover (2004) says that the disadvantage is that each optical path in transparent network must be individually transmission engineered or limits imposed on the length and routing of end to end paths contains transmission impairments and loss. On the same context fault isolation becomes more difficult and there are no signal associated overhead channels to support numerous operations and maintenance functions that employ payload associated adjacent node signaling.
Opaque networks are on the other hand the opposite of transparent networks. Grover (2004) says that opaque networks are defined as an end to end optical path that may use more than one optical wavelength along its route. Wavelength assignment to opaque networks can be changed as needed through wavelength conversion at cross connect nodes. Grover (2004) also says that “in future all optical wavelength conversion may be possible in which case an opaque network could thus remain optical end-to-end but at present wavelength conversion technically implies optical to electrical conversion and re-modulation onto a new laser” (p. 69).
Conditioning operations, capacity design, fault location, transmission engineering and capacity requirements are all more favorable with opaque networking. Grover (2004) says that “specifically wavelength blocking is eliminated so that problems of routing and capacity design are all logically similar to that of a SONET environment” (p. 70). Studies however show that a fully opaque network implies O/E and E/O transponders at each node and large electronic switching cores which are expensive and power consuming operating at 10 to 40 Gb/s (Grover, 2004).
In addition, Stern, Ellinas & Bala (2008) says that transparent networks are faced by the lack of adequate control and management and the inflexibility of configurations due to the nonlinear limitations that appear in large scale transparent optical networks. The absence of wavelength conversion can be dealt with in regeneration sites. Stern, Ellinas & Bala (2008) also indicated that sparse wavelength conversion in all optical express networks is achieved by utilizing the wavelength conversion that is already available at regeneration sites. Stern, Ellinas & Bala (2008) say that “the main problem of providing the control and management functions is still present, because access to electrical signals is not possible in transparent networks whereas these signals are readily available in opaque networks” (p. 858).
Research shows that in optical network where there is no wavelength conversion or limited conversion abilities individuals face the problem of routing and wavelength assignment (Grover, 2004). This implies that there are two ways that blocking can occur that is either through capacity blocking or wavelength mis-match blocking.Philipp (2000) says that an optical transport network is composed of interconnected nodes that is network elements, software and operation processes that must function together to provide services. He further says that the optical communication network extends over several point to point links and comprises several nodes that are divided into different coverage areas, represented by network planes operating on top of each other (Philipp, 2000).
In addition Downing (2004) says that the photonic signal found in optical fiber is fundamentally different from the electrical one. This is because prior to transmitting the signal, it must be converted from electrical form, modulated appropriately, and formatted for multiplexing and network transport (Downing, 2004). Downing (2004) noted that “WDM takes the upper hand of the big bandwidth of optical fiber through assigning a unique wavelength to each channel and allowing signals of different optical wavelengths to be transmitted down the same fiber” (p. 236). Besides that studies indicate that dense wavelength division multiplexing (DWDM) systems were designed to obtain many wavelength capabilities in the C-ans L-bands (Downing, 2004). It is important to note that current optical amplifier technologies allow for DWDM throughout the entire fiber transmission window (Downing, 2004).
Moreover, Faulkner & Harmer (2000) say that wavelength routing has recently seen a remarkable upsurge in interest as a potential transport technology for IP traffic since the traffic demand of the internet and other data services is on the upward growth. Wavelength Division Multiplexing is known to offer a great deal of solution for solving the fiber exhaustion. Faulkner & Harmer (2000) established that “WDM technology can carry over 100 signals at 2.5 or 10Gb/s and therefore these signals are terminated at an electrical node a very high switching capacity is needed” (p. 85). Optical technology provides an elegant solution by providing a possibility of optical bypass for the transit traffic by dropping the fractional traffic that is needed at a particular point (Faulkner & Harmer, 2000).
In opaque networks the physical layer security relies on the information obtained at opaque nodes where the signal is electronically regenerated (Mitrou, 2004). On the other hand, Mitrou (2004) says that “transparent optical networks, security is even more complex since the optical signals are not regenerated and therefore faults and attacks are more difficult to be detected and isolated” (p. 1395). Opaque networks give allowance of supervising the signal at every node of the opaque node in which the optical signal is normally converted to the electrical domain. Mitrou (2004) says that in transparent networks the data remains in the optical domain all long its path which implies that it does that without passing through any optical-to-electrical conversion but rather through optical amplification and optical switching in the near future, and optical regeneration and conversion further in the future. Also the optical signal is more exposed to degradation without being noticed by the network management system (Mitrou, 2004).
In conclusion, Mitrou (2004) argues that in transparent optical networks the signal remains in the optical domain along its path without going through any optic-to-electric conversion. He further says that transparent networks are very promising because they reduce unnecessary, expensive optoelectronic conversions besides offering high data-rate, give flexible switching, and support multiple types of clients. According to Mitrou (2004) transparent optical networks are known to contain two classes of network components that include; optical components that take care of the optical signal transmission and are not able to generate alarms. The second components are the monitoring equipment that is able to generate alarms and notifications when the optical signal is not the expected one. Failure of the monitoring equipment results in the loss of alarms that will be considered in the algorithm as lost alarms.
In this context, transparent optical networks are more susceptible to failures than opaque networks because the quality of the optical signal is not evaluated at each node and also because a single failure can affect more channels than in opaque networks as there are no transparency boundaries supported by optoelectronic regenerators Mitrou (2004).