1

Physical layer impairments (PLIs) are incurred by non- ideal optical transmission medium, accumulate along an optical path, and determine the feasibility or transmission quality of the light paths. If the received signal quality is not within the receiver sensitivity threshold, the receiver may not be able to correctly detect the optical signal, causing the light path (and the corresponding reserved resources) to be useless. Hence, it is important for network designers and operators to know 1) various important PLIs; 2) their effects on lightpath feasibility; 3) PLI analytical modeling, and monitoring and mitigation techniques; 4) various techniques to communicate PLI information to network layer and control plane protocols; 5) finally, how to use all these techniques in conjunction with control and management plane protocols to dynamically set up and manage optically feasible lightpaths. The PLIs and their significance depends on: network typeopaque, translucent, or transparent; the reachaccess, metro, or core/long-haul; the number and type of network elementsfiber, wavelengths, amplifiers, switching elements, etc.;

the type of applicationsreal-time, non-real time,

Types Of PLIs: Linear PLIs: intensity- independent, static in nature Non-Linear PLIs: intensity-dependent, dynamic in nature

Linear PLIs: Power Losses Power loss can be defined as the optical loss that is accumulated from source to destination along fiber-links and is normally made up of intrinsic fiber losses and extrinsic bending losses. Let P in be the power launched at the input of a fiber of length L; then the output power Pout is given by Pout = Pin eL, where is the fiber attenuation coefficient. Chromatic Dispersion The degradation of an optical signal caused by the various spectral components traveling at their own different velocities is called dispersion. CD causes an optical pulse to broaden such that it spreads into the time slots of the other pulses. Most commonly deployed compensation techniques are based on DCF Polarization mode Dispersion Polarization mode dispersion (PMD) is a form of modal dispersion where two different polarizations of light in a waveguide, which normally travel at the same

speed, travel at different speeds due to random imperfections and asymmetries, causing random spreading of optical pulses. Unless it is compensated, which is difficult, this ultimately limits the rate at which data can be transmitted over a fiber. Figure shows the pulse broadening of a pulse due to the fact that different polarization components traveling at different speeds. Polarization Dependent loss The two polarization components along the two axes of a circular fiber suffer different rates of loss due to irregularities in the fiber, thereby degrading signal quality in an uncontrolled and unpredictable manner and introducing fluctuations in optical signal to noise ratio (OSNR) The combined effect of PMD and PDL can further degrade the optical signal quality. PDL is a measure of the peak-to-peak difference in transmission of an optical component/system w.r.t. all possible states of polarization Amplifier Spontaneous Emission The primary source of additive noise in optically amplified systems is due to the ASE produced by the optical amplifiers used as intermediate repeaters and as preamplifiers at the receiver end. This noise is often quantified with noise figure (NF). Excess ASE is an unwanted effect in lasers, since it dissipates some of the lasers power. In optical amplifiers, ASE limits the achievable gain of the amplifier and increases its noise level. The ASE noise is very broadband ( 40 nm) and needs to be carefully analyzed to evaluate its degrading effect on system performance. CrossTalk Linear crosstalk arises due to incomplete isolation of WDM channels by optical components such as OADMs, OXCs, multiplexers/demultiplexers, and optical switches, i.e., the effect of signal power leakage from other WDM channels on the desired channel.

Non-Linear PLIs: Fiber nonlinearity originates from the intensity dependence of the fiber refractive index. The impact of nonlinearity increases as the optical power increases. Thus it enforces the upper limit to launch power. The importance of non-linear effects is growing due to 1) increase in optical power levels to increase the optical reach, 2) recent developments in optical components such as EDFA and DWDM systems to build more flexible networks, 3) increase in channel bit-rate to increase the traffic carrying capacity of wavelengths, 4) decrease in channel spacing to increase the number of wavelengths and overall network capacity.

Self-Phase Modulation (SPM) SPM refers to the self-induced phase shift experienced by an optical pulse during its propagation in optical fibers. An ultra-short optical pulse, when traveling in a medium, will induce a time varying refractive index of the medium, i.e., the higher intensity portions of an optical pulse encounter a higher refractive index of the fiber compared with the lower intensity portions. This results in a positive refractive index gradient (dn/dt) at the leading edge of the pulse and a negative refractive index gradient (dn/dt) at its trailing edge. This temporally varying refractive index change results in a temporally varying phase change leading to frequency chirping, i.e., the leading edge of the pulse finds frequency shift towards the higher side whereas the trailing edge experiences shift towards the lower side. Hence, the primary effect of SPM is to broaden the pulse in the frequency domain, keeping the temporal shape unaltered. Cross-Phase Modulation (XPM) The non-linear re- fractive index seen by an optical pulse depends not only on the intensity of the pulse but also on the intensity of the other co-propagating optical pulses, i.e., the non-linear phase modulation of an optical pulse caused by fluctuations in intensity of other optical pulses is called XPM Four Wave Mixing (FWM) FWM originates from third order non-linear susceptibility in optical links. If three

optical signals with carrier frequencies 1, 2 and 3, co-propagate inside a fiber simultaneously, it generates a fourth signal with frequency 4, which is related to the other frequencies by 4 = 1 2 3 In general for W wavelengths launched into a fiber, the number of FWM channels produced is M = W 2(W 1)/2) The FWM effect is independent of the bit-rate and is critically dependent on the channel spacing and fiber dispersion. Decreasing the channel spacing increases the four-wave mixing effect. Stimulated Brillouin Scattering (SBS) SBS occurs when an optical signal in fiber interacts with the density variations such as acoustic phonons and changes its path. In SBS, the scattering process is stimulated by photons with a wavelength higher than the wavelength of the incident signal. Stimulated Raman Scattering (SRS) In WDM systems, if two or more optical signals at different wavelengths are injected into a fiber, the SRS effect causes optical signal power from lower wavelength optical channels to be trans- ferred to the higher wavelength optical channels. This can skew the power distribution among the WDM channels reducing the signal-to-noise ratio of the lower wavelength channels and introducing crosstalk on the higher wavelength channels

PLI-Aware Service Level Agreements (SLA) RWA algorithms need to consider SLAs that are specific to the optical layer in order to realize dynamically reconfigurable generalized multi-protocol label switching (GMPLS)-based WDM optical networks Optical Power: The optical power at the end of a light path has to be within the dynamic range of the receiver. An optical receiver needs a minimum power, called receiver sensitivity, to distinguish between 1s and 0s. In addition to the receiver sensitivity, the minimum optical power required also depends on the type of forward error correction (FEC) used. Minimum Optical Signal to Noise Ratio (OSNR) To correctly decode and interpret the received signal, it is important for the received signal to be above the minimum OSNR level. OSNR depends on several impairments such as ASE, CD, PMD, etc.

Bit-Error Rate (BER) BER is a measure of service degradation in optical networks and should be below some threshold level; otherwise false alarms may be sent to higher layers indicating a failure which eventually may lead to setup of alternate lightpaths or rerouting of traffic. Q-factor This method can determine error ratios faster than the traditional BER test. Q-factor measures the quality of an analog transmission signal in terms of its signal-to-noise ratio (SNR). It takes into account physical impairments to the signalfor example, noise, chromatic dispersion and any polarization or non-linear effectswhich can degrade the signal and ultimately cause bit errors. In other words, the higher the value of Q-factor, the better the OSNR and therefore the lower the probability of bit errors. Q-factor is the difference between the mean values of the signal levels for a 1 and a 0 (1 and 0), divided by the sum of the noise standard deviation values (1 and 0) at those two signal levels assuming Gaussian noise and the probability of a 1 and 0 transmission being equal, i.e., Q = (10)/(1+0)

RWA algorithms can be defined as a procedure to establish a route and assign a wavelength for each connection request. This algorithms try to satisfy the agreed parameters on the SLA when they are calculating the lightpath. They can be classified in five approaches according to: 1st The constrains used to verify the feasibility of a lightpath such as OSNR, BER or Q-factor. 2nd The physical impairments that are considered in the feasibility evaluation. 3rd The type of RWA algorithm that it is used. It can be: Integrated: Route and wavelength are computed at the same step. 2-step: Route and wavelength are computed separately, one after the other. 4th The network scope that is used, such as centralized or distributed. 5th The PLI scope -> If PLIs are estimated using analytical models

either in a centralized server or in a distributed manner; or measure in real time using monitors.

General flow chart for PLI-aware RWA algorithms is shown on the figure. PLI-aware RWA algorithms establish the lightpaths following these procedure: 1. A new lightpath request arrives. 2. For each connection request a route and a wavelength are found using a RWA algorithm. It can be 2-step or integrated procedure. 3. The lightpath is tested for its feasibility. Then the OSNR, BER and Qfactor is estimated on the lightpath. If the estimated OSNR, BER and Q-factor satisfy the threshold requirement, then the lightpath is feasible and is admitted into the network. Otherwise, the RWA algorithm may select other route and/or wavelength and repeat the procedure to check their feasibility.

10

Translucent networks: are optical networks where at some intermediate nodes along the paths OEO conversion is done. The main RWA challenges on translucent networks are: Handling the physical layer impairments and the wavelength contention for all possible lightpaths. The optical-layer constraints and available resources may change dynamically. When there are a large number of nodes, it is necessary to use regeneration resources inside of a domain. Due to availability of network resources will change according to the change of traffic. In order to cope with this variation, the resource should be managed in a dynamic manner. In consequence, their availability status should be known by the whole network and been dynamically updated.

11

Intradomain Dynamic Balancing Routing (IDDBR) is a RWA algorithm use on translucent networks. In order to search different routes between source (S) and destination (D) the IDDBR uses the OLC-BFS procedure. In addition, in order to achieve a possible path, this algorithm calculates the Impairment Parameter Triplet (IPT) and the wavelength continuity constraint to achieve the possible paths. The IPT is composed by: Chromatic Dispersion (CD). Amplified Spontaneous Emission (ASE). Crosstalk.

12

OLC-BFC in order to calculate the different routes follows this steps: 1st Discards the possible routes that traverse more optical links. 2nd It looks if there are possible routes that traverse the same number of links. If there are it discards the routes that have higher values of IPT. Then, it selects the route which traverses less number of optical links. If there are not any possible routes that traverse the same number of optical links, it selects directly the route which traverses less number of optical links.

13

The IDDBR RWA algorithm follows 5 steps to get the path between the source (S) and the destination (D). Step 1: Trying optical bypass at all nodes 1.It checks if the wavelength of the source and the destination are the same, if it is it tries to find a route between them using the OLC-BFS procedure. This route has the same wavelength Ws along all the path. 2.If not, it passes to the step 2. Step 2: Trying regeneration either at S or at intermediate interior node. It tries to find a route which has an intermediate regenerator using OLC-BFS. If it found it, the route will use wavelength Ws from the source (S) to the regeneration resource (R) and wavelength WT from the R to the destination (D). If it fails, it goes to step 3. Step 3: Trying regeneration both at S and at an intermediate node R. 1.It identifies all possible intermediate regenerators that can be used between S

14

and D. 2. It uses OLC-BFS to select which route and which wavelength will be used between S and R. It will use WT between R and D. If it fails, Step 4 will be executed. Step 4: Trying regeneration at D. In this step the algorithm checks if there are any R available for this signal at node D. If it is the regeneration of the signal is carried out at node D. Then, OLC-BFS is used to find the different wavelengths that will be used along the path. If it fails, it goes to step 5. Step 5: Choice among candidate routes. If Step 5 is executed means that all previous steps had failed and there are a group of candidate routes. The preference to select a route from the group of the possible candidates the following step order has to be followed: 1)>2)>3)>4). If after executing again all the steps it still results more than one route, the one with the lowest values of IPT will be chosen.

14

Transparent networks are optical networks where the lightpaths are switched completely in the optical domain. The lack of optical regeneration increases the impact of the Physical Layer Impairments. PLI-aware RWA algorithms on transparent networks have 2 different approaches: Centralized: The RWA is done in a centralized manner. But, the wavelength assignment can be done either centralized or distributed. Distributed: The RWA is done using extensions of the GMPLS protocol.

15

The example of PLI-aware RWA algorithms that uses the centralized approach which is going to be explained uses the Network Management System.

16

The main elements that will be used to compute the route in this example of the centralized approach are: Network Element (NE): A single NE which is connected to all NE play the role of a server. Traffic Engineering Database (TED): It stores information about the network topology, resource availability and physical parameters. Network Management System (NMS): It is responsible for the administration, computation and provisioning functionalities of the whole network.

17

This algorithm in order to compute the path follows these steps: 1st A connection request is received by the NMS. 2nd The routing computation process is launched considering both the current TED information and the requirements of the connection request. 3rd The NMS configures all the NEs optical switching equipment involved in the path in a parallel way to set up the lightpath.

18

The distributed approach works in the same way than the centralized approach, but it has 2 differences in respect of the other approach: Each NE is responsible to setup the lightpaths. A GMPLS control plane is needed to establish the end-to-end path.

19

Traditional transport networks primarily consist of a transport plane and a management plane. Two sides of the boxes shown in the above slide. Here, the transport plane is responsible for carrying the user data across the network and comprises of network equipment such as, line interface cards, switch fabrics, backplanes and fiber plants. In this scenario all the (operations, administration, maintenance and provisioning)OAM&P are handled by the management plane. Now, we have an addition of optical control planes that essentially lie between management plane and the transport planes. The function of the control plane is to synchronize the intelligence between NEs. This paves way for NEs to have the information about the complete network topology and resource information which can be used to plan, establish and maintain user services. So, the introduction of control plane is required for fast and flexible resource provisioning, high reliability and scalability. In general, ITU-T has defined the architecture and requirements for an optical control plane. However, IETF is more focused on developing the protocols for an OCP. A third, standard body OIF (Optical Internetworking Forum) is deploying GMPLS protocols into ASON architecture. .

20

The main application of GMPLS in the context of optical networks is the dynamic establishment of lightpaths. Moreover, its drawback is lack of physical layer information details such as PLI, transponder characteristics and availability, regenerator/WC avail. Information etc. GMPLS serves to support switching techniques in multi-layered network. IP at L3 to OTN at L1

21

See the next slides for description.

22

The extension of routing protocols are used to carry the wavelength availability and PLI information. The route computation is based on constrained shortest path first (CSPF) algorithm at the source node. Moreover, a wavelength is selected as well. During the CSPF calculation, PLIs can be considered and hence no extra modifications are required for signaling protocol like, resource reservation protocol(RSVP-TE) in order to guarantee the optical feasibility of the lightpath. Modeling of PLIs should take into consideration the type of network elements and vendors. Several issues and challenges must be explored: most important parameters to carry, representation of these parameters, delay..etc..

23

This extension calculates the route at the source node using CSPF (of course using TED). Then, RSVP-TE is used to select an optically/physically wavelength before setting up the lightpath. Then, the source node launches the SETUP request on the network. This SETUP message accumulates the estimated PLI performance between any traversed link from source to the destination. The admission control interface at the destination checks whether the required QoS is met by analyzing the accumulated PLI performance. If the accumulated PLI performance is compliant with the receiver sensitivity, the a positive response is sent back and the lightpath is established. Otherwise, a negative message is sent back and the process can be repeated using another route. In this method, no modifications are required in the routing protocol. In the above process, each node keeps and updated local PLID and models to evaluate PLI performance. Issues: bi-directional path feasibility checking because non-linear impairments are asymmetric. Best wavelength to minimize future connection requests.

24

Routing protocol to carry wavelength availability information and linear impairments as these are relatively static in nature.

25

PCE is a new standard from IETF. It is capable of computing a network path or route taking in account the network graph and PLI constraints. Using this method the path computation can be either centralized or distributed. In the centralized version, PCE is aware of the whole network topology, resource availability, and physical parameters in a central repository called Traffic Engineering Database(TED). PCE also has another database, i.e. PLID, obtained from either NMS or PMS(performance monitoring system). When a request arrives at NE, it sends query to the PCE, which computes the required path using TED and PLID information. Next, PCE sends back the explicit path to the source node. Then, the source node establishes the path using signaling protocol (PATH/RESV/RSVP-TE). More traditional approach would be NMS based. Where NMS receives a connection request and it calculates the route using TED and received set of requirements. Then, NMS configures all the optical switching elements involved in the calculated path in parallel [6]. But, in [5] NMS works exclusively with PCE, and uses signaling protocol to establish path. SDN/OpenFlow based approach is more appreciated in closed network platforms like Campus networks and data center where IP/MPLS protocols are not used heavily.

26

PCE calculates the path in a centralized or distributed manner taking into account the global network topology, resources, and physical parameters etc. Which makes it preferable to guarantee the optimum resource utilization.

26

 PLI has to be considered in designing optical networks and managing off-line design network. Different approaches for RWA algorithms has to be taken in consideration depending on the type of the network. Because, PLI have more influence on transparent networks.

1. PLIs are incurred due to the non ideal optical transmission media and when ever there is an optical switching in optical domain PLIs cause severe damage to the received signal. 2. PLI has to be considered while designing and maintaining the optical network. 3. Monitoring and modeling of both Linear and Non linear PLIs in the optical layer are significant.

27

28

29