« Home « Kết quả tìm kiếm

Optical Switching Networks


Tóm tắt Xem thử

- Maier is the author of the book Metropolitan Area WDM Networks – An AWG Based Approach..
- 15.1 State of the art 238.
- Eytan Modiano of the Massachusetts Institute of Technology and Prof.
- We have seen in the previous section that the use of wavelength converters helps improve the performance of wavelength-routing AONs and adds to the flexibility of the network..
- Let us briefly explain the various switching granularities in the following, beginning at the top of the table.
- The control packet is used to configure intermediate OBS nodes on the fly along the path between the ingress node and egress node of the OBS network.
- As a consequence, only customers (and not carriers) have visibility and control of the optical network..
- To avoid channel collisions at the output port of the combiner, the collected wavelength channels must be different.
- Thus, a given wavelength channel can be used only at one of the S input ports at any time..
- This period of the wavelength response is called free spectral range (FSR).
- Generally, the FSR of a D × D AWG consists of D contiguous wavelengths, i.e., the physical degree of an AWG is identical to the number of wavelengths per FSR.
- The waveplate at the symmetry line of the device eliminates the polarization dependence..
- Each wavelength gives routing instructions that are independent of the input port.
- Typical output powers are of the order of −10 dBm.
- The tuning time is limited only by the speed of the electronics.
- Thus, dispersion limits the minimum bit spacing (i.e., the maximum transmission rate).
- The amount of accumulated dispersion depends on the length of the link.
- Thus, different polarizations of the signal travel at different velocities.
- As long as the optical power within an optical fiber is small, the fiber can be treated as a linear medium (i.e., the loss and refractive index of the fiber are independent of the signal power).
- Other characteristics of SBS are that the Stokes wave propagates in the opposite direction of the input light.
- The resulting ASE appears as noise at the output of the EDFA..
- is merely a convenient representation of the variability in the generated photocurrent as a separate component.).
- We also elaborated on the need and importance of the control plane which is responsible to.
- At the end of the book, optical packet switching (OPS) and optical burst switching (OBS) are explained..
- Finally, Chapter 10 describes the state of the art of optical packet switching (OPS) net- works.
- The realization of the label depends on the link technology in use.
- r Fiber Switch Capable (FSC) Interfaces: Interfaces that switch data based on a position of the data in the physical space.
- Also note that, in GMPLS networks, an LSP can be established only between and through interfaces of the same type.
- As a result, the number of links whose state information is flooded by the IGP can be reduced significantly, resulting in an improved scalability of the GMPLS network (Banerjee et al., 2001b).
- independently of the TE links they manage.
- The exact implementation of the control channel(s) is not specified in Lang (2005).
- This picture of the network is referred to as the link state database..
- In GMPLS networks, the fault management procedure of the LMP can be used to localize failures (Lang and Drake, 2002).
- Upon receipt of the fault notification, fault mitigation is initiated.
- 6.1 consists of the fiber cross-connect (FXC), band cross-connect (BXC), and wavelength cross-connect (WXC) layers.
- Such a condition can be practically obtained by nominating one of the PSR ring nodes as master.
- Besides synchronization, PSR requires additional processing at the periphery of the network to place packets in the appropriate photonic slots destined for the corresponding node.
- In our example, a packet destined for node B is stored in the transmission (TX) buffer of the source node.
- The destination address of a given photonic slot is determined by the destination address of the first packet transmitted in the photonic slot.
- If the arriving photonic slot is not assigned a destination, the packet selected for transmission is the oldest packet (i.e., the packet with the longest queueing delay) among the head-of-the-line packets that do not cause any wavelength collision in the photonic slot..
- The length of the fixed-size packets equals one time slot.
- The photodiode converts the optical signal into an electrical signal irrespective of the wavelength..
- The node starts unstacking the photonic slot at the beginning of the next cycle by setting switch R in the cross state..
- One of the main bottlenecks in today’s Internet is (electronic) routing at the IP layer..
- The value of the parameter BAT must be set carefully.
- Alternative optical contention resolution schemes (e.g., the use of bulky fiber delay lines [FDLs.
- At the edge of the OBS network, one or more users are attached to an OBS node.
- The four functions executed by edge OBS users can be categorized into (1) burst assembly, (2) signaling, (3) routing and wavelength assignment, and (4) the computation of the offset time for the control packet which is sent prior to the data burst.
- The choice of the burst assembly algorithm can have a significant impact on the performance of the OBS network.
- burst size parameters determine the length of the assembled burst.
- Clearly, the estimation and setting of the offset time is crucial.
- A wavelength channel is reserved and the optical switching fabric is configured at the OBS node right before the estimated arrival time of the burst..
- r Estimated Release: In estimated release, an OBS node estimates the end of the burst by using the offset and size information carried in the preceding control packet.
- Note that the choice of the resource scheduling scheme in OBS networks also depends on the burst assembly algorithm used by the OBS users at the OBS network edge.
- However, if the control packet is sent before the corresponding burst is completely assembled, OBS nodes have to deploy explicit release schemes (i.e., the edge OBS user has to send another trailing control packet which indicates the end of the burst transmission)..
- Contention resolution is one of the main design objectives in OBS networks.
- In the following, we describe the functional blocks of the OBS MAC and optical layers of Fig.
- r Time-Based Assembly Algorithms: The first category contains time-based burst assembly algorithms where a fixed time threshold T is used as the criterion to send out a burst (i.e., the burst is transmitted after T time units)..
- Based on a linear prediction method, the OBS user predicts the length of the new burst..
- Among other information, the control packet contains the predicted length of the corresponding burst.
- In doing so, part of the burst assembly delay can be effectively masked to higher layers, resulting in a decreased latency in OBS networks..
- Thus, resources at the OBS node are made available before the actual arrival time of the burst.
- To avoid both shortcomings of the extra-offset-based QoS scheme, wavelength preemption techniques can be deployed instead in order to provide service differentiation in OBS networks (Phuritatkul et al., 2006).
- To resolve contention at OBS nodes, several techniques were proposed and investigated (e.g., the aforementioned use of FDLs).
- Conversely, for large B most of the contending bursts can be temporarily stored in SDLs.
- Burst segmentation as a means to resolve contention in OBS networks was first considered in the context of the so-called optical composite burst switching (OCBS) paradigm (Detti et al., 2002).
- free at the output port on which the remainder of the burst can be forwarded.
- In the event of con- tention, an OBS node must know which of the contending burst segments will be dropped.
- In the first approach, tail dropping, the tail of the original burst is dropped (i.e., the last one or more segments of the previously arrived burst are dropped when another contending burst arrives).
- Based on the traffic classes of the assembled packets, bursts are assigned different priorities.
- The first strategy, called perfect overlap, groups all multicast sessions with the same multicast group members (i.e., the same destination OBS users).
- Specifically, a lightpath pre- booking request is initialized by a source OBS user at the beginning (instead of end) of the burst aggregation.
- The processing determines the required resources of the corresponding burst.
- Otherwise, it decreases the priority of the wavelength.
- The research and experimental demonstration of the feasibility of OPS networks al- ready started in the mid 1990s.
- (1) extraction of the label from the optical packet, (2) electronic processing of the label to obtain routing information, (3) optical routing of the payload and resolving contention if necessary, and (4) rewriting of the label and recombining it with the optical payload..
- output of the optical correlator the packet’s destination address can be recognized and the packet can be forwarded to the appropriate OPS output port..
- r Switching Matrix: The switching matrix performs the switching operation of the payload in the optical domain according to the commands received from the control unit.
- Initial research on OPS networks focused on slotted networks using synchronous OPS nodes for switching fixed-size packets (e.g., the ACTS KEOPS project) (Gambini et al., 1998).
- According to the position of the optical buffer, OPS nodes can be classified into four major configurations: (1) output buffering, (2) shared buffering, (3) recirculation buffer- ing, and (4) input buffering (Hunter et al., 1998).
- Once a packet has entered the optical delay line, it can be retrieved only a fixed time period later equal to the propagation delay of the delay line.
- In a feed-forward configuration, an FDL feeds forward to the next stage of the OPS node..
- and (3) the third block comprises M FDL arrays, one for each of the M output ports.
- The proposed schemes focused on per-class classification of the traffic.
- operation of the switch is slotted.
- Each of the N input and N output fibers carries only one wavelength.
- The operation of the optical gates is managed by the control unit of the OPS node..
- At the output ports of the extended broadcast-and-select OPS.
- The control packet contains the destination address of the optical data packet.
- The control information includes the state (empty or occupied) of each wavelength in the slot and the destination ring of the entire WDM slot.
- MANs are found at the metro level of the network hierarchy between wide area networks (WANs) and access networks.
- Various aspects of the aforementioned ring networks can be found in RPR.
- Typically, a source node selects the ringlet that provides the shortest path to the destination node in terms of hops (i.e., the source node applies shortest path routing).
- In general, the cleave point can be put on any span of the ring..
- either direction, enabling spatial reuse on the remaining part of the ring (Maier et al., 2006)..
- At system initialization, all nodes broadcast topology discovery control packets on both ringlets with an initial TTL value equal to 255 (i.e., the maximum number of nodes).
- The following description of the RPR fairness control is explained in greater detail in Yuan et al.
- the remaining 1/4 of the bandwidth of link 1 is assigned to flow (1,5).
- The degradation of the network performance due to channel or receiver collisions can be mitigated or completely avoided at the architecture and/or protocol level.
- This can be done a priori without or a posteriori with taking into account the status of the slots

Xem thử không khả dụng, vui lòng xem tại trang nguồn
hoặc xem Tóm tắt