Abstract: A label switching method is provided. When a Passive Optical Network (PON)-based Label Switching Path (LSP) is established, a PON logical service transmission channel is established between an Optical Line Terminal (OLT) and an Optical Network Unit (ONU). According to an identifier (ID) of the PON logical service transmission channel as a PON label, a PON-based Forwarding Information Base (FIB) table on the ONU is updated, and a PON-based Label Forwarding Information Base (LFIB) table on the OLT is updated, where the PON-based LFIB table records a forwarding relationship between an ingress port plus an ingress label and an egress port plus an egress label, and the PON-based FIB table records a forwarding relationship between the ingress port plus a destination address and the egress port plus the egress label. Therefore, the problems between network segments of different forwarding characteristics, protocol variation, inter-segment conversion, and mapping and control complexity are avoided.

Abstract: The present invention discloses a method and an apparatus for adjusting a service of optical synchronous digital hierarchy network, and the method comprises: at first configuring the input time-slot space-division cross path to the time-division module after adjustment in space-division module and the time-division cross path of the time-division module after adjustment; and configuring the output time-slot space-division cross path from the time-division module after adjustment in space-division module; deleting the original cross path formed by the time-division module before adjustment.

Abstract: An optical packet switching apparatus includes an optical coupler, an optical switch unit, and an optical switch control unit. The optical switch control unit includes an optical-to-electrical converter, a serial/parallel converter, an arrangement detector, a rearrangement unit, a frame synchronization unit, a route detector, a control signal generator, and an adjustment unit for adjusting the timing with which to output an optical switch control signal to the optical switch unit, based on arrangement information on a frame synchronization pattern fed from the arrangement detector.

Abstract: A multi-chassis network device includes a plurality of nodes that operate as a single device within the network and a switch fabric that forwards data plane packets between the plurality of nodes. The switch fabric includes a set of multiplexed optical interconnects coupling the nodes. For example, a multi-chassis router includes a plurality of routing nodes that operate as a single router within a network and a switch fabric that forwards packets between the plurality of routing nodes. The switch fabric includes at least one multiplexed optical interconnect coupling the routing nodes. The nodes of the multi-chassis router may direct portions of the optical signal over the multiplexed optical interconnect to different each other using wave-division multiplexing.

Abstract: An apparatus comprising a path computation element (PCE) configured for at least partial impairment aware routing and wavelength assignment (RWA) and to communicate with a path computation client (PCC) based on a PCE protocol (PCEP) that supports path routing, wavelength assignment (WA), and impairment validation (IV). Also disclosed is a network component comprising at least one processor configured to implement a method comprising establishing a PCEP session with a PCC, receiving path computation information comprising RWA information and constraints from the PCC, establishing impairment aware RWA (IA-RWA) based on the path computation information and a private impairment information for a vendor's equipment, and sending a path and an assigned wavelength based on the IA-RWA to the PCC. Disclosed is a method comprising establishing impairment aware routing and wavelength assignment for a plurality of network elements (NEs) in an optical network using routing and combined WA and IV.

Abstract: A network including nodes configured to provide auto-discovery and those that do not are provided in a network. The nodes that do not provide auto-discovery may be provided as end points or terminating nodes in the network. A path or circuit identifier is entered into a database at those nodes and communicated to a network management system. The network management system compares the path identifiers, and, if the two match, the network management system designates the nodes associated with the path identifiers as terminating nodes. A path through the network connecting these nodes can then be determined and monitored. In addition, fiber connection information may further be stored at each node and communicated to the network management system, such that links along the path can also be identified by the network management system. Thus, path determination and monitoring in a network including non-auto-discovering nodes can thus be obtained.

Abstract: A method and system for encoding and determining labels in a Dual Polarization (DP) Quaternary Phase Shift Keying (QPSK) signal is provided. A label frame, signature sequence, and data payload are combined using a complementary constant-weight code encoding (CCWC) encoder, the output of which is deinterleaved and differentially precoded to generate a polarized tributary of a DP-QPSK signal. This encoding can be duplicated for a second tributary of the DP-QPSK signal. The label can be determined using one or more polarizers and corresponding low-speed photodetectors, each applied to a copy of the DP-QPSK signal. The strongest output of the photodetectors is then used to determine the label. Alternatively, the DP-QPSK signal can be viewed as having XI, XQ, PH, and PV tributaries. These tributaries can then be translated into XI, XQ, YI, and YQ tributaries are encoded into a standard DP-QPSK signal.

Abstract: A packet-optical integrated switch without an optical transponder, includes: a packet line card configured to output an Ethernet packet signal to a pre-set output port; a packet switch fabric configured to transfer the packet signal from the packet line card to the output port previously set in a destination address included in the packet signal; a 10 gigabit Ethernet (10 GbE)/optical transport unit level 2 (OTU2) integrated line card configured to convert the packet signal from the packet switch fabric into an OTU2 optical signal having a pre-set wavelength; and a wavelength selection switch fabric configured to allocate the optical signal from the 10 GbE/OTU2 integrated line card to a pre-set wavelength division multiplexing (WDM) port by pre-set wavelength to exchange the optical signal to each port by wavelength, wherein the packet line card, the packet switch fabric, the 10 GbE/OTU2 integrated line card, and the wavelength selection switch fabric perform the reverse operations of the process, respectivel

Abstract: An apparatus comprising a path computation element (PCE) configured for at least partial impairment aware routing and wavelength assignment (RWA) and to communicate with a path computation client (PCC) based on a PCE protocol (PCEP) that supports path routing, wavelength assignment (WA), and impairment validation (IV). The PCEP comprises at least one operation selected from the group consisting of a new RWA path request operation and a path re-optimization request operation. Also disclosed is a network component comprising at least one processor configured to implement a method comprising establishing a PCEP session with a PCC, receiving path computation information comprising RWA information and constraints from the PCC, and establishing impairment aware RWA (IA-RWA) based on the path computation information and a private impairment information for a vendor's equipment.

Abstract: An optical network is disclosed comprising one or more photonic switching nodes is disclosed. Each of the switching nodes comprises a plurality of input ports; at least one output port; and a switch configured to route messages between the plurality of input ports and the at least one output port and provide bufferless resolution of contention between messages for a common output port.

Abstract: In asynchronous optical packet switches, scheduling packets from a buffer randomly will cause less efficient utilization of the buffer. Additionally, reordering of packets may cause problems for service quality demanding applications. According to the present invention a new electronic buffer scheduling algorithm is proposed and a switch utilizing this algorithm is disclosed. The algorithm is designed for utilizing the buffer resources efficiently, still avoiding serious packet reordering.

Abstract: A distributed packet switch to control data packet flows in a network is described. The distributed packet switch includes a means for operating over an asynchronous burst optical closed fiber medium. The distributed packet switch also includes at least one control system provided at a node to control data packet flow characteristics, such that the switch is configured to operate dependent on at least one efficiency parameter.

Abstract: An optical packet switching system includes optical packet switching apparatus and an optical packet transmitting apparatus. The optical packet switching apparatus includes client optical delay units for delaying optical packet signals, network optical delay units for delaying one of the network optical packet signals, the network optical delay unit having a longer delay time than the client optical delay unit, an optical switch unit for switching the route of the inputted client optical packet signal so as to be sent out, an optical switch control unit for controlling the optical switch unit. The optical switch control unit is configured in such a manner as to detect a free time slot. The optical packet transmitting apparatus adjusts transmit timing, with which the client optical packet signal is sent out, in such a manner that the client optical packet signal is inserted into the free time slot.

Abstract: An optical packet switching device is provided with: a first input unit and a second input unit for receiving an optical packet signal having destination information and priority information; a first demultiplexer and a second demultiplexer for branching the optical packet signal; an optical switch unit for routing one of branched optical packet signals; a first analyzer unit and a second analyzer unit for analyzing the header of the other branched optical packet signal so as to detect the destination information and the priority information; and an output competition determination unit for checking for temporal competition of a plurality of optical packet signals based on destination information and for determining whether the optical packet signals should be transmitted or discarded based on priority information when there is competition.

Abstract: Disclosed is an apparatus and method for medium access control (MAC) in an optical packet-switched network. The MAC apparatus may comprise a bandwidth allocation module and an MAC processor. The bandwidth allocation module determines a data transmission limit based on a probabilistic quota plus credit mechanism for each node of the network, dynamically informs all downstream nodes of unused quota and allows the downstream nodes to use remaining bandwidths of the upstream node. Through a control message carried by a control channel, the MAC processor determines uploading, downloading and data erasing for a plurality of data channels, and updates the corresponding contents in the control message.

Abstract: An optical switch controller controls an optical interconnection network that variably connects at least one input data channel to a plurality of outputs channels via at least one switching element. An address reader module has at least one semiconductor optical amplifier optically processes an optical signal. The address reader module obtains information by reading a data tag from the input data channel, and outputs an address control signal based on the information. The address control signal can be used to control switching elements in the optical interconnection network.

Abstract: An example method includes receiving at a scheduling server status information concerning at least one transmit queue for a plurality of optical pass-through switches (OptSws). A schedule for transmissions to take place is determined based on the status information for the OptSws. A control message is forwarded to the plurality of OptSws in order that the schedule may be implemented. In an embodiment, a schedule is determined for a wavelength available to transmit a transmission by removing unsupportable transmissions from a candidate transmission list, determining a first transmission that is the one of the transmissions of the candidate transmission list which most preferably satisfies a given parameter, updating a usage list to reflect determination of the first transmission; and replacing the candidate transmission list with the residue set of the first transmission. The usage list reflects transmissions determined by the scheduler.

Abstract: Provided is a burst scheduling method in an Optical Burst Switching (OBS) system in which a plurality of nodes are connected through a mesh-type network. When a TDB which has used many network resources via a plurality of nodes and an SHG burst generated in a previous node, among bursts including BCPs transmitted from the previous node, compete in a current node so as to occupy a specific output channel, scheduling is performed to cause the TDB to have a higher priority than the SHG burst such that the corresponding output channel is occupied. Therefore, it is possible to minimize a burst loss in a network node, thereby enhancing the overall system performance.

Abstract: An optical switching device the size and costs of which are reduced by decreasing the number of switching elements and which can flexibly accommodate the expansion of the number of ports. An optical demultiplexing section has 2n (n=1, 2, 3, . . . ) input ports and 2m (m>n) output ports and includes demultiplexing couplers for demultiplexing input optical packets. A switch fabric section includes optical gate elements for switching optical packets outputted from the optical demultiplexing section by switch drive control. An optical multiplexing section has 2m input ports and 2n output ports and includes multiplexing couplers for multiplexing the optical packets which pass through the optical gate elements. A scheduler exercises control over an entire optical packet switching process.

Abstract: An optical power information receiving unit receives optical power information transmitted from a power monitor. When it is judged that a new IT apparatus is connected to an optical switch, an optical switch control unit controls the optical switch to connect the IT apparatus and an optical switch control device. A control device port setting unit changes an IP address of an apparatus information acquiring port based on a control device setting address of an address information storing unit. An apparatus information acquiring unit acquires an IP address of the IT apparatus.

Abstract: A network element of a transport network has three fabrics housed within a single shelf of a telco rack, namely a packet fabric, an electrical fabric and an optical fabric. The shelf also includes inter-fabric circuitry, to bridge between the fabrics, e.g. packet-electrical and/or electrical-optical and/or packet-optical. The inter-fabric circuitries switchably transmit packets across fabrics in intermediate nodes of the transport network. The single shelf of some embodiments has a wall (“midplane”) between a rear region that holds the three fabrics and a front region that holds inter-fabric circuitry, and external interfaces to optical trunks and/or packet services and/or optical tributaries.

Abstract: A backbone network, comprising a network switch configured to communicate data over Ethernet and SONET/SDH interfaces without encapsulating the data. Also disclosed is a backbone network, comprising a plurality of synchronized network switches, wherein the switches are configured to communicate a plurality of time division multiplexed data streams across at least part of the network via a plurality of Ethernet interfaces and a plurality of SONET/SDH interfaces, and wherein the switches are configured to communicate the data streams without encapsulating the data streams.

Abstract: In a dual rate gigabit passive optical network, an optical line termination (OLT) transmits a first rate (GPON) message frame interleaved with a second rate (NGPON) message frame. An unused ATM partition of the GPON message frame is provided with a header and payload portion of the NGPON message frame so that the message frame remains at a predetermined length required by the network.

Abstract: Optical routers are currently unable to do packet switching except by translating the data to electronic data and then back, which is very inefficient. The present invention solves this problem by optically marking and detecting the packet headers or parts of them, translating at most only the headers or parts of them to electronics for making packet switching decisions, and keeping the rest of the packets in optical delay lines, and solving response-time problems. Another optimization described in this invention is improving routing efficiency and bandwidth utilization by grouping together identical data packets from the same source going to the same general area with a multiple list of targets connected to each copy of the data and sent together to the general target area. Another important optimization is a new architecture and principles for routing based on physical geographical IP addresses.

Abstract: Methods and apparatuses for dispatching OTN signals are disclosed. The method includes receiving an OTN signal frame; determining in sequence whether each byte in the OTN signal frame is an overhead byte; determining, if the byte is an overhead byte, whether the byte includes frame alignment data; and assigning, if the byte includes frame alignment data, the byte in sequence to a corresponding channel based on a sequence number of an OTN signal frame period, or storing, if the byte does not include frame alignment data, the byte in a buffer; or assigning, if the byte is not an overhead byte, the byte in sequence to a corresponding channel based on a location of the byte in the OTN signal frame.

Abstract: A multi-chassis network device includes a plurality of nodes that operate as a single device within the network and a switch fabric that forwards data plane packets between the plurality of nodes. The switch fabric includes a set of multiplexed optical interconnects coupling the nodes. For example, a multi-chassis router includes a plurality of routing nodes that operate as a single router within a network and a switch fabric that forwards packets between the plurality of routing nodes. The switch fabric includes at least one multiplexed optical interconnect coupling the routing nodes. The nodes of the multi-chassis router may direct portions of the optical signal over the multiplexed optical interconnect to different each other using wave-division multiplexing.

Abstract: A photonic network packet routing method includes the steps of optically encoding destination address information attached to an IP packet using light attributes, discriminating the encoded address information of the IP packet by optical correlation processing, switching to an output path for the IP packet based on a result of the discrimination, and outputting the IP packet labeled with prescribed address information on the output path selected by the switching step.

Abstract: An optical switch section that switches the path of an optical packet transmitted thereto and outputs the optical packet, and a control section that generates multiple switch control signals in accordance with the destination of an optical packet transmitted thereto, transmits the multiple control signals to the optical switch section through multiple signal transmission lines and controls the ON/OFF states of multiple optical switches are included, and delay measuring means for measuring a difference in delay time among the multiple signal transmission lines is provided. In this case, the control section adjusts the transmission times of the multiple switch control signals such that the difference in delay time measured by the delay measuring means can be cancelled and that the multiple switch control signals transmitted through the multiple signal transmission lines can arrive at the multiple optical switches simultaneously and transmits the multiple switch control signals.

Abstract: An arrangement includes a photonic band-gap assembly comprising at least one input wave guide and at least one output wave guides, and at least one routing element responsive to signals to selectively route a signal from the input wave guide to one or more of the output wave guides.

Abstract: An arrangement includes a photonic band-gap assembly comprising at least one input wave guide and at least one output wave guides, and at least one routing element responsive to signals to selectively route a signal from the input wave guide to one or more of the output wave guides.

Abstract: A method and a device are provided for swapping optical labels in an optical communication network. Optical information, including payload data and label data digitally encoded into the optical information, is received. At least one group of bits within the optical information is selectively inverted to rewrite the label data with new label data without changing the payload data. Each of the at least one group of inverted bits includes at least two bits and all bits of each of the at least one group of inverted bits are contiguous bits.

Abstract: A photonic network packet routing method includes the steps of optically encoding destination address information attached to an IP packet using light attributes, discriminating the encoded address information of the IP packet by optical correlation processing, switching to an output path for the IP packet based on a result of the discrimination, and outputting the IP packet labeled with prescribed address information on the output path selected by the switching step.

Abstract: An optical data storage system and method of use thereof are presented. The optical data storage system includes one or more optical buffer modules connected in series. Each optical buffer module includes a cross connect. Each cross connect is connected, by a pair of inputs and outputs, to an optical data storage unit, for example, a fiber delay line, by a pair to either an optical packet network or a cross connect of a first adjacent buffer module in the series, and by a pair to a cross connect of a second adjacent buffer module in the series. The buffer module also includes a read signal output line which is connected to a read signal input line of the second adjacent buffer module for transmitting a read signal. A control module within each buffer module directs the passage of data through the cross connect.

Abstract: A photonic network packet routing method includes the steps of optically encoding destination address information attached to an IP packet using light attributes, discriminating the encoded address information of the IP packet by optical correlation processing, switching to an output path for the IP packet based on a result of the discrimination, and outputting the IP packet labeled with prescribed address information on the output path selected by the switching step.

Abstract: Various embodiments of the present invention are directed to photonic-based interconnects for transmitting data encoded in electromagnetic signals between electronic mosaics. In one embodiment of the present invention, a photonic-based interconnect comprises a first photonic node coupled to a second photonic node via a waveguide. The first photonic node is coupled to a first electronic mosaic and is configured to transmit electromagnetic signals encoding data generated by the first electronic mosaic to a second electronic mosaic and receive electromagnetic signals encoding data generated by the second electronic mosaic. The second photonic node is coupled to the second electronic mosaic and is configured to transmit electromagnetic signals encoding data generated by the second electronic mosaic to the first electronic mosaic and receive electromagnetic signals encoding data generated by the first electronic mosaic.

Abstract: A photonic network packet routing method includes the steps of optically encoding destination address information attached to an IP packet using light attributes, discriminating the encoded address information of the IP packet by optical correlation processing, switching to an output path for the IP packet based on a result of the discrimination, and outputting the IP packet labeled with prescribed address information on the output path selected by the switching step.

Abstract: A wavelength-division-multiplexed based photonic burst switched (PBS) network, which includes edge and switching nodes, optically communicate information formatted into PBS control and data burst frames. Each PBS data burst frame is associated with a PBS control burst frame. A PBS burst frame includes a PBS burst header and burst payload having fields to indicate whether: (a) the PBS burst frame is a PBS control burst; (b) the control burst is transmitted on a wavelength different from that of the associated PBS data burst; and (c) the PBS burst frame has a label for use in a generalized multi-protocol label swapping (GMPLS)-based control system. The PBS burst payload frame includes fields to indicate (a) specific PBS payload information; (b) PBS data payload; and (c) an optional PBS payload frame check sequence (FCS) for error detection.

Abstract: Provided is a TDMA (time division multiple access) PON (passive optical network) OLT (optical line terminal) system for a broadcast service, including packet processor determining information according to types of frames (unicast, multicast, and broadcast frames) and a switch output port using header information (an IP address of a packet header, MAC (medium access control) addresses of Ethernet frames, and the like) of data received from an external node or the ONT (optical network terminal) and attaching the information to header parts of the frames to generate second data, a switch copying the second data by a predetermined number of times according to a transmission method and transmitting the second data to a corresponding destination output port according to the identification codes, M TDMA PON MAC processors removing the identification codes added to the second data and converting the second data into TDMA PON frames, and M optical transceivers converting the TDMA PON frames into optical signals and tr

Abstract: The Fibre Channel Credit Extender (FCCE) (600) is a network device that is disposed between and connected to an end node (210) and an optical repeater (220). The FCCE (600) contains as many buffer credits as necessary, to solve bandwidth problems in a network. In a situation where maximum bandwidth is required in both directions of a link, the FCCE (600) breaks a single logical link into three physically separated “linklets.” The short-distance linklets attain maximum bandwidth by use of the existing buffer credits of the end nodes. The long-distance linklet attains maximum bandwidth by use of very high receive buffer credits in the FCCEs (600). In this way, only those links that need maximum bandwidth over distances not covered by end-node credit counts need be attached to an FCCE (600). The FCCE (600) contains the optical repeater to gain distance on that link, and contains high credit count receive buffers to gain bandwidth on the link.

Abstract: There is described a transmission in a burst switching network, whereby data packets are collected to form a burst which is provided with a header. A loop-free common control channel is provided in the optical burst switching network to which all network nodes have access. Headers are transmitted by the loop-free control channel such that a header is distributed virtually in parallel to all network nodes and processed virtually in parallel in the network nodes.

Abstract: An optical device is provided with first and second inputs. A first coupler coupled is coupled to the first input and produces at least a first and second output. A second coupler is coupled to the second input and produces at least a first and second output. A third coupler is coupled to the first output of the first coupler and to the first output of the second coupler. A fourth coupler is coupled to the second output of the first coupler and to the second output of the second coupler. First and second crossing waveguides are provided with an angle selected to minimize crosstalk and losses between the first and second cross waveguides. The first crossing waveguide connects one of the first or second outputs from the first coupler with an input of the fourth coupler. The second crossing waveguide connects one of the first or second outputs from the second coupler with an input of the third coupler. A first phase shifter is coupled to the first and second waveguides.

Abstract: An apparatus and method for effectively transmitting a packet data and a control data of path information about the packet data in an optical packet data switching network. The method includes grouping a plurality of wavelengths into at least two wavelength bandwidths, each wavelength bandwidth being composed of neighboring wavelengths, and transmitting the optical packet data and the control data with wavelengths of different wavelength bandwidths respectively. Since the wavelength to transmit the optical packet data and the wavelength to transmit the control data has a difference from each other, a node receiving optical data easily divides the packet data and the control data.

Abstract: An optical apparatus is disclosed wherein increase of the number of wavelength selective switches provided for a standby system can be suppressed. The optical apparatus includes a plurality of upstream side optical devices and a plurality of downstream side optical devices configured such that a plurality of output ports to be set as output destinations of light from a plurality of input ports can be changed over for each wavelength. An upstream side standby switch connected at a plurality of inputs thereof individually to the input ports and can change over an output of light from the inputs for each wavelength. A downstream side standby switch is connected at an input thereof to the output of the upstream side standby switch and at a plurality of outputs thereof individually to the output ports and can output light from the input to the plural outputs thereof.

Abstract: A traffic signal node cross scheduling method includes: decomposing a multi-wavelength signal into multiple first single-wavelength signals; decomposing a first single-wavelength signal into multiple small-granularity wavelength signals; scheduling the multiple small-granularity wavelength signals; and converging the multiple small-granularity wavelength signals after scheduled into a second single-wavelength signal. A traffic signal node cross scheduling system is also provided by the embodiments of the present invention. According to the traffic signal node cross scheduling method and system provided by the embodiments of the present invention, one scheduling matrix may be used to implement the scheduling of the wavelength and sub-wavelength traffics, which improves the integration of the network node cross scheduling system and the utilization rate of the scheduling matrix, and reduces the hardware cost of the network node cross scheduling system.

Abstract: An optical transmission system capable of time difference correction without lengthening guard times, thereby improving optical packet transmission efficiency. A switching processor sets identical switching timing for input ports thereof such that signals input from the input ports are switched at the same timing. When an optical dummy packet is received, the switching processor switches the optical dummy packet to be returned to the originating node. A time difference corrector detects synchroneity of the switched and looped-back optical dummy packet and, if asynchronism is detected, varies readout timing until synchroneity is attained, thereby correcting the time difference so that the optical dummy packet may fit in the switching time range and thus can be switched normally.

Abstract: An address recognition apparatus may include a first normalizing unit, a second normalizing unit and an address determination unit. The first normalizing unit normalizes a first electric signal and generates a first normalized signal, wherein the first electric signal is associated with a first divided set of optical packet signals. The second normalizing unit normalizes a second electric signal and generates a second normalized signal, wherein the second electric signal is associated with a second divided set of optical packet signals. The address determination unit refers to the first and second normalized signals and determines whether a destination address of a set of optical packet signals is identical to or different from an address allocated to a self-station associated with the address recognition apparatus, wherein the set of optical packet signals has been divided into the first and second divided sets of optical packet signals.

Abstract: An optical wiring switching unit includes a first connection interface. A packet switch includes a second connection interface. A resource includes a third connection interface, and is formed with at least one of a server and a storage device. A control device controls the optical wiring switching unit to perform a switching of connection between the connection interfaces. The packet switch and the resource are connected to the optical wiring switching unit via the connection interfaces. A network is configured by connecting the packet switch and the resource with a control of the optical wiring switching unit.

Abstract: A system and method for a cell switching optical network. A block of data is buffered at an edge node and sliced into data slices. Slice headers are pre-appended to each data slice. The data slices are then scheduled for transmission onto an optical switching network during fixed time slots defined on a per carrier wavelength basis.

Abstract: An optical packet switch switches optical packets according to bit-rates at which the optical packets are provided. For example, optical packets that are received at similar bit-rates may be routed to a destination at separate time slots over a single channel wavelength, and optical packets that are received at different bit-rates may be routed to the destination over separate channel wavelengths. When optical packets are provided at different bit-rates on a plurality of input paths, optical packets provided at low bit-rates may be compacted before switching to the destination. Alternatively or additionally, the bit-rates of the optical packets may be balanced before switching to the destination. Bandwidth contention among optical packets may be resolved by polarizing optical packets originating from separate input paths in different polarization directions, and merging optical packets having different polarization directions onto a single switched channel wavelength.

Abstract: Various embodiments of the present invention are directed to photonic-interconnection-based compute clusters that provide high-speed, high-bandwidth interconnections between compute cluster nodes. In one embodiment of the present invention, the compute cluster includes a photonic interconnection having one or more optical transmission paths for transmitting independent frequency channels within an optical signal to each node in a set of nodes. The compute cluster includes one or more photonic-interconnection-based writers, each writer associated with a particular node, and each writer encoding information generated by the node into one of the independent frequency channels. A switch fabric directs the information encoded in the independent frequency channels to one or more nodes in the compute cluster.