High-capacity WDM-TDM packet switch

A method of scheduling performed by a controller of a core module haying at least one link from each of a plurality of ingress modules, said core module comprising a plurality of S≧1 space switches, each space switch having a plurality of input ports and a plurality of output ports, the method comprising steps of:

• creating a data structure comprising
• a first three dimensional matrix having a space dimension s representing space switches associated with the core module, a space dimension p representing space-switch input ports, and a time dimension r representing rime slots in a slotted time;
• and
• a second three-dimensional matrix having said space dimension s, a space dimension π representing space-switch output ports, and said time dimension t;
• receiving capacity-allocation requests from the ingress modules, each request specifying an input port p, an output port π and a number K of time slots per time frame;
• and
• selecting a space switch s and a time slot t and, if both entries {s, p, t} of the first matrix and {s,π, t}of the second matrix arm free, allocating space-switch s and timeslot t and marling entries {s, p, t} and {s,π, t} as busy.

A distributed packet switch comprising:

• a plurality of channel cross connectors each having a plurality of outer links and a plurality of inner links;
• a plurality of core modules each of said core modules comprising a number of space switches;
• a plurality of edge modules each including an edge controller and
• a plurality of core controllers, one core controller associated with each of said core modules;
• wherein
• each of said outer links connects to an edge module and includes a first number of A channels in each direction to and from said edge module;
• each of said inner links connects to a core module and includes a second number of channels in each direction to and from said core module,
• said can modules and said core modules are spatially distributed over a wide geographical area and
• each of said edge modules exchanges timing packets with each of said core modules to accomplish time coordination.

A packet switch comprising:

• a plurality of egress modules, each for transmitting packets to subtending packet sinks
• a plurality of ingress modules each ingress module having a plurality of ingress ports and an ingress controller, each of said ingress ports including:
• an associated ingress buffer for receiving packets from subtending packet sources;
• and
• means for sorting packets arriving in the ingress buffer into ingress queues each ingress queue corresponding to an egress module from which packets in said each queue are to egress from the switch for delivery to the subtending packet sinks,
• a plurality of core modules each connecting to each ingress module and to each egress module and having its own controller for allocating and scheduling data paths of sufficient capacities to accommodate connection requests received from any ingress module to transfer packets to any egress module, said own controller operating independently of and concurrently with controllers of other core modules;
• and
• at least one cross-connector connection a subset of the ingress modules to the core modules,
• wherein each ingress module includes means for routing a connection through at least one core module, and wherein said data paths carry data blocks of equal durations.

A method of switching a packet through a switch comprising a plurality of ingress modules each having at least one ingress port, a plurality of egress modules each having at least one egress port, and a plurality of core modules, wherein each ingress module is coupled to each core module, each core module is coupled to each egress module, the method comprising the steps of:

• one of said ingress modules receiving a packet from a subtending traffic source;
• said one of said ingress modules sending a connection request to a selected one of the core modules, requesting a connection of a specified capacity to one of said egress modules;
• said selected one of the core modules:
• determining available capacity:
• updating said connection request according to said available capacity;
• and
• returning the connection request to said one of the ingress modules;
• and
• said one of the ingress modules sending the connection request to another of the core modules if said available capacity is less than said specified capacity.