Method for constructing an AWG based N×N non-blocking optical multicast switching network

Patent metadata

Field	Value
Publication number	US-9854337-B1
Application number	US-201615344891-A
Country	US
Kind code	B1
Filing date	Nov 7, 2016
Priority date	Jul 28, 2016
Publication date	Dec 26, 2017
Grant date	Dec 26, 2017

Abstract

A method for constructing an AWG based non-blocking optical multicast switching network, comprising constructing a non-blocking optical copy network via a wavelength replication module and an arrayed waveguide grating recursively and constructing a non-blocking optical multicast switching network via cascading a data copy network with a point-to-point switching network. The number of active optical devices required for constructing an N×N optical switching network with r input/output ports and with each port carrying m wavelengths is just O(N log m N), realizing system scalability and saving hardware cost and power consumption. By splitting the routing path of the multicast network into a routing path with O(1) complexity in the copy network and a routing path in a point-to-point unicast switching network, the routing complexity of the multicast switching network is equivalent to that of a unicast switching network.

First claim

We claim:

1. A method for constructing an Arrayed Waveguide Grating (AWG) based N×N non-blocking optical multicast switching network, comprising
(1) constructing an AWG based N×N copy network A (r,m) having r input ports and r output ports, each of the input ports and output ports carrying m wavelengths,
arranging an m×m wavelength replication module (WR-module) on each port so as to have r m×m WR-modules on both an input stage and an output stage, with each consecutive m m×m WR-modules on either the input stage or the output stage being connected with an m×m AWG, resulting in r′ m×m AWGs on both an input side and an output side, r=r′m,
arranging m middle stage copy networks A (r′,m) to be in-between the r′ m×m AWGs on the input side and the r′ m×m AWGs on the output side, each output port of each AWG on the input side being connected with an input port of one of the middle stage copy networks A (r′,m), each input port of each AWG on the output side being connected with an output port of one of the middle stage copy networks A (r′,m),
constructing the middle stage copy network A (r′,m) recursively in a same manner;
(2) constructing an middle stage network cell B (r i ,m) by i recursive construction of subnetworks of the copy network A (r,m), followed by constructing the middle stage network cell B (r i ,m) to an AWG based three-stage copy network comprising three stages of WR-modules and two AWGs, r i m×m WR-modules on an middle input stage, m r i ×r i WR-modules on an middle stage, and r i m×m WR-modules on an middle output stage, the middle input stage being connected with the middle stage via an r i ×m AWG, and the middle stage being connected with the middle output stage via an m×r i AWG;
(3) constructing AWG based N×N copy networks C (r,m) by i=log m r−1 times of recursive decomposition of subnets of A (r,m), each subnet being comprised of an m×m WR-module and an m×m AWG and having an middle stage network B (m,m);
(4) constructing an AWG based N×N multicast network (r,m) by cascading the two copy networks C (r,m) by combining an output stage WR-module of one C (r,m) for performing replication on the input side with an input stage WR-module of the other C (r,m) for performing point-to-point switching on the output side.

2. The method of claim 1 , further comprising
splitting up an optical multicast request into a copy request on the copy network on the input side and a point-to-point unicast switching request on the copy network on e output side,
wherein the copy network on the input side replicates data according to a number of required data copies of the multicast request by allocating a set of consecutive output ports of the copy network on the input side for each multicast request according to a size of a label of an input channel thereof, a number of the ports of the copy network on the input side allocated for the multicast request is determined by the number of required data copies so as to constitute a copying request on the copy network of the input side; and
the copy network on the output side conducts a point-to-point switching for the data copy by routing each data copy on the input port of the copy network on the output side according to an actual destination address thereof and a unicast routing algorithm so as to switch the data copy to a corresponding destination output channel.

3. The method of claim 1 , wherein a non-blocking optical copy network comprises a WR-module and an AWG,
one of the m×m WR-modules copies a signal on one or multiple channels of the m input wavelength channels onto the copy module on one or multiple channels of the m output wavelength channels,
the copy module comprises a 1×m optical coupler connected with an m×1 multiplexer Mux via m wavelength selective converter WSCs,
each WSC comprises a tunable optical filter TOF and a fixed wavelength converter FWC,
the input signal is broadcasted from the 1×m optical coupler to the m WSCs,
a wavelength signal to be copied or converted for each WSC is selected by the TOF and converted by the FWC, and
a converted signal is multiplexed and outputted via the m×1 Mux.

4. The method of claim 1 , further comprising
successively labeling in the A (r,m), from top down, the WR-modules on the input stage, the AWGs on the input side, the middle subnets, the AWGs on the output side, and the WR-modules on the output stage,
connecting the WR-module on the input stage with a label α with the middle subnet with the label γ via a wavelength λ x , and
connecting the WR-module on the output stage with the label β with the middle subnet with the label γ via the wavelength λ y ,
wherein x=[[α] m +γ] m and y=[[β] m +γ] m .

5. The method of claim 1 , further comprising
successively labeling in the E (r i ,m) from top down the WR-modules on the input stage, the AWGs on the input side, the middle WR-modules, the AWGs on the output side, and the WR-modules on the output stage,
connecting the WR-module on the input stage with the label α with the middle WR-module with the label γ via the wavelength λ x′ , and
connecting the WR-module on the output stage with the label β with the middle WR-module with the label γ via the wavelength λ y′ ,
wherein x′=[α+γ] |Λ| , y′=[β+γ] |Λ| , and |Λ|=max{r i ,m}.

6. The method of claim 1 , wherein the two AWG based copy networks are successively cascaded,
the output stage WR-module of the first copy network is combined with the input stage WR-module of the second copy network to be a WR-module in one column,
the first copy network generating a required data copy and the second copy network switching the data copy to a final destination output channel.

7. The method of claim 1 , wherein a non-blocking optical copy network is constructed by an m×m WR-module and an m×m AWG, and the numbers of the WR-modules and the AWGs remain unchanged irrespective of a scale of the optical multicast switching network.

8. The method of claim 1 , wherein the AWG based copy network on the input side is routed by
successively labeling all the input channels and the output channels respectively as 0, 1, . . . , N−1;
ordering all the requests according to the labels of the input channels thereof as 0, 1, 2, . . . ;
labeling the middle stage subnetworks from 0 to m−1 successively; and
allocating the middle stage subnetwork with the label [i] m to the request i.

Assignees

• Univ Shanghai Jiaotong

Inventors

• Ye Tong
• Lee Tony T
• Hu Weisheng

Classifications

• H04Q2011/0016

  using wavelength multiplexing or demultiplexing · CPC title

• H04Q2011/0047

  Broadcast; Multicast · CPC title

• H04Q2011/0022

  using fibre gratings · CPC title

• H04Q11/0005 Primary

  Switch and router aspects · CPC title

• H04Q2011/0058

  Crossbar; Matrix · CPC title