Data neural network system and method

• US 10,462,039 B2
• Filed: 12/28/2016
• Issued: 10/29/2019
• Est. Priority Date: 04/06/2006
• Status: Active Grant

• Current Assignee: Samuel Bennett Wood, Samuel Frederick Wood
• Original Assignee: Samuel Bennett Wood, Samuel Frederick Wood
• Inventors: Wood, Samuel Frederick, Wood, Samuel Bennett
• Primary Examiner(s): Orgad, Edan
• Assistant Examiner(s): Maglo, Emmanuel K
• Application Number:
  US15/530,345
• Publication Number:
  US 20170161609A1
• Time in Patent Office: 1,035 Days
• Field of Search: None
• CPC Class Codes: G06N 3/06 Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons
• CPC Class Codes: H04L 45/04 Interdomain routing, e.g. hierarchical routing
• CPC Class Codes: H04L 45/08 Learning-based routing, e.g. using neural networks or artificial intelligence
• CPC Class Codes: H04L 45/121 by minimising delays
• CPC Class Codes: H04L 45/125 based on throughput or bandwidth
• CPC Class Codes: H04L 45/20 Hop count for routing purposes, e.g. TTL
• CPC Class Codes: H04L 45/28 using route fault recovery
• CPC Class Codes: H04L 45/302 Route determination based on requested QoS
• CPC Class Codes: H04L 45/50 using label swapping, e.g. multi-protocol label switch [MPLS]

Abstract

This application discloses a neural network that also functions as a packet data network using an MPLS-type label switching technology. The neural network uses its intelligence to build and manage label switched paths (LSPs) to transport user packets and solve complex mathematical problems. This architecture is well suited to interconnect large numbers of processors or computers into a neural network exhibiting advanced intelligence which can be used for complex activities such as managing the power grid. However, the methods taught here can be applied to other data networks including ad-hoc, mobile, Information Centric, Content Centric, Sensor, and traditional IP packet networks, cell or frame-switched networks, time-slot networks and the like.

Claims

1. A method of synthesizing a neural network from a communications network containing a plurality of interconnected nodes, where at least one node contains one or more transmitters and one or more receivers, and where each node corresponds to one or more neurons of the neural network comprising:

receiving at a first node a communications request from another node of the network;

initiating a time delay, not associated with network congestion, or the processing or emulation of time varying signals, in response to the reception of the communications request, where the time delay is used as a weight function for the corresponding neural network, and is also used to contribute to the overall latency of a possible path traversing multiple nodes across the communications network, where the path would be used for the transport of signals unaltered through modulation or demodulation;

transmitting after the time delay, at least a portion of the communications request to at least a second node of the network to expose the possible path for use by the corresponding neural network; and

selecting by the neural network, a path for the transport of communications, where multiple paths exist, by combining the weights associated with each node of the network traversed by the path, where the nodes only use knowledge of their immediate neighbors and not of the network beyond.

2. The method of claim 1 where the communications network is a radio communications network.

3. The method of claim 1 where the communications network is an optical communications network.

4. The method of claim 1 where the communications network is a packet data network.

5. The method of claim 4 where the packet data network is connection oriented.

6. A communications system comprising a neural network synthesized from a communications network that includes a plurality of interconnected nodes, where at least one node contains one or more transmitters and one or more receivers, and where each node corresponds to one or more neurons of the neural network comprising:

a first node with software or circuitry capable of receiving a communications request from another node of the network;

the first node further capable of initiating a time delay, not associated with network congestion, or the processing or emulation of time varying signals, in response to the reception of the communications request, where the time delay is used as a weight function for the corresponding neural network, and also used to contribute to the overall latency of a possible path traversing multiple nodes across the network, where the path would be used for the transport of signals unaltered through modulation or demodulation;

the first node further capable of transmitting, after the time delay, at least a portion of the communications request to at least a second node of the network to expose the possible path for use by the corresponding neural network; and

the neural network capable of selecting a path for the transport of communications, where multiple possible paths exist, by combining the weights associated with each node of the network traversed by the path, where the nodes only use knowledge of their immediate neighbors and not of the network beyond.

7. The system of claim 6 where the communications network is a radio communications network.

8. The system of claim 6 where the communications network is an optical communications network.

9. The system of claim 6 where the communications network is a packet data network.

10. The system of claim 9 where the packet data network is connection oriented.