# US 2019/0356595 A1 · Global Optimization and Load Balancing in Networks

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2019/0356595 A1
- Document ID: US20190356595A1
- Status: Published application
- Kind: A1
- Application number: 16/531,077
- Filed: 2019-08-04
- Published: 2019-11-21
- Earliest claimed: 2016-04-05
- Assignees: Teridion Technologies LTD
- Applicants: Teridion Technologies LTD
- Inventors: Jacob Avidar; Elad Rave; Lior Musman
- Web page: https://patentgrep.com/patent/US20190356595A1

## CPC

- H04L 47/125: by balancing the load, e.g. traffic engineering
- G01C 15/004: {Reference lines, planes or sectors}
- G06Q 10/047: Optimisation of routes or paths, e.g. travelling salesman problem
- H04L 45/12: Shortest path evaluation
- H04L 45/124: {using a combination of metrics}
- H04L: TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION (arrangements common to telegraphic and telephonic communication H04M)
- G01C: MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY (measuring liquid level G01F; radio navigation, determining distance or velocity by use of propagation effects, e.g. Doppler effects, propagation time, of radio waves, analogous arrangements using other waves G01S)
- G06Q: INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR

## Abstract

Method and apparatus for globally optimizing a weighted flow network, including the procedures of collecting a plurality of metrics in the weighted flow network, receiving a plurality of route requests from a first node to a second node in the weighted flow network, determining K best routes out of N possible routes from the first node to the second node according to the plurality of metrics, balancing a load in the weighted flow network by distributing the plurality of route requests over the K best routes from the first node to the second node, and updating the plurality of metrics.

## Claims

### Claim 1

A method for load balancing a data network, comprising the procedures of:
collecting, during a first period of time, training data that includes a plurality of metrics of said data network for connections between nodes of the data network, the data network including a plurality of network servers connected over a network and operable to route requests, the plurality of network servers corresponding to nodes;
training a machine model, by the plurality of network servers, using the training to generate a trained machine model operable to process metric data associated with the data network to determine a set of routes out of a plurality of routes to transfer data between nodes of the data network and a distribution metric to assign a percentage of a load to individual routes of the set of routes;
receiving a plurality of route requests to transfer data from a first node of the nodes to a second node of the nodes;
obtaining, during a second period of time, a set of metrics associated with the data network, the second period of time being different from the first period of time and occurring when receiving the plurality of route requests;
processing the set of metrics with the trained machine model during the second period of time to determine a set of routes from said first node to said second node and a distribution metric for the set of routes;
assigning a percentage of a load to be handled by individual routes of the set of routes based at least in part on the distribution metric; and
transmitting information as computer readable instructions to control transfer and movement of the plurality of route requests over the set of routes between the first node and the second node based on the percentage of the load assigned to the individual routes.

### Claim 2 (depends on 1)

The method of claim 1, further comprising:
obtaining, during a third period of time, additional metrics subsequent the plurality of route requests, the third period of time being different from the first period of time and the second period of time and occurring concurrently with controlling the transfer and the movement of the plurality of route requests;
updating the trained machine model during the third period of time to generate an updated set of routes and updated distribution metric based at least in part on the additional metrics; and
maintaining a load distribution based at least in part on the updated set of routes and the updated distribution metric including reassigning the percentage of the load to be handled by individual routes of the updated set of routes based at least in part on the updated distribution metric.

### Claim 3 (depends on 1)

The method according to claim 1, further comprising:
iteratively determining an optimized load distribution based at least in part on the trained machine model and the set of metrics; and
preconfiguring said optimized load distribution of said plurality of route requests regarding a number of said set of routes and the percentage of the load in said data network transferred per the individual routes.

### Claim 4 (depends on 3)

The method according to claim 3, further comprising:
dynamically modifying preconfigured optimized load distribution of said plurality of route requests at predetermined time intervals via human intervention.

### Claim 5 (depends on 1)

The method according to claim 1, wherein collecting the set of metrics further comprises culling data traffic patterns from websites that monitor a dynamic flow of how the data traffic patterns change over a period of time.

### Claim 6 (depends on 1)

The method according to claim 1, wherein said data network comprises a virtual flow network.

### Claim 7 (depends on 1)

The method according to claim 1, wherein said data network comprises a controlled weighted flow network.

### Claim 8 (depends on 1)

The method according to claim 1, wherein said data network comprises a plurality of nodes and edges.

### Claim 9 (depends on 8)

The method according to claim 8, wherein said plurality of nodes and edges are selected from a list consisting of: physical entities and virtual entities.

### Claim 10 (depends on 1)

The method according to claim 1, wherein said plurality of metrics are represented by a plurality of weights.

### Claim 11 (depends on 1)

The method according to claim 1, wherein said plurality of metrics comprises a plurality of derived metrics.

### Claim 12 (depends on 1)

The method according to claim 1, wherein said plurality of metrics are stored in a data structure.

### Claim 13 (depends on 12)

The method according to claim 12, wherein said data structure is selected from a list consisting of: a hash table; a database; a linked list; and an array.

### Claim 14 (depends on 1)

The method according to claim 1, wherein said procedure of determining the set of routes out of possible routes is executed using a complex shortest path algorithm.

### Claim 15 (depends on 14)

The method according to claim 14, wherein said complex shortest path algorithm is a K-shortest path routing algorithm.

### Claim 16 (depends on 15)

The method according to claim 15, wherein said K-shortest path routing algorithm is selected from a list consisting of: an extension to Dijkstra's algorithm; an extension to a Bellman Ford algorithm; and Yen's algorithm for K-shortest loopless paths.

### Claim 17 (depends on 1)

The method according to claim 1, wherein said procedures of collecting, learning, receiving, processing, selecting, updating and maintaining are executed iteratively to determine a plurality of sub-routes in a sub-network of said data network.

### Claim 18 (depends on 1)

The method according to claim 1, wherein said procedures of collecting, learning, receiving, processing, selecting, updating and maintaining are executed recursively to determine a plurality of sub-routes in a sub-network of said data network.

### Claim 19 (depends on 3)

The method according to claim 3, wherein said preconfigured optimized load distribution and said percentage is hardwired.

### Claim 20

Apparatus for load balancing a data network, comprising:
a metric collector, for collecting training data, during a first period of time, that includes a plurality of metrics in said data network including collecting at least a capacity of data transfer, a latency, a round trip time, and an actual monetary cost for using a particular connection, for connections between nodes of the data network, the data network including a plurality of network servers connected over a network and operable to route requests, the plurality of network servers corresponding to nodes;
the plurality of network servers comprising operative to train a machine model using the training data to generate a trained machine model operable to process metric data associated with the data network to determine a set of routes out of a plurality of routes to transfer data between nodes of the data network and a distribution metric to assign a percentage of a load to individual routes of the set of routes;
a route request receiver, for receiving a plurality of route requests to transfer data from a first node of the nodes to a second node of the nodes; and
a processor, for:
obtaining, during a second period of time, a set of metrics associated with the data network, the second period of time being different from the first period of time and occurring when receiving the plurality of route requests;
processing the set of metrics with the trained machine model during the second period of time to determine a set of routes from said first node to said second node and a distribution metric for the set of routes;
assigning a percentage of a load to be handled by individual routes of the set of routes based at least in part on the distribution metric;
transmitting information as computer readable instructions to control transfer and movement of the plurality of route requests over the set of routes between the first node and the second node based on the percentage of the load assigned to the individual routes.
