US 11,706,078 B1
Internet disruption detection
Chae A. Clark, Arlington, VA (US); Robert P. Gove, Jr., Reston, VA (US); Violet Lingenfelter, Arlington, VA (US); and Anthony Wong, Arlington, VA (US)
Assigned to Two Six Labs, LLC, Arlington, VA (US)
Filed by Two Six Labs, LLC, Arlington, VA (US)
Filed on Mar. 22, 2021, as Appl. No. 17/207,985.
Int. Cl. H04L 41/06 (2022.01); H04L 41/0677 (2022.01); H04L 41/12 (2022.01); H04L 43/50 (2022.01)
CPC H04L 41/06 (2013.01) [H04L 41/0677 (2013.01); H04L 41/12 (2013.01); H04L 43/50 (2013.01)] 19 Claims
OG exemplary drawing
 
1. A method of identifying and reporting network anomalies, comprising:
receiving a plurality of routing path messages of advertised routes in a network, the routing path messages indicative of available network paths between network entities;
identifying an anomaly in an available network path indicative of a change to an advertised route, the anomaly causing network traffic between autonomous systems (AS) to follow a different path;
aggregating the plurality of routing path messages defining an anomaly received during a plurality of time intervals;
comparing the anomalies in each time interval of the plurality of time intervals to the anomaly in the others of the plurality of time intervals by:
building a node structure defining an isolation forest representative of the paths depicted in the routing path messages;
traversing the node structure; and
concluding, based on the comparison, whether a difference in the anomalies is indicative of a network disruption by:
defining a plurality of decision trees based on routing paths indicative of a sequence of autonomous systems for satisfying a plurality of routes to a destination;
splitting the decision trees until each respective decision tree has a terminal leaf node; and
identifying, based on an average number of splits until a terminal leaf node is attained, whether the decision tree is indicative of an anomaly.