US 12,278,831 B2
System and method for real time distributed adaptive cube aggregation, heuristics based hierarchical clustering and anomaly detection framework for high volume streaming datasets
Arun Kumar Krishna, Bangalore (IN); and Pramod Konandur Prabhakar, Bangalore (IN)
Assigned to PELATRO PTE. LTD., Singapore (SG)
Filed by Pelatro Pte. Ltd., Singapore (SG)
Filed on Dec. 30, 2022, as Appl. No. 18/148,594.
Claims priority of provisional application 63/295,084, filed on Dec. 30, 2021.
Prior Publication US 2023/0216874 A1, Jul. 6, 2023
Int. Cl. H04L 9/40 (2022.01); G06F 16/22 (2019.01); G06F 16/2455 (2019.01)
CPC H04L 63/1425 (2013.01) [G06F 16/2264 (2019.01); G06F 16/24556 (2019.01); G06F 16/24568 (2019.01)] 18 Claims
OG exemplary drawing
 
1. A method for aggregating an incoming data stream containing a telecommunication subscriber activity data at a computing device having a processor, a non-transitory computer readable medium, and a memory, the computing device capable of a deep packet inspection (DPI) during consumption by the computing device, the method comprising the steps of:
establishing the incoming data stream to the computing device;
parsing on-the-fly said DPI data on a per-subscriber basis based into a plurality of dimensions, said plurality of dimensions correspond to a plurality of subscriber session attributes and a plurality of values corresponding to said plurality of subscriber session attributes;
separating on-the-fly a plurality of volatile datasets of said plurality of subscriber session attributes, a plurality of semi-volatile datasets of said plurality of subscriber session attributes, and a plurality of non-volatile datasets of said plurality of subscriber session attributes into a plurality of distinct memory bins for each plurality of datasets;
expanding an at least one of said plurality of distinct memory bins when said at least one of said plurality of distinct memory bins is consumed;
generating an in-memory hypercube having a plurality of hypercube dimensions each corresponding to said plurality of dimensions;
numerically translating an input query into an access query for said in-memory hypercube, said input query comprising an at least one subscriber attribute; and
returning a list of subscribers in response to said input query;
wherein said plurality of subscriber session attributes comprise a download volume, an upload volume, a traffic type, a session duration, a geolocation, and a quality of service.