US 12,438,796 B2
Device type identification method and apparatus, computer device, and storage medium
Zhicheng Dong, Shenzhen (CN); and Chenchen Peng, Shenzhen (CN)
Assigned to Tencent Technology (Shenzhen) Company Limited, Shenzhen (CN)
Filed by Tencent Technology (Shenzhen) Company Limited, Shenzhen (CN)
Filed on Nov. 18, 2022, as Appl. No. 17/990,553.
Application 17/990,553 is a continuation of application No. PCT/CN2021/120676, filed on Sep. 26, 2021.
Claims priority of application No. 202011106539.7 (CN), filed on Oct. 16, 2020.
Prior Publication US 2023/0083404 A1, Mar. 16, 2023
Int. Cl. G06F 15/16 (2006.01); H04L 9/40 (2022.01); H04L 43/10 (2022.01); H04L 43/12 (2022.01)
CPC H04L 43/10 (2013.01) [H04L 43/12 (2013.01); H04L 63/1433 (2013.01)] 17 Claims
OG exemplary drawing
 
1. A device type identification method performed by a computer device, the method comprising:
transmitting a port open/closed state probe packet to at least two target devices;
receiving port open/closed state information returned by the at least two target devices in response to the port open/closed state probe packet, wherein the port open/closed state information comprises port open/closed state vectors indicating whether each port of the at least two target devices is open or closed;
composing the port open/closed state vectors of the at least two target devices into a port state matrix;
determining a mean of data in each row or column of the port state matrix;
calculating a difference between the data in each row or column of the port state matrix and a corresponding mean, to obtain a difference matrix;
determining an eigenvalue of the difference matrix and an eigenvector corresponding to the eigenvalue;
extracting a sub-eigenmatrix from an eigenmatrix composed based on the eigenvector according to a target dimension, and using the sub-eigenmatrix as the port open/closed state feature;
performing decision classification on the port open/closed state feature through a decision tree classifier; and
determining a device type of each of the at least two target devices according to a result obtained from the decision classification.