Media session identification method for IP networks

1. Electronic network apparatus for identifying media sessions over IP and network packets as belonging or not belonging to a media session, comprising:

a packet characteristic extraction unit configured for obtaining traffic packet characteristics from a succession of packets, said packets respectively belonging to various unknown sessions;

a grouping unit associated with said packet characteristic extraction unit, configured to compare packets with other packets according to said extracted traffic characteristics to group together those packets whose extracted packet traffic characteristics are similar as probably belonging to a common session, and

a session analyzer configured for analyzing said grouped packets for session characteristics, therewith to identify a session type of said probable common session.

2. The apparatus of claim 1 , further comprising a resource allocator for allocating at least one member of the group consisting of resources, provisioning features and control features to said session.

3. The apparatus of claim 1 , wherein said session type identification comprises identifying a session managing entity generating or managing said session.

4. The apparatus of claim 1 , wherein said session analyzer comprises a neural network, using characteristics automatically built into said neural network during a neural network training phase.

5. Apparatus according to claim 1 , further comprising a thresholder for thresholding said traffic characteristics, and a packet session identifier configured to identify those packets whose characteristics are mutually within said thresholds for said grouping unit to group said packets as belonging to said common session.

6. The apparatus of claim 1 , wherein said traffic characteristics comprise at least one member of the group consisting of packet length, inter-arrival period, and average bandwidth.

7. The apparatus of claim 1 , wherein said grouping unit comprises a packet marker, associated with said packet session identifier, configured for inserting into a header of a respective packet an indication of said common session with which said respective packet has been identified.

8. The apparatus of claim 7 , further comprising a session analyzer associated with said packet marker, configured to obtain packets marked as belonging to said common session and to determine, from said packets, characteristics belonging to said common session, therefrom to identify a session type.

9. The apparatus of claim 8 , wherein said session analyzer is configured to obtain said characteristics belonging to said common session from parameter patterns common to said packets.

10. The apparatus of claim 8 , wherein said session analyzer is configured to use at least one member of the group consisting of:

packet length

inter-arrival period

variance of inter-arrival period

variance of length

covariance of inter-arrival period

covariance of length

a statistical derivative obtained from multi records of packet length

a statistical derivative obtained from multi records of inter-arrival period

a histogram built of at least one of the following;

i. packet length

ii. inter-arrival period

iii. variance of inter-arrival period

iv. variance of length

v. covariance of inter-arrival period

vi. covariance of length

vii. a statistical derivative obtained from multi records of packet length

viii. a statistical derivative obtained from multi records of inter-arrival period

a matrix histogram built of at least one of the following;

i. packet length

ii. inter-arrival period

iii. variance of inter-arrival period

iv. variance of length

v. covariance of inter-arrival period

vi. covariance of length

vii. a statistical derivative obtained from multi records of packet length

viii. a statistical derivative obtained from multi records of inter-arrival period.

11. The apparatus of claim 8 , wherein said session analyzer comprises a neural network, using characteristics automatically built into said neural network during a neural network training phase.

12. The apparatus of claim 1 , further comprising a resource allocator configured to allocate resources on a network according to said identified common session.

13. The apparatus of claim 8 , further comprising a resource allocator, configured to allocate resources on a network according to said identified session type.

14. The apparatus of claim 1 , wherein at least one of said packet session analyzer, said extraction unit and said grouping unit comprises a neural network.

15. A method of identifying session type, comprising

obtaining passing packets of respectively unknown sessions and unknown session types;

obtaining traffic packet characteristics of said passing packets of respectively unknown session types;

comparing said obtained packets with each other using respectively obtained traffic packet characteristics;

grouping together those packets having similar values of said traffic packet characteristics into a presumed session; and

analyzing said grouped packets of said presumed session for session characteristic;

using said session characteristics to identify a session type of said presumed session.

16. The method of claim 15 wherein said identifying a session type comprises identifying a session managing entity generating or managing said session.

17. The method of claim 15 wherein said identifying a session type is followed by at least one of the group comprising: providing provisioning and providing control actions to the session.

18. Method according to claim 15 , further comprising thresholding said traffic characteristics, and identifying those packets whose characteristics are mutually within said thresholds for said grouping.

19. The method of claim 15 , wherein said characteristics comprise at least one member of the group consisting of packet length, inter-arrival period, and average bandwidth.

20. The method of claim 15 , further comprising inserting into a header of a respective packet an indication of said presumed session with which said respective packet has been identified.

21. The method of claim 15 , comprising obtaining said characteristics belonging to said presumed session from parameter patterns common to said packets.

22. The method of claim 15 , comprising using at least one member of the group consisting of:

packet length

inter-arrival period

variance of inter-arrival period

variance of length

covariance of inter-arrival period

covariance of length

a statistical derivative obtained from multi records of packet length

a statistical derivative obtained from multi records of inter-arrival period

a histogram built of at least one of the following;

i. packet length

ii. inter-arrival period

iii. variance of inter-arrival period

iv. variance of length

v. covariance of inter-arrival period

vi. covariance of length

vii. a statistical derivative obtained from multi records of packet length

viii. a statistical derivative obtained from multi records of inter-arrival period

a matrix histogram built of at least one of the following;

i. packet length

ii. inter-arrival period

iii. variance of inter-arrival period

iv. variance of length

v. covariance of inter-arrival period

vi. covariance of length

vii. a statistical derivative obtained from multi records of packet length

viii. a statistical derivative obtained from multi records of inter-arrival period.

23. The method of claim 15 , comprising using a neural network and characteristics automatically built into said neural network during a neural network training phase.

24. The method of claim 16 , further comprising using a neural network and characteristics automatically built into said neural network during a neural network training phase.