Method and system for profiling teleconference session quality in communication networks

1. A method for profiling teleconference session quality, comprising:

receiving a network traffic flow from a network device connected to a network, wherein the network traffic flow is received during normal operation of the network;

determining, in response to receiving the network traffic flow, whether the network traffic flow is associated with a teleconference session based on a source or a destination of one or more packet data units of the network traffic flow being associated with a known teleconference service;

inspecting, in response to determining that the network traffic flow is associated with the teleconference session, a first teleconference session segment of the teleconference session to obtain a first feature tuple, wherein the first feature tuple is generated using information included in a header of one or more network traffic included in the network traffic flow, and wherein the network device is in a path of the teleconference session between the source and the destination of the network traffic flow, and wherein the first feature tuple is a collection comprising a plurality of values corresponding to different features of the teleconference session pertinent to determining a session quality of the teleconference session;

processing the first feature tuple using a classification learning model to obtain a first quality classification of a plurality of configured quality classifications, wherein the processing of the first feature tuple is based on using only the first feature tuple and without being provided an explicit indication of the session quality of the teleconference session, and wherein the first quality classification provides a single label summarizing the session quality of the teleconference session; and

providing, to an administrator device separate from the source device and the destination device, overall teleconference session metrics for the teleconference session reflecting the session quality of teleconference session based at least the processing of the first feature tuple, wherein the overall teleconference session metrics comprise the first quality classification.

2. The method of claim 1 , wherein

the features of the teleconference session comprises at least one selected from a group consisting of a network traffic bit rate, a network traffic packet rate, a network traffic bit rate jitter, a network traffic packet rate jitter, and a transport layer protocol.

3. The method of claim 1 , wherein the classification learning model is a support vector machine.

4. The method of claim 1 , further comprising:

prior to providing the overall teleconference session metrics to the device:

aggregating a set of quality classifications, obtained throughout the teleconference session, comprising the first quality classification; and

deriving a negative performance index for the teleconference session based on the set of quality classifications,

wherein the overall teleconference session metrics comprise the negative performance index.

5. The method of claim 4 , wherein the overall teleconference session metrics further comprise a root cause analysis result indicative of a root cause for the negative performance index.

6. The method of claim 4 , further comprising:

prior to aggregating the set of quality classifications:

inspecting a second teleconference session segment of the teleconference session to obtain a second feature tuple; and

processing the second feature tuple using the classification learning model to obtain a second quality classification,

wherein when aggregated, the set of quality classifications further comprises the second quality classification.

7. The method of claim 5 , wherein the root cause analysis result is obtained by:

mapping a plurality of network traffic flow directionalities of the network traffic flow to the set of quality classifications;

making a first identification, based on the mapping, of a portion of the network traffic flow directionalities associated with the negative performance index, wherein the portion of the network traffic flow directionalities is less than an entirety of the plurality of network traffic flow directionalities and includes only network traffic flow directionalities, among the plurality of network traffic flow directionalities, labeled with the negative performance index; and

making a second identification, based on the first identification, of a mode of the portion of the network traffic flow directionalities,

wherein the mode is one or more communications network connection types comprising a wired portion of a communications network and a wireless portion of the communications network.

8. The method of claim 5 , wherein

the root cause analysis result associates one or more communications network connection types with the root cause for the negative performance index, and

the one or more communications network connection types comprises a wired portion and a wireless portion.

9. A system, comprising:

an administrator device comprising a first computer processor; and

a teleconference analytics service operatively connected to the administrator device, and comprising a second computer processor programmed to:

receive a network traffic flow from a network device connected to the teleconference analytics service, wherein the network traffic flow is received during normal operation of the network;

determine, in response to receiving the network traffic flow, whether the network traffic flow is associated with a teleconference session based on a source or a destination of one or more packet data units of the network traffic flow being associated with a known teleconference service;

obtain, in response to determining that the network traffic flow is associated with the teleconference session, a feature tuple for a teleconference session segment of the teleconference session, wherein the feature tuple is generated using information included in a header of one or more network traffic included in the network traffic flow, and wherein the network device is in a path of the teleconference session between the source and the destination of the network traffic flow, and wherein the feature tuple is a collection comprising a plurality of values corresponding to different features of the teleconference session pertinent to determining a session quality of the teleconference session;

process the feature tuple using a classification learning model to obtain a quality classification of a plurality of configured quality classifications, wherein the processing of the feature tuple is based on using only the feature tuple and without being provided an explicit indication of the session quality of the teleconference session, and wherein the quality classification provides a single label summarizing the session quality of the teleconference session; and

provide, to the administrator device, overall teleconference session metrics for the teleconference session reflecting the session quality of teleconference session based at least on the processing of the feature tuple, wherein the overall teleconference session metrics comprise the quality classification, and the administrator device is separate from the source device and the destination device.

10. The system of claim 9 , further comprising:

a sampling network device that operatively connects to the teleconference analytics service or comprises at least a portion of the teleconference analytics service, and is programmed to inspect the teleconference session segment of the teleconference session to obtain the feature tuple.

11. The system of claim 10 , further comprising:

a teleconference service operatively connected to the sampling network device, and comprising a third computer processor programmed to manage real-time exchange of information during the teleconference session.

12. The system of claim 11 , further comprising:

a communications network comprising the teleconference service and the sampling network device.

13. The system of claim 12 , further comprising:

a pair of client devices operatively connected to one another through the communications network,

wherein communications exchanged between the pair of client devices comprise the network traffic flow for the teleconference session.

14. The system of claim 10 , wherein the sampling network device is a switch, a router, or a multilayer switch.

15. A non-transitory computer readable medium (CRM) comprising computer readable program code, which when executed by a computer processor, enables the computer processor to:

receive a network traffic flow from a network device on a network, wherein the network traffic flow is received during normal operation of the network;

determine, in response to receiving the network traffic flow, whether the network traffic flow is associated with a teleconference session based on a source or a destination of one or more packet data units of the network traffic flow being associated with a known teleconference service;

inspect, in response to determining that the network traffic flow is associated with the teleconference session, a first teleconference session segment of the teleconference session to obtain a first feature tuple, wherein the first feature tuple is generated using information included in a header of one or more network traffic included in the network traffic flow, and wherein the network device is in a path of the teleconference session between the source and the destination of the network traffic flow, and wherein the first feature tuple is a collection comprising a plurality of values corresponding to different features of the teleconference session pertinent to determining a session quality of the teleconference session;

process the first feature tuple using a classification learning model to obtain a first quality classification of a plurality of configured quality classifications, wherein the processing of the first feature tuple is based on using only the first feature tuple and without being provided an explicit indication of the session quality of the teleconference session, and wherein the first quality classification provides a single label summarizing the session quality of the teleconference session; and

provide, to an administrator device separate from the source device and the destination device, overall teleconference session metrics for the teleconference session reflecting the session quality of teleconference session based at least on the processing of the first feature tuple, wherein the overall teleconference session metrics comprise the first quality classification.

16. The non-transitory CRM of claim 15 , wherein the features of the teleconference session comprises at least one selected from a group consisting of a network traffic bit rate, a network traffic packet rate, a network traffic bit rate jitter, a network traffic packet rate jitter, and a transport layer protocol.

17. The non-transitory CRM of claim 15 , comprising computer readable program code, which when executed by the computer processor, further enables the computer processor to:

prior to providing the overall teleconference session metrics to the device:

aggregate a set of quality classifications, obtained throughout the teleconference session, comprising the first quality classification; and

derive a negative performance index for the teleconference session based on the set of quality classifications,

wherein the overall teleconference session metrics comprise the negative performance index.

18. The non-transitory CRM of claim 17 , wherein the overall teleconference session metrics further comprise a root cause analysis result indicative of a root cause for the negative performance index.

19. The non-transitory CRM of claim 17 , comprising computer readable program code, which when executed by the computer processor, further enables the computer processor to:

prior to aggregating the set of quality classifications:

inspect a second teleconference session segment of the teleconference session to obtain a second feature tuple; and

process the second feature tuple using the classification learning model to obtain a second quality classification,

wherein when aggregated, the set of quality classifications further comprises the second quality classification.