Analysis of arbitrary wireless network data using matched filters

• US 20070243864A1
• Filed: 04/13/2006
• Published: 10/18/2007
• Est. Priority Date: 04/13/2006
• Status: Active Grant

• Current Assignee: AT&T Mobility IP LLC (AT&T, Inc.)
• Original Assignee: Carrier Iq
• Inventors: Jaquet, James
• Granted Patent: US 7,764,959 B2
• Time in Patent Office: Days
• US Class Current: 455/423
• CPC Class Codes: H04W 24/08 Testing, supervising or monitoring using real traffic

Abstract

Analyzing an arbitrary data set to determine the presence or absence of a defined signature element, in order to understand an event or detect a root cause of an event in a wireless network. Signatures are generated and include signature elements. The signature elements in a given signature are associated with a coverage model. A data set is processed and values that correspond to each signature element are computed from the data set. The probability of detection for a given signature can then be computed based on the values computed from the data set, the coverage model and a statistical factor included in the signature. Different events typically have different signature elements in the associated signature.

Claims

1. A method for analyzing network data using matched filters, the method comprising:

obtaining a data set from a source;

computing one or more values from the data set, each value corresponding to a particular signature element;

evaluating the one or more values based on one or more signature elements included in a particular matched filter; and

computing a probability of detection for the particular matched filter based on the one or more signature elements.

2. A method as defined in claim 1, wherein computing a probability of detection further comprises computing the probability of detection based on posterior probabilities.

3. A method as defined in claim 2, wherein computing the probability of detection based on posterior probabilities further comprises identifying one or more conditions for the particular matched filter

4. A method as defined in claim 1, wherein obtaining a data set from a source further comprises obtaining a data set from one or more of a wireless device, a server log, network log, or application stream.

5. A method as defined in claim 1, wherein computing one or more values from the data set further comprises identifying the one or more signature elements included in the particular matched filter.

6. A method as defined in claim 1, wherein evaluating the one or more values based on one or more signature elements further comprises providing a coverage weighting factor for each signature element.

7. A method as defined in claim 1, wherein evaluating the one or more values based on one or more signature elements further comprises providing a normalized weighting factor for each signature element.

8. A method as defined in claim 1, wherein evaluating the one or more values based on one or more signature elements further comprises identifying a contribution to the probability of detection for each value computed from the data set.

9. A method as defined in claim 1, wherein evaluating the one or more values based on one or more signature elements further comprises identifying a match in a coverage model for each value.

10. A method as defined in claim 1, wherein evaluating the one or more values based on one or more signature elements further comprises providing a graph plotting one or more of the signature elements over a usage of the wireless device.

11. A method for analyzing a data set to determine a root cause of an event in a wireless network, the method comprising:

identifying one or more signature elements to be included in a signature for a particular event;

establishing a coverage model for each of the one or more signature elements;

computing a value for at least one of the signature elements using a data set obtained from a source;

identifying a range in the coverage model for each value; and

generating a probability of detection for the particular event based at least on each value and on each range associated with each value.

12. A method as defined in claim 11, wherein identifying one or more signature elements to be included in a signature for a particular event further comprises one or more of:

adding new signature elements to the signature;

or

removing signature elements from the signature.

13. A method as defined in claim 11, wherein the particular event is one or more of:

a dropped call;

a blocked call;

a neighbor list problem;

an unpromoted pilot problem;

a coverage problem;

a sector (PN) overshoot problem;

a hard hand-down problem;

an incomplete handoff;

a weak RF condition;

a device entering or exiting roaming conditions;

a device in the vicinity of a switch boundary;

a device in the vicinity of a frequency boundary;

a device in the vicinity of a coverage boundary;

a call initiated by the user;

a call initiated by the network;

a call initiated by a specific application on a device;

a call aborted by the user due to poor service;

mobile too far from base station to initiate call;

evaluation of the RF performance of the device;

statistics on network access attempts;

data transfer rates, in aggregate or by application;

voice quality;

time spent in soft-handoff;

impact area of PCF boundaries;

movement of a device within the wireless network;

determining if the device is inside a building;

identifying usage of the device;

determining if metrics are correctly submitted;

or

user visible errors.

14. A method as defined in claim 11, wherein establishing a coverage model for each of the one or more signature elements further comprises establishing one or more ranges in the coverage model for each of the one or more signature elements.

15. A method as defined in claim 14, wherein identifying one or more signature elements to be included in a signature for a particular event further comprises:

defining a coverage weighting factor for each of the one or more signature elements;

defining a normalized weighting factor for each of the one or more signature elements;

computing a contribution to the probability of detection for each of the one or more signature elements based on the value for the at least one of the signature elements.

16. A method as defined in claim 15, further comprising determining the contribution based on one or more of the range in the coverage model associated with the value computed from the data set, the coverage weighting factor, and the normalized weighting factor.

17. A method as defined in claim 14, further comprising tuning the signature for the particular event.

18. A method as defined in claim 14, wherein the data set is incomplete such that values are only computed for signature elements having associated data in the data set.

19. A method as defined in claim 13, wherein generating a probability of detection for the particular event based at least on each value and on each range associated with each value further comprises generating a probability of detection based on one or more conditions that occur prior to the particular event, wherein the posterior probability of the one or more conditions is include in the probability of detection.

20. A signature for analyzing a data set to identify an event in a wireless network, the signature comprising:

one or more signature elements for an event;

a coverage model for each of the one or more signature elements;

a statistical factor for each signature element;

a contribution for each signature element, the contribution based on the statistical factor; and

a probability of detection that indicates whether the event occurred in the wireless network.

21. A signature as defined in claim 20, wherein the probability of detection is a probability.

22. A signature as defined in claim 20, wherein the statistical factor comprises:

a coverage weighting factor for each signature element; and

a normalized weighting factor for each signature element.

23. A signature as defined in claim 20, wherein the event is one or more of:

a dropped call;

a blocked call;

a neighbor list problem;

an unpromoted pilot problem;

a coverage problem;

a sector (PN) overshoot problem;

a hard hand-down problem;

an incomplete handoff;

a weak RF condition;

a device entering or exiting roaming conditions;

a device in the vicinity of a switch boundary;

a device in the vicinity of a frequency boundary;

a device in the vicinity of a coverage boundary;

a call initiated by the user;

a call initiated by the network;

a call initiated by a specific application on a device;

a call aborted by the user due to poor service;

mobile too far from base station to initiate call;

evaluation of the RF performance of the device;

statistics on network access attempts;

data transfer rates, in aggregate or by application;

voice quality;

time spent in soft-handoff;

impact area of PCF boundaries;

movement of a device within the wireless network;

determining if the device is inside a building;

identifying usage of the device;

determining if metrics are correctly submitted;

or

user visible errors.

24. A signature as defined in claim 20, wherein the one or signature elements comprise one or more of:

an average RSSI;

an average RSSI for a given time period;

a slope of an RSSI average for a given time period;

a frame error rate;

a frame error rate for a given time period;

a slope of a frame error rate for a given time period;

an average of transmit power;

an average of transmit power for a given period of time;

an average of transmitter gain state;

an average of transmitter gain state for a given period of time;

a statistic on protocol message acknowledgement;

a change in frequency used by the device;

the geo-location of the device;

a statistic based on the geo-location of the device;

or

a statistic on the identity and strength of sectors visible to the device.