Abstract: A method of homogenizing expected data-harvest yields, the method comprising: selecting a population of data-sources; selecting a quantity of data as a data-harvest maximum, Hmax; and dividing the population into a set of groups according to the data-harvest maximum Hmax such that: each group includes at least one data-source; and each group generates an expected data-harvest yield, Hyield, that is less than or equal to the data-harvest maximum Hmax, Hyield?Hmax.
Type:
Grant
Filed:
September 17, 2015
Date of Patent:
November 8, 2016
Assignee:
GLOBAL TECHNOLOGY ASSOCIATES, LLC
Inventors:
Sandip Sandhu, Behrouz Heshmatipour, Mark Roger Patterson, Vincent Carl Star
Abstract: A method of identifying idiosyncratic performance-outliers amongst a plurality of cell-sites in a wireless network, the method comprising: selecting cell-sites; selecting, for each cell-site, at least one sector-carrier; selecting, for each sector-carrier, one or more KPIs; determining, for each KPI, a historical value and a recent value based on samples of the KPI taken during a historical period and a recent period, respectively, the recent period being a subset of the historical period; determining, for each KPI, a delta representing a difference between the historical value and the recent value thereby determining KPI-specific sets of deltas, respectively; normalizing each set of deltas; selecting, for each set of deltas, a KPI-specific threshold; comparing, for each set of deltas, members thereof against the corresponding threshold; and identifying, for each set of deltas, which combinations of cell-site and sector-carrier are KPI-specific performance outliers based on the comparison, respectively.