# US 12,192,806 B2 · Predictive modeling of energy consumption in a cellular network > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 12,192,806 B2 - Document ID: US12192806B2 - Status: Granted - Kind: B2 - Application number: 17/531,847 - Filed: 2021-11-22 - Published: 2023-05-25 - Granted: 2025-01-07 - Earliest claimed: 2021-11-22 - Assignees: Verizon Patent and Licensing Inc. - Applicants: Verizon Patent and Licensing Inc. - Inventors: Harish Arunachalam; Neil Singh Sethi; Martin Konkel; Robert Belson; Patrick Wagstrom - Web page: https://patentgrep.com/patent/US12192806B2 ## CPC - H04W 24/08: Testing, {supervising or monitoring} using real traffic - H04W 24/02: Arrangements for optimising operational condition - H04W 24/04: Arrangements for maintaining operational condition - H04W 84/042: {Public Land Mobile systems, e.g. cellular systems} - H04W: WIRELESS COMMUNICATION NETWORKS (broadcast communication H04H; communication systems using wireless links for non-selective communication, e.g. wireless extensions H04M1/72) ## Abstract A predictive modeling approach to managing cellular network infrastructure is disclosed. In an embodiment, a method can include receiving raw data from a plurality of data sources populated while operating a cellular network. The method can then generate per-logical cell site data by normalizing the raw data based on a set of LCSs in the cellular network to generate per-LCS data. The method can then generate an example from the per-LCS data and generate a predicted energy consumption value for the given LCS by inputting the example into a predictive model (e.g., a decision tree-based model, such as an XGBoost model). From this output, the method can determine if the predicted energy consumption value is higher than an expected energy consumption value (e.g., a historical range of consumption). If so, the method can then label the given LCS as an outlier. ## Claims ### Claim 1 A method comprising: receiving raw data from a plurality of data sources, the raw data collected while operating a cellular network; normalizing the raw data based on a set of logical cell sites (LCSs) in the cellular network to generate per-LCS data; generating an example from the per-LCS data, the example associated with a given LCS in the set of LCSs; generating a predicted energy consumption value for the given LCS by inputting the example into a predictive model; determining that the predicted energy consumption value is higher than an expected energy consumption value associated with the given LCS; and labeling the given LCS as an outlier. ### Claim 2 (depends on 1) The method of claim 1, wherein receiving raw data comprises receiving data comprising one or more of network consumption data, network equipment data, non-network equipment data, and external data. ### Claim 3 (depends on 2) The method of claim 2, wherein the network consumption data comprises one or more of cellular traffic volume, number of active connections, total number of carriers, and coverage radius of a given LCS; the external data comprises one or more of a location, temperature, precipitation, wind, gust, heating degree days, cooling degree days, and nearest weather station of a given LCS; the network equipment data comprises data related to one or more of radios, antennas, and baseband units of a given LCS; and the non-network equipment data comprises data related to one or more of heating, ventilation, and air conditioning (HVAC) units and batteries in a given LCS. ### Claim 4 (depends on 1) The method of claim 1, wherein normalizing the raw data based on a set of LCSs in the cellular network to generate per-LCS data comprises aggregating raw data associated with finer granularity to a given LCS based on a mapping of finer grained equipment to the given LCS. ### Claim 5 (depends on 1) The method of claim 1, wherein inputting the example into a predictive model comprises inputting the example into a decision tree-based model. ### Claim 6 (depends on 1) The method of claim 1, further comprising: retrieving a transmission loss for the given LCS; adding the transmission loss to the example to generate an augmented example; and predicting a likelihood that the given LCS includes an identified item of network equipment by inserting the augmented example into a second predictive model. ### Claim 7 (depends on 6) The method of claim 6 wherein inserting the augmented example into a second predictive model comprises inserting the augmented example into an ensemble model, the ensemble model including a sequential combination of machine learning models selected from the group consisting of a gradient boosting model, support vector machine model, neural network model, adaptive boosting, and logistic regression model. ### Claim 8 A non-transitory computer-readable storage medium for tangibly storing computer program instructions capable of being executed by a computer processor, the computer program instructions defining steps of: receiving raw data from a plurality of data sources, the raw data collected while operating a cellular network; normalizing the raw data based on a set of logical cell sites (LCSs) in the cellular network to generate per-LCS data; generating an example from the per-LCS data, the example associated with a given LCS in the set of LCSs; generating a predicted energy consumption value for the given LCS by inputting the example into a predictive model; determining that the predicted energy consumption value is higher than an expected energy consumption value associated with the given LCS; and labeling the given LCS as an outlier. ### Claim 9 (depends on 8) The non-transitory computer-readable storage medium of claim 8, wherein receiving raw data comprises receiving data comprising one or more of network consumption data, network equipment data, non-network equipment data, and external data. ### Claim 10 (depends on 9) The non-transitory computer-readable storage medium of claim 9, wherein the network consumption data comprises one or more of cellular traffic volume, number of active connections, total number of carriers, and coverage radius of a given LCS; the external data comprises one or more of a location, temperature, precipitation, wind, gust, heating degree days, cooling degree days, and nearest weather station of a given LCS; the network equipment data comprises data related to one or more of radios, antennas, and baseband units of a given LCS; and the non-network equipment data comprises data related to one or more of heating, ventilation, and air conditioning (HVAC) units and batteries in a given LCS. ### Claim 11 (depends on 8) The non-transitory computer-readable storage medium of claim 8, wherein normalizing the raw data based on a set of LCSs in the cellular network to generate per-LCS data comprises aggregating raw data associated with finer granularity to a given LCS based on a mapping of finer grained equipment to the given LCS. ### Claim 12 (depends on 8) The non-transitory computer-readable storage medium of claim 8, wherein inputting the example into a predictive model comprises inputting the example into a decision tree-based model. ### Claim 13 (depends on 8) The non-transitory computer-readable storage medium of claim 8, the steps further comprising: retrieving a transmission loss for the given LCS; adding the transmission loss to the example to generate an augmented example; and predicting a likelihood that the given LCS includes an identified item of network equipment by inserting the augmented example into a second predictive model. ### Claim 14 (depends on 13) The non-transitory computer-readable storage medium of claim 13 wherein inserting the augmented example into a second predictive model comprises inserting the augmented example into an ensemble model, the ensemble model including a sequential combination of machine learning models selected from the group consisting of a gradient boosting model, support vector machine model, neural network model, adaptive boosting, and logistic regression model. ### Claim 15 An apparatus comprising: a processor configured to: receive raw data from a plurality of data sources, the raw data collected while operating a cellular network; normalize the raw data based on a set of logical cell sites (LCSs) in the cellular network to generate per-LCS data; generate an example from the per-LCS data, the example associated with a given LCS in the set of LCSs; generate a predicted energy consumption value for the given LCS by inputting the example into a predictive model; determine that the predicted energy consumption value is higher than an expected energy consumption value associated with the given LCS; and label the given LCS as an outlier. ### Claim 16 (depends on 15) The apparatus of claim 15, wherein receiving raw data comprises receiving data comprising one or more of network consumption data, network equipment data, non-network equipment data, and external data. ### Claim 17 (depends on 15) The apparatus of claim 15, wherein normalizing the raw data based on a set of LCSs in the cellular network to generate per-LCS data comprises aggregating raw data associated with finer granularity to a given LCS based on a mapping of finer grained equipment to the given LCS. ### Claim 18 (depends on 15) The apparatus of claim 15, wherein inputting the example into a predictive model comprises inputting the example into a decision tree-based model. ### Claim 19 (depends on 15) The apparatus of claim 15, the processor further configured to: retrieve a transmission loss for the given LCS; add the transmission loss to the example to generate an augmented example; and predict a likelihood that the given LCS includes an identified item of network equipment by inserting the augmented example into a second predictive model. ### Claim 20 (depends on 19) The apparatus of claim 19 wherein inserting the augmented example into a second predictive model comprises inserting the augmented example into an ensemble model, the ensemble model including a sequential combination of machine learning models selected from the group consisting of a gradient boosting model, support vector machine model, neural network model, adaptive boosting, and logistic regression model.