GB2632641A - Reserve Power Optimisation - Google Patents

Abstract

An altered configuration for a base station is generated 220 in response to a received notification 210 that the base station is switching from a utility power supply to an alternative power supply, and the base station is changed 230 from an original configuration to the altered one. The altered configuration powers down one or more radio units within the base station, and the altered configuration is generated to maintain one or more performance indicators of the base station when using the alternative power supply. A further notification 240 may be received indicating a return to the utility power supply, in response to which the configuration is changed 250 to the original one. The performance indicators may relate to cellular services provided to user equipment (UEs) by the base station, such as a minimum data throughput or a minimum number of UEs to be supplied with cellular services, which may be determined using a machine learning algorithm such as a k-nearest neighbour (KNN) regression model.

Description

Reserve Power Optimisation

Field of the Invention

The present invention relates to a method for managing the power for one or base stations, and in particular for managing how base stations operate when they switch to alternative sources of power, such as battery backup systems.

Background of the Invention

Cellular telecommunications networks use base transceiver stations (BTS) or base stations to communicate between the telecommunications network and individual user equipment (UE). The majority of the time, base stations are powered using a utility power supply such as a three-phase commercial power source. In order to maintain cellular coverage during power outages, base stations may be provided with an alternative power supply, such as secondary (rechargeable) batteries or backup generators.

When the utility power supply is available then the secondary batteries may be charged using rectifiers to convert AC to DC. The battery chargers also ensure that the secondary batteries are not overcharged and so may have voltage regulators and associated circuitry.

When a power cut occurs then circuitry may switch from the utility power supply to the alternative power supply. However, this may only continue for a limited time until the secondary batteries are depleted and the base station can no longer function.

The base station may issue one or more notifications and alerts during this process to inform the telecommunications network that a power cut has occurred and that the alternative power supply is being used. Figure 1 illustrates a schematic diagram of a timeline describing the various stages of a power cut. At the start of the power cut, a power alarm may be triggered. The voltage of the secondary batteries may be monitored, and various voltage drop alarms may be issued at certain thresholds of the DC battery voltage. These alarms may include a critical threshold alarm and the secondary batteries may be disconnected when they reach a certain DC voltage to prevent damage to their cells.

Should this occur (and before the utility power is restored) then the base station will go offline and be totally powered down. When the utility power is restored then the alarms may be cleared and service from the base station may resume. The secondary batteries may be recharged so that they may be used again for further power cuts, should they occur.

Extending the time available for powering the base station on the alternative power supply will require larger secondary batteries or backup generators. Alternatively, particular services supplied by the base station may be shut down or throttled when using the alternative power supply, but this may affect customer satisfaction. Furthermore, repeated discharging of batteries reduces their lifetime, and they may need to be replaced more often. Therefore, there is a compromise between utilising base station resources more efficiently and providing customers with telecommunications services for as long as possible operating under an alternative power supply, but these compromises are not always achievable.

CN217133557 describes saving energy by monitoring the electrical consumption of remote radio units (RRUs) and shutting them down or powering them up according to demand from mobile devices.

CN104969607 describes a control device for controlling RRUs at a granular level to save energy.

Therefore, there is required a method and system that overcomes these problems.

Summary of the Invention

A telecommunications system may have one or more performance indicators or key performance indicators (KPI) describing a minimum acceptable throughput for different cellular services provided by each base station. Each base station can contain one or more radio units that typically operate on different frequencies or radio bands. Furthermore, each radio unit or remote radio unit (RRU) may provide cellular services for different cellular network technologies (e.g., 2G, 3G, 4G, and/or 5G or beyond).

When the base station switches from a utility power supply (that can provide more than enough electrical power to power all radio units within the base station) to a utility power supply then the base station can be reconfigured so that the performance indicators of the base station are met even when the alternative power supply is used. This altered configuration powers down one or more of the radio units within the base station in such a way that the performance indicators are maintained. Therefore, fewer radio units are powered to increase the time that the base station can operate on the alternative power supply (or reduce wear of the alternative power supply) whilst maintaining the required performance (meeting the performance indicators) for the base station.

For example, if more than one of the radio units provides a particular type of cellular network technology then shutting down one of the radio units may leave another radio unit to provide required technology services meeting the performance indicator or indicators.

A machine learning model may be used to predict a number of users that may wish to use the cell serviced by the base station and so configuration of the base station can be determined at a particular time when an alternative power supply is required, whilst maintaining the performance indicators for the telecommunications network. The machine learning model may continuously calculate the predicted number users and so the new configuration (for use during a power cut) can be determined immediately.

In accordance with a first aspect there is provided a method for managing one or more base stations, each one or more base station comprising a plurality of radio units, the method comprising the steps of: receiving a notification that a base station is switching from a utility power supply to an alternative power supply; in response to the notification generating an altered configuration for the base station, wherein the altered configuration is generated to maintain one or more performance indicators of the base station when operating using the alternative power supply; and changing the base station from an original configuration to the altered configuration, wherein the altered configuration powers down one or more of the radio units within the base station. Therefore, as many radio units may be shut down when the base station is using an alternative power source as possible to increase the time that the base station can operate on this alternative power source without degrading the services provided by the base station beyond that which is acceptable. The alternative power source can be put under less stress and so prolong its life and longer power cuts can be accommodated when necessary. The method may be executed within the base station, partially within the base station or within another telecommunications component. For example, a telecommunications component may manage one or more base stations by executing the method.

Preferably, the method may further comprise the step of receiving a further notification indicating that the base station has returned to the utility power supply and in response changing the base station from the altered configuration to the original configuration. Therefore, the altered configuration may only need to be used whilst the base station operates using the alternative power supply and so maintain full service as soon as possible.

Optionally, the alternative power supply may be a secondary (rechargeable) battery or an electrical generator. Other alternative power supplies may be used. For example, this may include supercapacitors or combinations of different types of power supplies. Where a secondary battery is used, the utility power supply may also charge the battery using a charging circuit including rectifiers, capacitors and voltage and current regulators,

for example.

Preferably, the one or more performance indicators of the base station may define performance indicators for cellular services provided to a plurality of user equipment, UE, by the one or more base stations.

Optionally, the one or more performance indicators of the base station may define a minimum data throughput for cellular services provided to the plurality of UEs. For example, the one or more performance indicators may state a minimum number of UEs that can be served and/or a particular volume or rate of data.

Optionally, the performance indicators may be used to determine a minimum number of user equipment, UE, to be supplied by cellular services at each different cellular network technology by the base station. Therefore, the performance indicators may be divided into cellular technologies rather than a total number of UEs that may be served.

This allows different levels of service for each technology type to be defined.

Optionally, the different cellular network technologies comprise any two or more of: 2G, 3G, 4G, and/or 5G. These types of technologies may be added to as new technologies emerge.

Advantageously, the altered configuration may be generated by determining a maximum number of radio units to power down whilst the base station can supply the minimum number of UE with cellular services at each different cellular network technology. For example, separate radio units (e.g., operating at different frequencies or in different bands) may provide services to more than one cellular technology (e.g., 2G and 3G simultaneously). They may also have individual capacities for each cellular technology. Therefore, generating the altered configuration may determine if an entire radio unit can be shut down with the remaining radio units able to service the different cellular technologies to the extent required by the performance indicators.

Optionally, the minimum number of UEs to be supplied by cellular services at each different cellular network technology by the base station to maintain the performance indicators may be determined using a machine learning, ML, algorithm. The ML algorithm may be trained over time using data provided by the telecommunications network and different base stations. This may be used to predict the number of UEs and their service (e.g., data) requirements at particular locations and at different times and so enable the performance indicators (e.g., key performance indicators or KPIs) to be met.

Optionally, the ML algorithm may be based on a regression model. Other models may be used.

Optionally, the regression model may be a k-nearest neighbour, KNN, regressor.

Preferably, the base station configuration may be changed (e.g., to and from the altered configuration) by sending commands to an operations support system.

In accordance to a second aspect, there is provided base station comprising: a plurality of radio units; a utility power supply input; an alternative power supply; and means adapted to execute the steps of: switching from a utility power supply to an alternative power supply; transmitting a notification that the base station is switching from the utility power supply to the alternative power supply; in response to the notification, receiving an altered configuration for the base station, wherein the altered configuration maintain one or more performance indicators of the base station when operating using the alternative power supply; and changing the base station from an original configuration to the altered configuration, wherein the altered configuration powers down one or more of the plurality of radio units within the base station.

Preferably, the alternative power supply may be a secondary battery and/or an electrical generator. Other alternative power supplies may be used such as generators (e.g., petrol or diesel) that automatically start, solar panels, etc. In accordance with a third aspect there is provided a telecommunications system comprising one or more base stations described above; and a data processing system comprising means for carrying out the steps of any of the above methods. The data processing system may be located within one of the base stations or be separate to it. The data processing system may also be a distributed or cloud computing system, for example.

The methods described above may be implemented as a computer program comprising program instructions to operate a computer. The computer program may be stored on a computer-readable medium, including a non-transitory computer-readable medium.

The computer system may include a processor or processors (e.g. local, virtual or cloud-based) such as a Central Processing Unit (CPU), and/or a single or a collection of Graphics Processing Units (GPUs). The processor may execute logic in the form of a software program. The computer system may include a memory including volatile and nonvolatile storage medium. A computer-readable medium may be included to store the logic or program instructions. The different parts of the system may be connected using a network (e.g. wireless networks and wired networks). The computer system may include one or more interfaces. The computer system may contain a suitable operating system such as UNIX, Windows (RTM) or Linux, for example.

It should be noted that any feature described above may be used with any particular aspect or embodiment of the invention.

Brief description of the Fiaures

The present invention may be put into practice in a number of ways and embodiments will now be described by way of example only and with reference to the accompanying drawings, in which: Fig. 1 shows a schematic diagram of a timeline for operating a base station under various power conditions; Fig. 2 shows a flowchart of a method for managing one or more base stations; Fig. 3 shows a schematic diagram of a computer system for managing one or more base stations; Fig. 4 shows a schematic diagram of a base station and different power supplies; Fig. 5 shows a schematic diagram of radio units within the base station of the Figure 4; Fig. 6 shows a schematic diagram of further radio units of the base station of Figure 4; Fig. 7 shows a graphical representation of battery life when powering the base station of Figure 4; Fig. 8 shows a schematic diagram indicating how the base station of Figure 4 can be managed; Fig. 9 shows a graph of throughput against users of a particular cell served by the base station of Figure 4; Fig. 10 shows an example output from a machine algorithm used to manage the base station of Figure 4; Fig. 11 shows a table indicating the operation of radio units within the base station of Figure 4; Fig. 12 shows example results for operating the base station of Figure 4 according to the method of Figure 3; Fig. 13 shows a graph of the backup time available when powering the base station of Figure 4 using an alternative power source; Fig. 13A shows a flowchart of an example method used to implement machine learning within the method of Figure 2; Fig. 14 shows a schematic diagram of a system for executing the method of Figure 3; and Fig. 15 shows graphical results indicating performance of the method of Figure 3.

It should be noted that the figures are illustrated for simplicity and are not necessarily drawn to scale. Like features are provided with the same reference numerals.

Detailed description of the preferred embodiments

Existing RAN vendor energy saving features are generally based on vendor equipment type (Vendor-based). Furthermore, such a strategy does not coordinate between different technologies working on the same radio unit (RRU) or units. A lack of coordination between different technologies reduces energy saving.

Figure 2 shows a flowchart of a method 200 for managing one or more base stations or base transmitter stations (BTS). The method is computer implemented and may be executed at one or more base stations within a telecommunications network or within a different component of the telecommunications network. For example, the method may be carried out by a separate computer system.

At step 210, a notification is received that one of a plurality of base stations has switched to an alternative power supply from a utility power supply (e.g., supplied by a power grid or power station). In response to this notification and at step 220, an altered configuration for the base station is generated. The altered configuration is generated so that one or more performance indicators or key performance indicators (KPI) of the base station is maintained when operating using the alternative power supply. The performance indicators may be specific to a particular base station or common across a set of base stations or all base stations in the telecommunications network.

At step 230, the original configuration of the base station is changed to the altered configuration. This may be achieved using an operations support system (OSS) that controls the configuration or a plurality of base stations (or gNodeBs), for example.

When changing the base station from an original configuration to the altered configuration, the altered configuration powers down one or more radio units or remote radio units (RRU) within the base station. When utility power is restored then a notification may be received at step 240 indicating that the base station is no longer running on the alternative power supply. This may trigger a change in configuration from the altered configuration back to the original configuration at step 250. Again, this may be controlled by the OSS in real time.

As shown in Figure 3, the computer system 100 that may execute the method of Figure 2 includes a number of components including communication interfaces 120, system circuitry 130, input/output (I/O) circuitry 140, display circuitry and interfaces 150, and a datastore 170. The system circuitry 120 can include one or more processors or CPUs 180 and memory 190. The system circuitry 130 may include any combination of hardware, software, firmware, and/or other circuitry. The system circuitry 130 may be implemented, with one or more systems on a chip (SoC), application specific integrated circuits (ASIC), microprocessors, and/or analog and digital circuits.

The display circuitry may provide one or more graphical user interfaces (GUIs) 160 and the I/O interface circuitry 140 may include touch sensitive or non-touch displays, sound, voice or other recognition inputs, buttons, switches, speakers, sounders, and other user interface elements. The I/O interface circuitry 140 may include microphones, cameras, headset and microphone input /output connectors, Universal Serial Bus (USB) connectors, and SD or other memory card sockets. The I/O interface circuitry 140 may further include data media interfaces (e.g., a CD-ROM or DVD drive) and other bus and display interfaces.

The memory 190 may include volatile (RAM) or non-volatile memory (e.g., ROM or Flash memory). The memory may store the operating system 192 of the computer system 100, applications or software 194, dynamic data 196, and/or static data 198. The datastore or data source 170 may include one or more databases 172, 174 and/or a file store or file system, for example.

The method and system may be implemented in hardware, software, or a combination of hardware and software. The method and system may be implemented either as a server comprising a single computer system or as a distributed network of servers connected across a network. Any kind of computer system or other electronic apparatus may be adapted to carry out the described methods.

-10 -Radio base stations or BTS nodes are generally powered using three-phase commercial power sources (utility power) to power on their internal equipment. Rectifiers and other circuitry are used to convert the AC supply to a DC voltage, which is at least partially used to charge a secondary (rechargeable) battery. Batteries or other alternative power supplies are used, in case a power grid outage occurs. This enables base stations to remain powered for a greater proportion of time.

The batteries will provide an average backup time depending on how fast power is drained by the various components of the base station. If the utility power supply fails for too long and beyond the time where the secondary batteries are able to provide backup power (e.g., when a safety limit or low voltage limit is reached) then the base station will completely power down and customers will lose all network coverage in the area or cell covered by the base station.

Figure 4 shows a schematic diagram of a system 400 for powering a base station 410. A utility or commercial power source 420 powers the base station 410 for a majority of the time. The three-phase AC power from the utility power source 420 feeds into a main distribution panel 430. The AC power is converted to a regulated DC voltage using a rectifier circuit 440, which is suitable to drive a plurality of RRUs 460, each operating at different frequencies and providing different cellular technology services. The DC voltage is also provided to battery strings 450. This may be further regulated to prevent overcharging and/or over-discharging, for example.

The system 400 of Figure 4 is illustrated during a time when a power cut takes place. Therefore, no power is (temporarily) provided by the utility power supply 420.

Instead, the battery strings 450 are discharged through the rectifier circuit 440 (which may also include a DC power regulator) and into the RRUs 460 so that they may continue to operate in the absence of the utility power supply 420.

In certain countries, power availability problems are ranked as the number one cause of network loss. Such problem causes severe customer dissatisfaction in regions that are impacted. Repeated discharging of batteries reduces their lifetime. This can cause high operating expenses and can also lead to a loss of cellular service when malfunctioning batteries are not replaced in time. A challenge is how to best utilise the resources within the base station to increase service availability during utility or public power cuts and provide cellular users with the basic services for as long as possible.

As described with reference to the method of Figure 3, the system 100 generates an altered configuration for the base station 410 and in particular a new configuration for managing power to the RRUs 460 when operating on an alternative power supply, such as the battery strings 450 of Figure 4. An example altered configuration is shown schematically in Figure 5. RRU 500 provides cellular services at 2100kHz. RRU 510 provides cellular services at 1800kHz. RRU 520 provides cellular services at 900kHz.

Furthermore, RRU 500 provides LTE (4G) services (L21) and 3G services (U2100).

RRU provides GSM (2G) services (DCS) and also provides LTE (4G) services (L18). RRU 520 provides 3G (UMTS) services (U900) and GSM (2G) services. Therefore, there is some overlap between the different services supplied by the different RRUs 460. The performance indicators for the cellular network may specify that all cellular services are proved at all cell locations. Therefore, only RRU 500 may be shut down fully and still meet the performance indicators. Shutdown of the RRUs is indicated by the crosses in Figure 5. The GSM functionality of RRU 510 may be shut or powered down as RRU 520 provides GSM services and can take over to some extent. The system may generate such an altered configuration based on the requirements defined by the performance indicators. If a particular cellular service level cannot be provided by a single RRU at that particular time then another RRU providing the same cellular service will remain powered, even when the base station 410 is powered using the alternative power supply. The altered confirmation may itself be replace by a further altered and different configuration and sent to the base station 410, should the performance indicators change over time. For example, it may be predicted that few active users will require a particular cellular service (e.g., 2G). This may allow further shutdowns of RRUs.

Figure 6 shows a different example altered configuration for a different example base station 410 (at a particular time with particular performance indicators that may be static or dynamic). In this base station 410, RRU 500 and RRU 510 are the same as those used in the example of Figure 5. However, the 900MHz RRU 620 is configured differently and can provide all three cellular technologies (GSM/2G, UMTS/3G and LTE/4G). Therefore, when operating on the alternative power supply, both RRU 500 and RRU 510 can be powered down and the base station 410 can maintain all required cellular technologies for the given performance indicators. This may assume that a single RRU -12 -can supply the expected number of mobile devices accessing the cell site serviced by the base station 410.

In both example scenarios (Figure 5 and Figure 6), the maximum energy saving is achieved at the time whilst maintaining the services defined by the static or dynamic performance indicators.

The present system may be described as a multi radio access technologies network availability optimizer. This provides a solution to mitigate against power cuts by activating a power saving mode in the base station 410, which may be vendor independent and as soon as a power cut is detected. Resource usage may be optimised as some resources may be shut down to prolong the lifetime of batteries without adversely affecting mobile users by increasing base station availability to meet performance indicators.

The particular altered configuration that is required for a particular cell site at a particular time of day or week may be generated using a machine learning algorithm. For example, a regression model may be used to ensure that the altered configuration (e.g., which RRUs are shut down) doesn't affect customer key performance indicators (KPI). For example, these may be defined as a minimum throughput that must be achieved at all times but the demands of this may vary from cell site to cell site and over time.

Only basic services need to be provided to cellular customers during a power cut. The level of basic service required may be defined using the performance indicators, that may be predetermined, static, and/or dynamic. The performance indicators may be static but based on demand. Therefore, in order to maintain a static performance indicator, different resource levels may be required (e.g., a maximum latency is require for each mobile device). The system and method may monitor the power source status (e.g., receiving notifications or polling a data source to determined how one or more base stations are currently being powered). Rollback to the original configuration may take place once the utility power supply is restored to a particular base station. This may ensure that all resources that have been shutdown (to preserve the alternative power supply for as long as possible) are then restored. Measurements may be made so that the effect of any optimisation or reconfiguration actions can be determined regarding the consumed power and enhancement to base station availability. Figure 7 shows schematically the enhanced -13 -effect on battery life when using the present method 200.

Figure 8 shows schematically the method 200 operating on RRUs within a base station 410. The three RRUs are shown operating in normal (with utility power) at the top of Figure 8 and with the RRUs operating in the altered configuration (using alternative power) at the bottom of Figure 8. Arrow 810 indicates when the utility power cut occurs. Arrow 820 indicates when the altered configuration is applied by an availability optimiser component 800. The original configuration is applied, as shown by arrow 830.

Within the availability optimiser component 800 operates, the following steps take place.

1. An analytics system obtains the base station (NodeB) configuration using an inventory system.

2. System generates an altered configuration, to apply to the base station 410. Running resources are shut down if possible, without breaching the performance indicators. A complete radio unit (RRU) is shut down if this is attainable under these conditions.

3. Resources are shutdown provided that: Customer basic KPIs (e.g., throughput) are maintained using an Al-empowered regression model (or otherwise) that defines the radio parameters threshold to ensure that the network provides minimum acceptable service KPIs are maintained. The base station maintains the cell site coverage layer, optimises the capacity layer to provide basic services and achieves improved energy saving.

4. The analytics system generates commands defining the altered configuration and sends these commands to the OSS to execute.

The following describes in more detail the use of machine learning to provide the altered confirmation and the parameters that may be applied to the base station 410 to implement this new configuration to be used during a (temporary) power outage. One of the important parameters that is used to activate a power saving mode is the number of -14 -camped users per cell. In a conservative mode, the reduction in power use may be typically set to a small value and fixed for all network cells to avoid any impact on mobile users. However, this leads to power saving optimisation actions that do not provide significant power saving (and overuse of batteries).

Therefore, it is necessary to find a solution that defines this important parameter (number of users) based on cell traffic and camped user trends. Figure 9 shows example graphical results indicating average user throughput against the number users within a cell. A machine learning model, e.g., a regression model, such as a k-nearest neighbour (KNN) regressor may be implemented by the Sklearn python library. This may be built and tuned to define a number of user devices per cell that can receive a guaranteed service so that throughput will not go below a certain minimum throughput (for this number of devices), as defined by performance indicators stored within the system.

The altered configuration may be provided in the form of a set of parameters or optimised parameters and may be generated using machine learning. The model predicts the optimum number of users per cell to achieve a minimum user throughput. This predicted value is then used to configure the cell and apply power saving features (an altered configuration, including RRU power states) accordingly. Figure 10 illustrates an example output from such a machine learning model for a plurality of cell sites and base stations 410. Therefore, the machine learning algorithm determines the maximum number of cellular devices that may be served by a particular base station 410 and still maintain the minimum throughput, as defined by the performance indicators of the telecommunication network.

Different optimisation scenarios have been simulated and investigated. The system and method use the machine learning algorithm, as implemented by an example Al module, to define the hardware that can be switched off and maintain the minimum throughput. Each scenario may differ from cell to cell based on its particular traffic profile and installed configuration to satisfy the target performance indicators (KPIs) and maximise energy efficiency based on the regression model output. Figure 11 shows a table indicating different scenarios for different cell sites that have been tested. Figure 12 shows the advantageous effect when applying the described method 200 to these different cell sites. Both results (1200, 1200) indicate an improvement in the amount of time that the alternative power supply can remain operational in the event of a power failure.

-15 -Resource blocking based on number of users may be carried out to improve alternative power use. This may involve automation checks that to determine the number of users camped to a cell. As can be seen from the graphical results of Figure 13, automation provides power saving actions only if the number of users is below a certain threshold (i.e., 30). It is also shown from these results that when the enhanced method 100 is activated, battery backup time increases from 1:23 hours to 2:44 hours, in this example.

The following describes in more detail the machine learning features and processes used to optimise parameters used within the method of Figure 2.

As described previously, to achieve enhanced battery backup time, it is necessary to reduce certain radio resources but without an unacceptable reduction in telecommunications services. Resources limiting actions may be implemented by passing parameters to the base stations (NodeB).

The main parameters that may be optimised include: The number of camped users per cell, which is inversely proportional to the user throughput.

The time window when an energy saving mode starts and ends.

RAN vendor cell carrier shutdown software features may have time windows and the number of users as input parameters. The time window may be automatically set whenever utility or grid power is lost. To set the number of users, thresholds for each cell can be selected or set so that each cell or radio resources within each cell can only be switched off (e.g., placed into sleep mode) if throughput of co-located cells or resources within the same cell site are higher than a certain threshold. For example, this may be 3Mbps but other values can be used. The number of users parameter may be fixed or dynamic. However, using a fixed number of users is not effective.

Therefore, a method for determining a dynamic number of users is required. This may be achieved by using a machine learning regression model, which can establish a relationship between a number of users camped per cell and an average required throughput per use. As mentioned previously, the average throughput should not drop -16 -below a predetermined threshold (e.g., 3Mbps). The regression model is built for each cell with a target to determine the number of users per cell that achieves at least the minimum average throughput per user (e.g., 3Mbps). This will provide a threshold of the overall system to aim for (i.e., by keeping radio resources powered) when operating from the alternative power source when the utility power is lost or unavaialbe.

In order to build the regression model, data may be collected for the average throughput per user against the active users per cell with the model trained accordingly.

Different regression models may be used. They may be assessed using R-Squared values, which represents error or model accuracy.

Preferably, the models may be selected from: 1. K Nearest Neighbors Regressor 2. Adaboost Regressor 3. Random Forest Regressor 4. Ridge Regression 5. Linear Regression The R-squared value evaluates the scatter of the data points around a fitted regression line. This provides a coefficient of determination, or the coefficient of multiple determination for multiple regression. For the same data set, higher R-squared values represent smaller differences between the observed data and the fitted values.

The following table provides example R-squared values for the above models using the same data set: gAlaighbditROgitatadie A8057450.Mr: Ridg ti55, f,,, , * R_Squar ti-402403 Random Barr fiR O.x55435591;s iLin Fi -17 -Therefore, the K Neighbors Regressor (k-nearest neighbour, KNN, regressor) provides the best performance in this example. ;Figure 13A shows a flowchart of an example method 1300 for using the regression model to predict the optimum number of users per cell, which achieves a minimum average user throughput. ;This method normalises cell information data, which are split into training and test data. A search grid is used to generate the best model parameters. The best model parameters are used in a prediction step and a database is built for each base station or cell and associated number of users. These data may change over time. ;The software code may be written using the Python programming language. The software code may be used to connect to data sources (e.g., the telecommunications network) with each cell in the network providing: Cell average user throughput; Cell number of camped users per cell; and Cell ID. ;Data normalsation is important and applied to raw cell data. This is because the KNN regressor depends on calculating a Euclidean distance to predict output values. A Scikit Learn Standard Scaler may be used to normalise the data, and in this example implementation, its equation for normalising data is: St aril" a r Z :r -ft lArit 1-1. rile Mr. ;1 *v--s,

_

st a nda rd de-vi Ettion: CT 0:" Where z is the normalised value -18 -Splitting or dividing the data into training and test data is also important to ensure that the model will generalise well in both training, validation, and testing data. Data may be split into test data, which represents 20% of the data samples and training data which represents 80% of the data, for example. Other percentage splits may be used.

The training data may be used to train the model, while the test data may be used to calculate the R-Squared of the model on data that wasn't exposed to the model during the training process.

The search grid is a mechanism used to identify the best values for model parameters. The search grid may be used to iterate for different values of: Number of estimators; and Number of levels of the decision tree.

Figure 10 provides example results and shows the output of the model after the best parameters were found and after the search grid iterations.

As shown, for each cell the model provides: The predicted number of users that achieves minimum user throughput; The R-squared value of using test data by the model; and The cell ID.

A database may be created that stores the optimum number of users that achieves minimum user throughput for each cell. This value may be used to set the threshold needed to take decisions powering down particular radio resources (e.g., radio units) or not in the event of the loss of utility power and reverting to the alternate power supply, as described previously.

As can been seen in the graph of Figure 13, at point A, B, the availability optimiser method was not implemented as the threshold of number of users was set conservatively to be 30 users per cell. The actual number of camped users at points A, B at the example times were 72, 98 respectively.

-19 -Upon applying the model to this cell site, the model predicted that the number of cell camped users that achieves a minimum acceptable average throughput per user should be 121 users, so this will allow the energy saving feature to be active and the cell to work more efficiently, whilst providing the minimum required service provision.

Figure 14 shows a schematic diagram of example software components that may be used to implement the method 100. A power alarm monitoring system 1410 provides alarm data to a data collection layer within an ipredictor 1420. The ipredictor 1420 may also include a processing layer and an action tracker. These may be implemented using Python but other languages may be used. The OSS 1430 receives the output from the ipredictor 1420 and applies the altered (and then original) configurations on the base stations 410. Figure 15 shows graphical results indicating the amount of back time available for two different cell sites implementing the method 100.

As used throughout, including in the claims, unless the context indicates otherwise, singular forms of the terms herein are to be construed as including the plural form and vice versa. For instance, unless the context indicates otherwise, a singular reference herein including in the claims, such as "a" or "an" (such as an ion multipole device) means "one or more" (for instance, one or more ion multipole device). Throughout the description and claims of this disclosure, the words "comprise", "including", "having" and "contain" and variations of the words, for example "comprising" and "comprises" or similar, mean "including but not limited to", and are not intended to (and do not) exclude other components. Also, the use of "or" is inclusive, such that the phrase "A or B" is true when "A" is true, "B is true", or both "A" and "3" are true.

The use of any and all examples, or exemplary language ("for instance", "such as", "for example" and like language) provided herein, is intended merely to better illustrate the disclosure and does not indicate a limitation on the scope of the disclosure unless otherwise claimed. No language in the specification should be construed as indicating any non-claimed element as essential to the practice of the disclosure.

The terms "first" and "second' may be reversed without changing the scope of the disclosure. That is, an element termed a "first" element may instead be termed a "second" element and an element termed a "second" element may instead be considered a "first" 35 element.

-20 -Any steps described in this specification may be performed in any order or simultaneously unless stated or the context requires otherwise. Moreover, where a step is described as being performed after a step, this does not preclude intervening steps being performed.

It is also to be understood that, for any given component or embodiment described throughout, any of the possible candidates or alternatives listed for that component may generally be used individually or in combination with one another, unless implicitly or explicitly understood or stated otherwise. It will be understood that any list of such candidates or alternatives is merely illustrative, not limiting, unless implicitly or explicitly understood or stated otherwise.

Unless otherwise described, all technical and scientific terms used throughout have a meaning as is commonly understood by one of ordinary skill in the art to which the various embodiments described herein belongs.

As will be appreciated by the skilled person, details of the above embodiment may be varied without departing from the scope of the present invention, as defined by the appended claims.

For example, the figures show only one or two base stations 410 being managed but any number may be present and managed by the system. Different computer languages and machine learning algorithms may be used. The machine learning algorithm may be trained on existing data or receive live data to enhance the model over time. The system and method use machine learning to estimate service degradation and a decision may be based on estimated results to switch off the cells or RRUs within the cells. The system and method may provide multi rate coordination between different technologies (2G, 3G and 4G) working on the same RRU. Network counters may be used to decide whether or not to switch off some resources. Components within the base station that are switch off or otherwise affected by the altered configuration may extend beyond the RRUs. The system and method may use OSS commands through automated command line access to execute decisions, which can be applied across the whole network without any required new hardware.

-21 -Many combinations, modifications, or alterations to the features of the above embodiments will be readily apparent to the skilled person and are intended to form part of the invention. Any of the features described specifically relating to one embodiment or example may be used in any other embodiment by making the appropriate changes.

Claims (15)

1. -22 -CLAIMS: 1. A method for managing one or more base stations, each one or more base station comprising a plurality of radio units, the method comprising the steps of: receiving a notification that a base station is switching from a utility power supply to an alternative power supply; in response to the notification generating an altered configuration for the base station, wherein the altered configuration is generated to maintain one or more performance indicators of the base station when operating using the alternative power supply; and changing the base station from an original configuration to the altered configuration, wherein the altered configuration powers down one or more of the radio units within the base station.
2. 2. The method of claim 1 further comprising the step of receiving a further notification indicating that the base station has returned to the utility power supply and in response changing the base station from the altered configuration to the original configuration.
3. 3. The method of claim 1 or claim 2, wherein the alternative power supply is a secondary battery or an electrical generator.
4. 4. The method according to any previous claim wherein the one or more performance indicators of the base station define performance indicators for cellular services provided to a plurality of user equipment, UE, by the one or more base stations.
5. 5. The method according to claim 4, wherein the one or more performance indicators of the base station define a minimum data throughput for cellular services provided to the plurality of UEs.
6. 6. The method of claim 4 or claim 5, wherein the performance indicators are used to determine a minimum number of user equipment, UE, to be supplied by cellular services at each different cellular network technology by the base station.
7. 7. The method of claim 6, wherein the different cellular network technologies comprise any two or more of: 2G, 3G, 4G, and/or 5G.
8. -23 - 8. The method of claim 6 or claim 7, wherein the altered configuration is generated by determining a maximum number of radio units to power down whilst the base station can supply the minimum number of UE with cellular services at each different cellular network technology.
9. 9. The method according to any of claims 6 to 8, wherein the minimum number of UEs to be supplied by cellular services at each different cellular network technology by the base station to maintain the performance indicators is determined using a machine learning, ML, algorithm.
10. 10. The method according to claim 9, wherein the ML algorithm is based on a regression model.
11. 11. The method of claim 10, wherein the regression model is a k-nearest neighbour, KNN, regressor.
12. 12. The method according to any previous claim, wherein the base station configuration is changed by sending commands to an operations support system.
13. 13. A base station comprising: a plurality of radio units; a utility power supply input; an alternative power supply; and means adapted to execute the steps of: switching from a utility power supply to an alternative power supply; transmitting a notification that the base station is switching from the utility power supply to the alternative power supply; in response to the notification, receiving an altered configuration for the base station, wherein the altered configuration maintain one or more performance indicators of the base station when operating using the alternative power supply; and changing the base station from an original configuration to the altered configuration, wherein the altered configuration powers down one or more of the plurality of radio units within the base station.
14. -24 - 14. The base station of claim 13, wherein the alternative power supply is a secondary battery or an electrical generator.
15. 15. A telecommunications system comprising the base station of claim 13 or claim 14; and a data processing system comprising means for carrying out the steps of and of claims 1 to 12.