IT201800007226A1 - Method and system for monitoring and optimizing the energy operation of at least one site in a telecommunications network. - Google Patents

Description

METHOD AND SYSTEM FOR MONITORING AND OPTIMIZING THE ENERGY OPERATION OF AT LEAST ONE SITE IN A NETWORK

OF TELECOMMUNICATIONS

The present invention relates to a method and a system for monitoring and optimizing the energy operation of at least one site in a telecommunications network.

The method and system described here are particularly, although not exclusively, useful in the field of telecommunications networks, that is to say the set of apparatuses and devices, connected together, in order to transmit information of various kinds at a distance, between two or multiple users.

Advantageously, the method and system object of the invention can be useful in the energy management of highly complex telecommunications networks, identifying the sites that need interventions or optimization procedures from an energy point of view.

In recent decades there has been a progressive evolution and complication of telecommunications networks towards highly complex technological structures, able to connect multiple users to each other and to cover otherwise unthinkable distances.

Today, telecommunications networks represent the technical infrastructure that allows the transmission and reception of information, such as an audio signal, a video signal, an image and so on, between two or more users located in geographically distinct locations.

Each telecommunications network, from the physical point of view, comprises a plurality of links (or branches) and nodes (or sites).

The branches represent the way for the transfer of signals that support the information while the nodes are the transmission and / or reception points of these signals.

In particular, these sites or nodes usually include structures containing, for example, equipment dedicated to the transmission of electromagnetic waves, electronic systems for data processing, air conditioning systems for maintaining the correct working conditions of the various electronic devices and so on. .

The individual sites of the telecommunications network, in fact, can be divided into at least three categories, according to the characteristics of the system and structure, such as: building, shelter and cabinet.

As is well known to those skilled in the art, one of the main drawbacks of telecommunications networks is the conspicuous consumption of electrical energy absorbed by the sites or nodes present in the network.

In fact, communication between the various sites included in the telecommunications networks often takes place through one-way radio signals whose consumption is significant in terms of the total consumption of each site.

In addition to the above, the energy consumption of each site is significantly affected by factors such as, for example:

- the location of the site (rural areas, urban areas, etc.);

- the number of users served by the telecommunications network;

- telecommunications standards (such as, for example, UMTS technology ("Universal Mobile Telecommunications System"), GSM ("Global System for Mobile Communications"), LTE ("Long Term Evolution"), etc.) used on the site;

- the number of carrier radio frequencies for each telecommunications standard.

A further drawback of these known solutions consists in the fact that they provide for the installation of multiple devices in order to ensure the proper functioning of the site included in the telecommunications network.

Another drawback of these known solutions consists in the fact that they do not allow any malfunctions relating to each site included in the telecommunications network to be identified in real time or in a timely manner.

A further drawback of these known solutions consists in the fact that they do not allow to optimize the energy operation of each site included in the telecommunications network.

It is clear that these solutions are costly in energy and economic terms and do not allow an energy characterization of the various sites included in the telecommunications networks.

Furthermore, the known solutions do not allow to use a decision support for the purposes of energy management of the telecommunications networks in order to define suitable types of intervention according to the needs of each site.

Currently there are techniques and procedures on the market that can reduce the energy consumption of a telecommunications network by acting solely on the parameters of the electronic devices included in each site such as, for example, the transmission power, the band used and so on.

In the light of the above, it is therefore an aim of the present invention to overcome the limits of the prior art set out above, by proposing a system and a method for monitoring and optimizing the energy operation of at least one site in a telecommunications network, which allow to distinguish each site from an energy point of view.

Another purpose of the invention is to propose a system and a method for monitoring and optimizing the energy operation of at least one site in a telecommunications network, which allow to adopt intervention techniques or procedures on each site according to the operating levels. of the same.

Another object of the invention is to provide a system and a method for monitoring and optimizing the energy operation of at least one site in a telecommunications network, which are highly reliable, relatively simple to build and at competitive costs if compared to the technique. Note.

A further object of the present invention is to provide the tools necessary for carrying out the method and the apparatuses that carry out this method.

Therefore, the specific object of the present invention is a method for monitoring and optimizing the energy operation of at least one site or node in a telecommunications network, said method comprising the following steps: A. receiving data from each site; B. read said data received from each site; C. process, for each site, said data through the following sub-phases: C1. perform a first process to estimate the energy consumption of each site and compare it with the real energy consumption of the respective site; and C2. performing a second process to determine a number of carrier radio frequencies associated with respective telecommunications standards for each site; and D. determining, for each site, a respective level of energy consumption and of said number of carrier frequencies by means of respective predetermined thresholds.

Still according to the invention, said phase B can further comprise the following sub-phase: B1. organize, for each site, said data in a respective data array, where = 1, ... is the index of each site included in said telecommunications network, and n is a positive integer.

Still according to the invention, said data array Di can comprise: a first subarray Di1 comprising real energy consumption values of said site, a second subarray Di2 comprising at least one telecommunications standard used in said site (21, ..., 2n), a third subarray comprising at least a number of radio frequencies associated with a respective telecommunications standard included in said second subarray Di2, a fourth subarray D i4 comprising a structural typology of said site, a fifth subarray

DÌS comprising a number of radio links of said site, a sixth subarray Di6 comprising topology / user data of said site, a seventh subarray D i7 comprising temperature and humidity data of said site, and an eighth subarray Di8 comprising time reference data associated with said site site.

Further according to the invention, said first process can process the data of said Di1 subarrays,

Di3, Di7 and Di8.

Advantageously according to the invention, said first process P1, associated with a respective first optimization step F1, can calculate a first index EE according to the following formula: where i = l, ... n is the index of each node or site included in said telecommunications network, and n is a positive integer, where said first subarray Di1 comprises values of said real energy consumption of said node, and D '1i is a further first subarray comprising values of said estimated energy consumption, determined by means of said first process P1.

Still according to the invention, said first process P1 can be achieved through a concatenation of a Long Short Term Memory Neural Network (LSTM) artificial neural network with a Feed Forward artificial neural network.

Still according to the invention, said artificial neural network FF comprises ten input nodes connected with four hidden layers, each of said four hidden layers comprising forty nodes. Advantageously according to the invention, three of said four hidden layers comprise an activation function of the Rectified Linear Units (ReLU) type and one of said four hidden layers comprises an activation function of the Sigmoid type.

Further according to the invention, said artificial neural network LSTM can comprise a single layer of sixty nodes or cells of the LSTM type connected to an output layer composed of a single node and to a synthesis layer in which said nodes of said node are also connected fourth hidden layer of said artificial neural network FF.

Still according to the invention, said second process P2 can process the data of said Di2 subarrays,

Di3, Di4 and Di6

Still according to the invention, said second process P2 can be carried out by means of a non-parametric decision tree method.

Further according to the invention, said phase C can further comprise the following sub-phase: C3. executing a third process P3 to estimate said energy consumption of each site and compare it with said real energy consumption of the respective site (21, ..., 2n), on the basis of the totality of said data included in said data array D i.

Advantageously according to the invention, said third process P3 can process the data of said subarrays

Di3, Di4, Di5, D i6, Di7 and Di8 in a second optimization phase P2, and processes the data of said subarray Di1, Di2, D i3 *, Di4, Di5, Di6, Di7e Di8 in a third optimization phase P3, wherein said subarray Di <*> 3 comprises at least a further number of radio frequencies associated with a respective telecommunications standard on the basis of said second process P2.

Preferably according to the invention, said third process P3, associated with said respective second optimization step P2, can calculate a second index EEI2 according to the following formula: where i = 1, ... n is the index of each site 21 ,. .., 2n included in said telecommunications network 2, and n is a positive integer, where said first subarray Di1 comprises values of said real energy consumption of said site (21, ..., 2n), and D'2i is a further second subarray comprising values of said estimated energy consumption, determined by said third process P3.

Still according to the invention, said third process P3 can be realized through a fully connected artificial neural network FF.

Still according to the invention, said artificial neural network FF fully connected can comprise eighteen input nodes, each of which connected with all the nodes of four hidden layers, each of said four hidden layers comprising fifty nodes, and a single output layer node.

Advantageously according to the invention, three of said four hidden layers can include an activation function of the Rectified Linear Units (ReLU) type, and one of said four hidden layers includes an activation function of the Sigmoid type.

Further according to the invention, a combination of said second process and said third process, associated with said respective third optimization step, can calculate a third index according to the following formula: where is the index of each site included in said telecommunications network , and n is a positive integer, where said first subarray comprises values of said real energy consumption of said site, and is a further third subarray comprising values of said estimated energy consumption, determined by the combination of said second process and said third process associated with called third phase of optimization.

Still according to the invention, said phase D can further comprise the following sub-phases: D1.

<verifying, for each site, that said first index> EEI1 is higher than a first predetermined threshold; D1.1 if said first index EEI1 is higher than said first predetermined threshold S1, determining, for each site, a first level of operation; D1.2 if said first index EEI1 is lower than said first predetermined threshold S1, proceed with step D2; D2. verify, for each site (21,…, 2n), that said second EEI2 index is higher than a second <predetermined threshold; D2.1 if said second index> EEI2 is higher than said second predetermined threshold S2, determine, for each site, a second operating level L2; D2.1 if said second index EEI2 is lower than said second predetermined threshold S2, proceed with step D3; D3. verify, for each site, that said third index EEI3 is higher than a third predetermined threshold S3; D3.1 if said third index EEI3 is higher than said third predetermined threshold S3, determine, for each site, a third level of operation; D3.2 if said third index EEI3 is lower than said third predetermined threshold S3, determine, for each site, a fourth level of operation.

The present invention also relates to a system for monitoring and optimizing the energy operation of at least one site or node of a telecommunications network, said system comprising a central control unit, configured to communicate, via a first telematic communication channel, with each site of said telecommunications network, characterized in that said central control unit is further configured to perform the method for monitoring and optimizing the energy operation of said at least one site or node (21, ..., 2n) in a telecommunications network such as defined above.

Still according to the invention, said central control unit may include transceiver means to communicate with at least one telematic device through a further telematic communication channel, and with each node via said first telematic communication channel.

Still according to the invention, said central control unit can comprise storage means for storing said data sent from each site to said central control unit.

A further object of the present invention is a computer program comprising instructions which, when said program is executed by a computer, cause the execution by said processor of phases A, B, C and D of said method described above.

The subject of the present invention also forms a storage medium that can be read by a computer, comprising instructions which, when executed by a computer, cause the execution by said computer of phases A, B, C and D of said method described above.

The present invention will now be described for illustrative but not limitative purposes, according to its preferred embodiments, with particular reference to the figures of the attached drawings, in which: Figure 1 is a block diagram which schematically shows an embodiment of a system for monitoring and optimizing the energy operation of at least one site in a telecommunications network, according to the present invention;

Figure 2 is a block diagram schematically showing an embodiment of a central control unit of the system according to Figure 1;

Figure 3 shows a flow diagram of the method for monitoring and optimizing the energy operation of at least one site in a telecommunications network, according to the present invention;

Figure 4 is a flow chart showing a possible embodiment of a first process in a first optimization step of the method according to Figure 3;

Figure 5 is a flowchart showing an embodiment of a concatenation of a Long Short Term Memory Neural Network (LSTM) artificial neural network with a Feed Forward (FF) artificial neural network according to the first process of Figure 4;

Figure 6 is a flowchart showing an embodiment of a second process in a third method optimization step according to Figure 3; And

Figure 7 is a flow chart showing a possible embodiment of a third process in a second optimization step and in a third optimization step of the method according to Figure 3.

In the various figures similar parts will be indicated with the same numerical references.

With reference to the aforementioned figures, the system for monitoring and optimizing the energy operation of at least one site in a telecommunications network according to the present invention, globally indicated with the reference number 1, mainly comprises a telecommunications network 2, constituted and schematized with a plurality of nodes 21, ......, 2n, a telematic communication channel 3, a central control unit 4, connected to the telecommunications network 2 by means of said telematic communication channel 3, and a telematic device 6 used by a user 7 and connected to the central control unit 4 through a further telematic communication channel 5.

The telecommunications network 2 is configured to allow the exchange of information between the nodes 21, ……, 2n. Advantageously, in an embodiment of the system 1, in each node 21, ..., 2n apparatus dedicated to the transceiving of radio signals, to the processing of data and so on can be arranged or installed.

Each node 21, ..., 2n of the telecommunications network 2, based on the characteristics of the system and structure, can preferably belong to different technical and structural categories such as, by way of example but not limited to, shelter, cabinet or building.

The telecommunications network 2 is further configured to transmit such information to said central control unit 4 via said telematic communication channel 3, such as for example the Internet network.

The central control unit 4 is configured to detect, control and process data from said telecommunications network 2.

With particular reference to Figure 2, the central control unit 4 essentially comprises: a processing unit 41, acquisition means 42, transceiver means 43 and storage means 44.

The processing unit 41 is the functional and main element of the central control unit 4 of the system 1 according to the present invention, and, for this reason, it is connected and in communication with the other elements of the central control unit 4 itself.

Processing unit 41 is equipped with calculation and processing means, configured to run software for energy monitoring and optimization, as well as to interface with the other elements of the central control unit 4.

Said processing unit 41 is also configured to control and coordinate the operation of the elements of the central control unit 4 with which it is connected and in communication.

Said processing unit 41 can be constituted by a computer or a plurality of computers, possibly connected in the cloud.

The acquisition means 42 of the central control unit 4 are configured to acquire data from each node 21, ..., 2n. Said acquisition means 42 are preferably connected to the storage means 44 through the processing unit 41.

The transceiver means 43 of the central control unit 4 are configured to receive data from the plurality of nodes 21, ..., 2n.

The transceiver means 43 are further configured to transmit data towards at least one telematic device 6.

The storage means 44 of the central control unit 4 comprise for example a database and are suitable for storing the data acquired by the acquisition means 42.

The telematic device 6, such as for example a desktop computer, a notebook computer, a tablet or a smartphone, operated by the user 7, is in communication with the central control unit 4 through said communication channel 5, such as the Internet network. .

The telematic device 6 is configured to receive data from the central control unit 4, such as audio signals, video signals, images and so on.

The telematic device 6 displays this information to the user 7 and may include a screen and a speaker.

The operation of the system 1 described above takes place as follows.

Said central control unit 4 is configured to receive data from each site or node 21, ..., 2n, store said data in said storage means 44, and organize said data in a respective data array, where = 1, ... is the 'index of each site 21,…, 2n included in said telecommunications network 2, and is a positive integer.

Advantageously, said data array comprises the following subarrays, each of which containing data belonging to a specific category:

- a first subarray comprising real energy consumption values of said site (21,…, 2n);

- a second subarray comprising at least one telecommunications standard used in said site (21,…, 2n);

- a third subarray comprising at least a number of radio frequencies associated with a respective telecommunications standard included in said second subarray;

- a fourth subarray indicating at least one structural type of said site (21,…, 2n); - a fifth subarray comprising a number of radio links of said site (21, ..., 2n),

- a sixth subarray comprising topology / user data of said site (21, ..., 2n), - a seventh subarray comprising temperature and humidity data of said site (21, ..., 2n), and

- an eighth subarray comprising time reference data associated with said site (21,…, 2n).

In particular, the time reference data included in said eighth subarray preferably comprise: date; Now; day of the month; day of the week; month; year.

Advantageously, the time reference is selected on the basis of the data available on a given site 21, ..., 2n and needs to be kept constant in order to build time relationships on this scale, as well as being associated with an estimated consumption consistent with the same. time scale.

In particular, said time reference data, depending on the sampling frequency of the input data to each site 21, ..., 2n, can take, for example, the following vector form:

[day of the month, day of the week, month, year];

[month year];

the first case indicates a daily time reference while the second case a monthly time reference.

According to the invention, the actual or estimated energy consumption data of each site 21, ..., 2n are associated with the time reference mentioned above on the basis of the timing of the input data to each site 21, ..., 2n.

The real energy consumption data included in said first subarray Di1, and the temperature and humidity data included in said seventh subarray Di7 have a time-varying nature, and allow an a posteriori evaluation of the deviation in terms of energy consumption between real values and expected values for each site 21,…, 2n included in the telecommunications network 2.

By way of example, the influence of temperature on energy consumption is represented by the Cooling Degree Days (CDD) parameter, which represents the sum of the differences in the external temperature, for all days of the month, with respect to a reference value equal to 20. ° C.

In addition to what has been said, according to the present invention:

- the second Di2 subarray includes one or more telecommunications standards used in site 21,…, 2n, such as GSM, UMTS, DCS, LTE, and so on;

- the third Di3 subarray comprises a number of carrier radio frequencies respectively associated with each telecommunications standard used in site 21,…, 2n;

- the fourth subarray Di4 includes the type of plant or structure associated with each site 21, ..., 2n, such as, for example, a shelter, a cabinet, a building and so on;

- the fifth subarray Di5 includes a number of radio links associated with each site 21,…, 2n;

- the sixth subarray Di6 includes topology / user data of said site 21, ..., 2n, such as, for example, the location and use of a radio base station associated with each site 21, ..., 2n;

Advantageously, the data included in said subarrays allow to uniquely identify each site 21, ..., 2n within said telecommunications network 2.

Furthermore, said central control unit 4 is further configured for:

<or select two or more of said subarrays>

from said data array

o determine for each site 21,…, 2n

a first process P1da called first subarray, <from said third subarray Di3, from seventh subarray> and from said eighth subarray Di8;

<a second process P2 from said second subarray>, from said third subarray Di3, from said quartosubarray Di4 and from said sixth subarray Di6, and a third process P3;

o determine for each site 21,…, 2n

a first optimization step F1 associated with said first process P1, a second optimization step F2 associated with said third <process P3, and a third optimization step> associated with a combination of said second process P2 and said third process P3;

or calculating for each site 21, ..., 2n a first index EEI1 on the basis of said first optimization step F1, a second index EEI2 on the basis of said second optimization step F2, and a third index EEI3 on the basis of said third optimization step F3;

or determining for each site 21, ..., 2n a first operating level L1 from a first comparison between said first index EEI1 and a first predetermined threshold S1, a second operating level L2 from a second comparison between said second index EEI2 and a second predetermined threshold S2, a third operating level L3 or a fourth operating level L4 from a third comparison between said third index EEI3 and a third predetermined threshold.

In an embodiment of the present invention, with particular reference to Figures 3 and 4, said first process P1, associated with a respective first optimization step F1, allows to generate a first index EEI1 ("Energy Efficiency Index") by statistically relating the time-varying characteristics of a specific site with its consumption.

Advantageously, said first index defines the ratio between estimated consumption and actual consumption for the site in question according to the following formula:

<where>

= 1, ... is the index of each node or site 21, ..., 2n included in said telecommunications network 2, and is a positive integer,

where is it

said first subarray Di1 includes values of consumption <real energy of said site 21, ..., 2n, and>

D′1i is a further first subarray comprising estimated energy consumption values, determined by said first process associated with said first optimization step.

In an embodiment of the present invention, said first process P1 is achieved through a concatenation of a Long Short Term Memory Neural Network (LSTM) artificial neural network with a Feed Forward (FF) artificial neural network.

The LSTM network allows you to analyze the temporal sequence of events by relating the past data with the current data during the training process (or training) of the network, which will be illustrated in detail in the continuation of the description.

By way of example, the data memory is defined on a monthly scale if daily data are available, or on a quarterly scale if monthly data are available.

It follows that each site is characterized by a different statistical model according to the structure assumed by the LSTM network as a function of the timing of incoming consumption.

In the case of daily time intervals, with time dependence of 30 days, the structure of said first subarray Di1 in said first process P1 is:

[Consumption (t0), ..., Consumption (t29)],

while in the case of monthly time intervals, with quarterly dependence, the structure of said first subarray Di1 is as follows:

[Consumption (t0), Consumption (t1), Consumption (t2)],

where t0,…, tn is a vector of time data respectively associated with the real energy consumption of site 21,…, 2n.

As previously said, the LSTM network is chained to a FF network, which takes into account the conditions of the site in question. The further input data of said first process P1 comprise the number of carrier radio frequencies, the time reference, and the temperature and humidity values, associated respectively with said subarrays Di3, Di7 and Di8.

In an embodiment of the present invention, the FF network is structured, in the case of a daily temporal frequency, by 10 incoming nodes connected with 4 hidden layers, each with 40 nodes.

On the first three layers, the Rectified Linear Units (ReLU) is implemented as an activation function, which takes the following form:

that is, a null value when the input is less than zero and a value equal to the input value when the latter is greater than zero.

On the fourth layer, on the other hand, a Sigmoid function described by the following expression is implemented as an activation function:

Finally, there is a last single neuron output layer with Sigmoid activation function in order to impose a constraint during the training phase.

The LSTM network is, in the case in question, structured in a single layer with 60 nodes (or cells) of the LSTM type. This layer is connected both to an output layer composed of a single node (with the function of optimizing the training on this specific part of the model), and to a synthesis layer in which the nodes of the fourth hidden layer of the network are also connected. FF.

LSTM cells are capable of storing and forgetting information using the following algorithms:

where they represent respectively the activation vectors of the forget gate, the input gate and the output gate of an LSTM cell, W and U represent the weight matrices, 5 represents the bias vector and 0 indicates the Sigmoid activation function.

The state vector Ct of each LSTM cell and the response ℎtd of the cell with respect to an input vector xts are determined by the following expressions:

In artificial neural networks, the estimation of parameters, that is the process that leads to the determination of the numerical values of the weights, is called learning (or training). This phase generally consists of an iterative procedure during which the value of the weights is adjusted step by step until the error between the measured data and the system outputs is reasonably low.

It is therefore a question of solving a minimization problem starting from a set of inputs and a loss function (or Loss Function), which is determined starting from the mean square error on the training values.

Among the possible methods of solving the aforementioned problem is the Adadelta algorithm. The latter allows you to vary the learning speed for updating the parameters of the artificial neural network by applying an exponential moving average:

where (indicates the correction to be applied to the parameters of said artificial neural network in order to obtain an update of the same with respect to the gradient. The correction (is given by the following expression:

where G / denotes the gradient of the loss function discussed above.

The mean square error of the average of the previous gradients weighed with a decay factor on the value of the current gradient, allows to take into account the variation of the gradient during the training process:

In an embodiment of the present invention, with particular reference to Figures 3 and 5, it can be observed how said second process P2 allows to obtain said further third subarray D <∗> 3 comprising an (optimal) number of carrier radio frequencies respectively associated with each standard of telecommunications of each site 21, .., 2n.

Advantageously, said second process P2 defines, for each site 21, ..., 2n a difference (or mismatch) between a number of carrier radio frequencies respectively associated with each telecommunications standard and a further number (or optimal number) of carrier radio frequencies respectively associated with each telecommunications standard, <the latter being estimated by said second process> P2.

In an embodiment of the present invention, said second process P2 is carried out by means of a non-parametric decision tree method. This approach allows to reduce the computational complexity allowing to maximize the information content in relation to the input data to said second process P2.

The input data of the second process P2 include, for each site 21, .., 2n, the telecommunications standards, the number of carrier radio frequencies, the type of system, the topology / user and, respectively, associated with said subarrays

By way of example, the data referring to the type of system indicates structures such as, for example, shelter, cabinet, building etc., the data associated with the topology / user refers, for example, to the location and user of the radio base station (cities with low or high user density, rural areas with negligible morphological reliefs with low or high user density, areas with considerable morphological reliefs with low or high user density, etc.), while telecommunication standards refer to technologies used in site 21, .., 2n under consideration, such as GSM, LTE, UMTS and so on.

In particular, during said second process P2, moreover, a subsequent subdivision of the data of said subarray (or training data) into two sets is carried out, whereby the variance of the elements falling therein is minimized.

The above is achieved through a function that minimizes the squared deviation between the average of the data of said Di3 subarray and their real values:

<where Y represents a datum of said subarray> Di3 associated with a specific technology (or specific telecommunication standard), ℎ indicates the mean squared deviations associated with a given subarray R, for a specific technology (or standard) P + Q, for a particular characteristic of the input vectors "of a given set.

It follows that the overall function S is <determined by the sum, for each technology, of the> ℎ related to it:

For all combinations of elements (+ R), a value is then calculated indicating the actual affinity between the elements falling on a single group. The best division between elements that have less affinity and that are able to maximize the information they contain is the minimum value:

By way of example, the subdivision proceeds until a minimum number of three samples falling into the group is obtained, which identify an indicative type of installation, or when the ratio between the maximum of the rejects between the number of carriers for a given technology falling into the group and the average of the number of these carriers is less than 0.5.

In an embodiment of the present invention, with particular reference to Figures 3, 6 and 7, said third process P3, associated with a respective second optimization step F2, and a combination of said second process P2 and said third process P3, associated with a respective third optimization step F3, allow to generate, respectively, a second EEI2 index and a third EEI3 index.

Advantageously, said second index defines the ratio between estimated consumption and actual consumption for the site in question according to the following formula:

<where>

= 1, ... is the index of each site 21, .., 2n including the said telecommunications network 2, and is a positive integer,

where is it

said first subarray comprises values of real energy consumption of said site 21, .., 2n, and>

D02i is a further second subarray comprising estimated energy consumption values, determined by said third process P3 associated with said second optimization step F2.

Advantageously, said third index EEI3 defines the ratio between estimated consumption and real consumption for the site in question according to the following formula:

d <where>

i = 1, ... n is the index of each site 21, .., 2n included in

said telecommunications network 2, and is an integer

positive,

where is it

said first subarray Di1 comprises consumption values

<real energy of said site 21, .., 2n, and>

D′3i is a further third subarray comprising values

of estimated energy consumption, determined by

combination of said second process P2 and said third

process P3 associated with said third phase of

optimization .

In an embodiment of the present

invention, said central control unit 4 is

configured to determine said third process P3per

<each site 21, ..., 2n from said subarrays Di1, Di3, Di4,> Di5, Di6, Di7 and Di8 in said optimization phase F2, and

from said subarrays Di1, Di4, Di5, Di6, Di7e Di8e by a

further third subarray D <∗>

i3 in said phase

<optimization F3, where said further third subarray> D ∗

i3 comprises at least one further number of frequencies

radio associated with a respective standard of

<telecommunications on the basis of said second process> P2.

The number of radio frequencies associated with a

respective telecommunications standard, in fact, in

this last case does not correspond to the real number but a

that estimated by said second process P2.

In an embodiment of the present invention, said third process P3 is carried out with the aid of a single fully connected artificial neural network FF having as input, the data included in said subarrays Di1, Di2, Di3, Di4, Di5, Di6, Di7e Di8 in the case of said optimization step F2, and the data <included in said subarrays Di1, Di2, D ∗>

<i3, Di4, Di5, Di6,> Di7 and Di8 in the case of said optimization step.

In an embodiment of the present invention, the fully connected network FF, said third process P3, is made up of 18 input nodes, each of which is connected to all the nodes of the hidden layers. The latter are 4, each having 50 nodes, but with different activation functions and a single neuron output layer.

The first three layers are characterized by said ReLU activation function, while in the case of the last hidden layer and the estimation neuron of the last layer, the Sigmoid activation function is used.

As in the case of the first process P1 discussed above, the configuration of the artificial neural network may be subject to structural changes or to the use of different activation functions, without compromising the functionality of the artificial neural network itself.

Again according to the invention, for each site 21, ..., 2n, said central control unit 4 is configured for:

- checking that said first index EEI1 is higher than a first predetermined threshold S1, and sending, via a further telematic communication channel 5, a respective first signal to a telematic device 6 used by a user 7;

- checking that said second index EEI2 is higher than a second predetermined threshold S2, and sending, through said further telematic communication channel 5, a respective second signal to said telematic device (6) used by said user 7;

- verify that said third index EEI3 is higher than a third predetermined threshold S3, and send, through said further telematic communication channel 5, a respective third signal or a respective fourth signal to said telematic device 6 used by said user 7.

In detail, said central control unit 4 <allows to compare said three indices EEI1, EEI2 and> EEI3 with said respective predetermined thresholds S1, S2 and S3 in order to highlight or not possible malfunctions (or operating levels) of each site 21, …, 2n and at the same time adopt appropriate types of intervention such as, for example, design decisions, repair of malfunctions, use of energy optimization techniques and so on.

In particular, said central control unit 4 is configured to determine:

- a first operating level L1 from a first comparison between said first index EEI1 and a first predetermined threshold S1,

- a second operating level L2 from a second comparison between said second index EEI2 and a second predetermined threshold S2,

- a third operating level L3 or a fourth operating level L4 from a third comparison between said third index EEI3 and a third predetermined threshold S3.

By way of example, the determination of these thresholds S1, S2 and S3 is carried out by evaluating the consistency, accuracy and reliability of the input parameters and the sensitivity of intervention on certain sites rather than others as a function of large-scale management strategies.

In an embodiment of the present invention, the values called predetermined thresholds are the following:

- Threshold 1 or S1: 1 / 0.60 (daily basis), 1 / 0.85 (monthly basis);

- Threshold 2 or S2: 1 / 0.80;

- Threshold 3 or S3: 1 / 0.75.

In the event that said first index EEI1 is higher than said first predetermined threshold S1, said central control unit 4 determines said first level of operation L1, associated with an energy malfunction of site 21, ..., 2n, or a significant deviation, in terms of energy consumption, between the real value and the estimated one.

If said second index EEI2 is higher than said second predetermined threshold S2, said central control unit 4 determines said second operating level L2, associated with a structural type malfunction of the site 21, .., 2n ,.

In the event that said third index EEI3 is higher than said third predetermined threshold S3, said central control unit 4 determines said third level of operation L3, associated with a malfunction of the telecommunications standards used in site 21, ..., 2n.

If said third index EEI3 is lower than said third predetermined threshold S3, said central control unit 4 determines said fourth operating level L4, associated with an energy efficiency of the site 21, ..., 2n.

In fact, as can be seen from Figure 3, if all the aforementioned three indices EEI1, EEI2 and EEI3 are lower than said respective thresholds S1, S2 and S3, the site 21, .., 2n is efficient and functioning.

In practice it has been found that the present invention fully achieves the intended aim and objects. In particular, it has been seen how the method and system for monitoring and optimizing the energy operation of at least one site in a telecommunications network thus conceived allow to overcome the qualitative limits of the known art, as they allow to optimize the energy operation of each node included in the telecommunications network.

Another advantage of the method and of the system according to the present invention consists in the fact that they allow to monitor the energetic functioning of each site included in the telecommunications network.

A further advantage of the method and of the system according to the present invention consists in the fact that they make it possible to identify any malfunctions relating to each site and to adopt corresponding types of intervention.

The present invention has been described for illustrative but not limitative purposes, according to its preferred embodiments, but it is to be understood that variations and / or modifications may be made by those skilled in the art without thereby departing from the relative scope of protection, such as defined by the attached claims.

Claims (24)

CLAIMS 1. Method for monitoring and optimizing the energy operation of at least one site or node (21, ..., 2n) in a telecommunications network (2), said method comprising the following steps: A. receive data from each site (21,…, 2n); B. read said data received from each site (21,…, 2n); C. process, for each site (21, ..., 2n), said data through the following sub-phases: C1. perform a first process P1 to estimate the energy consumption of each site (21,…, 2n) and compare it with the real energy consumption of the respective site (21,…, 2n); and C2. performing a second process P2 to determine a number of carrier radio frequencies associated with respective telecommunications standards for each site (21,…, 2n); And D. determine, for each site (21,…, 2n), a respective level of energy consumption and of said number of carrier frequencies by means of respective predetermined thresholds. Method according to the preceding claim, wherein said phase B further comprises the following sub-phase: B1. organize, for each site (21, ..., 2n), said data in a respective data array, where = 1, ... is the index of each site (21, ..., 2n) included in said telecommunications network (2) , and n is a positive integer. Method according to the preceding claim, wherein said array of data includes: - a first Di1 subarray including real energy consumption values of said site (21, ..., 2n), - a second Di2 subarray including at least one telecommunications standard used in said site (21, ..., 2n), - a third Di3 subarray comprising at least a number of radio frequencies associated with a respective telecommunications standard included in said second Di2 subarray, - a fourth Di4 subarray comprising a structural typology of said site (21, ..., 2n), - a fifth Di5 subarray comprising a number of radio links of said site (21, ..., 2n), - a sixth Di6 subarray including topology / user data of said site (21, ..., 2n), - a seventh subarray Di7 comprising temperature and humidity data of said site (21,…, 2n), and - an eighth subarray Di8 comprising time reference data associated with said site (21,…, 2n). Method according to any one of the preceding claims, wherein said first process processes the data of said subarrays Di1, Di3, Di7 and Di8. Method according to any one of the preceding claims, wherein said first process P1, associated with a respective first optimization step F1, calculates a first index EEI1 according to the following formula: where is it i = 1, ... n is the index of each node or site (21, ..., 2n) included in said telecommunications network (2), and is a positive integer, where is it said first subarray Di1 comprises values of said real energy consumption of said node (21,…, 2n), and D′1i is a further first subarray comprising values of said estimated energy consumption, determined by means of said first process. Method according to any one of the preceding claims, wherein said first process is carried out through a concatenation of an artificial Long Short Term Memory Neural Network (LSTM) with an artificial Feed Forward (FF) neural network. Method according to claim 6, wherein said artificial neural network FF comprises ten input nodes connected with four hidden layers, each of said four hidden layers comprising forty nodes. Method according to claim 7, wherein three of said four hidden layers comprise an activation function of the Rectified Linear Units (ReLU) type and one of said four hidden layers comprises an activation function of the Sigmoid type. Method according to any one of claims 6-8, wherein said artificial neural network LSTM comprises a single layer of sixty nodes or cells of the LSTM type connected to an output layer composed of a single node and to a synthesis layer in which said nodes of said fourth hidden layer of said artificial neural network FF are also connected. Method according to any one of the preceding claims, wherein said second process processes the data of said subarrays Di2, Di3, Di4 and Di6. Method according to any one of the preceding claims, wherein said second process P2 is carried out by means of a non-parametric decision tree method. Method according to any one of the preceding claims, wherein said phase C further comprises the following sub-phase: C3. perform a third process P3 to estimate said energy consumption of each site (21, ..., 2n) and compare it with said real energy consumption of the respective site (21, ..., 2n), on the basis of the totality of said data included in said data array . Method according to any one of the preceding claims, wherein said third process processes the data of said subarrays Di1, Di3, Di4, Di5, Di6, Di7 and Di8 in a second optimization step, and processes the data of said subarrays Di1, Di2, D <∗> i3, Di4, Di5, Di6, Di7e Di8 in a third optimization phase F3, where said subarray D <∗> i3 comprises at least a further number of radio frequencies associated with a respective telecommunications standard on the basis of said second process. Method according to any one of claims 12 and 13, wherein said third process P3, associated with said respective second optimization step F2, calculates a second index according to the following formula: where is it = 1, ... is the index of each site 21, ..., 2n included in said telecommunications network 2, and n is a positive integer, where is it said first subarray Di1 comprises values of said real energy consumption of said site (21,…, 2n), and D′2i is a further second subarray comprising values of said estimated energy consumption, determined by said third process P3. Method according to any one of claims 12-14, wherein said third process P3 is carried out through a fully connected artificial neural network FF. Method according to claim 15, wherein said artificial neural network FF fully connected comprises eighteen input nodes, each of which connected with all the nodes of four hidden layers, each of said four hidden layers comprising fifty nodes, and a layer of single node output. Method according to claim 16, wherein three of said four hidden layers comprise an activation function of the Rectified Linear Units (ReLU) type, and one of said four hidden layers comprises an activation function of the Sigmoid type. Method according to any one of claims 12-17, wherein a combination of said second process P2 and said third process P3, associated with said respective third optimization step F3, calculates a third index EEI3 according to the following formula: where is it = 1, ... is the index of each site (21, ..., 2n) included in said telecommunications network (2), and n is a positive integer, where is it said first subarray comprises values of said real energy consumption of said site (21, ..., 2n), and D′3i is a further third subarray comprising values of said estimated energy consumption, determined by the combination of said second process P2 and said third process P3 associated with said third optimization phase F3. Method according to claims 5, 14 and 18, wherein said step D further comprises the following sub-steps: D1. verify, for each site (21,…, 2n), that said first index EEI1 is higher than a first predetermined threshold S1; D1.1 if said first index EEI1 is higher than said first predetermined threshold S1, determine, for each site (21,…, 2n), a first level of operation L1; D1.2 if said first index EEI1 is lower than said first predetermined threshold S1, proceed with step D2; D2. verify, for each site (21,…, 2n), that said second index EEI2 is higher than a second predetermined threshold S2; D2.1 if said second index EEI2 is higher than said second predetermined threshold S2, determine, for each site (21,…, 2n), a second level of operation L2; D2.1 if said second index EEI2 is lower than said second predetermined threshold S2, proceed with step D3; D3. verify, for each site (21,…, 2n), that said third index EEI3 is higher than a third predetermined threshold S3; D3.1 if said third index EEI3 is higher than said third predetermined threshold S3, determine, for each site (21,…, 2n), a third level of operation L3; D3.2 if said third index EEI3 is lower than said third predetermined threshold S3determine, for each site (21,…, 2n), a fourth level of operation. 20. System (1) for monitoring and optimizing the energy operation of at least one site or node (21, ..., 2n) of a telecommunications network (2), said system (1) comprising a central control unit (4), configured to communicate, through a first telematic communication channel (3), with each site (21, ..., 2n) of said telecommunications network (2), characterized in that said central control unit (4) is further configured to perform the method for monitoring and optimizing the energy operation of said at least one site or node (21, ..., 2n) in a telecommunications network (2), according to any one of claims 1-19. System (1) according to claim 20, characterized in that said central control unit (4) comprises transceiver means (43) for communicating with at least one telematic device (6) through a further telematic communication channel (5) , and with each node (21,…, 2n) through said first telematic communication channel (3). 22. System (1) according to claims 20 or 21, characterized in that said central control unit (4) comprises storage means (44) for storing said data sent from each site (21, ..., 2n) to said unit control unit (4). 23. Computer program comprising instructions which, when said program is executed by a computer, cause the execution by said processor of phases A, B, C and D of said method according to any one of claims 1-19. 24. Computer readable storage medium, comprising instructions which, when executed by a computer, cause said processor to execute steps A, B, C and D of said method according to any one of claims 1-19 .