TECHNICAL FIELD

The invention relates to a method for the categorization of wireless devices through data associated with the signals sent from wireless devices to a radio access node in a radio access network, an apparatus operative to perform the method, a corresponding computer program and a corresponding computer program product.

BACKGROUND

Wireless communications networks, such as cellular networks, enable various human and machine centric services, including providing positioning measurement reporting of user devices for various purposes. Such positioning measurement data is sensitive both from a safety and privacy perspective, which means that mobile network operators have a responsibility and, in certain jurisdictions, a requirement to keep such specific user related information confidential. However, there are important insights to be had from positioning data more generally.

An example of this is information on how many users and their devices that are present in vehicles vs building vs are pedestrians. This type of information would be important for governmental organizations looking to gain further insight into the usage of infrastructure and how to plan further maintenance, improvement, or expansion of the infrastructure. This information would also help the mobile network operators in correctly dimensioning future radio access network resources and deployment of network infrastructure. In order to preserve confidentiality, such use of data would require an anonymized measure of user type that is both detailed enough to distinguish different types of users and general enough to not disclose the identify any individual user. This presents a problem that presently has no solution. Locating and identifying user type in sufficient detail currently requires the use of new dedicated environmental sensors such as cameras and ultrasonic signal-based detectors or detailed location data such as global positioning system coordinates and smartphone telemetry of users which present the problem of being individually identifiable given sufficient data even if anonymized.

Adaptive statistical learning of cellular users behavior by Yehonatan Broyde, Michael Livschitz, and Hagit Messer, Signal Processing, Volume 93, Issue 11, November 2013, discloses a method for dynamically estimating the percentage of indoor vs outdoor usage in cellular sectors, by applying an innovative mixed gaussian model for the accumulated pathloss measurements. SUMMARY

An object of the invention is to categorize wireless devices in a network using the data associated with the signals sent between the wireless devices and a radio access node in the radio access network.

According to a first aspect of the invention, there is an apparatus for categorizing of a plurality of wireless devices, WDs, through data associated with the signals sent between the WDs and a radio access node in a radio access network. The apparatus processor circuitry and a storage unit storing instructions which when executed by the processor circuitry causes the apparatus to become operative to: obtain the data associated with the signals, the data comprising data, L, indicative of received signal strength of the signals and data, T, indicative of time of flight of the signals. The instructions when executed by the process circuitry causes the apparatus to become operative to estimate the number of WDs, Ncategory, in the plurality of WDs that belong to a category of WDs, wherein the category of WDs is defined by a physical characteristic of a location of a WD belonging to the category. The estimating a number comprises determining an interval, TF _i , of time of flight values based on the data, T, indicative of time of flight of the signals. The estimating a number comprises determining, from the plurality of WDs, a number of WDs, N _i , being associated with time of flight values falling within the interval, TF _i , of time of flight values. The estimating a number comprises determining a maximum distance, ' associated with an upper bound of the interval, TF _i .

The estimating a number comprises determining a minimum distance, associated with a lower bound of the interval, TF _i . The estimating a number comprises determining a maximum average pathloss value, , of signals, the maximum average pathloss value associated with the maximum distance, . The estimating a number comprises determine a minimum average pathloss value, of signals associated with the minimum distance, The estimating a number comprises derive a set, Lp, of pathloss values from the data, L, indicative of the received signal strength of the signals. The estimating a number comprises determining a distribution mixture model, f(L\ θ), to fit the set of pathloss values, Lp, the distribution mixture model comprising a first set of distributions θm, thus providing, for each of the distributions. θk, in the first set, θm, a mean value, μ and contribution value, φk The estimating a number comprises determining a second set of distributions. θn, the distributions of which contribute to the distribution mixture model, f(L\ θ), within a first average pathloss interval defined by the determined maximum average pathloss value, and t e determined minimum average pathloss value, The estimating a number comprises determining a maximum average pathloss value, of the signals associated with WDs belonging to the category. The estimating a number comprises determining a minimum average pathloss value, of the signals associated with WDs belonging to the category. The estimating a number comprises determining a third set of distributions, θo, the distributions of which belong to the second set of distributions, θn, and contribute to the distribution mixture model, f(L\ θ), within a second average pathloss interval defined by the determined maximum average pathloss value, of signals associated with WDs belonging to the category and the determined minimum average pathloss value, of signals associated with WDs belonging to the category. The estimating a number comprises dividing a sum of the contribution values, (φo, of the distributions in the third set, θo, with a sum of the contribution values, (φn, of the distributions in the second set. θn, resulting in a contribution ratio. The estimating a number comprises multiplying the contribution ratio by the determined number of WDs to obtain the number of WDs belonging to the category. Hereby is achieved an estimation of the number of devices belonging to categories defined by pathloss in a way that requires no extra network resources or external equipment.

In an embodiment of the first aspect, the data associated with the signals containing only information that is unable to identify individual WDs from the plurality of WDs. Hereby is achieved an enablement of estimation of the number of devices belonging to categories defined by pathloss in an anonymous way. In an embodiment of the first aspect, the data indicative of a set of intervals of time of flight of signals, being derived from signals characterized by a timing advance. Hereby is achieved an enablement of estimation of the number of devices belonging to categories defined by pathloss through the use of less specific measurements. This may increase the efficiency.

In an embodiment of the first aspect, the data indicative of the number of WDs associated with a timing advance value, the WDs associated with signals characterized by a timing advance. Hereby is achieved an enablement of estimation of the number of devices belonging to categories defined by pathloss through the use of anonymous measurements defined by 3GPP standards.

In an embodiment of the first aspect, the set of times indicative of an interval of time of flight of signals, being defined by a timing advance value. Hereby is achieved an enablement of estimation of the number of devices belonging to categories defined by pathloss through the use of anonymous measurements defined by 3GPP standards.

In an embodiment of the first aspect, the distribution being a gaussian distribution. Hereby is achieved a reduction in complexity.

In an embodiment of the first aspect, the apparatus is operative to estimate the distribution mixture model by using the estimation maximization algorithm. Hereby is achieved a more accurate estimation of the number of devices belonging to categories defined by pathloss.

In an embodiment of the first aspect, the apparatus is operative to determine a maximum average pathloss using a pathloss model based on the physical characteristics of a location of the radio access node. In an embodiment of the first aspect, the apparatus is operative to determine a minimum average pathloss using a pathloss model based on the physical characteristics of a location of the radio access node. Hereby is achieved a more accurate estimation of the number of devices belonging to categories defined by pathloss.

In an embodiment of the first aspect, the apparatus is operative to obtain data indicative of a location of the radio access node. Hereby is achieved a more accurate estimation of the number of devices belonging to categories defined by pathloss. This is achieved by using the location of the radio access node to inform the categories of devices and the characteristics of the categories.

In an embodiment of the first aspect, the apparatus is operative to obtain data indicative of a direction of the signals sent from the WDs to the radio access node. Hereby is achieved a more accurate estimation of the number of devices belonging to categories defined by pathloss. This is achieved by using the direction of to inform the categories of devices and the characteristics of the categories. In an embodiment of the first aspect, the category being defined by the WD being located on a pedestrian whereby the pathloss experienced by a pedestrian is defined by open space surrounding the pedestrian. In an embodiment of the first aspect, the category being defined by the WD being located on a pedestrian whereby the pathloss experienced by a pedestrian is defined by open space surrounding the pedestrian. In an embodiment of the first aspect, the category being defined by the WD being located on a pedestrian whereby the pathloss experienced by a pedestrian is defined by open space surrounding the pedestrian. Hereby is achieved a more accurate estimation of the devices belonging to the category.

According to a second aspect of the invention, there is provided a method for categorizing of a plurality of wireless devices, WDs, associated with the signals sent between the WDs and a radio access node in a radio access network. The method comprises obtaining the data associated with the signals, the data comprising data, L, indicative of received signal strength of the signals and data, T, indicative of time of flight of the signals. The method comprises estimating the number of WDs, Ncategory, in the plurality of WDs that belong to a category of WDs, wherein the category of WDs is defined by a physical characteristic of a location of a WD belonging to the category. The estimating a number comprises determining an interval, TF _i , of time of flight values based on the data, T, indicative of time of flight of the signals. The estimating a number comprises determining, from the plurality of WDs, a number of WDs, N _i , being associated with time of flight values falling within the interval, TF _i , of time of flight values. The estimating a number comprises determining a maximum distance, ' associated with an upper bound of the interval, TF _i . The estimating a number comprises determining a minimum distance, ' associated with a lower bound of the interval, TF _i . The estimating a number comprises determining a maximum average pathloss value, of signals, the maximum average pathloss value associated with the maximum distance, .

The estimating a number comprises determine a minimum average pathloss value, of signals associated with the minimum distance, The estimating a number comprises derive a set, Lp, of pathloss values from the data, L, indicative of the received signal strength of the signals. The estimating a number comprises determining a distribution mixture model, f(L\ θ), to fit the set of pathloss values, Lp, the distribution mixture model comprising a first set of distributions θm, thus providing, for each of the distributions. θk, in the first set, θm, a mean value, μk and contribution value, (φk- The estimating a number comprises determining a second set of distributions. θn, the distributions of which contribute to the distribution mixture model, f(L\ θ), within a first average pathloss interval defined by the determined maximum average pathloss value, determined minimum average pathloss value, The estimating a number comprises determining a maximum average pathloss value, of the signaIs associated with WDs belonging to the category. The estimating a number comprises determining a minimum average pathloss value, , of the signals associated with WDs belonging to the category. The estimating a number comprises determining a third set of distributions, θo, the distributions of which belong to the second set of distributions. θn, and contribute to the distribution mixture model, f(L\ θ), within a second average pathloss interval defined by the determined maximum average pathloss value, of signa Is associated with WDs belonging to the category and the determined minimum average pathloss value, of signals associated with WDs belonging to the category. The estimating a number comprises dividing a sum of the contribution values, (φo, of the distributions in the third set, θo, with a sum of the contribution values, φn, of the distributions in the second set. θn, resulting in a contribution ratio. The estimating a number comprises multiplying the contribution ratio by the determined number of WDs to obtain the number of WDs belonging to the category.

In an embodiment of the second aspect, the data associated with the signals contains only information that is unable to identify individual WDs from the plurality of WDs.

In an embodiment of the second aspect, the data indicative of a set of intervals of time of flight of signals, being derived from signals characterized by a timing advance.

In an embodiment of the second aspect, the data indicative of the number of WDs associated with a timing advance value, the WDs associated with signals characterized by a timing advance.

In an embodiment of the second aspect, the range of times indicative of an interval of time of flight of signals, being defined as a timing advance value. In an embodiment of the second aspect, the distribution being a gaussian distribution.

In an embodiment of the second aspect, the method comprising estimating the distribution mixture model by using the estimation maximization algorithm.

In an embodiment of the second aspect, the method comprising determining a maximum average pathloss using a pathloss model based on the physical characteristics of a location of the radio access node. In an embodiment of the second aspect, the method comprising determine a minimum average pathloss using a pathloss model based on the physical characteristics of a location of the radio access node.

In an embodiment of the second aspect, the method comprising obtaining data indicative of a location of the radio access node.

In an embodiment of the second aspect, the method comprising obtaining data indicative of a direction of the signals sent from the WD to the radio access node.

In an embodiment of the second aspect, the method comprising the category being defined by the WD being located on a pedestrian whereby the pathloss experienced by a pedestrian is defined by open space surrounding the pedestrian. In an embodiment of the second aspect, the method comprising the category being defined by the WD being located inside a vehicle whereby the pathloss experienced within a vehicle is defined by a surrounding of either metal or glass equivalent to the structure of a vehicle. In an embodiment of the second aspect, the method comprising the category located inside a building whereby the pathloss experienced inside a building is defined by a surrounding of material equivalent to the structure of a building.

According to a third aspect of the invention, a computer program is provided. The computer program comprises computer readable instructions which is run on processing circuitry of an apparatus for categorizing of a plurality of wireless devices through data associated with the signals sent between the wireless devices and a radio access node in a radio access network. The computer readable instructions cause the apparatus to obtain the data associated with the signals, the data comprising data, L, indicative of received signal strength of the signals and data, T, indicative of time of flight of the signal. The computer readable instructions cause the apparatus to estimate (220) the number of WDs, Ncategory, in the plurality of WDs that belong to a category of WDs, wherein the category of WDs is defined by a physical characteristic of a location of a WD belonging to the category. The estimating a number comprises determining an interval, TF _i , of time of flight values based on the data, T, indicative of time of flight of the signals. The estimating a number comprises determining, from the plurality of WDs, a number of WDs, N _i , being associated with time of flight values falling within the interval, TF _i , of time of flight values. The estimating a number comprises determining a maximum distance, associated with an upper bound of the interval, TF _i . The estimating a number comprises determining a minimum distance, ' associated with a lower bound of the interval, TF _i . The estimating a number comprises determining a maximum average pathloss value, of signals, the maximum average pathloss value associated with the maximum distance, . The estimating a number comprises determine a minimum average pathloss value, of signals associated with the minimum distance, p The estimating a number comprises derive a set, Lp, of pathloss values from the data, L, indicative of the received signal strength of the signals. The estimating a number comprises determining a distribution mixture model, f(L\ θ), to fit the set of pathloss values, Lp, the distribution mixture model comprising a first set of distributions θm, thus providing, for each of the distributions. θk, in the first set, θm, a mean value, μk and contribution value, ( k The estimating a number comprises determining a second set of distributions,θn, the distributions of which contribute to the distribution mixture model, f(L\ θ), within a first average pathloss interval defined by the determined maximum average pathloss value, ' and he determined minimum average pathloss value, The estimating a number comprises determining a maximum average pathloss value, of the signals associated with WDs belonging to the category. The estimating a number comprises determining a minimum average pathloss value, of the signals associated with WDs belonging to the category. The estimating a number comprises determining a third set of distributions, θo, the distributions of which belong to the second set of distributions. θn, and contribute to the distribution mixture model, f(L\ θ), within a second average pathloss interval defined by the determined maximum average pathloss value, - of signals associated with WDs belonging to the category and the determined minimum average pathloss value, , of signals associated with WDs belonging to the category. The estimating a number comprises dividing a sum of the contribution values, φo, of the distributions in the third set, θ o, with a sum of the contribution values, φn, of the distributions in the second set, θn, resulting in a contribution ratio. The estimating a number comprises multiplying the contribution ratio by the determined number of WDs to obtain the number of WDs belonging to the category.

According to a fourth aspect of the invention a computer program product is provided. The computer program product comprises a computer program according to the third aspects of the invention. The computer program product comprises a computer readable storage medium on which the computer program is stored.

BRIEF DESCRIPTION OF DRAWING

The accompanying drawings, which are incorporated herein and form part of the specification, illustrate various embodiments.

Figure 1 is a diagram showing an example of a scenario where the embodiment may be used.

Figure 2a is a flow chart showing a process according to an embodiment.

Figure 2b is a flow chart showing a process according to an embodiment.

Figure 3 is a diagram showing an example of a gaussian mixture obtained by an embodiment.

Figure 4 is a diagram illustrating an example of a scenario according to an embodiment.

Figure 5 is a diagram illustrating a distribution of times of flight of signals according to an embodiment.

Figure 6 is a diagram illustrating a distribution of received signal strength according to an embodiment.

Figure 7 is a diagram illustrating an example of a scenario where an embodiment may be used.

Figure 8 is a diagram showing functional units of an apparatus according to an embodiment.

Figure 9 is a diagram showing functional modules of an apparatus according to an embodiment.

Figure 10 shows one example of a computer program product comprising computer readable means according to an embodiment. DETAILED DESCRIPTION

The invention will now be described more fully herein with reference to the accompanying drawings, in which certain embodiments of the invention are shown. This invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided by way of example so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.

The invention presented enables the capability to categorize devices according to their physical characteristics using data gathered in an anonymized fashion. This allows for network operators, governments, and business to understand the type of mobile devices, and by extension users, that are in an area without compromising the privacy or integrity of the users therein. This enables such use cases as knowing how many devices are traveling on a specific segment of road or how many pedestrians us a sidewalk without extra measuring equipment beyond the already existing base station.

At least some of the following variables may be used in this disclosure. If there is an inconsistency between variables, preference should be given to how it is used above. If listed multiple times below, the first listing should be preferred over any subsequent listing(s).

• T : data indicative of a set of times of arrival of signals sent between a plurality of devices and a radio access node

• N: number of devices

• TF: a set of intervals based on T

• i : an integer representative of an interval in the set TF of intervals

. TF _i : an interval of time of flight values based on T comprising TF

• N _i : number of wireless devices (WDs) associated with time of flight values falling with an interval of time of flight values indicated by i

• TA : a set of values corresponding to one or a plurality of an interval of times that may be subsets of T

. TA _i a value in the set TA corresponding to a range of times

• d: a distance between a WD and a base station

• : a maximum distance between a WD and a base station corresponding to an upper bound of an interval of time of flight values

• : a minimum distance between a WD and a base station corresponding to a lower bound of an interval of time of flight values • L : data indicative of a set of received signal strength of the signals sent between a plurality of devices and radio access node

• Lp : a set of pathloss values derived from L

• f(L\ θ): distribution mixture model of Lp

• θ: Set of component distributions of the mixture model

• k: integer representative of a component distribution in the set θ

• j : set of values representative of a plurality of component distributions in the set θ

• m : set of values representative of a plurality of component distributions comprising the first set of distributions θ

• n : set of values representative of a plurality of component distributions comprising the second set of distributions

• o: set of values representative of a plurality of component distributions comprising the third set of distributions

• o_category: set of values representative of a plurality of component distributions associated with devices belonging to a category in the set θ

• o_vehicle set of values representative of a plurality of component distributions associated with devices belonging to the vehicle category in the set θ

• o_building set of values representative of a plurality of component distributions associated with devices belonging to the building category in the set θ

• o_pedestrian \ set of values representative of a plurality of component distributions associated with devices belonging to the pedestrian category in the set θ

• θk. Set of parameters associated with component distribution k

• μk. mean of component distribution k

• φk. contribution value of component distribution kin the mixture model

• average pathloss for dmin average pathloss for dmax

• minimum average pathloss for a category

• maximum average pathloss for a category

• : ean average pathloss for signals associated with T Fi l _building : maximum pathloss attributable to a wall or structure of a building l _{min vehicle} : minimum pathloss attributable to a vehicle l _{max vehicle} : maximum pathloss attributable to a vehicle

Fig. 1 schematically illustrates an example of a scenario where the embodiment of the current disclosure may be used. Fig. 1 features a radio access network 100 wherein a radio access node, illustrated here as a base station 110, is connected to wireless device, WDs. The WDs comprising a WD 120 carried by a pedestrian, a WD 130 inside a vehicle, and a WD 140 inside of buildings. Fig. 1 further comprises an apparatus 115, for the categorization of WDs with data obtained from the radio access node.

In some embodiments the wireless communication network 100 is an LTE network, a 5G NR network, or a form of future wireless communication network such as, as yet unspecified, 6G. In certain embodiments, a radio access node 110, which in LTE is referred to as an eNB and in 5G NR is referred to as a gNB, may for example be a high-power macro base station, a low-power base station or a relay node.

The radio access node communicating via signals referred may act as a single transmitter receiver antenna or may have directional antennas or antenna arrays. The communication directional antenna or antenna array may direct signals in specific directions or area. These directions or areas may be referred to as both cells and beams which are both equally applicable to the concepts described herein. In certain embodiments, a WD is any type of device which has access to a cellular communication network by wireless transmitting and/or receiving signals to a base station. Some examples of such as WD include but are not limited to, a User Equipment device in a 3GPP network, and a Machine Type Communications device.

The apparatus 115 may be a part of the radio access node 110. The apparatus may be separate from the radio access node but may comprise part of the radio access network 100. The apparatus may be completely separate and comprise a communication interface for obtaining data from the radio access node.

Fig. 2 illustrates a flowchart showing an example of a method 200 performed by, but not limited to, an apparatus 115 for the anonymous and accurate categorization of WDs and their approximate locations. The apparatus may be located in the radio access node 115, a different part of the radio access network 100, or completely separate. A category of a WD is defined by a physical characteristic of a location of a WD. Such physical characteristics may include surrounding objects such as buildings. vegetation, or glass. Physical characteristics may also include but are not limited to, atmospheric phenomenon, terrain, altitude of the WD and altitude in relation to the radio access node, and distance from the radio access node. Examples of categories may be but are not limited to, WDs outdoors used by, for example, pedestrians or attached to, for example, environmental objects; WDs inside vehicles; or WDs inside of buildings or other non-mobile structures. Other potential categorizations of WDs maybe rapidly moving outdoor WDs or WDs attached to the outside of vehicles.

The method of Fig. 2, the apparatus begins with a first step 210, by obtaining data associated with the signals sent between the WDs and the radio access node. The data comprises data indicative of received signal strength of the signals and data indicative of time of flight of the signals. In the present embodiment, a pathloss is estimated from the received signal strength in a known manner. The measurements of the pathloss form the data set, L _p .

In one embodiment, the apparatus obtains data indicative of a set of times of flight of signals sent between a plurality of WDs and a radio access node, represented as the variable T. This data may also be derived from signals characterized by a shift in transmission time based on the time of flight in contrast to the time of flight of signals. This shift represents the required shift that must occur in the transmission of signals from the WDs to compensate the time of flight of the signals and synchronize with transmission time slots defined by the radio access network. The set of times of flight of signals, T may be divided into one or more intervals of time of flight. These one or more intervals of time of flight form the set, TF. The data may instead comprise TF along with an accompanying number of WDs, N ⁱ belonging to each interval of time of flight in the set. Each interval of time of flight in TF is represented as TFi where / is an integer representative a specific interval of time of flight in the set,

TF.

The data may also comprise shifts corresponding to a specific intervals of times of flights, TF _i i TF. These shifts correspond to a value of time that every WD may be assigned with which to shift forward its transmission timing based on the time of flight of the signal.

Knowledge of the value of the TFor TFi parameter enables the approximation of the distance or range of possible distances between a WD and the base station with which it is communicating. A specific time of flight corresponds to an approximate distance d A specific range of time of flight values would correspond to an approximate minimum distance, the smallest time of flight and a maximum distance, the largest time of flight that a WD could have had to the base station. at the time of transmission. Thus, knowledge of the time of flight or range of times of flight of a WD or a plurality of WDs allows for a grouping of WDs. This could also allow for the base station to obtain a value representing the number of WDs that share a range of times of flight. Given the times of flight described above, the WDs in a group of similar times of flight, or in a range of times of flight, will have been at a similar distance from the base station at the time of transmission. The further utility of these distances are described in steps 235 and 240.

In another embodiment, this data may derived by signals characterized by a timing advance, TA. The main difference between TF and TA is that while TF is a set of intervals of times, TA is a set of values corresponding to discrete sets that in total may comprise or exceed the range of time of flight of the signals corresponding to T. The timing advance, TA, therefore, comprises multiple timing advance values TA _i . TA _i is one of the values of discrete sets where the integer I may signify a specific value, TA from the set of values TA. For example, TA may correspond to the intervals of times {0:1 *10 ^-9 , 1 *10 ^-9 : 2*10 ^-9 , 2*10 ^-9 :3*10 ^-9 , 3*10 ^-9 : 4*10 ^-9 } and. TA2 corresponds the set {1 *10 ^-9 : 2*10 ^-9 }. In the example the TA would comprise the set of values {0.5*10 ^-9 , 1.5*10 ^-9 , 2.5*10 ^-9 , 3.5*10 ^-9 } corresponding to the set of sets of time above and TA2 equaling the value {1.5*10 ^-9 }. The time, TA _i , represents the required shift that must occur in thetransmission of signals from the WDs to compensate a time of flight of the signals and synchronize with transmission time slots. These shifts are discrete in nature and are defined in for example in 3GPP standard TS 38.211, V 15.2.0. The discrete shifts mean that every WD is assigned a fixed value, TA _i , with which to shift forward its transmission timing based on the time of flight of the signal. Knowledge of the TA _i , and the set of times it corresponds to, thus allows for an approximation of the range of possible distances between a WD and the base station with which it is communicating. A specific timing advance value corresponds to an approximate minimum distance and a maximum distance that a WD could have had to the base station, at the time of transmission. Thus, knowledge of the value of the TA parameter of a plurality of WDs allows for a grouping of WDs based on their timing advance. In one embodiment, TA and N ⁱ , is derivable from the data indicative of a set of time of flight of the signals, being derived by signals characterized by a timing advance ortiming advance value. In another embodiment, N ⁱ is derivable by counting the times of flight in the set of time of flight of the signals corresponding to each TA _i in TA . In a different embodiment, the data, T, may also contain and is a value representing the number of WDs that share a timing advance value. Given the discrete shifts described above, the WDs in a group will have been at a similar distance from the base station at the time of transmission.

One embodiment of the method 200 is when the data is associated with signals sent between the WDs and the radio access node over a longer period of time, for example in the order of minutes. Due to the likelihood of WDs traveling over a distance, the signals sent at different points in time likely have different times of flight. Signals from the same WD thereby risk corresponding to multiple values in the set TA or multiple intervals of times of flight of TF. In this embodiment, the method 200 is appropriate for determining the number of WDs in each category of WD in a single TA or TF ring or sector based on the determined and distances for the corresponding set in TA or TF. Rings within the distances would be in the case of an omni directional base station antenna and sectors of area within distances in terms of directional antennas of, or beams from, a base station.

Another embodiment of the method 200 is when the data is associated with signals sent between the WDs and the radio access node over a much shorter time frame, for example in the order of seconds. Devices are unlikely to have traveled very far and are therefore unlikely to have changed their time of flight of the signal so as to appear in multiple values of TA or multiple intervals of times of flight of TF. The WDs that do change their value in the set TA could also be compensated for by including a counter for the number of WDs that do change their TA value during the measurement interval and discounting a number of the WDs counted from the final values. This approach would allow for a base station wide, antenna wide, or beam wide accounting of the categories of WDs with a communication link.

Fig. 2 describes the embodiment of method 200 where data is associated with signals sent between the WDs and the radio access node taken over longer time duration, for example over a 15 minute time span.

Once the data is obtained, the apparatus then proceeds, in a second step 220, to estimate the number of WDs in a plurality of WDs that belong to a category. Th category is defined by a physical characteristic of a location of a WD belonging to the category. The number of WDs belonging to a category is estimated by conducting step 220, comprising steps 230 through 270, for a category.

A first step 230 of step 220 is performed by determining an interval, TF _i if times of flight to analyze from a set TF . This is done by first grouping the set of times of flight, T, of the signals sent during transmission from WDs to the base station into intervals of time, TF _i , forming the set of intervals TF. The determining of TFmay be performed either before, at the time of, or after obtaining of data associated with the signals. The determining may also be conducted in a WD agnostic fashion whereby only data indicative of average times of flight per device is provided as opposed to data associated with the signals. The intervals of time may be uniform in terms of the range of time of the interval or may vary for each interval. Furthermore, an interval of time may be predetermined, or determined after the data is obtained. The intervals of time may be determined based on the amount of data gathered and over how long. If more data was available, sets with smaller intervals of time may be set and visa versa for less data.

TF _i may then be chosen based on the interest of evaluating an area at a certain distance from the base station or may just be an average distance from the base station represented by TF _i or data corresponding to TFi may correspond to the most WDs and therefore contain the most data to analyze or may be chosen for some external reason or completely arbitrarily.

The apparatus also determines from the plurality of WDs, a number of WDs, Ni, being associated with the time of flight values falling within the interval TFi of time of flight values. N _i may be included with the data, T. N _i may be derived from the number of time of flight values falling within the interval TF . N _i may also be derived from data corresponding to each TA _i in TA. N _i may also be determined from data associated with the WDs and gathered from one or a plurality of different radio access nodes or devices.

The apparatus, in a second step 235, then determines the minimum distance, d _min , associated with the lower bound of the chosen interval TF _i of TF. This is done by determining the minimum distance traveled by signals with the times of flight of the lower bound, of the chosen interval of the signals. The apparatus, then determines the maximum distance, dmax, associated with the upper bound of the chosen interval TF _i of TF. This is done by determining the maximum distance that could be traveled by signals with the times of flight of the upper bound, ,of the chosen interval of the signals.

The apparatus enters a third step 240. At 220, for the time of flight interval i the apparatus determines a maximum average pathloss, using a pathloss model capable of estimating pathloss in the environment. The apparatus then determines minimum average pathloss, using a pathloss model capable of estimating pathloss in the environment. The embodiment used to generate the graph of Fig. 3, uses a pathloss model referred to as the COST HATA model, which is particularly suited to environments with a wireless communications network using a carrier frequency of between 800 and 2000 MHz and substantial urbanization. The minimum average pathloss for time of flight interval i, , is determined using the distance Similarly, the maximum average pathloss for time of flight interval i, is determined using the distance - Pathloss is the loss of power of a signal sent from the transmitter to the receiver over the channel between them. The data, Lp, indicative of pathloss is derived from the data, L, indicative of a set of received signal strength of the signals sent between a plurality of WDs and a radio access node. Lp , may be determined by subtracting a transmission power of the radio access node by the received signal strength derived from the data, L.

An average pathloss is defined as an average of pathloss experienced by signals sent between a single WD of the plurality of WDs and the base station. The average pathloss is dependent on a variety of factors including but not limited to terrain, objects in an environment comprising the base station and the plurality of WDs (Buildings, vegetation, vehicles, etc.), propagation medium, the distance between the transmitter and receiver and the location of the antennas. A pathloss model is typically used to consider one or several of these factors. A pathloss model typically uses values of the variety of factors including, but not limited to, height of the base station, height of the WD, carrier frequency, weather, building density and type of antenna. When considering one or several of factors, a WD has a maximum average pathloss, when the factors being considered have the greatest impact. A WD similarly has a minimum average pathloss when the factors being considered have the least impact.

The apparatus then, in a fourth step 245, determines a distribution mixture model, to fit the data, Lp, indicative of the pathloss of the signals. In other embodiments, the apparatus may determine a distribution mixture model, to fit the data, L, indicative of received signal strength of the signals. The distribution mixture model, comprises a first set of distributions, θm, comprising one or a plurality of component distributions θ. The component distributions may be in the form of gaussian distributions. The component distributions may also be in the form of other distributions such as log-normal distributions. A single component distribution, θk, comprises a mean μ , and a contribution of the component distribution, φk , in relation to the mixture model. The distribution mixture model may be determined using an estimation maximization (EM) on the data, L, indicative of received signal strength of the signals associated with the WDs associated with TF _i . The result of the EM algorithm is the mixture model, f(L\ θ), comprising the set of component distributions, θ. The data, Lp, can be represented by a gaussian mixture model, where the pathloss distributions of multiple categories can be approximated by a series of individual gaussian models forming the mixture. These categories may include WDs used by pedestrians, WDs in the vehicles, and WDs in the buildings. An example of the mixture model according to the embodiment is more clearly shown in Fig. 3. In other embodiments, other optimization algorithms may be used to generate a mixture model such as gradient descent and conjugate gradient.

Fig. 3 schematically illustrates an example of the mixture model obtained from an embodiment of the current disclosure where graph 300 illustrates a probability distribution function of a distribution mixture model vs pathloss derived from data, Lp. This mixture model, f(L _P \ θ) , 310 is created using an estimation maximization algorithm. The estimation maximization matches a set of component distributions, θ, in a way which minimizes error between the mixture of component distributions and the underlying data, Lp. The underlying data, Lp, is formed by the pathloss of signals between a WD or plurality of WDs and a base station, antenna, or beam. In the present embodiment, the error between the estimated distribution mixture model and the underlying data is determined using a Bayesian information criterion, but other error estimation functions found in the state of the art could be used, such as an Akaike information criterion. The linear combination of Gaussian distributions are Gaussian distributions which allow for their combination through addition and thereby are suitable for matching data in this way. The circles 320, 330, 340, 350, 360, and 370 represent the individual mean values, μk of the component distributions, θ, of the mixture model.

Once the pathloss distribution components are estimated according to estimation maximization, the apparatus then in a fifth step 250 determines a second set of distributions, θn, which contribute to the distribution mixture model, f(L\ θ) within a first average pathloss interval defined by the determined maximum average pathloss value, , and he determined minimum average pathloss value, - he pathloss interval is associated with the data indicative of interval TF _i ,

To reiterate, i is an integer representative of a interval of the set TF. kis an integer representative of a specific component distribution in θ of the distribution mixture model, f(L\θ). θk comprises the mean, μ k, and the weight, φ k, of the component distribution in relation to the mixture model. μ is defined as pathloss, φ is defined as being a percentage, j is a set of integer values used to represent a subset of component distributions of θ. n is a set of integer values used to represent the subset θn of component distributions of θ. Given that the categories may impact the pathloss of a subset of the WDs, the pathloss incurred by the category with the biggest impact must be taken into account when considering the maximum possible average pathloss, which may be beyond simply (305). This is due to the fact that the pathloss model used to determine assumes no walls or similar obstructions are between a WD inside a building and a radio access node. Depending on the categories chosen, other such factors impacting the maximum possible average pathloss not considered by the pathloss model should be compensated for. The same may be considered of within the minimum average pathloss, (301) if the pathloss model overestimates the minimum average pathloss when compared to a certain category. An example of this may be that the pathloss model assumes non line of sight transmission which would have a worse minimum average pathloss when compared to a category considered, which has line of sight as a physical characteristic and possibly a better minimum average pathloss. In the current embodiment, to account for a user, belonging to TF _i , being inside a building the following equation is used:

The value of is then increased by the maximum pathloss attributable to a wall or other structure of a building, I _{buil ding} , which may be determined empirically or set according to values well known in the art. This is shown as line 306 in Fig. 3. This results in the component distributions of θk being determined to contribute to the mixture model for pathloss values associated with TF _i An example of this is shown in figure 3, where the component represented by mean 370 is outside of the Possible for TF _i since at the distance between the base station and a WD, the pathloss does not show that a WD may have an average pathloss of mean 360 or greater ,where the greatest average likely, is modeled to be In a sixth step 255, the maximum average pathloss value, of signals associated with WDs belonging to the category is determined by the apparatus. The minimum average pathloss value, of signals associated with WDs belonging to the category is also determined by the apparatus. These boundaries are dependent on the category and the physical characteristic of the location of the WD belonging to the category.

A simple example of this may be a scenario with two categories of WDs. The first category of WDs exist in free space. The second category of WD exist behind an obstacle. The maximum average pathloss value of the first category is the greatest pathloss of a signal propagating in free space and the minimum average pathloss value of the first category is the smallest pathloss of a signal propagating in free space. The maximum average pathloss of the second category however is the same as the first but with the added pathloss incurred by the signal passing through the obstacle. The minimum average pathloss of the second category is the same as the first category but with the added pathloss incurred by the signal passing around the obstacle.

In some examples, the boundaries of other categories may need to be determined. In embodiment presented in figure 1, the categories considered are WDs used by pedestrians, WDs inside vehicles, and WDs inside buildings. The categories are expected to have a different pathloss given the difference in their physical location in relation to the base station. For example, a WD in a building will have a higher pathloss compared to a WD used by a pedestrian given the fact that the building's walls will be located between the WD and the base station. Similarly, a WD in a vehicle will have a higher pathloss compared to a WD used by a pedestrian. However, a WD in a vehicle will have a lower pathloss when compared to a WD inside a building, given the comparatively large window area in a vehicle. While there are certainly exceptions to the pathloss behavior of each of these physical characteristics of the locations of the WDs, the data, L, may assist with this if the data is indicative of an average of the received signal strength of the signals per WD. This would allow for the suppression of potential outlier pathloss data.

In the example of WDs belonging to the vehicular category, the average pathloss for the signals of a WD is below the average pathloss of all WDs given the large amount of metal and typically poor transmission angles inherent in with transmission from inside vehicles. To account for this the mean of the minimum and maximum values, and is taken to determine the mean average pathloss, • for a WD with signals associated with TF _i . can be seen with line 302. A maximum average pathloss value of the signals associated with WDs belonging to the category would be the mean average pathloss, • with the addition of the maximum pathloss associated with the window of the vehicle, vehicle- This result can be seen with line 304. A minimum average pathloss value of the signals associated with WDs belonging to the category would be the mean average pathloss, , with the addition of the minimum pathloss associated with the window of the vehicle, - This result can be seen with line 303.

The apparatus, in a seventh step 255, determines a third set of distributions, θo, the distributions of which belong to the second set of distributions, θn, and contribute to the distribution mixture model, f(L\ θ), within a second average pathloss interval defined by the determined maximum average pathloss value, of signals associated with WDs belonging to the category and the determined minimum average pathloss value, , of signals associated with WDs belonging to the category. In a continuation of the example of WDs belonging to the category of WDs inside a vehicle, the apparatus determines which component distributions, 3 from the larger set of component distributions do contribute to the mixture model. To determine the components distributions belonging to the category of WDs inside vehicles, the following range is used: where vehicle is the set comprising all means (340) of the component distributions, - that, for WDs in a vehicle associated with T F _i , contribute to the mixture model.

Both the distributions, and are now identified as contributing to the components of all WDs associated with TF _i and to the components of WDs belonging the vehicle cate associated with TF _i respectively.

The apparatus in an eighth step 265, is divides the sum of the contributions values, , of the distributions in the third set, θo, with a sum of the contribution values, , of the distributions in the second set, θn, resulting in a contribution ratio. In the example this done by determining the sum of and the sum of respectively. The two summations are then divided by each other to form the ratio.

The apparatus, in a ninth step 270, multiplies the contribution ratio by the determined number of WDs to obtain the number of WDs belonging to the category. In the example of the category of WDs in a vehicle, the determined ratio is multiplied by the total number of WDs, N ⁱ , to determine the number of WDs belonging to the vehicle category associated with TF _i . Steps 265 and 270 are shown in the equation: resulting in the total number of WDs inside vehicles associated with T F _i .

To calculate the number of WDs used by pedestrians, the same process described in relation to steps 255 to 270 can be used, where, instead of expression (2), the following is used: where pedestrian is the set comprising all means (320, 330) of the component distributions, pedestrian , that contribute to the mixture model for WDs used by a pedestrian associated with TF _i .

To calculate the number of WDs used by inside building, the same process described in relation to steps 255 to 270 can be used, where, instead of expression (2), the following is used: where building is the set comprising all means (350, 360) of the component distributions, builidng • that contribute to the mixture model for WDs inside a building associated with T F _i . Once the apparatus has estimated the distribution of WDs across the categories for WDs associated with TF _i , the apparatus may proceed to repeat step 220 for TF _i+x , where i + x equal an integer corresponding to a different subset of TF. This may be to estimate the distribution of WDs in categories for a different area defined by and according to T F _{i+ x} .

The values representing the number of WDs in each category result in an estimate for the number of WDs in each category for TF _i and corresponding to the area between the distances between d _min and d _max from the base station whilst maintaining anonymity for any of the WDs being measured, since the bulk data set contains no identifiable or traceable information. This is simply not possible with more WD focused tracking methods, such as cameras, global positioning system or smartphone telemetry.

Fig. 4 schematically illustrates an example of a scenario wherein embodiments of the current disclosure have been implemented. In this scenario, the apparatus obtains data from the base station 110 which the data comprises data, L, indicative of received signal strength of the signals and data, T, indicative of time of flight of signals. This data may be received signal strength of signals and time of flight of signals. The data may also be indicative of a determined timing advance set, TA , and N _i corresponding to TA _i of TA. The signals are to and/or from a series of WDs of various categories: WDs used by pedestrians, WDs in a vehicle, or WDs in a building at different ranges, both with line of sight and without. The lines are representative of the shortest path the most powerful transmission signal may take to reach the base station. This an example of how a distance between the base station and a WD may be estimated from a time of flight of a signal or an associated timing advance value. This is also an example of how the environment and the distance between the base station and a WD may impact the received signal strength. The white blocks are representative of buildings. Devices 402, 403, 408 and 409 are outdoor units either representing pedestrians or stationary machine WDs. Devices 401 and 405 are inside vehicles. Devices 404, 406, and 407 are all inside buildings. Devices 402 and 403 have line of sight while 401 has partial line of sight, with the vehicle structure partially obstructing the path of the signal, and the other WDs have non-line of sight. In other implementations, altitude also might have an impact, for example in relation to line of sight in dense urban environments. In this embodiment, all the named WDs are connected to the base station 110 through a single directional antenna pointed in the direction of the WDs. Fig. 5 illustrates a diagram illustrating an example of an embodiment of the present disclosure. In Fig. 5 signals from the WDs 401-409 in Fig. 4 are shown along the axis representing time of flight with appropriate timing advance values, TA i, 510, 520, and 530 displayed along the axis.

Fig. 6 illustrates a diagram illustrating an example of an embodiment of the present disclosure. In Fig. 6, signals from the WDs 401-409 in Fig. 4 are shown along the axis of received signal strength representing average received signal strength of the signals. As can be seen with particularly the second TA i value 520 and the WDs 404 and 405 inside. The received power of WDs 404 and 405 in the scenario illustrated in Fig. 4 are significantly different, the received power of vehicle WD 405 being closer to that of outdoor WDs 408 and 409. However, when compared with each other in terms of the second TA i value 520, it becomes very easy to see in the graph 404 and 405 respectively as they are far apart from each other in terms of power but closer together in terms of timing advance and therefore distance. Given the knowledge that the second TA i value 520 contains 2 WDs and using the pathloss characteristics described in method 200, the difference in power would distinguish them. An opposite example would be in regards to WDs 405, 408, and 409, all of whom share similar received signal strength. This would seem to indicate similar characteristics and therefore similar categories. However, taking into account the second and third TAi values 520 and 530, it is apparent that these WDs are different categories. 405 is a vehicle closer in distance and belonging to the second TAi.408 and 409 are used by pedestrians but farther away in distance from the base station and belonging in the third TA i This relationship between received signal strength and distance makes it possible to use method 200 considering that the timing advance values 540 and 530 will account for the WDs where without the timing advance consideration, these two values would be conflated into the same category.

Fig. 7 schematically illustrates an example according to an embodiment of the current disclosure. In the embodiment, base station 110 has an antenna array which has the ability to direct individual beams. Antenna arrays using beams and directional antennas both have to ability to focus their signals more effectively and thereby reduce the area covered by the antennas. This allows for a smaller physical area to be analyzed by the apparatus. This further al lows for statistics to be gathered in a more detailed fashion in terms of a smaller area on a per beam or per antenna basis. The smaller area defined by d _min and d _max and the antenna's coverage combined with the embodiment of obtaining data associated with signals sent between the WDs and the radio access node over a much shorter time frame could allow for near real time analysis of the categorization of WDs in the area around a base station while maintaining the anonymity for any single WD. This is simply not possible through other more WD specific means. Additionally, the antenna and/or beam placement may improve planning for an implementation of the method 200 . Due to the confined measurement area covered by a beam, data obtained may be more closely related to the area defined by d _min and d _max resulting in more location specific pathloss modeling. The smaller area defined by d _min , d _max , and the antenna's coverage may also result in certain contributions to the mixture model being included or not depending on aspects of the area. Certain contributions to the gaussian mixture model may not be included if for example, the smaller area is known to not include buildings for a specific T Fi. In the example of figure 3, this would result in the component distribution 360 to be determined to not contribute since l _building would not considered.

In Fig. 7, example of smaller areas defined by the coverage of the multiple beams (701, 702, 703, 704, 705, 706) from an array antenna (110) include: on a road covered by beams 702 and 703 or, a plaza with pedestrians covered by beam 704 or, buildings covered by beams 701 and 706. There may also be instances where there is a limited blend of WDs in an area covered by an antenna without one or a plurality of categories of WDs being present in the area. An example of a limited blend of WDs would be for beam 705 where there only exist WDs used by pedestrians and WDs in buildings.

Fig 8. is a block diagram of the apparatus 115 according to some embodiments. As shown in Fig. 8, the apparatus 115 may comprise: processing circuitry 810, which may include one or more processors such as general purpose microprocessors application specific integrated circuits (ASIC), field- programmable gate arrays (FPGA), or any other suitable processing circuitry.); a communications interface 820 for receiving and/or sending information; and a storage medium 830, which may include one or more non-volatile storage WDs and/or one or more volatile storage WDs such as random access memory (RAM). In embodiments where the apparatus includes a programmable processor 810, a computer program product comprising a computer readable medium 1020 may be provided, such as, but not limited to, the storage medium 830. The storage medium 830 may be a magnetic media, optical media, memory WD, or the like. The storage medium 830 may contain a computer program 840 containing computer readable instructions 1040 that when executed by the processor circuitry 810 causes the processor circuitry to perform operations according to embodiments disclosed herein. According to other embodiments, processor circuitry 810 may be defined to include a storage medium, hence, a separate storage medium is not required.

Fig. 9 is a diagram showing functional units of a network function in the apparatus 115 according to some embodiments. As shown in Fig. 9, the apparatus comprises a number of functional modules; an obtain module configured to perform step 210 and an estimate module configured to perform step 220. In general terms, each functional module may be implemented in hardware or in software. Preferably, one or more or all functional modules may be implemented by the processing circuitry, possibly in cooperation with the communications interface and/or the storage medium. The processing circuitry may thus be arranged to, from the storage medium, fetch instructions, thereby performing any steps of the network function 115 or the base station 110 as disclosed herein.

Fig. 10 is a diagram showing a computer program product 1010 according to an embodiment of the invention. As shown in Fig. 10, the computer program product 1010 comprises a computer readable medium 840 storing a computer program 1030 comprising computer readable instructions 1040. The computer readable medium may be, a magnetic media (e.g., a hard disk), optical media, memory WDs (e.g., random access memory, flash memory) and any other suitable computer readable medium.

While various embodiments of the present disclosure are described herein, it should be understood that they have been presented by way of example only, and not limitation. Thus, the breadth and scope of the present disclosure should not be limited by any of the above-described exemplary embodiments. Moreover, any combination of the above-described elements in all possible variations thereof is encompassed by the disclosure, unless otherwise indicated herein or otherwise contradicted by context.

Additionally, while the processes described above and illustrated in the drawings are shown as a sequence of steps, this was done solely for the sake of illustration. Accordingly, it is contemplated that some steps may be added, some steps may be omitted, the order of the steps may be re-arranged, and some steps may be performed in parallel.