US 2012007 1157A1 (19) United States (12) Patent Application Publication (10) Pub. No.: US 2012/0071157 A1 MARKOULDAKS (43) Pub. Date: Mar. 22, 2012

(54) METHOD FORMOBILE NETWORK (52) U.S. Cl...... 455/423:455/422.1 COVERAGE EXPERIENCE ANALYSIS AND MONITORING (57) ABSTRACT (76) Inventor: Yannis MARKOULIDAKIS. Nea In a telecommunications network including a plurality of Smyrni (GR) s mobile terminals, a system and method of providing an indi y cation of network coverage in relation to at least one of the plurality of mobile terminals, the method and system being (21) Appl. No.: 13/101479 configured to: determine the occurrence of incoming call events in relation to the at least one terminal; record the (22) Filed: May 5, 2011 occurrences together with an identity of at least one appli O O cable terminal, a location and/or a time associated with the (30) Foreign Application Priority Data communication event; and analyse the recorded incoming May 5, 2010 (GB) 1OO7507.5 call event data, particularly unreachable terminal events, to determine an indication of experienced network coverage for each of the at least one terminal. In this way, an indication of Publication Classification actual user experience can be determined, which is usable in (51) Int. Cl. indicatively treating customer complaints, making network H04/24/00 (2009.01) improvements and analysing mobile terminal performance.

&ixi: 388xx:k:::::::

Patent Application Publication Mar. 22, 2012 Sheet 1 of 4 US 2012/0071157 A1

FIGURE

Patent Application Publication Mar. 22, 2012 Sheet 2 of 4 US 2012/0071157 A1

(RE

888 & 38 3xx88 ::::::::::::::::::::: 8:38&

Patent Application Publication Mar. 22, 2012 Sheet 3 of 4 US 2012/0071157 A1

(RE3

&::::::::::::

::::::::::::::::::::::::::: is:::::::::: 33

8xxxxx xxxix: {}: i: ifk:x: 38:

8:::::::::::: {x : 38

Patent Application Publication Mar. 22, 2012 Sheet 4 of 4 US 2012/0071157 A1

FORE &

sessssssssssssssssssssssssss

3i due :: US 2012/007 1157 A1 Mar. 22, 2012

METHOD FORMOBILE NETWORK assumed distribution of customers according to their place of COVERAGE EXPERIENCE ANALYSIS AND residence. In mature networks (e.g. Western Europe), popu MONITORING lation coverage is generally above 99%. Geographic coverage is also based on propagation prediction tools and field mea FIELD OF THE INVENTION Surements, but provides the extent of network coverage over the physical geography of a country (e.g. taking into account 0001. The present invention relates to a cellular telecom factors like flat terrain, mountains, islands, Sea, etc.) Both munications network and the management thereof. More par population and geographical coverage estimates are based on ticularly, the present invention relates to a system and method signal strength thresholds corresponding to acceptable out for monitoring Voice call events, particularly unsuccessful door or indoor signal levels. For 3G networks, the 3G network call attempts to/from a mobile terminal, and even more par coverage threshold can also consider service related param ticularly unsuccessful call attempts due to the terminal not eters like Video-Telephony, HSDPA 1 Mbps throughput, etc. being reachable, such as would occur in out of coverage 0014 Typically in order to derive geographical coverage, situations. Even more particularly, the present invention the country map is divided into squared units called “pixels' relates to a system and method for providing network data and based on a propagation prediction model provides the analysis in relation to out of coverage situations. coverage in each pixel. To achieve this, the coverage predic tion tool estimates the signal strength occurring per pixel for BACKGROUND each base station of the radio network. As for any pixel there 0002 Unsuccessful call attempts encompass both missed may be multiple base stations which provide some level of calls terminating to a mobile and the inability of a mobile coverage, the propagation prediction tool assigns to this pic terminal to initiate calls. In mobile/cellular network perfor ture the maximum received signal strength from a single base mance analysis, the rate of missed calls for a user is a Key station. It should be noted that the pixel size can be variable as Performance Indicator (KPI), and there are a number of situ e.g. in urban areas a higher accuracy is required and thus a ations where a missed call may occur, including where: Smaller pixel size can be used versus the one used in rural 0003 (a) the called user's mobile terminal was switched areas. A specific signal strength threshold is defined (e.g., off: i.e. turned off by the user or by the terminal itself in the corresponding to “outdoor or “indoor coverage acceptable event of a depleted battery, both of which result in the termi signal levels) so as to filter the results and provide the per nal sending a network detach message before shutting down; centage of geographical area which is covered as follows: this case also includes the de-activation of mobile network access without shutting down the terminal (e.g., flight-mode). 0004 (b) the called user was busy: i.e. on another call or 1 A (1) not answering the ringing terminal; and Peo coy (A., 6) SA) XPess4 (i,0), S(j) 0005 (c) the called user was not reachable/non-con i=l tactable by paging procedure indicating out of coverage or other conditions with the same effect: i.e. terminal attached to the network, but non-responsive to the paging message sent Where: by the network on the occurrence of the incoming call. 0006 From a customer viewpoint, the first two missed call I0015 P., (A,0): The percentage (%) providing the situations, (a) and (b) listed above, are generally acceptable, predicted geographical coverage in area Abased on Received since they are within the user's control. The third point (c), Signal Strength (RSS) threshold 0 (indbm) corresponding to however, is out of the user's control, and therefore likely to certain coverage conditions (e.g. indoor); cause dissatisfaction and be viewed as a service inadequacy. 0016 A: The area A can be the whole country or any Similarly the inability to initiate a call is likely to be due to the region of the country; terminal entering a non-reachable state, and again be viewed 0017 M: The number of pixels into which the area A has as a service inadequacy. been divided into by the propagation prediction tool; 0007. The problem of the terminal entering a non-reach 0018 S(A): the surface area of area A (in square km); able state may be due to a number of issues, such as: 0019 S(): the surface of the pixel j (in square km). This 0008 i) network coverage extent or continuity problems approach allows for variable pixel size. If a fixed pixel size is (i.e. the user of the mobile terminal may have moved and used for the entire area then s()/S = 1/M in Eq. (1); and entered a region with no signal coverage or with only a weak 0020 Piss (i,0) is a coverage characterization function signal coverage); defined as follows: 0009 ii) mobile terminal associated problems (e.g. termi nal antenna sensitivity, performance of the terminal receiver, etc.); 1 if RSS(i) > 0 (2) 0010 iii) network protocols and standards associated PRSs (i. 6) = { () if RSS(i) < 0 problems (e.g. 3GPP idle mode terminal operation stan dards); 0011 iv) rarely occurring conditions e.g. local interfer (0021. Where RSS() is the predicted RSS level (indbm) in ence in paging channels; or pixelj. 0012 V) any combination of the above. 0022 Population coverage uses the same methodology of 0013 Currently the network coverage of a mobile network geographical coverage prediction, however, it additionally is described by KPIs such as Population Coverage and Geo takes into account the geographical distribution of the popu graphic Coverage. Population coverage is calculated using lation's residence as reported in the national demographic propagation prediction tools and field measurements for an statistical databases. US 2012/007 1157 A1 Mar. 22, 2012

0023 The population coverage can then be expressed as customer experiences based on their actual geographic posi follows: tion. In other words the actual user experience on coverage in a particular position is dependent upon various factors, including their actual physical situation (e.g. indoors/out MA (3) doors) and also the actual population distribution at that time Pop coy (A. 0) = Npop (A) 2.X PRSsRSSJ (i. 6). N,popj (i) (e.g. due to peak hour traffic or an entertainment event). Therefore, the factors to consider are more dynamic than those taken into consideration in the population and geo graphic coverage KPIs. Where: 0031. In other words, in practice these KPIs provide an indication of the expected customer experience, however, the I0024 P., (A,0): The percentage (%) providing the actual user experience in terms of coverage (i.e. the user is predicted population coverage in area Abased on RSS thresh making calls/sessions or receiving calls in his/her actual posi old 0 (in dbm) corresponding to certain coverage conditions tion during the day) is a more complicated issue since popu (e.g. indoor) and the population's residence distribution in lation distribution may vary by a set of factors including at area A. least the following: (0025 N. (A): The population with reported residence in area A. 0.032 (a) the time of the day (e.g. located at home, (0026 NG): The population with reported residence in travelling to work, located at work, travelling back pixelj. It should be noted that as the pixel size is relatively home, etc.); small (e.g. 50 mix50 m) the population with residence in this 0033 (b) working/non-working days (e.g. population specific pixel may not be available at this accuracy. Therefore distribution over the week-ends is typically different to a this is usually calculated based on the population data refer working day and it may also depend on weather condi ring to a wider area A, around this pixel (e.g. a city, a munici tions); pality, etc.) as follows: 0034 (c) holiday seasons: during these periods and depending on the country the population distribution may significantly change towards tourist resorts (ski S(i) (4) centre, sea-side, etc.) including remote and low popu No (j) s No.(A) lated areas in the rest period of the year. 0035. The actually experienced network coverage can be expressed as follows: 0027. Where j is a pixel in area A, S(i) the surface of the pixel, S(A,) the surface of the area A, and N.(A) the population of area A. Assuming that the area A is divided into West (A,t) (6) Sub-areas A, (n=1,2,..., N.) the population coverage can be 1 expressed as follows in relation to the geographical coverage Pacto (A, D = M.A., 2, PRSS (Gros (i, t), 6MT (i)) of area A:

Pop coy (A. 0) = (5) Where, NA 0036 P. (A,t): the actually experienced coverage by 1 1 MA customers located at time instance tin area A expressed as the No.(A) Xsc.) SA) 2. PRSs (i. 6). S(i) => percentage (%) of customers in area A which are in coverage; 0037 N(A,t): The number of customers located in area NA Popo (A,0) =XP (A): Poco (A,) A at the time instance t; =l I0038 G.(it): The actual position of customer i at time instance t; 0039 0 (i): the RSS threshold corresponding to the sen Where P(A)-N-(A)/N,(A) the portion of the popu sitivity of the Mobile Terminal (MT) receiver of customer i: lation in area A with residence in area A. 0028 Such KPIs as geographical and population coverage 10040 Pess(G.(it).0,(i)) is the coverage characteriza are exploited for understanding the extent of coverage pro tion function which in this case is defined as follows: vided by a network, in addition to being key drivers for prioritising investments in network coverage expansion both in terms of Volume as well as in terms of focusing investment 1 if RSSc (Gos (i, t)) > 6MT (i) (7) to certain geographical areas. PRSS (Gros (i, t), 6MT (i)) = { 0 if RSSc (Gos (i, t)) < 0 MT (i) 0029. As it can be observed both KPIs are static indica tions of network coverage. Moreover, their accuracy is also Subject to the geographical data available (e.g. terrain details, (0041) Where, RSS(G.(it)) is the actual RSS value (not a pixel size) as well as the accuracy of the propagation predic predicted one) for the actual geographical position of cus tion model used (e.g., Hata like empirical models, Cost-231 tomer i at time instance t (expressed as G.(it)). model, ray tracing, etc.) 0042. It should be noted here that this function can be 0030. Although these KPIs are a very useful reference for considered as the coverage experience of customer i at time understanding the extent of offered network coverage, they instance t P. (MT,t) . do not provide an accurate picture of the actual coverage the P. c. (MT,t)-PRSs (Ge(i,i),0ft(i)) (8) US 2012/007 1157 A1 Mar. 22, 2012

0043. In order to derive the actually experienced coverage 0050 P(A) is the penetration of mobile customers of for a time period T starting at time instance t and ending at the operator under study in the population of area A. In time instance t+T the following formula applies: relation to the operator customer penetration in the entire country the following applies:

1 t+T (9) Pact coy (A, T) = . Pact coy (A, t) di NA (14) T i Pur(A) =XPur(A). No (A,) 0044 And for the actually experienced coverage of a single customeri (due to Eq. (8)) we have: 0051. It is important to note here that the penetration of customers in an area A, is a function of the coverage provided 1 frt 1 +T (10) by the operator in this area. For example in a green-field Picco (MT, t) = . P. (MT, t) at network deployment the customers Subscribe only in areas T J. where there is coverage. This fact may create a variance in the customer penetration per area A. Once a network has 0045. From equations (6) to (10) it is evident that the reached a mature coverage level (e.g. population coverage estimation of the actually experienced coverage is a very hard higher than 80%) then this variance may get reduced. issue as it is not possible for the network operator to have 0.052 To further simplify the approach we assume that for exact information on the actual customer position at all time all areas a common distribution between environments instances as well as the actual RSS values at the specific applies and a common penetration rate also: position that each customer is located. 0046. In order to further understand the difference a (A.T.)-a(A,T), PT(A)=PT(A) (15) between KPIs like population coverage and actually experi 0053 Under these assumptions we can express the actu enced coverage the actual coverage per pixel unit used in the ally experienced coverage as follows: population coverage analysis can be adopted: Assuming that customers are distributed in a number of k (k=1,2,....K) different coverage environments (e.g., indoor, outdoor, deep K (16) indoor, etc.) each characterized by a known RSS threshold 0. Pact coy (A, Tr) & X. ak (A, Tr). Pop coy (A, 6k) common for all terminals. Moreover, we may assume that at k=1 a small area equal to a pixel unit the number of customers positioned in environment of type k is statistical stable during a time period T and equal to N.C.k.T). The same applies for 0054 From Eq. (14) it is evident that even under a signifi the total number of customers located in the area A (N, cant number of simplifying assumptions the approximation (AT)). In this case the actually experienced coverage can be ofactually experienced coverage from population coverage is expressed as follows: rather difficult to be estimated. 0055. There is therefore a need for an alternative approach to measure and analyze coverage, particularly where custom 1 K MA (11) ers experience problems. Such an approach would be also be Pico (AT) six A in XX Pess (i,0): Na?j, k, T) advantageous for better understanding the network coverage. 0056 Further, a modern trend in mobile network opera tion is the so called “customer centric management'. Cus 0047 Assuming a stable distribution of customers in sub tomer centric management is an area of network data analysis areas A, for time period Twe get: which aims to provide KPIs on a per MSISDN basis (i.e. a per user basis) in order to characterise individual customer expe rience. This analysis is useful in various applications such as (12) addressing customer complaints (e.g. those that might arise Pact coy (A. T) & NA. T) CiiSt due to a large number of missed calls outside their control), or tracking the Service Level Agreements (SLAB) offered to NA K 1 MA corporate customers. XX. Nes (j(j, k, T). Sch, 2. PRSS (i., 6). S(i) => 0057 Moreover, the cellular telecommunications industry does not take a single, common, approach to the assessment NA K of network coverage of different networks in the same or P A, T) as different regions. Thus while geographic/population cover act coy (A. T) Ncist (A, T) XXN (j, k, T); Poco (A,i,0) age KPIs are typical, it is often difficult to compare these models directly as they can be based on different assumptions (e.g. propagation prediction models, received signal strength 0048. In order to understand the relationship between the actual experienced coverage versus the population coverage thresholds, geographical database accuracy level, etc.) There we consider a time-period T, which corresponds to an ideal is therefore also a need to be able to compare the coverage situation which all customers are located very close to their offered by different networks in an analogous manner. residence (e.g., late evening hours) i.e., SUMMARY OF THE INVENTION Ns,(Ak, T-)-Plata,) N-(A)'a (A.T.) (13) 0.058 According to a first aspect the present invention Where: provides, in a telecommunications network including a plu 0049 a (A.T.) is the portion of customers in area A, rality of mobile terminals, a method of providing an indica located in environment k during time period T. tion of network coverage, the method including: US 2012/007 1157 A1 Mar. 22, 2012

0059 determining the occurrence of incoming call events 0069. This further aspect of the invention may also be in relation to at least one of the plurality of mobile terminals: implemented as a system, particularly as a network server 0060 recording a result of each incoming call event, implementing Such a method. This further aspect of the inven together with at least one call property; and tion, particularly when applied to a statistically significant 0061 analysing the recorded incoming call data to deter number of communication initiation attempts, advanta mine an indication of experienced network coverage for theat geously enables a realistic and comparable indication of least one terminal, a given terminal type, a given location actual network coverage to be obtained without requiring and/or a given time period. specific network information regarding reasons for call set up 0062 According to another aspect the present invention failures which is not always available at the signalling mes provides, in a telecommunications network including a plu Sages. rality of mobile terminals, a network element configured to 0070. Other aspects of the invention are set out in the provide an indication of network coverage in relation one or attached claims. more of the plurality of mobile terminals, a given mobile terminal type, a given location area and/or a given time period BRIEF DESCRIPTION OF THE DRAWINGS the network element configured to: determine the occurrence 0071. For a better understanding of the present invention a of incoming call events in relation to at least one of the number of embodiments will now be described, by way of plurality of mobile terminals; record a result of each incom example, with reference to the accompanying drawings, in ing call event, together with at least one call property; analyse which: the recorded incoming call data to determine an indication of 0072 FIG. 1 illustrates an example telecommunication experienced network coverage for the one or more mobile system architecture useful in describing embodiments of the terminal, a given terminal model type, a given location and/or invention; a given time period. 0073 FIG. 2 illustrates a terminal status state model useful 0063. These aspects of the invention can advantageously in describing embodiments of the invention; associate call events with customer experience. That is, the 0074 FIG. 3 illustrates a flow chart for a measurement analysis of the communication events may be utilised to procedure according to an embodiment of the invention; and assess, and thereafter improve, a user's mobile terminal usage experience by identifying unreachable terminal situations, 0075 FIG. 4 illustrates a flow chart for a measurement particularly out of coverage situations, and determining the procedure according to a further embodiment of the inven cause, or at least a likely cause. The network experience tion. indication can be used to initiate network improvements focused in areas where coverage problems are detected Such DETAILED DESCRIPTION as by optimizing base station radio parameters (e.g. antenna 0076 FIG. 1 illustrates an example telecommunications direction, tilt, power, etc.) in certain identified areas or network configuration in which embodiments of the present deployment of additional sites. Moreover, considering the invention may be implemented. customer centric model certain types of customers can be 0077. The mobile terminal (1) may be any suitable por targeted (e.g. customers complaining for coverage problems, table device, including but not limited to a handheld mobile high usage customers, etc.) telephone, a personal digital assistant (PDA), dedicated traf 0064. Advantageously, it has been recognised that users fic navigation device or a laptop computer equipped with a actual distribution and need for communication is dynamic, network connectivity datacard. and accordingly these aspects of the invention are capable of 0078. The telecommunications network illustrated capturing actual user experience. These aspects of the inven includes a number of base stations (BS) 3,4,5,7A,8,9, where tion are considered to be better correlated with users’ needs to each BS includes an antenna system which serves a number of make and receive calls than the population and geographic cells (e.g. three-sector cells) operating in one or multiple KPIs currently in use. spectrum bands (e.g. 900 MHZ, 1800 MHz, 2100 MHz, etc.) 0065. More specifically, these aspects of the invention and implementing one or multiple mobile network standards enable an indication of experienced network coverage to be (e.g. GSM/GPRS, 3G/HSDPA, LTE, etc.) of the cellular/ derived from unsuccessful call attempts to/from a mobile mobile telecommunication network. The mobile terminal terminal, particularly those due to out of coverage events. will synchronize with a BS, typically the one providing it with 0066. According to a further aspect, there is provided in a the strongest signal (usually, but not necessarily, the closest telecommunications network including a plurality of mobile BS) in order to be able to send and receive communications terminals, a method of providing an indication of network across the telecommunications network. Each base station coverage in relation to at least one of the plurality of mobile serves a cell, and a set of cells controlled by the same core terminals, the method including: determining the initiation of network switch node (e.g. the SGSN in UMTS RTM) is a call set up procedure from the at least one terminal; moni defined as a location area (LA). toring the duration of the call setup procedure; and identify 0079. When a terminal moves from one Location Area ing a failed call attempt when the call setup procedure dura (LA) to another, or is Switched on, its change in status is tion exceeds a predetermined threshold. updated through a Location Registration procedure to a home 0067 Preferably, where a failed call attempt is identified, location register (HLR) 10. This Location Registration pro the timed set up duration is used to determine areason for the cedure involves a core network node storing a unique terminal failed call Such as an unanswered paging event. identifier, typically the IMSI, along with the LA identity, so 0068. This approach can be achieved through monitoring that network traffic can be correctly directed towards the of the network protocols or signalling links and is applicable mobile terminal. to all types of mobile terminating calls (i.e., calls originating 0080. Then, when a terminating call is initiated towards from and terminating to mobile customers of the same mobile the mobile terminal, the setup phase for terminating the call to network and calls terminating to mobile customers registered the mobile terminal includes the following steps: to in a network different than the one that the originating party 0081 (a) HLR interrogation where the terminal status is registered to). (attached/detached) as well as the LA where the terminal US 2012/007 1157 A1 Mar. 22, 2012

is currently roaming are retrieved (note that in the gen 0091. There is also a possibility for a terminal to enter the eral case a mobile terminal may roam to another network state “out of coverage 23 from the “busy’ 22 state. Such an according to the roaming standards in which case the event will cause the so called dropped-call case during which a call is forced to be released due to lack of adequate cover HLR indicates the target network); and age. I0082 (b) the core network switching node serving the 0092. With this background in mind, a first embodiment of target LA triggers a paging procedure through the radio the invention will now be described. This embodiment of the access network that serves the target LA. During this invention serves to provide a more accurate indication of the paging procedure all cells covering the target LA broad network coverage actually experienced by a user. More spe cast a proper signalling message through the paging cifically, this embodiment of the invention provides an indi signalling channels with a proper terminal identity (e.g., cation of the likelihood that a terminal is in an “out of cover in GSM this identity is the Temporary Mobile Sub age 23 state. 0093. In this embodiment, the data is preferably gathered scriber Identity (TMSI)). The terminal is periodically at the MSC/VLR level. In this regard, the MSC and/or one or monitoring the paging channels of the cell to which it has more network servers associated with the MSC, are config synchronized with and in the case that it detects its own ured so as to gather data relating to mobile terminating calls identity it responds to the networkin order to identify the discriminating between the following cases: exact cell through which the call setup process can be 0094 unsuccessful paging events; completed. In the case that the network receives no 0.095 incoming calls towards terminals for which the periodical location registration timer has expired; response from the paged terminal after a time-out 0.096 successfully established incoming calls; and threshold there may be an option to repeat paging broad 0097 incoming calls to busy. cast for this terminal in the same LA. If there is still no 0098. This data is preferably gathered and recorded in a response from the terminal then the terminal is appar database associated with the core network (e.g. the MSC/ ently not reachable and the paging procedure is com VLR), together with indicators of the terminal/user involved, pleted with an “unsuccessful outcome. the location and/or the time of the event. The location indi cation could be an indicator of the serving cell in case of a 0083. Whether or not a terminal can make and receive Successful paging event or incoming call to busy. Alterna calls will depend upon its status. As illustrated in FIG. 2, the tively, the location indicator may be in regard to a LA i.e., a state model for a communication attempt will typically have group of base stations serving a particular region (e.g. a four different states: Location Area/Routing Area for MSC 2 or BS3-BS5). I0084. Idle, in coverage 20; 0099. The periodic location registration procedure is an 0085 Switched off 21; additional source of events which, at the occurrence of an I0086 Busy 22; and incoming call, can indicate out-of-coverage conditions. I0087 Out of coverage 23. According to this procedure each mobile terminal after the completion of an interaction with the network (e.g. a call, a 0088. When the terminal is idle and in coverage 20, it is session, a location update, etc.) resets a timer with a network able to perform all types of communication and signalling controlled time threshold which is known to both the terminal procedures (e.g., periodic location area update (LAU) regis and the network (e.g. 3 hours). If the timer exceeds the thresh trations, receive paging requests, make originating calls, etc.) old the terminal performs a periodic location registration and In other words, the standard cell selection and reselection resets the timer. If at the time that the timer expires the process applies. When a terminal transits from idle and in terminal is out-of-coverage then it is not feasible to perform coverage 20 to “switched off 21, a location deregistration the periodic location registration procedure. In Sucha case the mobile terminal performs a location registration once it enters occurs, so the network knows it is in this state. In the case of network coverage again. At the network side the periodic an incoming call towards a Switched off terminal the call location registration is exploited as follows: on the event of setup is terminated at the interrogation phase as the HLR the arrival of a call terminating to a mobile terminal the indicates that the terminal is Switched off (i.e., paging proce network estimates the time period between the current time dure is not employed in this case). This may occur when the and the time-stamp indicating the moment that the periodic terminal is manually switched off by the user or due to an location registration timer was reset (i.e., the most recent exhausted battery. When the terminal is subsequently interaction with the terminal took place). If this time period is Switched on, it transits to "idle and in coverage 20 again, and longer than the timer threshold then the network concludes a location registration is performed, so that the network that the terminal is out-of-coverage otherwise a periodic loca knows the terminal's status and current location area for any tion registration would have been performed on the expiration Subsequent paging requests. of the timer and does not proceed with the paging procedure. 0089. When a paging request to the mobile terminal is 0100. According to the above description on periodic Successful and there are available resources to serve a call, the location registration procedure it is evident that the network terminal transits from "idle and in coverage 20 to “busy' 22. can monitor at the arrival of an incoming call the cases where This transition will similarly occur where the terminal ini the terminal is not reachable due to unsuccessful periodic tiates a call which is successfully set up. The terminal will location registration procedure associated to the ability of the terminal to communicate with the network. In the vast major remain in the busy state 22 (also referred as dedicated mode in ity of the cases this phenomenon corresponds to out-of-cov the standards) until the call is released, thereby returning the erage conditions. It is to be appreciated that there are a set of terminal to "idle and in coverage 20. extraordinary cases where this may also apply such as the user 0090. The final state is that of “out of coverage 23 which extracting the terminal battery while the terminal is switched will typically occur when the terminal remains Switched on, on (and thus the terminal cannot perform a detach procedure) but the received signal strength level is below a relevant or the terminal having some malfunction, etc., but these cases threshold. This state will be unknownto the network, as when are not common and are unlikely to adversely affect the the terminal moves out of coverage, no signalling notification results. takes place. In the out of coverage 23 state, the terminal is not 0101 The following Tables 1 and 2 provide an example of able to perform location area updates, nor is the networkable the types of information that can be gathered in this embodi to reach the terminal via paging. The terminal will exit this ment of the invention. Table 1 refers to the collection of data state when it moves back into coverage, thereby being able to on a per MSISDN basis while Table 2 refers to the collection Successfully perform a location registration. of data in the form of event counters: US 2012/007 1157 A1 Mar. 22, 2012

TABLE 1.

A number B number (i.e. calling party) (i.e. called party) Date Time LA ID Ewent +301234567890 +301234567891 26 Dec. 2009 00:03:45 100A Busy 3O1234567892 - 301234567893 26 Dec. 2009 OO:12:01 102E Unanswered Paging +301234567894 +301234567895 26 Dec. 2009 01:24:38 100A. Unanswered Paging +301234567896 +301234567897 26 Dec. 2009 02:12:22 101B Successful Call +301234567898 +301234567899 26 Dec. 2009 02:13:45 100A Expired Periodic LR timer

TABLE 2 Expired Busy Unanswered Per. LR Answered LAID Date Time Period Attempts Paging Timer Paging 100A 26.12.2009 10:00-10:15 1253 458 S4 10834 100A 26.12.2009 10:15-10:30 142O S42 58 14231 100A 26.12.2009 10:30'-10:45 1325 488 43 12394 100A 26.12.2009 10:45'-11:00 1290 423 42 11034 100A 26.12.2009 11:OO'-11:15 1120 394 35 10235

0102) A procedure that may be used for obtaining this logged in any suitable manner, including as a event table, as information is illustrated in FIG. 3. The MSC/other network illustrated in the example of Table 1, or even as event counters server is configured to monitor network signalling (30) in on a per terminal basis (i.e. for each terminal being moni order to intercept the information regarding incoming call tored, have counters for successful calls, etc.) which are events (i.e. mobile terminating call set up). The calling party linked to the relevant locations of the events. Alternatively if identity (Party A), called party identity (Party B), date and the scope of the measurement is the network and not the time information is the standard data set transmitted and individual customer event counters per LA can be monitored shared in the network, and so is readily obtained. As an deriving statistics for specific time periods (e.g. every 15') as alternative to the MSC/server monitoring the network signal shown in the example of Table 2. ling for this data, however, the core network may be config 0111. The information compiled can be used in various ured to send the necessary information to the server compil ways in order to provide at least the following types of results: ing the data. 0112 Network coverage analysis of the entire network; 0103) From the network protocol operation at the network 0113 Network coverage analysis for part of the net nodes, the server will be able to determine the occurrence of work up to the level of Location Areas: different types of mobile terminating call setup events: 0114. User centric analysis: coverage experience of a 0104 Incoming call towards a switched-off terminal particular customer; (31) 0115 User Group centricanalysis: coverage experience 0105 Incoming call towards a mobile terminal for for a particular group of customers; which the periodic location registration timer has 0116 Terminal centric coverage experience analysis: expired (32) coverage experience associated to a specific terminal 0106 Incoming call towards a mobile terminal for type; which the paging procedure was unsuccessful (i.e., 0.117 Time based analysis of the network coverage: unanswered paging) (33) 0118 Daily analysis 0107 Incoming call towards a busy mobile terminal 0119 Time Zone analysis (e.g. morning hours, work (34) ing hours, evening hours, etc.) 0.108 Successful mobile terminating call (35) I0120 Week-ends vs. normal working days analysis 0109. Once a missed call event is detected, and the reason I0121 Holiday seasons (summer, Christmas, etc.) determined, the server will make a record of the event (36). I0122. It should be noted that the analysis of terminal per Alongside the record of the event, the server will also record formance mentioned in the above list requires the identifica an ID of the called party (37), and possibly also the calling tion of the device identity of the called mobile user. This party ID, and the time of the event (39). The server may also information is only available for the calling mobile user (apart record the called party location (38) in terms of the Location from where the MSISDN the International Mobile Equip Area (LA) recorded in the HLR as the current terminal loca ment Identity or IMEI is available as part of the call setup tion. An advantage of doing this is that this LA can then be signalling). However, typically only the MSISDN of the associated with the LA to which the mobile terminal registers called user is made available at call setup phase. To identify when it moves back into coverage, which may assist in iden the device used by a called mobile of the home mobile net tifying black spots in the network coverage. work the IMEI information provided in the CDRs (Call 0110. This information can be continually recorded and Detailed Records). To achieve this for a specific called logged as required. It is to be appreciated that the data may be MSISDN it is necessary to identify a CDR corresponding to US 2012/007 1157 A1 Mar. 22, 2012

a call at which the target MSISDN is the calling MSISDN. In such a CDR the IMEI of the target MSISDN will be recorded and thus retrieved. P. coy (MT, T) (18) 0123 Based on the following two facts it is possible to Putos, MT, T = Ti, r, derive a relation between the above mentioned measured data and the actually experienced network coverage: 0.124 (a) from a statistical viewpoint the arrival process Where: of calls terminating to mobile terminals is recognised to I0133. P., (MTT) is the steady state probability of be well approximated by a Poisson arrival process and the “switched on terminal MT, to be in the out of coverage 0.125 (b) the terminal state model presented in FIG. 2 is state during time period T. actually a Markovian state model (because the state at 10134) P(MTT) is the steady state probability of the which the mobile terminal (MT) transits to, depends terminal MT, to be in the “switched off state during time only on the previous state of the terminal). period T. 0126 From the above two facts it can now be derived that the portion of incoming calls corresponding to inability of the 0.135 The probability provided in Eq. (16) can be esti network to reach the terminal (i.e. unanswered paging and mated based on the network measurements as follows: expired periodic LR timer) can be used to estimate the Out of-Coverage probability assuming a steady state equilibrium Nu PAG (MT, T) + NE PLR(MT, T) (19) for the Markovian model of FIG. 2: Pout colson(MT-TsMT, T) as NMT, T-NMI, T,

At PAC (MT, T) + E PLR(MT, T) = (17) 10136. Where N(MTT) is the number of calls terminat cal in (MTT). P. coy (MT, T) => ing to terminal MT, during time period T for which the ter Au pac (MT, T)+AE PLR(MT, T) minal was in “switched off state. P. coy (MT, T) & call in (MT, T) I0137 The above KPI is meaningful for MTs which receive NupAG (MT, T) + NE PLR(MT, T)f T at least one incoming call while in Switched-on state. Notti, (MT, T)f T > 0.138. An issue that needs to be handled in such estimation Nu PAG (MT, T) + NE PLR(MT, T) is the calling user behavior in the case that a called mobile P. coy (MT, T) & Nitt (MT, T) terminal is not reachable due to lack of coverage. In Such a case the calling user receives an answer from an automatic response system that the called user is not reachable and depending on the case the call may be diverted to the called Where: user's Voice Mail. The network operators have introduced I0127 W. (MTT) is the rate and N. (MTT) the more advanced solutions for Such cases like the so called number of unsuccessful paging events for mobile terminal i “Call Catcher service. According to this service, as soon as (corresponding to a certain MSISDN) during the time period the terminal is again attached to the network, the called user T will receive an SMS with information on the missed calls I0128 (MTT) is the rate and N. (MTT) the during the time of not being reachable (switched off or out of number of terminating call arrivals to mobile terminal i dur coverage). In such a case the calling party may listen to a ing the time period T for which the periodic location regis message like “the called user will be notified of your call as tration timer has expired. soon as the terminal becomes reachable'. I0129 W. (MTT) is the rate and N. (MTT) the 0.139. In some cases, calling users may perform a number number of terminating call arrivals to mobile terminal i dur of repeated calls to the same unreachable user due to the need ing the time period T. for on-the-spot communication to the called party and possi 0130 P. (MTT) is steady state the probability that bly where they consider that the called user's unreachability is the MT, is out of coverage during time period T. In other due to a problem of a temporary nature. As a result of Such words an estimation of the portion of time during which the cases the main assumption of "Poisson arrival process for MT is out of coverage. incoming calls is violated and may distort the estimation of 0131 Actual Coverage Experience of Individual Custom out-of-coverage probability KPIs. To deal with this issue an CS appropriate filter can be applied in the MSISDN based data 0.132. In order to approach the estimation of the actually (see Table 1) where calls originating from a user (MSISDN1) experienced network coverage by the customer we should and terminating to the same user (MSISDN2) with called take into account the fact that we are probably interested only party being in out-of-coverage State (i.e. either unsuccessful for switched on terminals as switched off terminals in prin paging is performed or the periodic LR timer has expired) ciple correspond to a state where the user is aware of being not with a time difference less than a threshold (e.g. 1 minute) are reachable (either intentionally or due to battery outage which filtered out and ignored in the estimation of out-of-coverage apparently is not related to network coverage). In this context KPIs. Therefore the KPI of Eq. (19) can now be expressed as the following probability should be estimated: follows: US 2012/007 1157 A1 Mar. 22, 2012

0146 Actual Coverage Experience in specific Areas 0147 To derive the relevant KPIs for a specific area A (e.g. Nu PAG (MT, T) + NE PLR(MT, T) - (20) a Location Area) then the following KPI for the estimation of Nu REp(MT, . T) the out-of-coverage probability for switched on MTs in area Pout coyls (MT T) s A during time period T (P. (A.T.N)) can be applied: Nu REp(MT, . T)

1 Nd Ncust (A,id) (23) 0140. Where N. (MT.S.T) is the number of repeated Pucovis (A, T, Na) is s NX X. calls within the time thresholds (e.g. 1 minute) from the same MT to MT, when MT, was not reachable due to either unsuc Nu PAG (MT. Ta) + NE PLR(MT. Ta) - Nu REP (MT. S. Ta) cessful paging or due to expired periodic LR timer. Nail in (MT, T4) - Nef (MT, T) - Nu REp(MTig, T4) 0141. It is considered that the above estimation is accurate 1 Nd Ncist (A.T.d) provided that a substantial number of events are occurring so is NX X Pucovis (MT. Ta) as to have statistical stability of the estimated probability. This may be an issue where we consider the coverage expe rience of a particular mobile terminal as a limited number of 0.148. In terms of statistical stability of this KPI it can be incoming calls are taking place on a daily basis. In order to seen that as a high number of MTs is served by a LA this KPI achieve some statistical stability in the estimation of the cov can achieve Statistical stability even for a single day (N-1). erage experienced by a specific mobile terminal either a 0149. This KPI can be exploited so as to approximate the longer term period T should be considered or the experience KPI of Eq. (9) i.e., the actually experienced coverage in area during a time period T should be averaged over a number of A: days as follows: Poy acA, T-sl-Pou cows, (A, TNa) (24) 0150. From Eqs. (22) and (24) it is evident that the moni toring of events related to out-of-coverage conditions at the 1 (21) Port cols (MT, T, Nd) & arrival of calls terminating to mobile terminals can indeed Nd provide an estimation of the actually experienced network W coverage. S. Nu PAG (MT, T1) + NE PLR(MT, T1) - Nu REp(MT, , T: ) 0151. An alternative KPI to the one provided in Eq. (24) is 4. Neal in (MT, TA)- N. (MTTA)- Nu REp(MT, 5, T4) the following:

0142. Where N is the number of days considered and T P out coyls (A, T, Nd) s (25) refers to the time period T (e.g. 10:00-11:00') of day d. 0143. In order to make the above estimation more accurate l Nd NU PAG (A, Ti ) + NE PLR (A, Tl) - NU REP (A, . T) days can be split into working and non-working days as Na A. Neal in(A, TA)- Nf(A, TA)- Nu REp(A, 5, T4) different user mobility patterns and call patterns occur depending on the day type. 0144. Based on the analysis presented so far it is now evident that the actually experienced coverage by a specific Where: customer i (or MT) defined in Eq. (10) can be expressed as 0152 N. (A.T.) is the number of unsuccessful pag follows: ing events for area A during the time period T. I0153 N. (A.T.) is the number of terminating call oil-covis win (MT, TN (22) arrivals to area A during the time period T for which the 0145 The KPI can be used to identify problems for one or periodic location registration timer has expired. more users at particular time of the day (e.g. by identifying 0154 N. (A.T.) is the number of terminating call out-of-coverage problems in peak traffic hours) and can even arrivals to area A during the time period T. further identify locations where those out of coverage prob I0155 N (A.S.T) is the number of repeated calls lems occurat certaintimes. A wide variety of problems which within the time thresholds (e.g. 1 minute) from the same MT cannot be detected through standard network Statistics can be to the same MT, in area A when MT, was not reachable due to identified through Such an analysis. For example the analysis either unsuccessful paging or due to expired periodic LR can reveal the actual experience of particular customers com timer. plaining about coverage problems. The operator can analyze 0156 The benefit of this KPI (25) from the calculation the actual coverage experience for this customer and identify viewpoint is that with the exception ofN (A.S.T.) it can whether there is some unusual pattern in the recent customer be directly derived from the set of counters presented in the experience (e.g. related to a network outage) or overall the example of Table 2 (while KPI of Eq. (24) requires analysis of customer experience is poor compared to the average cus data per MSISDN presented in the example of Table 1). As the tomer experience. Moreover, the analysis can identify the parameter N. re-(A.S.T.) is dependent on the calling party areas (with an accuracy of a Location Area) where the user behavior which is not related to the area A where the called experiences problems helping thus the operator to propose party is located as well as the time period under study it is safe Some relevant Solution to the customer (e.g. provide a signal to consider the following assumption: booster for the user's house or office premises, introduce a NU Rep.(A.S.T.)sa Rep(s)|Nu Pag(A,T)+NE PLR(A, micro base station, tune a base station, etc.) T.) (26) US 2012/007 1157 A1 Mar. 22, 2012

0157. Where C(A.S.T.) is the portion of repetitive both from calculation simplicity and from the customer expe incoming calls to MTs with either unanswered paging (N, rience viewpoint. Moreover, this KPI analysis can provide an (AT)) or expired periodic LR timer (N. (A.T.)) overall network satisfaction estimation (i.e. not location spe (i.e., detected in out-of-coverage state) in area A during cific) if area A is selected to be the entire network area. The periodT, and according to threshold S. This parameter can be KPI can also be used to identify coverage problems in more estimated based on an off-line analysis of data with events at specialized areas with the accuracy of a Location Area (LA) the MSISDN level (see example of Table 1). Based on this (i.e. a set of cells) where, depending on the time Zone in which approach we have: a problem occurs, different conclusions on the coverage prob lems may result. For example if coverage problems occur during time Zones where customers are located indoors (ei Pw out coylswn (A, T, Nd) s (27) ther at home or at work environment i.e., late evening or W during working hours) then the problem can be related to the 1 X Nu PAG (A, Tl) + NE PLR(A, T,) (1 - a REp()) quality of indoor coverage. On the other hand if the problem Nd Nail in (A, Tl) - Nsf(A, Tl) - a REP (). occurs during the time that customers move from house to Nu PAG (A, Tl) + NE PLR(A, Tl) work and vice versa (e.g. 7:30-9:30' in the morning) then the coverage problem can be related to the quality of coverage in the city main highways. A similar analysis can be performed Where: for non-working days and holiday periods in order to identify coverage problems in areas visited by customers during Such (0158 N, (A.T.) is the number of calls terminating to area periods (e.g. a holiday resort). A during time period T for which the called terminals were in a “switched off state. 0166 Advantageously the network coverage approxima 0159. N. is the number of days for which samples over the tion provided by this KPI is not based upon expected location time period T are averaged in order to obtain a stable statis or the coverage provided at that expected location (i.e. as per tical estimation. population coverage), but instead gathers real data for mobile 0160. Now the following equation provides the relation terminals as they move about the network, and is therefore a between KPI of Eq. (27) with the one of Eq. (23): truer indication of their network experience. 0.167 Ranking of Device Models by Actual Coverage Per formance P. out coyls (A, T, Nd) s (28) 0.168. In another aspect of the invention the data monitored Nd Ncist (A.T.d) Nicai in (MTT) by the network can be exploited so as to characterize the N. (A. T.) Pout colson (MT, Ti, Nd) > coverage experience in association with a terminal type. To iX), achieve an association between the called MSISDN and the P. out coyls (A, T, Nd) s terminal type used by this MSISDN should be created. This is feasible through a terminal identity called IMEI (Interna 1. NdNcus (7d) tional Mobile Equipment Identity) as the first 8 digits of this W, X,d=1 Xi=1 wall (MT, Ta): Pos (MT, Ta Na) number represent the mobile terminal model. The IMEI infor mation is available in the network CDRs however, generally only for the calling party and not for the called party. There (0161 Where w, (MTT) is the portion of incoming fore, the CDRs can be used to associate the called MSISDNs calls in area A which are terminated to MT, during the time with the appropriate IMEIs (off-line processing). With the period T. From Eqs. (28) and (23) it can be seen that the KPI information of the device models it is then feasible to produce of Eq. (23) considers that each MT has the same significance a KPI related to the experienced coverage with the used while in the KPI of Eq. (28) the experience of each MT is device model (ME.): weighted with the number of incoming calls it receives or in other words with its communication needs. 0162 This provides the means to provide a weighted cov P. out colson (ME, T, Nd) & (31) erage experience KPI as follows: Nu PAG (ME, T4) + NE PLR(ME, TA) - Nd P. coy ac?4, Tl-P, out cows. (4, TNa) (29) 1 Nu REp(ME, . T) (0163 Where P, (AT) is defined as follows: Nd X. Nail in (ME, T4) - Nsf(ME, T4)- d=1 Nu REp(ME, &, T)

Pact coy (A, T) = (30) P out cols (ME. T. Nd) s 1 t+T 1 West (A,t) Nd T ? NA. T) 2, Wau in (i. T). PRSS (Gros (i, t), GMT (i)) l Nu PAG (ME. T.) + NE PLR(ME, Ti) (1 - a REp(g)) Nd Nau in (ME, T4) - Nsf(ME, T) - aREp(). Nu PAG (ME, Ti) + NE PLR (ME, T) (0164. Where the parameters N (A,t), G.(it), 0 (i) and Pess(G.(it), 0 (i)) are the ones defined in Eq.(6). 0.165. This KPI (30) expresses the actually experienced 0169. And the actual coverage associated with mobile coverage weighted by the incoming communication needs of device model ME, will be: the customers. As a consequence this KPI is quite attractive P. 1.act ca(MET)-sl-P, a cays,(ME, TNa) (32) US 2012/007 1157 A1 Mar. 22, 2012

0170 It should be noted that the time period T in this case 0.179 HLR interrogation: switched off terminals as well can be selected to be e.g. the busy hour or it may selected to be as terminals with expired periodic location registration a day, a week or a month as the terminal performance over timer are identified at this step. longer time periods may be of interest. 0180 VLR access: busy terminals are identified at this 0171 The results of such an analysis can be used in order step. to rank mobile device models according to their coverage 0181 Paging: Paging procedure exploits a timer in performance. Assuming a substantially high number of order to allow for the terminal enough time to respond sampled devices it is reasonable to use this benchmarking for and depending on the network configuration paging characterizing the device performance in reference to its abil broadcast can be repeated in order to conclude an unsuc ity to be reachable. Devices may have significant differences cessful paging event. in their receiver performance due to a number of factors like 0182. As a result of the above sequence of the call setup the antenna sensitivity, the device design, the signal process procedure an unsuccessful paging event exhibits longer call ing unit of the device, the implementation of standards (e.g. setup duration than other cases of unsuccessful call setup. idle mode operation, Switching from 2G to 3G), etc. As a Therefore if we define a threshold (e.g. 6 seconds) accord result this analysis may become helpful especially in identi ing to which calls terminating to mobile terminals are char fying problematic device models. Similar analysis is already acterized depending on whether the call setup duration available in reference to the performance of the devices vs. the exceeds this threshold can approximate the number of calls rate of dropped calls per device. However, the analysis of with unanswered paging. actual experienced coverage may capture different aspects of 0183 The number of failed call setup events for mobile the terminal performance. The identification of device per terminating calls monitored at the signaling links can be formance problems can be provided as a feed-back to the provided by the following equation: device manufacturer for corrective actions in the current or in future device models. 0172 Coverage Experience Analysis based on Signalling NE PLR(x, TN)+Ne.(x, TN.) (33) Links Monitoring Where: 0173 According to an additional embodiment of the invention, rather than using the MSC or a network element 0.184 X: a generic notion of three options of the analysis to associated therewith to gather the data, the data gathering may be performed: (a) x=MT, for defining the KPIs for a specific be performed by monitoring the signalling links of the net MSISDN, (b)x=A where A is a network area ranging from an work through appropriate probing systems (e.g. using the LA to the entire network and (c) x=ME, for KPI analysis for typical passive probing systems currently exploited in tele a specific mobile terminal model. communication networks). According to this technique the 0185 N. as defined already is the number of days for call setup messages terminating to mobile terminals can be which samples over the time period T are averaged in order to analysed in order to determine the following cases: obtain a stable statistical estimation. 0.174. Successful Call Setup: as indicated by the content (0186 N, (x.T): is the number of failed mobile ter of the signalling message which is part of the call setup minating call setup events during time period T for entity X. process. 0187 N, (x.T.C): is the number of unanswered pag 0.175 Calling party is busy: this case may also be indi ing events during time period T resulting based on the call cated in the contents of the call setup signalling mes setup duration threshold for entity X. Sages. (0188 NAx.T): is the number of unsuccessful mobile 0176 Failed Call Setup: In this case the reason of failure terminating call setups during time period T resulting due to is not evident from the signalling message content. It switched off called MT for entity X. Note this parameter may be due to the fact that the terminal was found to be cannot be measured from the signaling links monitoring. It is in Switched-offstatus at the HLR or the periodic location typically determined from a status in the HLR. registration timer was expired, there might be an unan (0189 N (x.T): is the number of unsuccessful mobile Swered page or even Some traffic channel congestion in terminating call setups during time period T resulting due to the cell of the called party. expired periodic LR timer for the called MT for entity X. Note 0177. It should be noted here that “successful call setup this parameter also cannot be measured from the signaling process” in the context of this analysis refers to the case that links monitoring. “ringing tone' phase is entered. Therefore a call which enters 0.190 N(x,T): is the number of unsuccessful mobile the ringing phase but not answered by the called party is not terminating call setups during time period T resulting due to considered unsuccessful in this analysis. some other reason than the ones provided above for entity X 0.178 A problem with analysing call set up messages is the (e.g. unanswered paging or traffic congestion). This is yet fact that the signalling messages do not provide information another parameter cannot be measured from the signaling that allows the identification of unsuccessful paging events. links monitoring. However, the current embodiment of the invention provides 0191 It is evident that in order to be in a position to an alternative method to detect such events. The way to iden provide Some estimation of the experienced coverage then the tify unsuccessful paging events is by monitoring the duration required parameters which are not directly monitored from of the call setup procedure. From the way that call setup the signaling links should be estimated. This is achieved as procedure is performed it has been realised that unsuccessful follows: paging events correspond to longer call setup duration versus other cases of failed call setup. Indeed the call setup proce dure steps include: NE PLR(x, TN.a)-ai.(T)'N,U call in (, TNa) (35) US 2012/007 1157 A1 Mar. 22, 2012

Where: 0201 This embodiment of the invention is particularly (0192 C.(T): is the portion of failed mobile terminating useful for benchmarking, i.e. gauging the efficiency of net call setup events due to the called MT being switched off for works operating adjacent to the primary network. Here, the entity X. term “adjacent” networks does not indicate networks that are (0193 Cr(T): is the portion of failed mobile terminating somehow literally adjacent, but rather those which are inter call setup events due to the expired periodic location regis connected with the primary network, Such as in other regions/ tration timer of the called MT for entity X. countries or operated by a separate network operator. Essen 0194 It should be noted here that such KPIs can be derived tially an adjacent network is therefore one from which based on off-line analysis of data derived from network ele specific data in relation to unsuccessful call attempts, busy ments. However, their statistical stability can prove relevant call attempts and the like are not directly obtainable. That is, for an analysis referring either to area level (x=A) or mobile they are networks that the primary network operates along device model (x=ME). In the case of analysis for specific side, and whilst the networks cooperate to allow interoper MTs (x-MT) these KPIs cannot be estimated directly from ability, not all data is shared. the signaling links. 0202 This embodiment of the invention enables this lack 0.195 Based on the information provided in the signaling of full data sharing to be overcome, by making intelligent links it is possible to perform an analysis of repetitive calls estimates of unsuccessful call attempts and the like. More following an unsuccessful paging attempt based on a thresh specifically, these intelligent estimates are based upon the old as described already (N(x.T.S). Alternatively the duration of a call set-up procedure. With reference to FIG. 4. parameter C(S) defined in Eq. (26) can be exploited as this the network server implementing the embodiment will moni parameter is related to calling party behavior. However, this tor calls (40) being initiated from its home network. When an approach is statistically relevant only for area level analysis outgoing call is detected, a timer is initiated to time the call set (x=A) or mobile device model analysis (x=ME). up procedure (41). This may be performed for all outgoing 0196. Based on the above assumptions the actually expe calls from the home network or only those which are being rienced network coverage can be expressed as follows: directed outside of the home network. This latter approach is possible using probes deployed over the signalling links serv ing the inter-connection of two networks. Poy act (X, T, Nd) & & (36) 0203 As it can be seen from FIG. 4, it is feasible to identify Successful call setup (42), unsuccessful paging Nu PAG (X, T4, 6) + (pLR(X, T4). Nu call in (X, Ta) events (by monitoring the call setup duration (41) and exam (1 - a REp(f)) ining whether the measured call setup duration is equal or Noat in (v, Ta) - Nu call in (v. Tid) longer than a pre-determined threshold T. c. (43)) and as f(x, T4) + ap R (X, T4) a REP (S)- other call setup non-completion cases (44. 45) (e.g. called NU PAG (X, Ti, ), a REP (8) terminal busy, Switched off expired periodic location regis tration timer, etc.) 0204 If the call is successful, as with the previous embodi 0.197 Where C(S) is the portion of repetitive incoming ments of the invention, a record or relevant counter can be calls to MTS with either unanswered paging or expired peri updated to record this event (46). In case plain counters are odic LR timer according to threshold S. implemented then the total number of call setup attempts and (0198 It is evident that the above KPI (36) is relevant for the unsuccessful paging events are counted per monitored area level and mobile device model analysis (i.e. x=A or adjacent network so as to provide an estimate of KPI (16). In x=ME). For analysis per MT (x=MT) and in the case that no other source of information is available an approximation the case a record is stored then information like the adjacent analysis can be used focusing only on unsuccessful paging network identity, a time-stamp of the event, the event type and analysis and ignoring the Switched off and expired periodic the called party MSISDN can be stored per record. location registration timer events parameters. 0205. In this regard, it has been found that an unsuccessful (0199. In another embodiment of the invention, the KPI call set up due to unanswered paging typically takes at least derived from signalling links monitoring may also be derived 2-4 seconds more time to complete versus other cases of for other networks provided that it is feasible to monitor the unsuccessful call setup. Moreover, a threshold of T, in signalling links which serve the interconnection traffic the order of 8 seconds appears to be appropriate for this between the two networks. In such a case it is possible to embodiment of the invention. However, this threshold should compare different networks against the same benchmark in be tuned for the specific network to be monitored as this is order to identify differences in the experienced coverage. Subject to network implementation and configuration aspects. This KPI therefore avoids incompatibilities between the cur 0206 Ideally the analysis of these embodiments of the rently used coverage estimates, and can even facilitate bench invention is performed on a regular basis in order to identify marking between different networks, including between peer problems quickly, and ideally as they arise. In a similar man networks and networks in other regions (e.g. between ner the data may be compared with networks operating in Vodafone Greece and Vodafone Italy). It is to be appreciated different localities in order to identify problems that may exist that this embodiment of the invention only considers outgo in theoretically equivalent environments. ing communications from one or more mobile terminals of 0207. In a further embodiment of the invention, the analy the home network towards the adjacent networks and not sis can simply be in regard to the mobile terminals (i.e. MSIS incoming communications. DNs) that experience the highest number of unsuccessful 0200. This embodiment of the invention will now be paging events. The rationale for this analysis is that regardless explained in more detail with reference to the flow chart of of the number of incoming calls per MSISDN, if the number FIG. 4. of missed calls exceeds a certain threshold, then the user will US 2012/007 1157 A1 Mar. 22, 2012

become concerned of the problems associated with network where: coverage or the terminal performance in idle mode operation. 0208. It is to be appreciated that whilst the embodiments of the invention have been described with particular reference Peys & -NU PAG + NE PLR to an unreachable event due to a network/terminal or standard Ncal in - Nsaf problem, an additional situation where such an event could apply is in the event of an accidental or intentional device and: termination, such as would occur where the battery is pulled P. relates to the out of coverage probability for out (i.e. in this situation the terminal does not have time to Switched on mobile terminals; initiate a network detach message). This situation is unlikely N. c. relates to the number of unsuccessful paging to greatly affect the analysis since it is a very rare occurrence. events to mobile terminals; 0209 Further, the embodiments of the invention have been N relates to the number of terminating call arrivals to described and illustrated through the specific figures with mobile terminals for which the applicable periodic loca particular reference to a network having 2G and 3G access tion registration timer has expired; and technologies. However, alternatively or in addition, other N relates to the number of terminating call arrivals to access technologies are envisaged such as LTE and MIMO mobile terminals; technologies (particularly 3G MIMO techniques such as N. relates to the number of calls terminating to mobile WCDMA and HSPA) or future terminal models supporting terminals for which the called terminal was in a switched these access technologies can be addressed accordingly. off state. 0210 Additionally, the embodiments of the invention 5. The method of claim 4 further including averaging the have been specifically described in relation to unsuccessful equation over one or more given time periods in order to call attempts, which includes unsuccessful voice and video achieve statistical stability. calls. The embodiments of the invention may also be adapted 6. The method of claim 4 further including filtering out for other services available on cellular networks, such as repeated calls to unreachable mobile terminals within a pre Instant Messaging, push-to-talk, etc. determined time threshold. 7. The method of claim 4 further including determining 1. In a telecommunications network including a plurality of No pao, NE PLR. N. and Neffrom: mobile terminals, a method of providing an indication of a) event counters on a per terminal basis; or network coverage, the method including: b) event counters on a location basis. determining the occurrence of incoming call events in rela 8. The method of claim 4 further including analysing the tion to at least one of the plurality of mobile terminals: data to perform at least one of the following applications: recording a result of each incoming call event, together identify network coverage problems in given areas or at with at least one call property; and given times; analysing the recorded incoming call data to determine an handle customer complains related to network coverage; indication of experienced network coverage for the at benchmark mobile terminal performance versus coverage least one terminal, a given terminal type, a given location experience; and/or a given time period. benchmark different mobile networks versus the offered 2. The method of claim 1 further including determining the network coverage experience. result of each incoming call event to be recorded from at least 9. The method of claim 1 wherein the analysis of the the following options: recorded data includes determining an indication of the net a) a Successful call including a Successfully established call work coverage experienced in a given area (A) during time and/or Successful terminating call attempt; period (T) using the following equation: b) an unanswered paging: c) called terminal busy: 1 Nd NU PAG + NE PLR - NU REP d) called terminal detached from network; and Pw out cowlswn e) a failed incoming call due to an expired periodic location N A Nau in - Nef - Nu REP registration update timer. 3. The method of claim 1 further including determining the where: at least one call property to be recorded from at least the P, , , is the out of coverage probability for following options: Switched on mobile terminals in area A during time an identity of the called party; period (T); a time period applicable to the call; N. c. is the number of unsuccessful paging events in a location applicable to the called party; area A to mobile terminals during the time period T: an identity of the terminal type for the called party; Ne e is the number of terminating call arrivals in area A to mobile terminals during the time period T for which the mobile network of the calling and the called party. the applicable periodic location registration timer has 4. The method of claim 1 wherein the analysis of the expired; and recorded data includes determining an indication of the net N is the number of terminating call arrivals in area A work coverage experienced by a given mobile terminal or to mobile terminals during the time period T: mobile terminal type (P co - e. c) according to the following equation: NS is the number of calls terminating in area A to mobile terminals during time period T for which the called coy acrl Fout cowswn terminal was in a Switched off state; and US 2012/007 1157 A1 Mar. 22, 2012

N is a portion of repetitive incoming calls in area A to determine the occurrence of incoming call events in rela mobile terminals within time threshold S. tion to at least one of the plurality of mobile terminals; 10. The method of claim 1 wherein the analysis of the record a result of each incoming call event, together with at recorded data includes determining an indication of the net least one call property; work coverage experienced in a given area (A) during time analyse the recorded incoming call data to determine an period (T) according to the following equation: indication of experienced network coverage for the one or more mobile terminal, a given terminal model type, a given location and/or a given time period. Pact coy (A, T) = 18. The network element of claim 17 further configured to analyse the recorded data by determining the indication of the 1 frt--T 1 West (A,t) i? NA, t) 2, Wall in (i. T). PRSS (Gpos (i, t), 64T (i)) network coverage using the following equation:

1 Nd Nu PAG + NE PLR - Nu REP 11. In a telecommunications network including a plurality Pw out Coylswn of mobile terminals, a method of providing an indication of N A Nau in - Nef - Nu REP network coverage in relation to at least one of the plurality of mobile terminals, the method including: determining the initiation of a call set up procedure from wherein: the at least one terminal; P out , relates to the out of coverage probability for monitoring the duration of the call setup procedure; and switched on mobile terminals; identifying a failed call attempt when the call setup proce N. c. relates to the number of unsuccessful paging dure duration exceeds a predetermined threshold. events to mobile terminals; 12. The method of claim 11 further including, where a N relates to the number of terminating call arrivals to failed call attempt is identified, using the timed set up dura mobile terminals for which the applicable periodic loca tion to determine a reason for the failed call Such as an tion registration timer has expired; and unanswered paging event. N relates to the number of terminating call arrivals to 13. The method of claim 11 further including using the mobile terminals; failed call set up events to obtain an indication of network N. relates to the number of calls terminating to mobile coverage for a given region, a given user terminal, a given terminals for which the called terminal was in a switched terminal type and/or a given time period. off state; and 14. The method of claim 11 further including, where the N is a portion of repetitive incoming calls to mobile call set procedure is not successful, and a reason cannot be terminals within a predetermined time threshold. determined from the content of signalling messages, then: 19. The network element of claim 18 further configured to where the timed set up equals or exceeds the threshold, the use the equation to determine the actual network coverage failed communication is determined to be due to an for: unanswered paging event; and/or i) the one or more mobile terminals during a time period T where the timed set up is less than the threshold, the failed ii) the given mobile terminal type during a time period T. communication is determined to be due to other reasons, and/or Such as the called terminal being detached/busy. iii) the given location area A. 15. The method of claim 11 further including: 20. The network element of claim 17 further configured to determining Successful call set up events and busy calling analyse the recorded data by determining an indication of the party events from the content of signalling messages; network coverage experienced in a given area (A) during time and period (T) according to the following equation: additionally using the Successful call setup events to obtain the indication of network coverage. 16. The method of claim 11 further including: using the unsuccessful call set up events in order to obtain Pact coy (A, T) = the indication of network coverage for a plurality of 1 t+T 1 West (A,t) different networks; and i? Neust (A, t) 2, wall ini (i.(i, T). PassPRSS (G.(it),(Gros (i,i), 6MT6 ft (i))(i comparing the indications for the different networks. 17. In a telecommunications network including a plurality of mobile terminals, a network element configured to provide 21. The network element of claim 17 further configured to an indication of network coverage in relation one or more of use the data analysis to identify network coverage problems the plurality of mobile terminals, a given mobile terminal in given areas or at given times. type, a given location area and/or a given time period the network element configured to: c c c c c