WO2015128855A1

WO2015128855A1 - Method and system for road traffic data collection

Info

Publication number: WO2015128855A1
Application number: PCT/IL2014/050193
Authority: WO
Inventors: Jonathan Silverberg; Raz STEINMETZ; Michael BEN-ASSOR; Daphna SHEZAF
Original assignee: Decell Technologies Ltd
Priority date: 2014-02-26
Filing date: 2014-02-26
Publication date: 2015-09-03

Abstract

Technique for modeling a road segment, comprising matching of routine cellular measurements with occasional location measurements. The cellular and the location measurements are gathered independently on the road segment. Based on the matching, the technique further creates a verified cellular map (model) of the road segment where locations of cellular events are verified by the collected occasional location measurements.

Description

Method and system for road traffic data collection Field of the invention

The present invention relates to the field of monitoring road traffic by collecting and processing traffic data for various purposes, for example for navigation. The suggested solution utilizes a mobile (cellular) network. Background of the invention

Presently, two main approaches exist for collecting data on vehicular traffic based on Floating Car Data (FCD): a first approach is based on utilizing information from accurate location sensors (say, GPS systems) placed on vehicles, while a second approach is based on utilizing information from cellular sensors, such as mobile phones in the vehicles, which information is accumulated by a cellular (mobile) network. Systems utilizing information from cellular networks may be called NetFCD or CFCD systems.

Practice shows that the two approaches may be used separately or in various combinations in modern traffic monitoring systems. Various systems for traffic monitoring and traffic data collection were described in the prior art, some of them will be mentioned below.

The principle lines of technology for collecting traffic data were proposed by several companies including by the Applicant's Mother company Decell Inc. and are as follows.

There are mobile network events that a vehicle driving along a road is expected to encounter in a relatively high consistency, namely these events are handovers of a cellular terminal of the driver or a passenger from the service area of one cell of the mobile network to the service area of another, adjacent cell. Some of such handovers happen in a very short stretch/segment of a road, for almost every car/mobile phone passing along the same road. Some of the handovers occur during a call and some during the idle mode. The mobile network has the ability to monitor these events in real time and to report such an occurrence to a processing entity, such as the traffic data collection server. The report usually includes a unique mobile phone/terminal ID, a time stamp, the source (first) cell and the destination (second) cell.

Given that the above-mentioned events occur in a very well pinpointed/known location along the road, further matching of two such events for the same user would represent a trajectory between two known points throughout a given time frame, hence enabling the calculation of the travel time, the average speed, a delay etc. (The problem of determining the accurate location will be discussed below.) The described procedure can be repeated for a large group (sample) of subscribers that travel along the same stretch of the road.

Performing the above mentioned procedure for a large road network would enable generating a high quality real time traffic map.

In the described general method, the main challenge is the ability to determine (pinpoint) exact locations of the above mentioned handovers. By today, two main techniques have been quite common in the art:

- Driving the road in a dedicated car to perform a so-called test drive, while recording the cellular activity of a set of phones and measuring the exact location where the handovers take place. This procedure is very resource demanding hence limited in scale. Yet another variation of this procedure is by monitoring both the exact location (e.g. GPS location) of a cellular device and its Cellular location (e.g. Cell-ID location) for a group of vehicles (e.g. Taxis) and accumulating the correlation between these two locations throughout a period of time. (Proposed, for example in WO2011/120193 which is mentioned further below).

Building a model of the anticipated location of the handovers based on the road topology and the cellular antenna attributes (location, direction, width, power etc.). The ability to model the signal propagation and derived coverage footprint is very limited and the results received have been of rather medium quality. Solutions proposed in the prior art utilize one or another of the above techniques.

As has been mentioned, Decell Inc. (the Mother company of the Applicant) filed one of the first patent applications in the field, which described the basis of a so-called CFCD (Cellular Floating Car Data) method.

US2003069683A (to Decell Inc.) describes a traffic monitoring system and methods for monitoring traffic including a population of users bearing a multiplicity of mobile communication devices. The system includes a mobile communication network interface which receives and stores location information from the multiplicity of mobile communication devices. The system also comprises a traffic monitor operative to compute at least one traffic -characterizing parameter on the basis of the location information. The entire description of US2003069683A is incorporated herein by reference.

Since then, several other companies have developed technological solutions in this field. ITIS Holdings company proposed the way handover events of mobile phones (i.e. the events when an individual mobile terminal passes from the service area of one cell to the service area of another cell) can be used as a basis for traffic data generation, while relying on geographical models that use the location of the antenna towers (cellular base stations) as references.

US2006009233A (To ITIS Global Services Limited) describes a method for geographically locating a cellular phone. The method comprises: determining an effective cell-area for each of a first cell and a second cell in a cellular network; and determining a handover area within which the cellular phone is likely to be located when control of the cellular phone is transferred from the first cell to the second cell; wherein the determination of the handover area and the effective cell-area for each of the first cell and the second cell are made based on a topological relationship between the first cell and the second cell. Some other technical solutions (below) proposed utilizing real, physical field measurements performed along all the covered roads of signal reception as the reference for their system. It is quite understood that the above mentioned field measurements are more accurate than the antenna based computer models. However, special vehicles or so-called test drives dedicated for such physical measurements seem to be a very expensive and time consuming solution.

US2003100317A (to O. Avni & Y. Kaplan ) describes a system that continuously extracts traffic load and speed on roads within the coverage area of a cellular network. The data is extracted directly from the higher level of communications in a cellular network without using any external sensors. The cellular data used by the system includes cell handover sequences as the major input. The method consists of a learn phase in which a dedicated vehicle with a location device (say GPS system) travels across the covered routes and collects the cellular data and location data in parallel. In the continuous data collection stage the cellular data is processed and correlated to the data collected in the learn phase to yield the route and exact location of the vehicle on it. Each such two locations yield the travel distance and time, and thus the driving speed in this route section. This procedure is performed continuously across the covered area to yield the traffic load and speed.

US2005227696A (also to O. Avni & Y. Kaplan ) describes a further developed technique that enables correlating a car to a road it travels on and determining its speed by using only the partial data that arrives to the cellular switch. The data accumulated in the learn stage is then analyzed and processed to create a reference database. In the operational stage, communications on the cellular network control channel are monitored continuously, and matched against the reference database in order to locate their route and speed. The route and speed data is used in order to create a traffic status map within the designated area, and to alarm in real time on traffic incidents. The data analysis and Database structure are done in a manner that enables a) fast, high reliability initial identification of the vehicle's route in the operational stage, based on the handovers' cell ID only; b) fast, high reliability follow up forward and backwards of the vehicle's route in the operational stage; c) real time, high reliability incident detection.

By now, it has been realized in the prior art, what are the advantages and the disadvantages of the main approaches to collecting traffic data. As mentioned above, the first approach is based on utilizing information from location sensors (say, GPS systems placed on vehicles). A known Floating Car Data (FCD) system processes commercial fleet GPS data locations available from a Fleet management system (such as a Taxi dispatching center). However, the appearance of fleet vehicles (which can be taken as a kind of traffic sensors) is not necessarily sufficient for continuous coverage of a whole road network. Therefore, it is not sufficient for the existing FCD systems to generate qualified traffic information according to the Fleet GPS locations.

As mentioned, the second main approach to collecting vehicular traffic information utilizes cellular phone subscribers as data source to provide ubiquitous coverage of traffic information. However, the cellular approach also has its disadvantages.

US5465289A describes a method and apparatus for providing vehicular traffic information using a cellular telephone system technology. Traffic sensors monitor the control and voice channel transmissions of cellular units within a cellular telephone system. Data from these transmissions is extracted and analyzed according to a statistical model, and derived vehicle geo-locating information to generate vehicular traffic information that is transmitted to a central control center. By combining the information from all of the traffic sensors and each individual cell within a cellular telephone system, a picture of the traffic conditions existing along major thoroughfares may be determined.

Cellular terminals (e.g. cellular phone, smart phone, or other wireless devices connected to a cellular network) carried by vehicles are widely distributed over the street network of the whole city and country. In recent years, a couple of ideas/systems were coming up with the similar basic principle of "using cellular terminals as traffic sensors, extracting their movement information from their operational data existed in the cellular network, and then generating traffic information." With the widely distributed cellular terminals, a potential coverage over the whole city, inside and outside, is obtained. Although good performance was validated in simple area (i.e. relative simple networks, typically in outer-city area and highways) for all these similar ideas/systems, the performance in dense area (typically in inner cities) is still a challenge.

In a dense area, the structure and the radio environment of the cellular network are complex and dynamic, which bring lots of dynamics on mobility management behaviors of the cellular terminals, e.g. the large dynamics of the handover area, the dynamics of the handover sequence when driving along a street etc. It should be noted that the dynamics on mobility management behaviors is part of operational data of the cellular terminals. With static- information (e.g. cell map) or historic statistic- information (e.g. pre -measured or defined pattern) on the cellular network, the matching of the traffic sensors' locations onto the street network and thereafter the traffic estimation will be a big challenge. That is, in a cellular-based system, a very rough pattern is used for mapping, which is created by e.g. a static cell map or a statistic method using dedicated cars to do some pre-measurement and hard to be up-to-date. Therefore, map matching seems to be a challenge.

WO11120193A (to SIEMENS AKTIENGES ELLSCH AFT) tries to overcome the above drawbacks of the both main approaches. It discloses a system for providing traffic information which utilizes three types of traffic sensors. The system is adapted for calculating appearance of one or more first type traffic sensors in a positioning system, and determining a middle area in a street network according to the appearance of the one or more first type traffic sensors. The system is also capable of selecting multiple third type traffic sensors within the middle area determined from the one or more first type traffic sensors; obtaining track reports of the positioning modules from the positioning system and obtaining operational data of the communication modules from a cellular system for the multiple third type traffic sensors, and forming mapping relation between road segments in the street network and traffic patterns; and extracting a handover sequence of a second type traffic sensor from its operational data and finding traffic information of the second type traffic sensor via comparing the handover sequence with the traffic patterns.

Object and summary of the invention

It can be stated that the main requirement and the main problem of the prior art techniques for monitoring and collecting traffic data is to determine exact geographical locations of cellular events (usually, of cellular handovers) on the roads, so as to have clear geographic references for information received from users of a cellular network. Attempts of directly determining such geographical locations occur to be very expensive, since they require either a dedicated vehicle for explicitly performing test drives (field measurements), or fleet vehicles for performing such tests implicitly. It is therefore the first object of the invention to propose a novel inexpensive technique for modeling a road segment, without performing preliminary test drives.. The modeling, first of all, allows producing effective raw data for traffic service providers, and secondly - allows creating a verified cellular model of the road segment for real time traffic motoring.

Based on the above, or independently it is an object of the present invention to provide a technique for real time monitoring, collecting and generating traffic data with accuracy equivalent and even better than field measurements provide. Such a second proposed technique may also be implemented without the need to actually perform any dedicated test drives in advance (in a so-called "modeling phase"). Another object is to perform that second technique without the need to obtain continuous, dynamic accurate location information in a working ("runtime" real time) phase.

Further, an object is also to propose a technique (method and system) for monitoring and collecting traffic data from known and relatively inexpensive traffic data sources (or sensors), which technique would allow obtaining at least one highly valuable traffic data product being for example: raw data for traffic service providers, a parameter of estimated travel time (ETT) on a road section, a verified cellular map/model of a road section, etc. In the frame of the present patent application, the verified cellular map of a road section should be understood as a topological map of a road section, being fulfilled by information about mobile network cell(s) covering that specific road section, with verified (i.e., checked for higher accuracy) geographical location of borders between said cells on that specific road section. There is still a further object of the invention, which is to propose a method for accurately determining geographical locations of cellular events.

An additional object of the invention may be set, which is to propose a new method of implementing a mobile network based traffic data collection system (aka, CFCD, NetFCD), for example by utilizing a known CFCD, NetFCD system to provide initial data (model) for the mentioned second technique. The Inventors are going to propose a technique - a method, a system and a software product - for

1 ) modeling a road segment, for example for providing raw data for traffic service providers,

2) traffic data monitoring,

3) determining accurate geographic locations of cellular events (such as cellular handovers, etc.) on a road segment.

All of the above techniques are performed in a manner different than those presently known in the prior art i.e., without performing test drives explicitly or implicitly, but by performing indirect/statistical matching of independently performed regular, routine cellular measurements with occasionally/randomly performed location measurements on the road segment. The Inventors also propose modeling the road segment based on said matching so as to create a geographically verified cellular map/model of the road segment where locations of the cellular events are verified by said occasional location measurements . Results obtained at any of the technique variations (1, 2, 3) can further be used to generate high quality traffic information.

The indirect matching should be understood so that the cellular measurements and the occasional location measurements are provided independently, i.e. regardless the fact whether they are taken by cellular and location sensors co-located (say, on a same vehicle) or not; in this situation the cellular and the location measurements cannot be matched directly, but only indirectly/statistically.

The technique of modeling may form part of a second novel technique proposed by the Inventions, namely - of a new method, system and software product for traffic data monitoring.

In turn, the second technique may form part of a third proposed technique which is determining accurate geographic locations of cellular events (such as cellular handovers).

The proposed novel modeling is based on principles different from those known in the prior art. It comprises indirect matching of regular cellular measurements and occasional/random location measurements, which is different than operations a) and b) respectively being

a) combined/synchronized cellular and location measurements - "test drive" or "field measurement" by a dedicated vehicle carrying both a cellular sensor and a location sensor,

b) combined/synchronized cellular and location measurements obtained from a group of non-dedicated vehicles, while each of them carrying both a cellular and a location sensor (i.e. such as a fleet vehicle at which a cellular sensor and a location sensor are permanently co-located).

As already mentioned, both a) and b) approaches are expensive and time consuming. The proposed novel modeling requires neither a), nor b) which is highly essential for providers of traffic services, who require verified road maps/models where cellular handovers have accurate geographical locations.

Based on the above, one of the main characteristic features and advantages of the invention is that the proposed technique utilizes continuously obtained cellular data from a cellular/mobile network ("the first traffic data source") but does not require obtaining continuous accurate location data. Instead, sporadic/occasional location data from the location sensors such as GPS devices occasionally passing via the area or, say, immobile photo/video cameras situated at road crossings have been found to be sufficient for the proposed technique to form "the second traffic data source", i.e. to provide reference location data on geographical location of a vehicle.

This reference location data is then matched with the cellular data obtained from cellular sensors.

The proposed technique may thus be used in places/circumstances where a cellular network exists but there is a limited number of mobile and stationary devices suitable for determining geographic position of vehicles (and thus for producing the location data). For example, we have a case where only a limited fleet of vehicles use active (transmitting) GPS devices, and/or there is a limited number of stationary devices such as photo/video cameras which produce information suitable for determining geographical location of vehicles.

More specifically, the first object of the invention can be achieved by providing:

A method for modeling a road segment based on gathering and indirect matching of traffic data received from cellular sensors and location sensors, wherein

said gathering of traffic data from cellular and location sensors is performed in a manner that includes

performing routine cellular measurements by cellular sensors during a calibration period, thus obtaining a flow of cellular measurements, and

obtaining occasional only, while accurate location measurements by location sensors during the same calibration period, thus obtaining a sample of accurate location measurements,

wherein said cellular and said location measurements are performed independently, i.e., regardless the fact whether any of the location sensors is co-located with any of the cellular sensors participating in said gathering during the calibration period;

and wherein

said indirect matching comprises matching, by time, the flow of cellular measurements with the sample of accurate location measurements, both related to said road segment during the calibration period.

The cellular measurements are usually provided at moments of cellular events taking place on the road segment. The cellular event should be understood either as a usual cellular handover or, for example, a sudden interruption of cellular connection which may occur at a place not covered by cellular services (say, a tunnel, a border of cellular coverage, etc.).

It is understood that the above method comprises gathering traffic data related to the road segment, from two (a first and a second) traffic data sources, wherein

a first traffic data source being a cellular (mobile) network capable of producing cellular data generated by a plurality of cellular sensors (in the frame of this description, cellular or mobile devices, phones, terminals) while moving in an area including the road segment and covered by the mobile network, and a second traffic data source comprising one or more location sensors being devices capable of generating, at least occasionally, accurate location data related to said road segment. The location sensors of interest generate location data related to said road segment only occasionally, thereby collection of the location data takes place sporadically or occasionally (from time to time).

Such location sensors may be: at least one transmitting GPS device of a vehicle passing the road section by chance/sporadically, at least one traffic observation camera or sensor capable to produce occasional location reports concerning a specific time period, and about specific vehicles passing in front of the camera during that specific time period.

Preferably, the step of performing routine cellular measurements to obtain the flow of cellular measurements may comprise:

- obtaining cellular measurements formed from the cellular data supplied by the plurality of cellular sensors at moments of cellular events (mainly handovers); each cellular measurement comprising ID of the cellular sensor, ID of two of the cellular events performed (optionally, including ID of the cells passed through) by said sensor during one travel, and two respective timestamps for said two cellular events.

Optionally, the method may also make use of other types of cellular event (e.g. In call handover, Idle handover, Periodic update etc.) Further preferably, the step of obtaining a sample of accurate location measurements may comprise:

- occasionally and independently obtaining location measurements on points along a road section overlapping (preferably for more than e.g. 50% ) with the road segment; each of said location measurements comprising a timestamp, and geographical location of said point, (wherein said points may be but are not obligatory co-located with said cellular events);

Still preferably, said indirect matching may comprise:

- combining (crossing) the flow of cellular measurements with the sample of said location measurements, thus obtaining combined data,

- processing said combined data to find those of said cellular measurements respectively matching with said sample of location measurements by time (such matching cellular measurements being called Verified Segment

Readings), thereby -producing a Verified database of traffic data for said road segment, from said Verified Segment Readings.

Actually, the Verified Database of traffic data per road segment already constitutes effective raw data, valuable for traffic data service providers.

By using the Verified Database, it will further be possible to create a complete model of the road segment, to perform real traffic data monitoring, as well as to allow determining, with accuracy close to that of field measurements, geographical locations of the cellular events.

Also, the indirect matching may comprise calculating, for each Verified Segment Reading, a location/geographically based Estimated Travel Time (ETTg), and a cellular based Estimated Travel Time (ETTc), both of which will further be used for fulfilling the model of the road segment.

The indirect matching can be also called statistical matching since it utilizes a statistical approach when matching/crossing by time the flow of the regularly collected cellular measurements with the sample of occasionally and independently collected accurate location measurements.

The raw data (the Verified Database) allows creating an

accurate/geographically verified cellular model and map of the road segment, when fulfilled by a calibration step which will be described further below. The principle of the calibration step is calibrating said verified Database of the road segment to obtain a cellular model of the road segment in the form of configuration data allowing calculation of Estimated Travel Time ETT of the road segment based on cellular measurements only. It should be noted that the proposed method preferably comprises a

preliminary, pre-modeling phase for determining potential cellular events (such as potential cellular handovers) actual for the road segment; the pre-modeling phase may be performed by one or another methods known from the prior art; however, an exemplary pre-modeling phase will be disclosed later on in the present description.

The proposed method consequently allows modeling of a road comprising multiple road segments, and further - modeling of an area comprising multiple roads, thereby obtaining a verified cellular map of a road/area of interest.

As mentioned above, the above-proposed method for modeling of a road segment may further serve a basis of a second proposed novel method - a method for monitoring traffic data.

In other words, the novel method of traffic data monitoring comprises:

creating a cellular model of the road segment by:

- occasionally gathering, from location sensors, accurate location

measurements on a road segment during a calibration period,

- verifying the occasionally gathered accurate location measurements by a plurality of cellular measurements gathered from cellular sensors on the road segment, during said calibration period,

wherein said cellular and said location measurements are performed independently, regardless the fact whether any of the location sensors is permanently co-located with any of the cellular sensors participating in said gathering during the calibration period;

- composing from said plurality of cellular measurements a verified Database for the road segment;

the method further comprises - performing a runtime phase of the method, by collecting cellular

measurements in real time and by calculating ETT for the road segment, based on the cellular measurements only and using the configuration data. The above calibrating step/phase may preferably comprise determining the following elements of the configuration data

reliable pairs of cellular events (for example, called Traffic Usable Handover Pairs list), selected as those demonstrating maximal correlation when calculating for such a pair: ETTg values using related to that pair accurate location measurements, and ETTc values using related to that pair cellular measurements; transformation coefficients and/or formula parameters and/or statistical model for each of the selected reliable pairs of cellular events, enabling to obtain highly accurate ETTg values from ETTc values calculated using cellular measurements only.

The generalized steps of the above-proposed methods will be further described in detail, to clarify the stages and steps, to explain how the idea of producing and utilizing the occasional location data (location measurements) affects results of collecting the data traffic and obtaining the traffic data product(s) - such as the ETT or the Verified cellular road segment map.

For simplicity, the methods will be now presented by the above-mentioned stages or phases, namely:

- a preliminary, pre-modeling phase,

- a data gathering and modeling phase (end of the method of obtaining a Verified Database of a road segment, forming raw data suitable for traffic data providers);

- a calibration phase (end of obtaining a working model of the road segment, allowing further traffic data monitoring),

- a runtime phase (end of the method for traffic data monitoring). The preliminary, pre-modeling stage/phase is generally known, for example, from the Applicant's publication US2003069683A .(The pre- modeling phase is necessary to obtain data on Potential handovers per specific road segment. To obtain the data on Potential handovers, the following steps may preferably be performed:

- obtaining a stream of cellular data from the mobile network, collected from the cellular sensors moving in the area covered by the mobile network, and extracting there- from information about pairs of cells between which handovers occur (logging and counting these pairs);

- selecting, from the information about pairs of handovers, data on such pairs of handovers that occur more frequently than others (e.g. more than 100 times during one day), and marking the selected data as Common handovers;

- obtaining an estimated segment cell coverage for said segment, by putting into mutual correspondence/crossing a cell map of the mobile network and a road segments map derived from a standard available digital road map . It should be emphasized that at this stage the estimated cell coverage considered is preferably chosen to be biased towards a broader rather than a narrower coverage; wherein each segment cell coverage being presented as information on a specific road segment reasonably covered/overlaid by a group of cells of the mobile network;

- crossing the information on segment cell coverage with the Common handovers data, thereby obtaining a list of Potential Segment handovers which are most expected in the specific road segment (the handovers belonging to the segment cell coverage i.e., taking place between the cells covering the road segment).

The pre-modeling phase may be performed separately from the main method (i.e., in advance), or immediately before a phase of gathering traffic data/modeling which opens the main method. The pre-modeling phase may be performed using systems known in the art.

The phase of gathering (traffic data) and modeling a road segment has been generally described before.

The phase of gathering data/modeling is relatively prolonged and (the calibration period) may take some days or even weeks. The prolonged period of the gathering phase is quite essential, since it allows collecting and duly utilizing the inexpensive occasional, sporadic location data.

The occasional location data has the advantage that it is inexpensive even if bought from its providers, and does not overload the processing system which will also be described. The phase of gathering traffic data and modeling performed for the road segment during the calibration period may preferably comprise the following operations (in case cellular events are Handovers):

gathering of cellular data and occasional gathering of location data.

The gathering of cellular data may comprise:

- collecting data items (cellular measurements) about any Handovers from the mobile network and logging those data items belonging to a specific segment's Potential Segment Handovers;

- in case that, for a particular cellular sensor (say, a mobile device), more than one Handovers are recorded that belong to the Potential Segment Handovers, logging the following information as a" Segment Reading" for any pair of

Handovers (a First Handover and a Second Handover) selected from said more than one Handovers:

a First Handover's cells, a First Handover's timestamp, a Second Hanover's cells, a Second Handover's timestamp, Travel time defined as a Second Handover time minus the First Handover time derived from respective timestamps thereof.

Actually, the "Segment Reading" is a "cellular measurement" of a specific cellular sensor/user, taken along the road segment (since this cellular measurement is taken at the Potential Segment Handovers being crossed quite often.).

In other words, in the step of collecting cellular data, the method selects among the Potential Segment Handovers such pairs of handovers (the handovers of the pair are not required to be adjacent, just any pair from available combinations) which take place quite often and for one and the same cellular sensor (in practice, for one and the same vehicle). The meaning of the selection is that these pairs of Handovers take place indeed in the segment of the road which has been pre-modeled. Since each data item/ cellular measurement about Handover is provided with a timestamp, the selected pairs of Handovers supply their timestamps which form the "Segment Readings" for the specific segment.

The occasional gathering of the location data is performed during the same predetermined period of time (the calibration period), and preferably comprises:

obtaining the Sample of accurate location measurements for the specific road section; each of said location measurements is performed by a specific location sensor, associated with a timestamp and a specific accurate location data (it would be ideal could the accurate location measurements be performed at Handover points, but they may be taken at different points along a road section which overlaps the road segment so that the length covered by the section should be significant compared to the segment length (e.g. above 50% or even 80%); optionally, it also comprises calculating ETTg (estimated travel time) for said Sample using pairs of accurate location measurements made by a specific location sensor. As can be understood, the method's accuracy increases as the road segment is uniform from the point of traffic speed/travel time. It goes without saying that the shorter the road segments, the more accurate the estimates. It is correct both in case the Sample is performed for a longer road section or for a shorter road section than the road segment, but the criterion is at least a partial overlap, say of about 60% between the sample section and the road segment.

The most preferred location sensors can be: active (transmitting) GPS devices in vehicles which occasionally move along the road segment, or immobile cameras for Radar/photo/video observation (for example, at dangerous road sections or at road crossings) which may provide a sample of measurements covering at least part of the road segment.

The location data (measurements) from such location sensors can be obtained occasionally, from time to time, either by a special request (say, in the case of an immobile camera, a special time window may be documented - and a sample of location measurements may be produced by such a camera which has quite a long observation distance), or sporadically (say, in the case of a

GPS sensor of a taxi, transmitting information when occasionally moving along the road segment -). The amount of data derived from such location sensors is not expected to be enough to cover the roads continuously but is expected to generate a sufficient number of samples of accurate traffic readings along the calibration period per road segment.

The phase of data gathering is terminated by the step of combining the cellular data with the location data, and is preferably performed by matching, by time, the flow of cellular measurements (said "Segment Readings") with said Sample of accurate traffic measurements/readings , wherein both the Segment Readings and the Sample are taken for the same road segment or portion thereof and being both performed around the same time; the matching thereby producing Verified Segment Readings forming in turn the Verified Database of said road segment.

The obtained information of Verified Segment readings, i.e., the Verified Database can be further used for creating the geographically verified cellular map (or a cellular-topologic map/model ) of the road segment.

The Verified Database may be fulfilled by estimates of ETT for the road segments. For each Verified Segment Reading two estimated ETT values can be calculated - one of the Sample (a geographic or GPS ETTg value) and one of the Segment Reading (a cellular ETTc value).

The Database of geographically verified cellular readings (i.e., the Verified Database) may be considered a preliminary product of the proposed technique, which actually constitutes raw data which may be separately sold to traffic data service providers. The Verified Database may further allow obtaining a cellular model which is one of the main traffic data products of the proposed technique.

The first and the second main traffic data products, namely the cellular model and the Segment ETT can be determined in the frame of the second inventive technique for monitoring traffic data. The second technique comprises a gathering/modeling phase (possibly but not obligatory the one described above), and additionally comprises the calibration phase and the runtime phase mentioned before in the description:

the calibration phase, for deriving coefficients, per cellular handover, between the ETTc values (obtained from cellular data) and the ETTg values (obtained from location data) , thereby forming a cellular model of the road segment, and

the runtime phase, wherein the cellular data only, being collected on the continuous basis, is being continuously processed using said coefficients, to calculate the Segment Estimated Travel Time ETT. The Segment ETT is the final ETT product which may be continuously supplied to providers/users of traffic information services. In more details, the calibration phase may comprise a correlation procedure in the Verified Database, including the following operations:

- for each road segment, grouping all the Verified Segment Readings by Handover pairs which respectively produced said Verified Segment Readings

- for each Handover pair, checking correlation of the value of ETTg (the geographic ETT) calculated for a specific Sample, with values of ETTc (the cellular ETT) calculated for the respective Verified Segment Readings,

- in case the cellular ETTc for a specific Verified Segment Reading correlates with the geographic ETTg of the Sample, the Handover pair of such a Verified Segment Reading is considered a reliable or Traffic Usable

Handover pair (TUHP) for further use in real time traffic monitoring, and

- per each said Traffic Usable Handover pair, a coefficient (a transformation coefficient) is derived between the cellular ETT and the geographic ETT, for further use of said coefficient in determining final ETT for the road segment - from cellular data in real traffic monitoring

Preferably, said correlation can be checked by conducting a correlation test between the geographical ETTg (say, GPS ETT) and the cellular ETTc (Segment Reading Travel time); for example, a confusion matrix of traffic loads may be built, or a regression may be conducted as the correlation test. Such regression may also derive the abovementioned transformation coefficients.

The calibration phase may be run from time to time (including in parallel to the runtime phase below) in order to keep the models up to date, as the cellular network and road network change with time. The runtime phase is actually the phase where real traffic monitoring takes place, and where, based on the obtained model of the road segment, the final ETT product per road segment is issued to a provider/user of suitable services.

It should be emphasized that the transformation coefficient which was derived per TUHP at the correlation phase is further used for calculating the traffic product of interest -final ETT - from cellular data obtained by said TUHP at the running phase of the method, and without further use of the location data source/sensors. The runtime phase may, for example, comprise the following:

- Continuously generating cellular/handover events, i.e. continuously receiving cellular measurements on cellular events from the cellular sensors;

- in case that, in a predetermined time frame, one cellular sensor generates two handover events matching one of said Traffic Usable Handover Pairs, generating a reading of the final ETT for that TUHP by using the

transformation coefficient for said TUHP,

- performing the above two steps for one or more different cellular sensors in the predetermined time frame (and probably for different TUHPs);

- Processing thus received two or more readings of the final ETT and generating a final estimated travel time - the Road Segment ETT.

The predetermined time frame should be understood as comparable with a period suitable for a typical cellular sensor for traveling along the road Segment, say between a pair of its handovers.

To achieve the object of determining accurate geographical locations of cellular events, such as handovers, the method may take into account traffic jams that occupy only part of the road Segment, in a way that one part of the segment is congested (e.g. until a location of a traffic accident or another so- called interruption point ) and the 2^nd part is free (e.g. after the interruption point) The exact location of said interruption point can be easily spotted from location/GPS data (by noticing the point in which the travel speed changes significantly).

Next, by cross referencing several different Traffic Usable Handover pairs for the road segment, both from the above mentioned traffic jam conditions and from regular conditions, it is possible not only to find coefficients for pairs of handovers, but also to accurately pinpoint the location of each handover, as ratios between the coefficients in the 2 scenarios (with and without the interruption) would correlate with the distances between the interruption point and the exact handover point.

According to a second aspect of the invention, there is also provided a system for implementing the proposed technique. For gathering traffic data and modeling a road segment, the system comprising: a Modeling Server adapted to receive:

cellular measurements and a cellular map of the road segment,

a digital geographic map of the road segment, and

occasional accurate location measurements on the road segment, the Modeling Server being capable of matching the cellular measurements with said occasional accurate location measurements taking into account the cellular map and the geographic map so as to obtain a Verified Database comprising geographically Verified Segment Readings.

In the system for modeling, the Modeling server may be further provided with functionality (software and/or firmware means) for calibrating data of the Verified Database of said road Segment, so as to produce Configuration Data for further calculating estimated travel time ETT of the road Segment, thereby forming a model of the road Segment

There is further provided a system for traffic data monitoring, comprising said system for gathering traffic data and modeling a road segment, wherein the Modeling server is further suitable for calibrating data of the Verified Database of said road Segment, so as to produce Configuration Data for further calculating estimated travel time ETT of the road Segment, thereby forming a model of the road Segment;

said system for traffic data monitoring further comprises a Real Time Server for continuously receiving cellular measurements from the mobile network, and processing the received cellular measurements using the Configuration Data received from the Modeling server, thereby producing the ETT for the road segment.

The system may comprise a location data source capable of collecting occasional location data from location sensors for providing occasional accurate location measurements for the Modeling Server. The system may be presented as an extended system further including a Mobile Network for collecting cellular measurements from cellular sensors and producing a cellular map of road segments, and a Road Network capable of producing the digital map of road segments.

In accordance with a third aspect of the invention, there is also provided a software product comprising computer implementable instructions and/or data for carrying out the method according to the above description, stored on an appropriate computer readable storage medium so that the software is capable of enabling operations of said method when used in a computerized system, for example in the system according to the invention. There is also provided a suitable computer readable storage medium (for example, one or more computer servers), accommodating the software product or portion thereof. The proposed technique has been proven to determine geographic location of cellular events and to perform the traffic data gathering and processing both very accurately (with accuracy equal or higher than the mentioned field drives provide) and by involving a very small effort and low costs, even if the occasional location data (forming part of the traffic data to be gathered) has to be bought. Brief description of the drawings

The invention will further be explained and illustrated with the aid of the following non-limiting drawings, in which:

Fig. 1 is a schematic illustration of a road segment where traffic is monitored using two traffic data sources: cellular source and location source.

Fig. 2 is a simplified block diagram of a pre-modeling phase which may precede the method of the invention.

Fig. 3 is a simplified block diagram of the data gathering and modeling phase according to the invention.

Fig. 4 is a simplified block diagram of the calibration phase according to the invention.

Fig. 5 - is a schematic illustration of a graph of regression being an exemplary result of a correlation test conducted for cellular and location measurements, to build a cellular model of a road segment.

Fig. 6 - is a simplified block diagram of the runtime phase of the proposed method.

Fig. 7 - is a simplified block diagram of an embodiment of the system proposed for implementing the method according to the invention.

Detailed description of preferred embodiments

Fig. 1 is a schematic illustration of a road segment, schematically marked 0-1 , of a road 10, where vehicular traffic is monitored using two traffic data sources: cellular data source (DSl) and location data source (DS2). The cellular data source DSl is formed in a cellular network 30 serving an area including, inter alia, also the road segment 0-1. The network 30 is in continuous communication with a plurality of cellular sensors; on the segment 0-1 we occasionally see sensors 14, 18, 22 which, for the purpose of our description, are cellular phones carried by moving vehicles 12, 16, 20. The cellular network 30 forms a number of cells (cellular clusters) A, B, C, D, E, F, etc. so that some of the cells cover the road segment 10. Borders between the cellular clusters are characterized in that a cellular sensor crossing such a border undergoes a cellular handover. For simplicity of the drawings , cellular events are presented as handover events only. Some handovers are shown in Fig. 1 , and approximate locations thereof are marked by thick lines AB, BC, etc. In practice, just some vehicles have GPS devices, and only some of them (for example, GPS devices of taxis in communication with a taxi station dispatcher, but not only) are the active, transmitting ones which can be considered location sensors. One such active location /GPS sensor 17 is shown on a car 16. Another active location/GPS sensor is shown on a car 19. In other words, such mobile location sensors appear in the road segment just occasionally. It should be emphasized that, in the new proposed technique, location sensors do not have to be respectively co-located with cellular sensors. It is quite an important detail of the technique, speaking for the independent manner of taking cellular and location measurements. To illustrate that, car 19 in Fig. 1 does not have a cellular sensor/phone. Note that the double sensor car type like car 16 is not required in the new technique and therefore it is shown by dotted lines. Further, the road 10 may be equipped with traffic observation cameras 24, 26 and 28, and some of them may be located within the road segment 0-1 as shown in Fig.1. The cameras constitute another kind of location sensors; it should be noted that they do not have co-located cellular devices. All location sensors are shown as thick black rings.

Fig. 1 may serve for acknowledgement of some known principles of traffic monitoring and traffic data collection; however, it will also be used for explanation of the newly proposed method of the invention.

Let vehicle 12 moves along the road 10, and carries a cellular phone being a cellular sensor 14. If vehicle 12 passes along the whole segment 0-1, sensor 14 crosses the handovers AB, BC and CD and at the moments of passing them may send to the network 10 respective Cellular data 1 , 2 and 3. If vehicle 12 takes a bypass road (not shown), its cellular sensor will send to the network 10 cellular data at the moments of crossing other handovers, for example AE and FD. Each cellular data event/item has its timestamp, ID of the sensor and ID of the handover. It is understood that a cellular measurement performed for any pair of the handovers belonging to the road segment 0-1 may produce a value of travel time and -if exact locations of the handovers are known, - also a value of speed for the current hour and for the current road conditions. A plurality of cellular measurements related to a plurality of vehicles may form a Database that, in case it is supported by exact location data, may allow calculating the Segment Estimated Travel Time (ETT). Many prior art solutions utilize continuous collection of location data from mobile GPS sensors. Indeed, when there are many vehicles like 16 (having an active GPS sensor 17) on the road segment 10, it is easy to collect multiple combined measurements, similar to the following one: associate Cellular data 1, 2 from the cellular sensor 18 with Location data GPS1, GPS2 collected from the GPS sensor 17 of the same vehicle 16. There is also a possibility to perform a preliminary test drive, using a vehicle like 16 to determine exact distances along the road section 0-1. However, in reality providing the drive tests for a large territory is very expensive. Also, the practice says that a huge number of road segments to be monitored are not sufficiently loaded with active GPS sensors.

To resolve this contradiction, the present invention comes with its proposed technology which is capable of providing accurate traffic estimates in the absence of continuous collection of location data, and without test drives.

The figures which follow will mainly refer to such a scenario of Fig. 1.

In the following exemplary description, cellular events are considered to be cellular handovers. The proposed inventive technique is perfectly applicable and advantageous when active location sensors are not co-located with cellular sensors, i.e. for example when occasional location data is obtained from car 19 and cameras 24, 26, 28. Fig. 2 shows an exemplary block-diagram of a preliminary, pre -modeling phase of the proposed method. It may take some hours/days. The pre -modeling phase is intended to determine potential handovers of a specific road segment. The pre -modeling phase utilizes: initial data formed in a cellular network 30 and initial data comprised in a road network (digitized map, or topological info ) 40 for processing in a processing unit 32.

1. Cellular sensors moving in the area served by the mobile network 30 create a continuous stream of cellular event data. Thus collected data can be extracted from the mobile network, and pairs of cells between which handovers occur can be logged and counted as" Handovers". All the handovers that occur more frequently than others (e.g. more than 100 times in one day) are marked as "common handovers" of the area (block 34 in the processing unit 32).

2. The geographic map of each road segment (see "Road obtained from road network 40) is overlaid/crossed with the mobile network cells that may reasonably cover it (part of information obtained from the network 30). This data is approximated based on the network cell plan/map (antenna coordinates, direction, width, approximate coverage radius etc.) received from the network 30. For this approximation the estimated coverage taken is positively biased, i.e. if we are not sure whether the cell covers the road or not we may assume that it does. At the end of this phase we receive a list of cells that may cover the relevant road segment ("Segment cell coverage" - block/list 36).

3. For each road segment, the Segment cell coverage (list 36) is crossed with the Common Handovers of the area (list 34). This then results in obtaining a list of possible handovers that may occur when traveling along the road segment ("Potential Segment handovers" - block/list 38).

Fig. 3 schematically illustrates the proposed modeling phase comprising data gathering phase, which aims at obtaining Verified Segment Readings (and there-from, a Verified Database of the segment); in the Verified Segment Readings, cellular data is verified by location data, which is preferably occasional/sporadic.

1. During a period of a few days/weeks (a calibration period), handovers events are collected from the mobile network 30 and if they belong to a Segment's of interest "Potential Segment Handovers", they are logged. Furthermore, if for a given mobile device/sensor more than one Handover is recorded that belongs to the Potential Segment Handovers, then the following is logged: First Handover cells, First Handover timestamp, Second Handover cells, Second Handover timestamp, estimated Travel time ETTcellular or ETTc (Second Handover time minus the First Handover time); this data is logged as an item "Segment Reading". The "Segment Readings" list is shown as block 42. Actually, selecting from the Potential Segment Handovers pairs of handovers (may be not adjacent pairs, any pair from available combinations) for a mobile device, has the meaning, that the handovers may have indeed taken place in the segment of the road we have modeled. Further actually, the term "Segment Readings" has the meaning of cellular measurements taken on the road segment.

2. During the same calibration period of time, a sample of accurate traffic location data is collected for the monitored roads in the area, at least (and preferably) from sporadic, occasional location sensors and accumulated in a database/center 44 of the location source data. The most common and wide spread suppliers of location data would be GPS devices in vehicles, (say, sensor 21 on car 19; but also cameras 24, 26, 28) sending GPS locations to the center 44. The amount of data derived from such devices is unpredictable and is not expected to be enough to cover the roads continuously, but is expected to generate in block 44 a limited sample of accurate traffic location measurements during the calibration period. An accurate GPS traffic reading is derived from some (block 48 in block 32) is formed by two location measurements made by one location sensor/user (GPS) along the road segment, has two timestamps and is preferably accompanied by its ETTg (estimated travel time to travel along the road section covered by the GPS traffic reading).

3. If it is possible, (when matching a plurality/flow of cellular Segment readings with a limited plurality/sample of location readings) to match a Segment reading with an accurate GPS reading (both on the same segment and around the same time), then a Verified Segment Reading is generated. A plurality of such readings form a Verified Database per road segment (see block 50).

In other words, if we have the location measurements (and thus GPS or camera traffic readings at two points for one and the same vehicle for a road segment), we may then adjust the "Segment Readings" to that data, i.e. to select the Segment Readings taken approximately in the same period of time. We thus obtain Verified Segment Readings. We keep in mind that for pairs of handovers, their "cellular ETT or ETTc" is calculated using timestamps of the two handovers. For each location (geographical) measurement there are also two timestamps , so the "geographical ETT or ETTg" may be calculated.

Such raw data comprising Verified Segment Readings and optionally even the two values of ETTc and ETTg is a valuable information which may be provided to traffic data service providers.

Fig. 4 shows a simplified scheme of the calibration phase of the proposed technique which logically terminates the modeling phase. The calibration phase comprises performing correlation in the Verified Database (see block 50 of Fig. 3). In a preferred example, a correlation check is conducted between the geographic ETTg calculated from the sporadic location data, and the cellular ETTc calculated from the cellular data, in order to obtain the following configuration data, namely: -to obtain so-called traffic usable (reliable) handover pairs and transformation coefficients for further calculating the product of interest -Segment ETT -using only cellular data, in a running phase of the technique. The correlation is performed for the same road segment, per handover pair, for example as follows:

1. For each road Segment, all Verified Segment Readings are grouped by the first and second Handovers (i.e., by Handover pairs) - see block 52.

2. For each Handover pair a correlation test is conducted between the geographic ETT or ETTg (obtained from GPS and/or camera readings) and the Segment Reading Travel time (cellular ETT or ETTc). To do that, a regression may be conducted as a correlation test - shown as a block 54.

3. Those Handover pairs that result with high correlation (over varying traffic situations), would represent cellular events that correlate with the actual traffic and will be called Traffic Usable Handover Pairs (reliable handover pairs). Only these pairs will then be used for the real time traffic monitoring. Once the configuration data is obtained for all TUHPs of the segment, the model ( a verified cellular model of the road segment) can be considered to be complete. Upon combining the information on configuration data from several such cellular handover pairs along the road, a verified cellular model/map of the road can be built. During the runtime phase ( will be shown in Fig. 6) of the technique for traffic monitoring, the obtained cellular model - i.e., the obtained configuration data including coefficients will be used to transform and size the Cellular reading travel time (cellular ETT or ETTc of a specific reliable handover pair) to the final value of road Segment ETT.

The results of the correlation test are registered in the list of block 56.

By cross referencing several different Traffic Usable Handover pairs for a segment, and by using Accurate GPS traffic readings for traffic jams that occur along different parts of the segment, it is possible not only to find coefficients for pairs of handovers, but also to accurately pinpoint the location of each handover.

If traffic jams are utilized for more accurate allocation of cellular

events/handovers, even more accurate/checked information than that of block 56 will be found in an optional block 57. Actually, block 57 comprises data with the accurate geographic locations of handovers in the road segment.

Fig. 5 schematically shows a graph of correlation and regression which may be built for obtaining the transformation coefficients.

Let curve 58 is the geographical ETT (ETTg) changing during the daytime, built using a number of location (GPS or camera) measurements performed on the road segment 0-1 (see Fig. 1).

Curve 60 is the cellular ETT (ETTc) changing during the daytime, built using a number of cellular measurements (readings ) which are performed for the pair of handovers AB and CD on the road segment 0-1.

Curve 62 is the cellular ETT (ETTc) changing during the daytime and built using a number of cellular measurements (readings ) being performed for the pair of handovers EF and HG. .

As can be seen, there is a strong correlation between the curves 58 and 60 and no correlation between the curves 58 and 62. The handover pair AB+CD will be therefore considered the Traffic Usable Handover pair, and the pair EF+HG will be dropped from the Database. The desired regression/transformation coefficients may be calculated for the Handovers pair AB+CD, in order to arrive to the real (geographic) ETTg values of curve 58 from cellular ETTc values of the curve 60 .

Whenever a coefficient of correlation coefficient (a transformation coefficient) is derived, then the cellular readings of the Handovers pair AB+CD can be used for estimating the ETT of the road segment 0-1 without further use of location sensors, and provided that the road segment is considered homogenous.

Fig. 6 schematically illustrates the real time, runtime phase of the method, where location sensors are not used and location data is not required at all and the main product of the technique - ETT - is received using the obtained Cellular Model of a road segment/road.

Cellular measurements related to cellular events are continuously received from the mobile network 30.

If a user (cellular sensor) generates 2 handover events (i.e., two cellular measurements on such events) that relate to a Traffic Usable Handover pair (crossing with the TUHP list of block 56 or 57), then a Traffic Reading is generated and stored in block 64. The traffic reading may be generated by using the regression/transformation coefficients calculated as described above and stored in the block 56 (57). Several traffic readings from different users in a given time frame are processed and, based on the Model (more specifically - using TUHPs and the stored transformation coefficients), several ETT values are generated in 64 and finally, a final estimated travel time Segment ETT is generated (block 66).

Fig. 7 schematically illustrates a simplified block-diagram of one embodiment of an extended system for implementing the proposed method of traffic monitoring.

Fig. 7 shows a first system 100, marked by a dotted contour, for modeling a road segment and possibly, for determining accurate geographic locations of cellular events (such as handovers). The system comprises a mobile network 102 being a source of Cellular data collected from cellular sensors (not shown) and capable of creating data on cellular events/handovers and a cellular map of an area comprising a road segment of interest. System 100 further comprises: a Road Network 104 (such as a digital road map available from a suitable network),

a Location data source 106 suitable for providing occasional/sporadic though accurate location measurements (for example, a center for accumulating GPS data measurements which may be occasionally collected from various GPS sensors, from immobile cameras, etc.);, and

a Modeling Server 108.

Based on the information received from blocks 102, 104 and 106, the Modeling Server 108 is at least capable of building a Verified Database of segment readings at least for one road Segment, according to the algorithm briefly shown in Fig. 3 and described above. In other words, Server 108 accommodates relevant portions of a software product responsible at least for gathering cellular data and occasional location data, partially modeling the road segment and producing the Verified Database..

Generally, the modeling server 108 should be capable of modeling a number of road Segments and consequently a road and a road area) so as to obtain a Cellular model 107 and, optionally, a verified cellular Map 109 thereof.

To this end, the modeling server 108 is provided suitable hardware/software for performing calibration of the Verified Database according to the brief algorithm shown in Fig. 4 and disclosed in appropriate portion of the description. The calibration will result in obtaining the Configuration Data (that comprises reliable handover pairs TUHPs and transformation coefficients) and which actually constitute building blocks of the cellular model 107 of interest. Another important product generated by the first system 100 after the modeling may be a Verified cellular Map 109. The Map 109 may further comprise accurately determined locations of cellular events (e.g., handovers), if the modeling server 109 is also adapted to process data on traffic jams ( not shown in the drawing).

It should be noted that the server 108 may optionally be adapted to perform the pre-modeling phase (for example, according to the operations shown in the flow chart Fig. 2) for ensuring the modeling stage. Alternatively, results of the pre-modeling phase may be obtained by 108 from an external source (not shown). Fig. 7 also indicates a full, second system 112 for traffic data monitoring; the second system 112 comprises the first system 100 with the Modeling server 108 capable of performing the above -described correlation functions and producing the Cellular Model , which manifests itself by issuing the Configuraiton Data (both marked 107). The second system 112 further comprises a Real Time Server 110 capable of continuously receiving cellular measurements (in this example, data on handover events) from the mobile network 102. The Server 110 is responsible for processing the received cellular measurements using the Cellular Model 107 supplied by the Modeling server 108 (for example, according to the algorithms schematically illustrated and described with reference to Fig. 6). The processing performed at the Real Time Server 110 corresponds to the runtime phase of the method for traffic data monitoring.

It should be noted that the Real Time Server 110 does not need and does not receive any location measurements, and produces its main product - estimated travel time ETT per road segment (marked 111), based only on cellular measurements received from the mobile network 102 (the measurements being collected at the network 102 from mobile sensors), and using the Cellular Model 107. As mentioned above, the main components of the Cellular Model 107 are items of Configuration data, each comprising a Traffic Usable Handovers Pair (TUHP) and a transformation coefficient/formula which can be used, in the runtime phase of the method, for transforming an ETTc value, obtained for the pair of TUHPs, into an ETT reading. ETT per road segment is further formed in the server 110 from a number of ETT readings obtained for different TUHPs of the segment during a predetermined time frame.

It goes without saying that in the complete system 112, the servers 108 and 110 may be joined in a common server accommodating all necessary portions of a complete software product responsible for performing the inventive method for traffic monitoring. Component algorithms of the proposed software product are schematically illustrated in Figs. 2 to 6.

The compact system 100 may comprise only the modeling server 108 being capable of communicating with sources of cellular data and location data, such as networks 102, 104 and the GPS source data 106. The compact system 112 may comprise only the Modeling Server 108 and the real time server 110 being in communicaiton with one another and capable of receiving the necessary cellular and the location data. It should be appreciated that the proposed method and system may be implemented in a number of differing versions and embodiments which, though not described in detail in the present description, should be considered part of the invention whenever defined by the claims which follow.

Claims

Claims:

1. A method for modeling a road segment, wherein the modeling comprises:

- matching of routine cellular measurements with occasional location measurements, the cellular and the location measurements being gathered independently on the road segment;

- based on the matching, creating a verified cellular map or model of the road segment where locations of cellular events are verified by the collected location measurements.

2. A method for modeling a road segment, comprising gathering traffic data received from cellular sensors and location sensors, and further performing matching thereof, wherein

obtaining occasional accurate location measurements by location sensors during the same calibration period, thus obtaining a sample of location measurements,

and wherein

said matching comprises matching, by time, the flow of cellular measurements with the sample of accurate location measurements, both related to said road segment during the calibration period.

3. The method according to claim 2, wherein the step of performing routine cellular measurements to obtain the flow of cellular measurements comprises:

- obtaining cellular measurements formed from the cellular data supplied by the plurality of cellular sensors at moments of cellular events ; each cellular measurement comprising ID of the cellular sensor, ID of two of the cellular events performed by said sensor during one travel and two respective timestamps for said two cellular events.

4. The method according to any one of claims 2 to 3, wherein the step of obtaining a sample of accurate location measurements comprises:

- occasionally and independently obtaining location measurements on points along a road section overlapping with the road segment by more than 50%; each of said location measurements comprising a timestamp, and geographical location of said point.

5. The method according to any one of claims 2 to 3, wherein said matching comprises:

- combining the flow of cellular measurements with the sample of said location measurements, thus obtaining combined data,

- processing said combined data to find those of said cellular measurements respectively matching with said sample of location measurements by time, thereby producing a Verified database of traffic data for said road segment, the Verified database being formed by a plurality of Verified Segment Readings.

6. The method according to claim 5, wherein said matching comprises calculating, for each Verified Segment Reading, a location based Estimated Travel Time (ETTg), and a cellular based Estimated Travel Time (ETTc).

7. The method according to any one of the preceding claims, comprising preliminary determining potential cellular events being potential handovers, actual for the road segment.

8. The method according to any one of claims 5 or 6, wherein the modeling further comprises:

- calibrating said verified Database of the road segment thereby obtaining a cellular model of the road segment in the form of configuration data to allow further calculation of Estimated Travel Time ETT of the road segment based on cellular measurements only.

9. The method according to Claim 8, for traffic data monitoring on the road segment, the method further comprises performing a runtime phase by collecting cellular measurements from the road segment in real time and by calculating ETT for the road segment based on the cellular measurements only and using the configuration data.

10. The method according to Claim 1 , for traffic data monitoring on the road segment, wherein the modeling comprises:

- occasionally gathering, from location sensors, said accurate location measurements on the road segment during a calibration period,

- verifying the occasionally gathered accurate location measurements by a plurality of said routine cellular measurements gathered from cellular sensors on the road segment during said calibration period,

- calibrating said verified Database of the road segment thereby obtaining said cellular model of the road segment in the form of configuration data allowing calculation of Estimated Travel Time (ETT) of the road segment based on cellular measurements only;

the method further comprises

- performing a runtime phase of the method, by collecting cellular

measurements in real time and by calculating ETT for the road segment, based on the cellular measurements only and using the configuration data.

11. The method according to any one of Claims 8, 9 or 10, wherein the calibrating comprises determining the following elements of the configuration data:

- reliable pairs of cellular events, selected as those demonstrating maximal correlation when calculating for any one of said pairs: ETTg values using related to that pair accurate location measurements, and ETTc values using related to that pair cellular measurements;

- transformation coefficients for each of the selected reliable pairs of cellular events, determined as enabling to obtain highly accurate ETTg values from ETTc values calculated using cellular measurements only.

12. The method according Claim 7, wherein said determining of the Potential handovers comprises the following steps:

- obtaining a stream of cellular data from the mobile network, collected from the cellular sensors moving in the area covered by the mobile network, and extracting there-from information about pairs of cells between which handovers occur;

- selecting, from the information, data on such pairs of handovers that occur more frequently than others, and marking the selected data as Common handovers;

- obtaining an estimated segment cell coverage for said segment, by crossing a cell map of the mobile network and a road segments map derived from a digital road map; - crossing the information on segment cell coverage with the Common handovers data, thereby obtaining a list of Potential Segment handovers which are most expected in the road segment.

13. The method according to Claim 12, wherein the modeling performed for the road segment during the calibration period comprises gathering of cellular data and occasional gathering of location data,

wherein the gathering of cellular data includes:

- collecting data about any Handovers from the mobile network and logging said data belonging to a specific segment's Potential Segment Handovers;

- in case that, for a particular cellular sensor more than one Handovers are considered the Potential Segment handovers, for any pair of the Potential Handovers logging the following information as a" Segment Reading", wherein any said pair comprises a First Handover and a Second Handover:

a First Handover's cells, a First Handover's timestamp, a Second Handover's cells, a Second Handover's timestamp, Travel time defined as a Second Handover time minus the First Handover time derived from respective timestamps thereof.

14. The method according to Claim 13, wherein the occasional gathering of the location data is performed during the same predetermined period of time being the calibration period, and includes:

obtaining the Sample of accurate location measurements for a specific road section covering said road segment by at least 60%, wherein each of said location measurements is performed by a specific location sensor, and is associated with a timestamp and a specific accurate location data.

15. The method according to any one of Claims 2 to 14, wherein the road segment model is fulfilled by estimates of ETT, for each Verified Segment Reading two estimated ETT are calculated - one of the Sample, being a geographic or GPS ETTg value, and one of the Segment Reading, being a cellular ETTc value.

16. The method according to any one of claims 8 to 15, wherein

the calibration comprises a correlation procedure in the Verified Database, including the following operations:

- for each Handover pair, checking correlation of the value of ETTg calculated for a specific Sample, with values of ETTc calculated for the respective Verified Segment Readings,

- in case the cellular ETTc for a specific Verified Segment Reading correlates with the geographic ETTg of the Sample, the Handover pair of such a Verified Segment Reading is considered a Traffic Usable Handover pair (TUHP) for further use in real time traffic monitoring, and

- per each said Traffic Usable Handover pair, a transformation coefficient is derived between the cellular ETT and the geographic ETT, for further use of said coefficient in determining final ETT for the road segment from cellular data during real traffic monitoring.

17. The method according to any one of claims 9 to 16, wherein the runtime phase comprises :

- continuously generating handover events by continuously receiving cellular measurements on cellular events from the cellular sensors;

transformation coefficient for said TUHP, - performing the above two steps for one or more different cellular sensors in a predetermined time frame;

- processing thus received two or more readings of the final ETT and generating a final estimated travel time being the Road Segment ETT.

18. The method according to any one of Claims 2 to 17, collecting said occasional location measurements performed in a traffic jam occupying a part of the road Segment, and utilizing said traffic jam location measurements for accurately determining location of cellular events.

19. The method according to any one of the preceding claims, comprising gathering of said occasional location measurements from location sensors being either active GPS devices or immobile traffic observation cameras.

20. A system for implementing the method according to any one of Claims 1 to 19.

21. A system for gathering traffic data and modeling a road segment, comprising:

a modeling server adapted to receive

-cellular measurements and a cellular map of the road segment,

a digital geographic map of the road segment, and

- occasional accurate location measurements on the road segment, the Modeling Server being capable of matching the cellular measurements with said occasional location measurements taking into account the cellular map and the geographic map, and verifying locations of cellular events by said location measurements

so as to obtain geographically Verified Segment Readings for respective road segments, forming together a Verified Database.

22. The system according to Claim 21, wherein the Modeling Server is designed for calibrating data of the Verified Database of said road segment, so as to produce configuration data for further calculating Estimated Travel Time ETT of the road segment, thereby forming a cellular model of the road segment.

23. The system according to Claim 22, intended for traffic data monitoring, wherein the Modeling server is suitable for calibrating data of said Verified Database so as to produce Configuration Data for further calculating estimated travel time ETT of the road Segment, thereby forming a cellular model of the road Segment;

said system for traffic data monitoring further comprises a Real Time Server for continuously receiving cellular measurements from the mobile network, and processing the received cellular measurements using the Configuration Data received from the Modeling server thereby producing the ETT for the road segment.

24. The system according to any one of Claims 21 to 23, further comprising a location data source capable of collecting occasional location data from location sensors for providing the occasional accurate location measurements to the Modeling Server.

25. A software product comprising computer implementable instructions and/or data for carrying out the method according to any one of Claims 1 to 19, stored on an appropriate computer readable storage medium so that the software is capable of enabling operations of said method when used in a computerized system.

26. A computer readable storage medium such as a server, accommodating the software product according to Claim 25, or a portion thereof.