Page 327 - Big data - Concept and application for telecommunications
P. 327

Big data - Concept and application for telecommunications                       5


            7.2.2   Users' behaviour data

            It is required to choose main services both in packet switch (PS) and circuit switch (CS) domains to monitor
            the users' behaviour.
            For the PS domain the data is collected through the Gn/S1-U interface [b-3GPP TS29.060] and the S1-U
            interface [b-3GPP TS36.413], including but not limited to: who are the users, when, where and what they are
            doing on the Internet, how much traffic they consume and what the result is.

            For the CS domain the data is collected through a mobile switching centre (MSC) in CN, including but not
            limited to: who are the users, when, where and to whom they are using voice call, how long the voice call
            lasts and what the result is.

            7.2.3      Common parameter data
            7.2.2.1    Common parameters in networks

            Common parameter data in networks mainly includes the distribution of base transceiver stations and cell
            key parameters. Users' behaviour data and signal data should be computed for every cell according to a set
            of indexes which can represent the running status of the mobile network. For base transceiver stations these
            indexes can be summarized by their own cells.
            It is required to collect data items such as the cell information, the base station which the cell belongs to, the
            network type, the location of the base station (i.e. the longitude and the latitude of the base station or the
            administrative region that the base station belongs to).
            7.2.2.2    Common parameters in applications

            Common parameter data in applications mainly includes the classification and names of applications that
            users enjoy. Users' behaviour data and signal data should be computed for every application classification or
            popular application according to a group of indexes which can represent the service sensing. For every cell
            or base transceiver station these indexes can be summarized by users' accesses.
            7.2.2.3    Common parameters in users

            Common parameter data in users mainly includes the mobile station international ISDN number (MSISDN)
            and the terminals they use. The users' behaviour data and signal data should be computed for every place
            the MSISDN belongs to or the terminal type that users have according to a group of indexes which can
            represent the users' experience in the network and service. For every cell or base transceiver station these
            indexes can be summarized by users' accesses.
            It is required to collect data items such as the user identity, the MSISDN, the international mobile subscriber
            identity  (IMSI)  and  the  terminal  information,  i.e.  the  terminal  brand,  the  terminal  model,  the  terminal
            operating system.


            7.3     Data collection methods
            There are different methods to collect different data categories.

            The  users'  behaviour  data  is  collected  by  deep  packet  inspection  (DPI)  from  the  Gn/S1-U  interface.  For
            information on DPI requirements, see [ITU-T Y.2770] and for information on the DPI framework, see [ITU-T
            Y.2771].
            The signal data in the RAN is collected from network management elements through files and the signal data
            in CN is collected by DPI from the corresponding interfaces.

            The common parameter data is collected from external databases.

            7.4     Requirements for big data pre-processing
            The  data  collected  through  different  methods  is  stored  in  files,  messages,  queues  or  databases.  Before
            analysis and storage, the data needs to be pre-processed. There are three stages of data pre-processing; data
            extraction, data transformation and data load.

                                                                           Network and infrastructure    319
   322   323   324   325   326   327   328   329   330   331   332