pilot pollution

21
3G Optimisation Workshop Pilot Pollution London, 28 th of April 2003

Upload: adawit

Post on 17-Sep-2015

239 views

Category:

Documents


4 download

TRANSCRIPT

  • 3G Optimisation Workshop

    Pilot Pollution

    London, 28th of April 2003

    CoE Radio Meeting, Paris 01/10/02

  • DefinitionA pilot pollution area is an area where an excessive number of scrambling codes are received and lead to a degradation of qualityIn other words Pilot Pollution corresponds to DL interferencePractical condition:

    Best server CPICH_Ec is goodAndBest server CPICH EcNo is bad

    CoE Radio Meeting, Paris 01/10/02

  • ThresholdsA single EcNo threshold is not recommended :when Ec decreases, even with a single cell and no traffic, the EcNo naturally degrades, therefore at low Ec level it is less straightforward to relate a weak EcNo to pilot pollutionWhen the Ec level is low, the main problem is coverage, not quality

    Recommended thresholds: Best server CPICH_Ec > -100 dBmAndBest server CPICH EcNo < -10 dBCaution:for unloaded network only

    CoE Radio Meeting, Paris 01/10/02

  • EcNo: theory Formula:Ec: Energy_chip, received level on the CPICHRSSI: Received Signal Strength IndicatorIn detail:

    CoE Radio Meeting, Paris 01/10/02

  • EcNo: Single cell no traffic

    CoE Radio Meeting, Paris 01/10/02

  • EcNo: Handover, no traffic

    CoE Radio Meeting, Paris 01/10/02

  • EcNo: typical values EcNo degrades with Ec levelEcNo degrades with overlaps. BLER degradation may occur only when 15 cells are received with the same level (lab test, field test to be performed). In the field it seems that the BLER starts degrading when EcNo < -10 dB.EcNo is almost similar for all overlap situations at lowest Ec levels, meaning that a pilot pollution area has a low impact on cell edge.

    CoE Radio Meeting, Paris 01/10/02

  • Main reasons for pilot pollution Radio design (F factor) High sites Large antenna beamwidth Non tilted antennasTraffic (P_bs) P_bs = DL power used by the Base Station. The higher the number of users, the stronger P_bs More interferences at busy hoursA closer look at the formula gives reasons for a bad EcNo:

    CoE Radio Meeting, Paris 01/10/02

  • Solution How can we improve Pilot Pollution areas?2 ways:1) Improve the Ec level of the best server

    2) Reduce the Ec level from some of the interferers

    Either way we need to identify and sort: the dedicated best serving cells the interfering cells

    CoE Radio Meeting, Paris 01/10/02

  • Generic Process: display First step is to display the pilot pollution areasWith either scanner or trace mobile data, this can be easily done using the condition:Best server CPICH_Ec > -100 dBmAndBest server CPICH EcNo < -10 dB

    CoE Radio Meeting, Paris 01/10/02

  • Generic Process: pre-check Before optimising a pilot pollution area, basic questions shall be answered: Were there maintenance problems? - check if all sites were transmitting properly Is there a site planned to be on air in a near future? Is the area worth optimising? (Secondary road, hill top)

    CoE Radio Meeting, Paris 01/10/02

  • Generic Process: identify cells Reporting levels from cells involved in a pilot pollution area is much easier than tracking an interferer in GSM

    With a scanner, all possibly decoded SCs are reportedWith a trace mobile, all SCs from the monitored set are reported meaning that no missing neighbours shall remain before analysing pilot pollution

    CoE Radio Meeting, Paris 01/10/02

  • Serving and interfering cells One basic difficulty when tackling a pilot pollution area is to define among all reported cell:

    1) Which cells are the proper dedicated best servers of the area

    2) Which cells are the interferers to optimise

    Both local and large scale analysis are needed

    CoE Radio Meeting, Paris 01/10/02

  • Serving cells The wished best serving cells are not obviously the measured best cells:in the area, top cells may fluctuatebest or 2nd best measured cell may be a remote cell that we would rather optimise

    A practical approach is necessarydefine as best cells those from the closest physical siteskeep a remote cell only if absolutely necessary (bringing useful service that closer cell could not provide)

    How many cells?3 cells is a typical number, as it is the size of the active set

    CoE Radio Meeting, Paris 01/10/02

  • Interfering cells Basically all non best server cells are the interferersWhich cells to optimise?It is necessary to get the big picture of the area and identify the worst interferers

    Worst interferers can be defined as:Cells involved in a high number of pilot pollution areasCells received with interfering levels without being best server. The following layer shall be displayed:

    CoE Radio Meeting, Paris 01/10/02

  • Optimisation actions We shall specify which levers may be used to optimise a pilot pollution area. Ideally we should be able to optimise antennas, namely: Tilt modification Azimuth change Change of antenna (to get a more narrow horizontal beam-width, or an electrical tilt)

    We do not recommend CPICH power changesThe best option is tilt modification

    CoE Radio Meeting, Paris 01/10/02

  • Optimisation actions Ideally the Ec level reduction on the interferer shall be such that:

    Caution when optimising the interferer:Do not reduce useful coverageDo not create new pilot pollution area by reducing the Ec level where this interferer is best server

    Use the field measurements to assess what dB reduction is needed, but use planning tool for a full simulation of the optimisation

    CoE Radio Meeting, Paris 01/10/02

  • Summary: flow-chart Tools required:Agilent ScannerNATSuite (post-process)NetAct planning tool

    Note: A different process will be required with a loaded network (to be studied)Updated thresholdsFocus on busy hoursTraffic related optimisation

    CoE Radio Meeting, Paris 01/10/02

  • Risks specific to UMTS In GSM: Frequency Planning High number of frequencyFew interferencesHigh sites and large antenna beam-width may not be a problem In UMTS No Frequency Planning All cell use the same frequency interferences depending on radio design and trafficsame radio design as for GSM may not be suitable

    CoE Radio Meeting, Paris 01/10/02

  • Expectations Moreover interferences will arise quickly with load and can take a few months or more to be optimisedtherefore pre-emptive actions are advised before the network gets uncontrollableIn the early days, few pilot pollution areas are expected (e.g. EN in Bristol), Without load we should not drop calls due to pilot pollution

    Yet feedback from areas with a higher site density of than EN Bristol will be useful

    CoE Radio Meeting, Paris 01/10/02

  • 3G Optimisation Workshop

    Pilot Pollution

    London, 28th of April 2003

    CoE Radio Meeting, Paris 01/10/02