An Analysis of GSM Handover based On Real Data by ijcsiseditor1


More Info
									                                                              (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                        Vol. 9, No. 7, 2011

    An Analysis of GSM Handover based On Real Data

Isha Thakur                                        Bhavana Jharia                                           Gopal Chandra Manna
ME Student, Communication System                   Associate Professor, Depptt. Of EC                       Sr. General Manager
Engineering Branch                                 Jabalpur Engineering College                             BSNL, Jabalpur
Jabalpur Engineering College, M.P., India          M.P, India                                               M.P, India                                

                                                                          the RxLev and RxQual ,alone, is not sufficient to provide the
Abstract—Handover decisions in GSM networks are based on the              accurate result for optimum handover solution .So, there is a
difference in received signal strength, between the serving cell          need of a new handover scheme which not only consider
and the neighboring cells. But in a practical scenario,                   RxQual and RxLev ,but also some other important parameters,
particularly in city area ,considering difference in signal level         for a better handover process .
strength alone , is an inferior criteria to decide handover issue
,because the towers are in close proximity & the absolute signal
strength is quite good to continue the communication ,without                 In the present work ,we focus our attention on
much difficulty .Also, in these environments, multipath                   incorporating some more decision criterion in the handover
reflections, scattering due to moving vehicles & diffraction from         algorithm .After an extensive study of the GSM measurement
multiple building edges ,contributes to poor signal quality, hence        reports obtained from a telecom company, it has been
forcing the mobile to transmit more power to continue the                 validated that the transmit power (TxPower) ,aggregate C/I &
communication.                                                            the FER ,should be given due importance in the handover
                                                                          decision ,along with RxLev and RxQual. The reason behind
           Continuation of an active call is one of the most              incorporating these parameters is explained ahead.
important quality measures in the cellular systems. Handover
process enables a cellular system to provide such a facility by               In the dedicated mode, TxPower consumes the battery
transferring an active call from one cell to another. Different
                                                                          power of the mobile handset .Normally the acceptable range
approaches have been proposed and applied in order to achieve
better handover service, by various researchers. The principal            of the TxPower is between 5 and 15, where 5 is the desired
parameters considered in the present work, which are used to              value. .Hence, TxPower has been proposed to be an important
evaluate handover techniques are: Received signal quality                 parameter in the handover decision process. The FER may
(RxQual), FER, Received signal level, MS-BS distance, transmit            increase in two cases (1) If the complete frame is lost or
power (TxPower) & aggregate C/I.                                          destroyed in transmission and (2) Frame could not be obtained
                                                                          because error correcting code is destroyed .Hence, FER is a
         In the present work, thorough analysis has been done for         considerable parameter in handover decision. Similarly the
the received signal strength difference threshold, along with             interference level received from all the interference sources in
other RF quality parameters. To ensure best performance to all            the system should be given due importance in the criterion list
mobile users at all times and all locations an active set of
                                                                          for the handover decision .The desired carrier level and the
parameters has been calculated for critical values along with
signal strength difference threshold.                                     interfering carrier level are calculated and measured in
                                                                          dBm.For convenience, we normally use the C/I ratio to
     Keywords: Received signal quality (RxQual), FER, Received            determine whether an interference case is acceptable or not.
signal level (RxLev on uplink and downlink), MS-BS distance,
transmit power (TxPower) & aggregate C/I.                                      Since in the real time cellular systems, handover failure
                                                                          may occur due to a number of practical issues, by introducing
                                                                          additional criteria for handover decision making, spurious
                                                                          handover can be avoided to a large extent. Conventional
                    I. INTRODUCTION                                       techniques suffer from inefficiencies caused by the fact that in
   Traditional handover algorithms are based on relative                  the practical scenario, particularly in city area, difference in
signal strength, relative signal strength with threshold, relative        signal level strength has proved to be an inferior criteria to
signal strength with hysteresis, relative signal strength with            decide handover issue .To overcome these limitations, the
hysteresis and threshold [1] [2]. Handover analysis uses fuzzy            authors has proposed an active set of parameters along with
logic based prediction techniques also [3] [4].Later an                   their optimum values which can be used to provide better
extensive study found that the received signal strength                   handover decision efficiency.
(RxLev) & the received signal quality (RxQual), are the prime
parameters in the handover decision. However ,considering

                                                                                                     ISSN 1947-5500
                                                             (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                       Vol. 9, No. 7, 2011

  The rest of the article is organized as follows. Section II                         III. HANDOVER ANALYSIS
reviews related works. Section III gives analysis of the
handover algorithm. In Section IV, results are discussed, as                   Handover initiation criteria analyzed in the present
obtained from the model. Finally, conclusions and future                 paper is based essentially on five variables: the received signal
perspectives are discussed in Section V.                                 level (RxLev), received signal quality (RxQual), FER,
                                                                         transmit power (Txpower) & aggregate C/I value. In order to
                                                                         study the effect of the above mentioned variables on the
               II. LITERATURE REVIEW                                     handover decision, extensive analysis of the GSM
                                                                         measurement data has been carried out .Out of total 21
    Several aspects of the analytical handover model have been           handovers, few handovers were like transit entry into cell, few
investigated in the previous works. An analytic model of                 were false handover triggers and only in 10 cases, the call
handover algorithm has been presented in [5, 6, 7] based on              continued for appreciable time i.e. the handover was stable
the level crossings of the difference between the signal                 .The data collected before and after handover were reliable
strengths received from two base stations in a log normal                and taken into consideration for detailed study.
fading environment. The basic model has the route of the
mobile chosen to be the straight line between two BSs. Two               A. Parameter Evaluation
important performance indicators of a handover algorithm are
the mean number of handoffs for this route and the delay in                     We start the analysis by studying the behavior of
handing off, both of which need to be minimized. The tradeoff            various performance metrics with respect to the distance
curve between these two conflicting indicators was drawn in              between serving base station & the mobile unit. The line of
order to determine the amount of hysteresis and averaging to             sight distance is calculated for a number of cells using the
be used in the algorithm.                                                distance Haversine formula as under

    In [5], the validity of the Poisson model has been                         Dist_Los= SQRT (POWER (F, 2) +POWER (K, 2)
demonstrated for the specific case where the signal strength is           Where,
stationary. The handover process was studied in terms of                  F=height of the BTS antenna in meter
certain level crossings of the difference between the received            K=non-line of sight distance in meter
signal strength from two BSs; the model works well where it
is most needed, in the range of optimal parameters. This work            Where,
has been extended in [6, 7] for the no stationary case, in which         K= (ACOS (SIN(C)*SIN (H) +COS(C)*COS (H)*COS (J-
the level crossings are modeled as Poisson process with time-            E)))*6371*1000
varying rate functions. Further, theoretical analysis using level
crossings is given in [8]. In [9], the model was applied to               Where,
obtain certain criteria for designing practical handoff                   C=latitude of BTS antenna
algorithm, especially for designing algorithms that are robust            E=longitude of BTS antenna
with respect to variations in the radio propagation                       H=latitude of mobile station
environment. This includes extensions of the model to take                J=longitude of mobile station
into account the absolute value of the signal strength from the           Radius of earth 6371 Km
current BS to avoid handoffs when the weaker signal is strong                Once the distance values are obtained, the plots between
enough, has been shown in [10].                                          this distance & the respective parameter is plotted.

     It has been observed that the analysis done in the previous             1) RxLev vs. Distance
work on handover ,has been validated by simulation results
only .None of the work has been done on the practical data to                 Ideal plot of the RxLevel verses distance ,will be the one
validate their findings .However in the present paper ,the               in which the RxLevel value should exhibit a downtrend with
effect of several performance metrics (RxLev,RxQual,FER,                 an increasing distance .One of the plots shown in figure.1
Aggregate C/I &TxPower ) on the handover decision has been               ,exhibits this behavior .As the distance between the mobile
validated by analyzing the measurement data ,as obtained                 station & the serving base station increases ,the received
from the drive test results in GSM network, from Katni town              signal level decreases .The entry to this cell occurred at a
of Madhya Pradesh state ,India.                                          distance about 320m, from 320 m to ~340m, the signal
                                                                         strength varies heavily from -60 dam to -77dbm which
                                                                         indicates that the recording are done at shade coupled with
                                                                         heavy transient reflections form neighboring moving vehicles .
                                                                         As the distance is increased, line of sight is available, signal
                                                                         strength was stable and there were gradual fall with distance
                                                                         .The points where the handover situation occurs, are identified
                                                                         by observing the sudden downtrend in the RxLev .Once these

                                                                                                    ISSN 1947-5500
                                                           (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                     Vol. 9, No. 7, 2011

points are known, the behavior of the RxLevel over these
points is recorded for observation.
                                                                           3)   FER vs. Distance

                                                                          In the cellular communication, not only the continuation of
                                                                       the call is necessary but also quality of speech is an essential
                                                                       parameter in analyzing the performance of handover
                                                                       algorithms. The Frame Error Rate (FER) measurement is used
                                                                       by the mobile to detect bad frames. The mobile starts the
                                                                       substitution and muting process, and within 300 ms of bad
                                                                       frame reception it completely mutes the speech. Out of 104,
                                                                       FER measurements are done over 100 frames, which
                                                                       correspond to ~2 s of speech. The variation of %FER with the
                                                                       distance between the serving BTS & the MS is as shown in
                                                                       figure 3.In contrast to observations in case of RxLevel and
                                                                       RxQual ,FER shows better performance at the near region and
                                                                       even at the far region with exception in the middle. This
                                                                       shows strong immunity of GSM system from frame errors.
                                                                       But, the overall trend was in upward direction indicating
                                                                       contribution of this parameter for handover decision.
                 Figure 1. RxLev vs. Distance

    2)   RxQual vs. Distance

   RxQual is a value between 0 and 7, where each value
corresponds to an estimated number of bit errors in a number
of bursts. Each RxQual value corresponds to the estimated bit-
error rate according, which varies from BER <0.2% for
RxQual 0, 0.8 %< BER<1.6% for RxQual 3 and BER >12.8%
for RxQual 7.
  The RxQual value showing an increase contributes to the
handover decision making .The variation of this parameter
with the distance ,one of which is shown as under in Figure.2
.At the entry into the cell, RxQual had wide variation which
shows presence of strong interference .Slowly this situation
improves as the vehicle goes slightly away where a dominant
part in line of sight signal .At the end ,few observations show                             Figure 3. FER vs. Distance
RxQual>4 which indicated requirement of handover .Overall
positive slope indicated healthy situation for handover
prediction .                                                           4) Aggregate C/I vs. Distance
                                                                           The aggregate carrier-to-Interference (C/I) ratio is the ratio,
                                                                       expressed in dB, between a desired carrier (C) and an
                                                                       interfering carrier (I) received by the same receiver. The
                                                                       variation of the aggregate C/I with the distance (Figure
                                                                       .4),better than 15dBm in most cases shows that it has only a
                                                                       minor effect on the handover decision, but considered here as it
                                                                       has positive slope.

                 Figure 2. RxQual vs. Distance

                                                                                                   ISSN 1947-5500
                                                             (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                       Vol. 9, No. 7, 2011

                                                                         value – initial value) are taken into account . The distance of
                                                                         the mobile unit from the serving station & the target station is
                                                                         also examined .This approach is adopted to identify the role of
                                                                         each parameter in the handover, more clearly & accurately.
                                                                         The handover cases which exhibit the near ideal situation are
                                                                         identified & are taken into consideration for deriving the
                                                                         optimized handover situation.


                                                                         The handover position is identified in the excel sheet of the
                                                                         drive test report ,by looking at the „event type‟ column .After
                                                                         selecting the location ,a set of about 50 observations ,before &
                                                                         after the handover event are considered for evaluating the
                                                                         average values of the before & after values of each parameter
                                                                         .A sample sheet is shown in APPENDIX to demonstrate method
                 Figure 4. Aggregate C/I vs. Distance                    of calculation has been done (sheet 1).The transition
                                                                         (handover) had taken place from cell id 509 to 619 at the
                                                                         position highlighted in the sheet .Before values pertains to
5) TxPower vs. Distance                                                  values (and average thereafter) before transition in old cell and
                                                                         after values after transition to new cell .This method is carried
   Transmit power plays a very important role in sustaining              out for each parameter. The calculations & the respective
higher battery life of the mobile handset .In the dedicated              sheets are obtained by performing the calculations in the
mode, TxPower is monitored constantly by the serving station             similar manner .The delta (∆) values of the five parameters are
.Normally the acceptable range of the TxPower is between 5               obtained by performing the subtraction of the final & initial
& 15, where 5 is the desired value..Higher TxPower is                    values respectively. The two other distance calculations (from
unacceptable not only because it consumes the battery power              serving cell to target cell & from serving cell to mobile
of the mobile, but also because it may adversely affect the              station) is performed by the distance haversine formula as
mobile user‟s health. The variation of the TxPower with the              mentioned in previous section.
distance is given in figure 5.
                                                                         C. Optimization

                                                                         The optimum situation for handover is identified by
                                                                         comparative analysis. The comparison of the parameter values
                                                                         at the time of handover is done with respect to the
                                                                         recommended range of values & the ideal values respectively
                                                                         .The tabulated form of the values obtained is given in sheet2.
                                                                         It has been found that the handover situation in the 5, 8 & 9 th
                                                                         cell case is exhibiting „near ideal‟ situation. These 3 cases are
                                                                         then scrutinized to obtain the optimum condition for handover.


                                                                         (a)RxLev and ∆RxLev: The 4.3dbm increase in the RxLev
                                                                           after the handover has taken place, averaged for all 3 cells,
                                                                           is a sure sign of a successful handover.
               Figure 5. Txpower vs. Distance                            (b)RxQual and ∆RxLev: The performance of the RxQual
                                                                            value ~=1.46 is most appropriate in the cell after
                                                                         (c)FER & ∆FER: The optimum performance of the FER is
 B. Handover Cell Analysis                                                 fulfilled by the cell of the serial number 8.

   Once the relationship between the distance & the respective           (d)TxPower: The cell of the serial number 5 is exhibiting the
parameters is drawn, the analysis of handover cells is studied             best case of the TxPower based decision criteria.
exclusively. The cells in which the handover has occurred are
first identified & then the parameter values are studied for             (e)Aggregate C/I: The aggregate C/I criteria is fulfilled by the
each cell separately .Not only the absolute values of the                  cell of the serial number 5.
various parameters is studied but also the relative values (final

                                                                                                    ISSN 1947-5500
                                                               (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                         Vol. 9, No. 7, 2011

We have validated the role of various parameters on the
handover decision making in this paper. It has been found that          [1] Gregory P. Pollioni, “Trends in Handover Design”, IEEE
the behavior of the respective parameters & the role of each            Communications Magazine, vol. 34, March 1996, pp. 82-90
chosen parameter on the handover decision making, is
satisfying the ideal cases to a close extent .Also the                  [2]    P. Marichamy, S. Chakrabati and S. L. Maskara,
optimization has contributed to obtain a set of values of the           “Overview of handoff schemes in cellular mobile networks
respective parameters, which serve as the best case to decide           and their comparative performance evaluation”, IEEE
the handover ,as given table1 .                                         VTC‟99, vol. 3, 1999, pp. 1486-1490

TABLE I RESULT                                                          [3] M. Chiu, M. Bassiouni, “Predictive Schemes for HandoN
                                                                        Prioritization in Cellular Networks Based an Mobile
                                                                        Positioning”, EEE Joumal on selected amas in
           RxQual         FER      Tx Power(dBm)                        communications, Vol. 18, No. 3, March 2000
(dBm)                                              Aggregate C/I
 -72         5                          15            15.82             [4] M.S.Dang, A Prakash, D.K. Anvekar, D. Kapoor,
∆RxLev                             ∆Tx             ∆Aggregate           R.Shorey, "Fuzzy Logic Based Handoff in Wireless
          ∆RxQual         ∆FER                                          Networks", in Proceedings of the 51st Vehicular Technology
(dBm)                              Power(dBm)         C/I
                         -0.3794                                        Conference (VTC 2000 Spring), Tokyo,
 4.3        -1.5                      -6.537          1.1066
                                                                        15-18 May 2000, pp.2375-2379

                                                                        [5] R. Vijayan and J. M. Holtzman, “The dynamic behavior of
                    V.   CONCLUSION                                     handoff algorithms,” in Proc. Ist Internat. Con$ UniversG 1
                                                                        Personal Commun., Dallas, TX, Sept. 1992.
   The aim of this investigation was first to define some
appropriate performance measures for inter-cell handovers.              [6] “Analysis of handoff algorithm using nonstationary signal
The obtained results showed the outperformance of handover              strength measurements,” in Proc. GLOBECOM ‟92, Orlando,
algorithm based on multiple parameters (i.e. RSS, BER etc). In          FL., Dec. 1992.
this paper, we have extended the model for analyzing the
performance of the handoff algorithm based on signal strength           [7]      “A model for analyzing handoff algorithms,” IEEE
measurements. This model enables us to achieve good                     Trans. Veh. Technol., vol. 42, no. 3, Aug. 1993.
analytical approximations easily and fast. Therefore, this
model can be used by the network designer to help optimize              [8] R. Vijayan and J. M. Holtzman, “Foundations for level
the behavior of the handover strategy by setting appropriate            crossing analysis of handoff algorithm,” in Proc. ICC ‟93,
hysteresis, absolute threshold, and other parameters such as            Geneia, Switzerland, May 1993.
the averaging length for different propagation environments.
Handover condition for at least 3 of the 5 parameters should            [9] R. Vijayan and J. M. Holtzman, “Sensitivity of handoff
be met to take handover decision while 4 conditions meeting             algorithms to variations in the Drooagation environment.” in
will be sufficient.                                                     Prcc 2nd Internat. Conf I s -Universal Personal Commun.,
       As a future course of work, more importance can be               Ottawa, Canada, Oct. 1993.
given to the QoS issues where in more number of radio and
network parameters are taken into consideration for averaging           [10] R. Beck, F. W. Ho, “Evaluation and performance of field
the threshold values. This ensures that a handover can be               strength relatedhandover strategies for micro-cellular systems,”
hastened or delayed as the situation requires and also prevent          in Proc. 3rd Nordic Sem. Digital Land Mobile Radio Commun.,
unnecessary handover that may take place due to momentary               Copenhagen, Denmark, 1988.
fading of any one of the parameter. Hastening the handover
ensures that a call is not dropped due to non availability of
resources. Handover delayed ensures that unnecessary
handover does not take place leading to loading of the base

                                                                                                    ISSN 1947-5500
                                                                 (IJCSIS) International Journal of Computer Science and Information Security,
                                                                                                                           Vol. 9, No. 7, 2011


                           Isha Thakur received her B.E. (Hons.)             From 1997 to 2002, Dr. Manna has worked as Deputy General
                           degree       in     Electronics      and          Manager in a Telecommunication Training Centre of DoT. He was
                           Telecommunication Engineering from                first to install live training node for Internet Service Provider (ISP),
                           Takshshila Institute of Engineering and           designed training schedules and prepared handbook and lab practice
                           Technology Jabalpur (M. P.) in 2008.              schedules. He had conducted training programs for 5 batches of
                           Currently, she is pursuing her M. E.              participants deputed by Asia Pacific Telecomm unity (APT) and 3
                           from the Department of Electronics and            more exclusive batches for Sri Lankan Telecom. He had also
                           Telecommunication          Engineering,           conducted several seminars with international experts through
                           Government Engineering College                    UNDP/ITU projects. In 2000, he had delivered distinguished speech
                           Jabalpur. Her research interest includes          on ADSL in a seminar organized by ITU. During 1995 and 1996, Dr.
                           Computer networks and Future                      Manna was posted in Telecommunication Engineering Centre (TEC)
generation in mobile communication System.                                   and developed Artificial Intelligence (AI) based software for E10B
                                                                             telephone exchanges named E10B Maintenance Advisor
                                                                             Dr. Manna had worked as Development Officer in WEBEL
                             Bhavana Jharia received her B.E.                (erstwhile PHILLIPS) Telecommunication Industries during 1983-
                             degree      in     Electronics      and         1984 after which he joined DoT and worked in different executive
                             Telecommunication Engineering from              capacities up to 1994.He was awarded National Scholarship in 1973
                             Govt. Engineering College Jabalpur (M.          based on school level examination and silver medal for performance
                             P.) in 1987. She did her M.E. (Solid            in college. He had both graduated and post graduated in Radio
                             State Electronics) from University of           Physics and Electronics Engineering from University of Calcutta and
                             Roorkee, Roorkee in 1998 and Ph.D.              undergone trainings at Beijing University of Post and Telecom China
                             (VLSI Technology) from I.I.T. Roorkee           in 1990 and DARTEC, Montreal, Canada in 1999.
                             in 2005. She joined the Department of
                             Electronics and Telecommunication
                             Engineering, Govt. Engineering College
Jabalpur (M. P.) as faculty in 1990, where at present she is working
as an Associate Professor. She has 25 publications in National,
International referred Journals and Conferences. Her research
interests are in Electronics Design and Simulation and Low Power
VLSI Technology. She is a member of IE (I), CSI, VLSI Society of
India, senior member of IACSIT and Life Member of ISTE.

                             Dr.     Gopal Chandra Manna is
                             working as Senior General Manager
                             (Head Quarters), Inspection Circle,
                             BSNL, a wholly owned Company
                             under           Department            of
                             Telecommunications (DoT), Govt. of
                             India. Dr. Manna has carried out
                             extensive research on coverage issues
                             of GSM, CDMA, WCDMA and
                             WiMAX radio access. Study of
                             Wireless Traffic and QoS estimation of
                             Cognitive Radio are his current areas of
research. In Addition, he has written several articles on advanced
telecommunications which has been published in national and
international journals and symposiums. Dr. Manna is regularly
invited as a panel expert, invited speaker, session chair etc. in
seminars and conferences.
Dr. Manna has developed and conducted one week course on Quality
of Service Monitoring at Information and Communication
Technologies Authority, Mauritius as International Expert through
Commonwealth Telecom Organisation London during August 2010.
He had also delivered a speech on WiMAX coverage Evaluation at
International Conference on Advanced Communications Technology
2011 at Seoul, Korea and chaired a session on Network Management.
He had also delivered speech on ADSL at International
Telecommunication Union seminar in 2000 at Bangalore, India.

                                                                                                           ISSN 1947-5500
              (IJCSIS) International Journal of Computer Science and Information Security,
                                                                        Vol. 9, No. 7, 2011

[1] SHEET 1


                                                   ISSN 1947-5500

To top