QoS ENHANCEMENT IN 4G HETEROGENEOUS NETWORKS USING KALMAN FILTER & EWMA

Transcription

1 International Journal of Electronics and Communication Engineering and Technology (IJECET) Volume 8, Issue 3, May-June 2017, pp , Article ID: IJECET_08_03_004 Available online at ISSN Print: and ISSN Online: IAEME Publication QoS ENHANCEMENT IN 4G HETEROGENEOUS NETWORKS USING KALMAN FILTER & EWMA Bibin Mathew, Winner George and Mildred Pereira Department of Electronics & Telecommunication Engineering St.John College of Engineering & Technology, Palghar, Maharashtra, India ABSTARCT With the advent of 4G technology interoperability between different wireless technologies can be easily achieved. In Vertical Handoff (VHO), connection transfer switching takes place from one Radio Access Network (RAN) to another. Now a day s numerous methods are available to improve the Quality of Service in heterogeneous networks. This project concerns about enhancement of heterogeneous networks using filtering technique methods. Two technologies, UMTS and WLAN is considered for this project. The filtering techniques includes calculation of path loss and then filtering received signal strength under various propagation environments such as urban, suburban and rural. COST 231 Hata, COST Walfish-Ikegami, ECC 33, Ericsson path loss models are considered for this project. In order to improve the throughput and vertical handoff performance, appropriate filtering methods such as Kalman filtering and Exponentially Weighted Moving Average (EWMA) methods are used. Both the Kalman and EWMA filtering techniques shows an corresponding improvement in RSS and throughput. Key words: 4G Heterogeneous Networks, QoS, Pathloss, Vertical Handoff, RSS, Kalman filter, EWMA. Cite this Article: Bibin Mathew, Winner George and Mildred Pereira, QoS Enhancement in 4G Heterogeneous Networks Using Kalman Filter & EWMA, International Journal of Electronics and Communication Engineering and Technology, 8(3), 2017, pp INTRODUCTION The technology of connection transfer switching from one Radio Access Network (RAN) to another RAN is known as vertical handoff whereas in horizontal handoff switching takes place between same networks. Almost all studies on vertical handoff are using Received Signal Strength (RSS) as the basic handoff decision indicator, in which handoff decisions are made by comparing the RSS with the preset threshold values. The RSS shows the average signal power between the transmitter and the receiver. In order to predict the basic parameter, 28 editor@iaeme.com

2 QoS Enhancement in 4G Heterogeneous Networks Using Kalman Filter & EWMA mean RSS, we can use different propagation models. Propagation models have traditionally focused on predicting the received signal strength at a given distance from the transmitter, as well as the variability of the signal strength in a close spatial proximity to a particular location. Propagation models that predict the signal strength for an arbitrary transmitterreceiver (T-R) separation distance are useful in estimating the radio coverage area of a transmitter. Propagation models are useful for predicting signal attenuation or path loss. Path loss is defined as unwanted reduction in the transmitted signal power. This path loss information may be used as a controlling factor for system performance or coverage so as to achieve perfect reception. We measure this path loss in different area like rural, urban, and suburban with the help of propagation path loss models. The filtering techniques like Kalman filter and EWMA are used to reduce the noise from the RSS computed from path loss models. The various path loss models considered for this study are Free Space Path Loss Model, COST 231 Hata Model, ECC-33 Model, Ericsson Model and COST 231 Walfish- Ikegami Model. The remainder of this paper is organized as follows: Related and existing works are discussed in section 2. Overview of proposed algorithm is presented in section 3. Section 4 comprises various modules of proposed algorithm. Performance assessment is carried out in section 5 and finally conclusion is given in section RELATED WORKS Path loss is the reduction in power of an electromagnetic wave as it propagates through space. It is a major component in analysis and design of link budget of a communication system. The concept of various path losses and its comparison for various environments are described in [2]-[5]. It also describes the operating frequencies and the maximum distance that each model supports. The [6] gives an idea about how the RSS prediction can be made from path loss measurements. This predicted signal strength at a point will vary according to various path loss models and various propagation environments. As the signal strength gets faded and goes below the RSS threshold value with respect to distance, then there is a need for handoff. The handoff can be done to same network if it is available or it can handoff to the other existing networks which is also known as vertical handoff. Various algorithms to avoid unnecessary handoff is specified in [7] For each network, there is a threshold value below which connection break with active station. Therefore the signal strength must be greater than threshold point to maintain the connection with serving network. The signal become weak as mobile moves far away from serving station and gets stronger signal towards new station as it move closer. There is a need for Handoff if RSS of active station decreases below threshold level to maintain the connection. The aim of this work includes comparison of various path loss models under different propagation environments, calculation of RSS from path loss models and attaining QoS enhancement in terms of RSS and throughput using Kalman filter and EWMA. All the three technologies do not support these applications with equal QoS. 3. PATHLOSS MODELS, RSS, FILTERING TECHNIQUES The various path loss models considered for this study are Free Space Path Loss Model, COST 231 Hata Model, ECC-33 Model, Ericsson Model and COST 231 Walfish- Ikegami Model editor@iaeme.com

3 Bibin Mathew, Winner George and Mildred Pereira 3.1. Free Space Path Loss Model (FSPL) Path loss in free space defines how much strength of the signal is lost during propagation from transmitter to receiver. FSPL is diverse on frequency and distance. The calculation is done by using the following equation PL FSPL = ( 20log10( d) ) + 20log10 ( f ) (3.1) Where, f: Frequency [MHz] d: Distance between transmitter and receiver [m] Power is usually expressed in decibels (dbm) COST 231 Hata Model The Hata model is introduced as a mathematical expression to mitigate the best fit of the graphical data provided by the classical Okumura model. Hata model is used for the frequency range of 150 MHz to 1500 MHz to predict the median path loss for the distance d from transmitter to receiver antenna up to 20 km, and transmitter antenna height is considered 30 m to 200 m and receiver antenna height is 1 m to 10 m. To predict the path loss in the frequency range 1500 MHz to 2000 MHz. COST 231 Hata model is initiated as an extension of Hata model. It is used to calculate path loss in three different environments like urban, suburban and rural (flat). This model provides simple and easy ways to calculate the path loss. Although our working frequency range (2 and 2.4 GHz) is outside of its measurement range, its simplicity and correction factors still allowed to predict the path loss in this higher frequency range. The basic path loss equation for this COST-231 Hata Model can be expressed as PL = log10 ( f ) log10 ( hb ) ahm + ( log10 ( ht )) log10 d + Cm (3.2) d: Distance between transmitter and receiver antenna [km] f: Frequency [MHz] h t : Transmitter antenna height [m] The parameter C m is defined as 0dB for suburban and rural environments and 3 db for urban environments. The parameter ah m is defined for urban environments as The value for ah m in suburban and rural (flat) areas is given as 2 ah 3.20 log for f (3.3) m h ( ( )) 400MHz = 10 r h r is the receiver antenna height in meter Hata-Okumura Extended Model or ECC-33 Model One of the most extensively used empirical propagation models is the Hata-Okumura model, which is d on the Okumura model. This model is a well-established model for the Ultra High Frequency (UHF) band. Recently, through the ITU-R Recommendation P.529, the International Telecommunication Union (ITU) encouraged this model for further extension up to 3.5 GHz. The original Okumura model doesn t provide any data greater than 3 GHz. Based on prior knowledge of Okumura model, an extrapolated method is applied to predict the model for higher frequency greater than 3 GHz. The tentatively proposed propagation model of Hata-Okumura model is referred to as ECC-33 model. A different approach was taken by the Electronic communication Committee (ECC) which extrapolated the original 30 editor@iaeme.com

4 QoS Enhancement in 4G Heterogeneous Networks Using Kalman Filter & EWMA measurements by Okumura and modified its assumptions. The path loss equation for ECC-33 model is defined as PL = A (3.4) fs + Abm Gt - Gr Where, A fs : Free space attenuation [db] A bm : Basic median path loss [db] G t : Transmitter antenna height gain factor G r : Receiver antenna height gain factor These factors can be separately described and given by as A (3.5) fs = log 10 ( d ) + 20 log 10 ( f ) (3.6) A [ ] 2 bm = log10 ( d) log10 ( f ) log10 ( f ) h G [ [ ] ] 2 (3.7) t = log b log10 ( ) d When dealing with gain for medium cities, the G r will be expressed in [ ][ ] Gr = log10 ( f ) log10 ( hr ) (3.8) For large city, the G r will be expressed in Where, G r = 0.759h r (3.9) d: Distance between transmitter and receiver antenna [km] f: Frequency [GHz] ht: Transmitter antenna height [m] hr: Receiver antenna height [m] This model is the hierarchy of Okumura-Hata model and is defined for urban and suburban areas. This model is not suitable for rural environment COST 231 Walfish-Ikegami (W-I) Model This model is a combination of J. Walfish and F. Ikegami model. The COST 231 project further developed this model. Now it is known as a COST 231 Walfish-Ikegami (W-I) model. This model is most suitable for flat suburban and urban areas that have uniform building height. Among other models like the Hata model, COST 231 W-I model gives a more precise path loss. This is as a result of the additional parameters introduced which characterized the different environments. It distinguishes different terrain with different proposed parameters. Cost 231-WI takes the characteristics of the city structure into account : Heights of buildings h Roof Widths of roads w Building separation b Road orientation with respect to the direct radio path ϕ The equation of the proposed model is expressed in For LOS condition as, PL LOS = log10 ( d) + 20log10( f ) For NLOS condition as, PL NLOS = LFSL + Lrts + Lmsd for urban and { } suburban (3.10) (3.11) 31 editor@iaeme.com

5 Bibin Mathew, Winner George and Mildred Pereira L FSL = Free space loss L rts = Roof top to street diffraction L msd = Multi-screen diffraction loss Free space loss is given by L FSL = log ( d ) + 20 log ( ) (3.12) f log10 ( w) + 10log10( f ) + 20log10 ( H mobile ) + Lori L rts = hroof > hmobile (3.13) ϕ for 0 ϕ 35 L ori = ( ϕ 35) for 35 ϕ 55 (3.14) ( ϕ 55) for 55 ϕ 90 The multi-screen diffraction loss is given by Lbsh + ka + kd log10 ( d) + k f log10 ( f ) 9log10 ( f ) 9log10 ( B) for Lmsd > 0 L msd = 0 for Lmsd < 0 (3.15) 18log10 (1 + h ) for h > hroof Lbsh = 0 for h hroof 54 for h > hroof ka = h for d 0.5km and h hroof h ( d ) for d < 0.5km and h hroof 0.5 k a 54 = h h h h mobile ( d ) 0.5 = h = h roof for h > h roof for d 0.5km and h for d < 0.5km and h h h roof mobile h h roof roof (3.16) k d 18 = h h roof for h for h > h h roof roof (3.17) f 1 for medium sized city 925 k f = f 1 for metropolitan center 925 (3.18) d : Distance between transmitter and receiver antenna [m] f : Frequency [GHz] B : Building to building distance [m] 32 editor@iaeme.com

6 QoS Enhancement in 4G Heterogeneous Networks Using Kalman Filter & EWMA w : Street width [m] ϕ : Street orientation angel w.r.t. direct radio path [degree] The following data is used for this simulation purpose, i.e. building to building distance 50 m, street width 25 m, street orientation angel 30 degree in urban area and 40 degree in suburban area and average building height 15 m, station height 30 m Ericsson Model To predict the path loss, the network planning engineers are used a software provided by Ericsson company is called Ericsson model. This model also standss on the modified Okumura-Hata model to allow room for changing in parameters according to the propagation environment. Path loss according to this model is given by: In eqn(3.19), g(f) is defined by, g( f ) = 44.9 log10 ( f ) 4.78( log ( )) 2 10 f (3.19) 2) PL = a0 + a1 log10 ( d) + a2 log10 ( h ) + a3 log10 ( h ).log10 ( d) 3.2(log10 (11.75 h r ) + g( f ) t b (3.20) f : Frequency [MHz] h t : Transmission antennaa height [m] h r : Receiver antenna height [m] The default values of thesee coefficient parameters (a0, a1, a2 and a3) for different terrain are given in Table 3.1. These values are obtained by least square methods. Table 3.1 Values of coefficients for Ericsson model 3.6. RSS Measurements In order to add channel fading characteristics, log-distance model is used to find out the RSS. The Instantaneous Received Signal Strength (IRSS) at a distance d from the AP/BS is given by IRSS( d) = Pt P0 10n lo ( d ) + S( d) + R( d) (3.21) og 10 P t = Transmitted power in db P 0 = Path loss in db n = Path loss exponent S(d) = Large scale fading factor R(d) = Small scale fading factor For the large scale fading a correlated log-normally distributed random variable is used and is given by 33 editor@iaeme.com

7 Bibin Mathew, Winner George and Mildred Pereira S( d) = N(0, σ ) + exp( vt ) * S( d 1) (3.22) d 0 where σ is the standard deviation of the normal distribution and is given by = 1 exp( 2vT d σ (3.23) ( ) s σ 0 σ s is the standard deviation of log-normal fading V=velocity in m/s T=sampling period in seconds d 0 =reference distance in meter An Anti-log Rayleigh Distribution (ALR) is used to simulate the small scale fading effect and is given by 10γ R( d) = R 10 log( lnu ) (3.24) m + + ln(10) R m = Mean of ALR distribution γ = is Euler s Gamma constant U = random variable uniformly distributed between 0 and Kalman Filter The Kalman filter is a recursive predictive filter that is d on the use of state space techniques and recursive algorithms. It estimates the state of the dynamic system. the Kalman filter minimizes the mean square error of the estimated parameters. It is called a recursive filter because it doesn t need to store all previous measurements and reprocess all data each time step. The Kalman filter equations are given by Table 3.2 Kalman filter time update ( predict ) equations. ) ) X ( k k 1) = σ X ( k 1 k 1) (3.25) P( k 2 k 1) = σ P( k 1 k 1) + σ s (3.26) Table 3.3 Kalman filter measurement update ( correct ) equations. K( k) = P( k k 1) /( P( k k 1) + σ RD (3.27) ) ) ) X ( k / k) = X ( k k 1) + K( k)[ IRSS( k) X ( k k 1)] P ( k / k) = (1 K( k)) P( k k 1) (3.29) (3.38) Here, X ) ( k k 1) is the estimated value of X(k) d on IRSS values. P ( k / k 1) is the one step minimum prediction at state k and P ( k / k) is the MMSE at state k. K (k) is the Kalman gain. σ RD represents the variance of the multipath process editor@iaeme.com

8 QoS Enhancement in 4G Heterogeneous Networks Using Kalman Filter & EWMA σ is the variance of the zero mean Gaussian noise of the Anti Rayleigh AR(1) model of the shadow process. σ s is the standard deviation of the fading process Exponentially Weighted Moving Average (EWMA) EWMA is one of the simplest filters for reducing noise. It is the most simplest and efficient filter we can implement. The name comes from the concept that the filter s output, Y is the weighted average of current input and each previous input, with weighting decreasing exponentially. A filter that places more emphasis on the most recent data would be more useful. Such a filter can be designed by following the importance to more recent data by discounting older data in an exponential manner. The EWMA which is a low pass filter can be applied to smooth out the RSS. ( ) 1 EWMAi = ( 1 α) EWMAi 1 + α RSSi 0 < α < (3.30) Where, EWMA i = Current estimated EWMA value. EWMA i-1 = Prior estimated EWMA value. RSS i = Current measured RSS value. α = Smooth factor which can filter the noise of RSS and can keep the EWMA stable. A Kalman Filter (KF) also known as Linear Quadratic Estimation (LQE), is an algorithm which uses a series of measurements observed over time, containing noise and other inaccuracies and produces estimates of unknown variables that tend to be more precise than those would be d on a single measurement alone. Using prediction and correction, KF operates recursively on streams of noisy input data to produce a statistically optimal estimate of underlying system noise. So it can be used to reduce the mean square error from the given measured RSS values. EWMA is one of the simplest filters for reducing noise. It is the most simplest and efficient filter we can implement. The name comes from the concept that the filter s output is the weighted average of current input and each previous input, with weighting decreasing exponentially. A filter that places more emphasis on the most recent data would be more useful. Such a filter can be designed by following the procedure used in developing the moving average filter. Thus the Quality of Service (QoS) in terms of RSS and throughput can be improved by using these two filtering methods. 4. OVERVIEW OF PROPOSED ALGORITHM Due to the heterogeneity and the diversity of access networks, various user applications with different QoS requirements pose new challenges in designing optimal network selection algorithm for guaranteeing seamless QoS support to the users. Thus, VHO is necessary to provide uninterrupted services to mobile users anywhere and anytime in 4G Networks editor@iaeme.com

9 Bibin Mathew, Winner George and Mildred Pereira Figure 4.1 Block Diagram of Proposed Algorithm The proposed block diagram is given in fig.4.1. The block diagram shows the QoS enhancement from UMTS to WLAN. This is simulated for all the three environments by using different propagation models specified in the introduction section. From all the three environments, the propagation model which performs best is selected and its RSS is computed. It will then goes to the network condition check stage and will stay in the same network if the current RSS is less than the RSS threshold value otherwise it goes to the dwell timer block. Dwell timer will calculate the time over which the call is maintained in the cell without handoff. After the specified time if still RSS is larger than the RSS threshold value it moves to the filtering stage which consists of either Kalman filter or EWMA filter otherwise the user will stay on the same network. The filtering techniques such as Kalman and EWMA improves the QoS of the signal in terms of RSS and throughput. The final condition is checked after the filtering stage which allows the user to stay on the same network 1 if the RSS is below the threshold level otherwise vertical handoff will occur to network 2 which is WLAN. 5. RESULTS AND DISCUSSION The path loss models considered for this study are Free Space Path Loss Model, COST 231 Hata Model, ECC-33 Model, Ericsson Model and COST 231 Walfish- Ikegami Model. For simulation purpose we have considered that UMTS is having a radius of 2 km and WLAN is having 150 meters. The operating frequencies are fixed at 2Ghz for UMTS and 2.4Ghz for WLAN. Transmitter antenna height is 30 m in urban and suburban area and 20 m in rural area where as receiver antenna height is taken as 3m for all the three propagation environments. The average building height is considered to be 15m, building to building distance as 50m and street width as 25m. The FSPL is used as a reference model for comparison purposes. The comparison of these models under various propagation environments for a 2 km distance is given in Figure editor@iaeme.com

10 QoS Enhancement in 4G Heterogeneous Networks Using Kalman Filter & EWMA Figure 5.1 Path loss comparison for urban environment Figure 5.2 Path loss comparison for suburban environment Figure 5.3 Path loss comparison for rural environment 37

11 Bibin Mathew, Winner George and Mildred Pereira From the figure it is clear that, for urban environment, Ericsson model show the least path loss whereas, for sub urban and rural areas COST-231 Walfish Ikegami model performs the best. The RSS computation graphs are shown below. Figure 5.4 RSS vs Distance using Ericsson model for urban environment Figure 5.5 RSS vs Distance using Wlafish-Ikegami model for suburban environment

12 QoS Enhancement in 4G Heterogeneous Networks Using Kalman Filter & EWMA Figure 5.6 RSS vs Distance using cost -231 Walfish Ikegami model for rural environment. Table 5.1 Network Unavailable Distance Values Environment Path loss models COST-231 W-I normal - COST-231 W-I KF - COST-231 W-I - EWMA Ericsson normal Urban Signal drops below threshold - at approximately 550m away suburban Signal drops below threshold at approximately 700m away from BS. Handoff to WLAN at 70m away from AP.. Total unusable distance is 1230 meters Signal drops below threshold at approximately 1100m away from BS.Handoff to WLAN at 175 m away from AP. Total unusable distance is 725 meters Signal drops below threshold at approximately 900m away from BS. Handoff to WLAN at 150 m away from AP. Total unusable distance is 950 m Rural Signal drops below threshold at approximately 1500m away from BS. Handoff to WLAN at 40m away from AP. Total unusable distance is 460meters. Signal strength is not falling below the threshold value. WLAN available at a distance of 100m away from AP. Total unusable distance is 450 meters editor@iaeme.com

13 Bibin Mathew, Winner George and Mildred Pereira Ericsson-KF Ericsson-EWMA from BS.Handoff to WLAN at 50 m away from AP. Total unusable distance is 1400 meters Signal drops below threshold at approximately 1200m away from BS.Handoff to WLAN at 100 m away from AP. Total unusable distance is 700 meters Signal drops below threshold at approximately 600m away from BS. Handoff to WLAN at 150 m away from AP. Total unusable distance is 1250 meters For the simulation purposes, the threshold value of RSS for UMTS and WLAN is fixed as -110 dbm and -90 dbm respectively. It is also assume that the user is travelling away from UMTS BS to WLAN AP which is present at the UMTS cell boundary. The RSS is plotted for all the selected path loss models which are Ericsson in urban and COST 231 W-I model for suburban and rural, are shown in Figures (5.4)-(5.6) (a). The corresponding RSS filtering using Kalman and EWMA filters are shown in (5.4)-(5.6) (b). The Figure (5.4) shows RSS computation and filtering of path loss models under urban environment. In COST 231 W-I model, due to large scaling and small scaling fading effects the signal strength falls below the UMTS threshold value at 700 meters away from BS. After that the connection to WLAN takes place at 70m away from AP. Thus a total of 1230m unusable distance is there. From the table 5.1, it is well clear that Ericsson shows minimum unusable distance for Urban areas whereas COST-231 Walfish-Ikegami model shows the minimum network unavailable distance in the suburban and rural propagation environments. The Kalman filter and EWMA further reduces this distance to larger extent thereby improving the connectivity and QoS. The Figures (5.7)-(5.9) shows the graph between distance and throughput for different path loss models under various propagation environments. The Figures (5.7)-(5.9)(a) shows throughput of normal RSS whereas, Figures (5.7)-(5.9)(b) shows throughput of RSS filtered by Kalman filter and EWMA. This graphs are purely a numerical one that utilizes the throughput formula given by Shannon s capacity theorem. Throughput is defined as the average rate of successful message delivery over a communication channel. Throughput = B log 2 (1+(S/N) (5.1) B = Bandwidth S= Signal power N= Noise power 40 editor@iaeme.com

14 QoS Enhancement in 4G Heterogeneous Networks Using Kalman Filter & EWMA Figure 5.7 Ericsson model: Distance vs Throughput plot for urban environment Figure 5.8 COST-231 W-I model: Distance vs Throughput plot for suburban environment 41

15 Bibin Mathew, Winner George and Mildred Pereira Figure 5.9 COST-231 W-I model: Distance vs Throughput plot for rural environment From the Figures (5.7)-(5.9), it is clear that the throughput is improved by using Kalman filter and EWMA. Both the techniques performs better than the normal RSS computed from path loss models. The throughput is calculated from Shannon s channel capacity theorem. Nearer to stations and access points, the EWMA gives a higher throughput than the Kalman filter. That means it gives higher quality signal strength at a distance closer to BS and AP s. As the distance increases, KF outperforms EWMA with its higher signal strength and throughput making it useful for large distance propagations. Thus the QoS is very much improved by using these filtering techniques there by reducing the network unavailable distance. 6. CONCLUSION This paper proposes a QoS enhancement d network selection for 4G wireless networks. The QoS enhancement is achieved in terms of RSS and throughput. The best pathloss models in each environment is selected and RSS is computed under various propagation environments. The signal strength is further enhanced by using filtering techniques such as Kalman filter and EWMA. The simulation results clearly shows that as the distance between two networks increases the QoS enhancement in terms of RSS and throughput can also be achieved. REFERENCES [1] E.Gustaffson and A..Jonsson, Always Best Connected, IEEE wireless Communication, vol.10, no.1, pp.49-55,feb.2003 [2] Imranullah Khan, Tan Chon Eng, Shakeel Ahmed Kamboh Performance Analysis of Various Path Losss Models for Wireless Network in Different Environments, International Journal of Engineering and Advanced Technology (IJEAT) Volume-2, Issue-1, ISSN:pp , October 2012 [3] Noman Shabbir, Muhammad T. Sadiq, Hasnain Kashif and Rizwan Ullah Comparison of Radio Propagation Models for Long Term Evolution (LTE) Network, International Journal of Next-Generation Networks (IJNGN) Vol.3, No.3, September editor@iaeme.com

16 QoS Enhancement in 4G Heterogeneous Networks Using Kalman Filter & EWMA [4] Purnima K Sharma Comparative Study of Path Loss Models Depends on Various Parameters, International Journal of Engineering Science and Technology (IJEST) Vol. 3 No. 6 June [5] Moses Ekpenyong, Joseph Isabona, Etim Ekong, On Propagation Path Loss Models For 3-G Based Wireless Networks: A Comparative Analysis, Georgian Electronic Scientific Journal: Computer Science and Telecommunications Vol. No.2, 2010 [6] K.Ayyappan and P.Dananjayan, Propagation Model For Highway In Mobile Communication System, Ubiquitous Computing and Communication Journal,Vol.3 N0. 4, pp 61-66, 2008 [7] S.v.surwase and S.s.sambare. Article: Practical Analysis Vertical Handover Decision (VHD) Algorithm for WIMAX and WLAN. International Journal of Computer Applications 108(16):21-25, December [8] Rachel Kleinbauer, Kalman filtering implementation with matlab, study report in the field of study Geodesy and Geoinformatics, University Stuttgart, Helsinki November, 2004 [9] C.Amali,Bibin Mathew, B. Ramachandran, Intelligent Network Selection Using Fuzzy Logic for 4G Wireless Networks, International Journal of Electronics & Telecommunication Engineering (IJECET),Volume 4, Issue 2, March April, 2013, pp [10] Settapong Malisuwan and Wassana Kaewphanuekrungsi. A Review of Spectrum Auction In 4G Lte 1800 Mhz: The First Transition of Telecommunications Industry from Concessions to Licensing Regime in Thailand, International Journal of Advanced Research in Management, 6 (3), 2015, pp editor@iaeme.com