Fuzzy Logic Based Traffic Balancing in a GSM Network

Transcription

1 Open Access Journal Journal of Research in Engineering 1 (2) 2014, Journal homepage: Fuzzy Logic Based Traffic Balancing in a GSM Network Solomon T. Girma 1*, Dominic B. O. Konditi 2 and Edward N. Ndungu 3 1 Department of Telecommunication Engineering, Pan African University Institute of Science, Technology and Innovations, AICAD, JKUAT, or (Ethiopia) 2 Department of Electrical and Electronics Engineering, Multimedia University of Kenya, Nairobi, Kenya 3 Department of Telecommunication and Information Systems, Jomo Kenyatta University of Agriculture and Technology, Juja, Kenya * Corresponding Author - Solomon.tshm@gmail.com Abstract It is commonly accepted that the problem of network congestion control remains a critical issue and a high priority, especially given the growing size, demand, and speed of the increasingly integrated service networks. This is where the need for an intelligent multi criteria handoff algorithm becomes apparent. Efficient load balancing algorithm is important in serving more mobile stations in the wireless networks. This paper presents the design and implementation of a fuzzy multi-criteria handoff algorithm based on signal strength, path-loss and traffic load of base stations and the received signal to interference ratio as to balance traffic in all the neighboring sites at any time. This can be achieved by using Fuzzy Logic. The proposed algorithm can balance load of the base station by handing off some ongoing calls on cell edge in highly loaded cells to migrate to overlapping under-loaded cells, such that the coverage area of the loaded BTS virtually shrunk towards cell center of a loaded sector. In case of low load scenarios, the coverage area of a BTS is presumed to be virtually widened to cover up to the partial serving area of neighboring BTS. This helps a highly loaded neighboring BTS. Keywords Congestion, Fuzzy Logic, Handoff, Path loss, Received Signal Strength, Interference 63

2 S. Girma et al., Fuzzy Logic Based Traffic Balancing in a GSM Network 1. Introduction When a Mobile Station (MS) wants to communicate with another MS or Base Transceiver Station ( BTS), MS must first obtain Traffic Channel (TCH) from one of the BTS that hears it best. If there is a TCH available, it is given to the requesting MS. In case all the TCH are occupied, the new call is blocked. This kind of blocking is called new call blocking. The user releases the TCH under either of the following scenarios [1]: (a) the user completes the call; (b) the user moves to another cell before the call is completed. The process of moving from one cell to another, while a call is ongoing, is called handoff. During this handoff process, the mobile user requires that the BTS in the new cell assign to it a TCH [1]. If no TCH is available in the new BTS, the handoff call is blocked. This kind of blocking is called handoff blocking and it refers to blocking of ongoing calls due to the mobility of the users which leads a call to terminate prematurely due to dropout [1]. more efficient across the whole cluster network [3].There are a number of methods to balance traffic load among BTS [4]. Traffic load of a BTS can be shared among the neighboring BTS either by physical optimization or network parameter optimization Physical Optimization Antenna Tilting One of the most important physical optimization methods is based on antenna down tilting [5]. Antenna tilt is defined as the angle between the main beam of antenna and horizontal plane [5]. It is measured in degree and can have positive and negative values. The positive value is referred to as down-tilting and negative value means as up-tilting. A tilt value of 0' shows that the direction of the main beam is parallel to the ground plane. Antenna tilt can be achieved either mechanically or electrically. The mechanical tilt is achieved by manually tilting an antenna for the required orientation. Whereas, the electrical tilt is done by changing the characteristics of signal phases on each element of the array. The electrical down tilt provide a smooth reduction in the coverage when compared to mechanical down tilt [2]. Fig. 1. New call and handoff call Call blocking rate is one of the key performance indicators (KPI) which influence the network performance and customer satisfaction in cellular networks. The degree of TCH congestion in the network results in large number of TCH blocking which greatly affects negatively the subscriber satisfaction and revenue of a mobile service provider [2]. The available congestion relief methodologies such as cell splitting, aggressive frequency re-use pattern, microcells and expanding frequency band depend on time and cost factors for implementation [2]. The additional investment on capacity methods would be beneficial only if there is congestion on the whole network and proportional increase in traffic throughout the year. In the following section we will discuss load balancing methodologies used in cellular networks, without additional investment on capacity, to relief the problem of blocking rate Existing Blocking Relief Methodologies in Cellular Systems Load balancing can be used to explain any mechanism whereby over-loaded BTS distribute some of its traffic to its less-loaded neighbors to make the radio resource Fig.2. Load balancing using antenna tilting With the down tilt, one can direct the antenna radiation further down to the ground. The down tilt is advisable when one wishes to decrease interference and coverage in some specific areas, each BTS to meet only its designated area. When selecting the optimum tilt angle, the goal is to have as high signal strength as possible in the area where the BTS should be serving [5]. Beyond this serving area of the BTS, the signal strength should be as low as possible. Therefore, down tilting effectively reduces the coverage area of the BTS and disallows distant MS to access the BTS. A too aggressive down tilting strategy however leads to an overall loss of coverage and creates coverage holes which eventually lead to call drop [5]. 64

3 Journal of Sustainable Research in Engineering Vol. 1 (2), Antenna Height The antenna height is fundamental to BTS coverage area. If the antenna height is increased, path-loss will reduce. The relation between antenna height and coverage area is stated on two-ray models, Hata, Okumura and COST 231 [6]. If antenna height is doubled, then the coverage will be increased by 6db in two ray model [6]. To share load on a congested cluster, one can apply either of the following methods; (a) lowering antenna height of over-congested BTS, (b) raising antenna height of under-congested neighboring BTSs. This method is at risk of creating coverage hole or interference in the whole cluster Network Parameter Optimization Transmitter Power The most efficient method of varying the coverage area of a BTS is to adjust the BTS s transmitter power. A minimal transmit power effectively disallows more distant mobiles to access the BTS, thereby decreasing the coverage area and prohibits distant MS to access the BTS [7]. Hence, reducing BTS coverage area runs at the danger of creating coverage holes. Having the coverage holes on the cellular system adversely affect the performance of cellular system leading to call drop. Conversely maximum transmitter power allows distant MS to access the BTS, at the cost of producing interference on those BTS using the same frequency [7] CBQ and RXLEV_ACCESS_MIN The parameter such Cell Bar Quality (CBQ) and receiving access minimum (RXLEV_ACCESSMIN) can be adjusted to control TCH congestion [2]. The CBQ is represented as a character, with value of YES or NO. Default a value is NO. If CBQ of a BTS is equal to NO, the BTS selection priority is normal. If CBQ is equal to YES, the selection priority is low. By setting the value of CBQ as YES in over-loaded BTS and NO in under-loaded BTS, traffic loading is balanced in the wireless cluster. RXLEV_ACCESS_MIN is a threshold that allows MS to select a BTS only if the receiving signal level of the BTS is greater than a certain specified value [2]. The RXLEV_ACCESS_MIN is an integer number within the range of -47 to -110dbm. Its default value should be close to the reception sensitivity of the MS (for example this parameter is -102dbm in Ethio-Telecom for GSM 900MHZ and-100dbm in GSM 1800MHZ). The increase in RXLEV_ACCESS_MIN decreases its logical coverage area of BTS. The traffic in the network is achieved by adjusting the logical coverage area of over-loaded and under-loaded BTS. RXLEV_ACCESS_MIN cannot be set beyond certain value else result it will in a blind area on the boarder of BTS Half Rate The traffic channel in GSM network can carry user speech either in full-rate (TCH/F-13kbits/s) or in halfrate (TCH/H-5.6kbits/s) channel mode [2]. When TCH/H is in use, one time slot is shared by two connections thus doubling the capacity of the BTS. However the use of half-rate has associated degradation in speech quality and hence it is not recommended to solve the issue of congestion in mature mobile service providers Timing Advance Timing advance (TA) is a parameter used to indicate how far the MS is from the BTS. The actual measure is time-how much time it takes for the signal to travel the distance from the BTS to MS. The speed of the Radio Frequency (RF) is equal to speed of light, and in this way the distance can be calculated. The maximum radius of a normal GSM is 35km and this radius is divided into 64 equal TA steps [8]. Each step is therefore approximately 550m. For example, the TA value for a mobile that is 1500m from the BTS is 2. This parameter can be used to limit the maximum coverage, for example if the TA value is set to 2, only MS within the region of 1500m can be served. Hence from the network side if lower TA is set for over-congested BTSs and higher TA for under-congested BTS, traffic can be balanced in the cluster Handoff Process An MS in the presence of overlapping cellular coverage can connect to any BTS. MS in a wireless network, switches its current Point of Attachment (PoA) to a new wireless network using a process called Handoff [9].To have global connectivity, Handoffs are extremely important in cellular communication because of the cellular architecture employed to maximize spectrum utilization. When a mobile terminal moves away from a base station, the signal level drops and there is a need to switch the communications channel to another base station. That time there is a need for a handoff to be executed. Handoff is the process of changing the communications channel associated with the current ongoing connection while a call is in progress [10]. Many metrics have been used to support handoff decisions, including received signal strength (RSS), signal to noise ratio (SIR), path-loss, and distance between the MS and BTS, traffic load, mobile velocity and among others. The single criteria handoff decision compares one of the metric from the serving BTS with 65

4 S. Girma et al., Fuzzy Logic Based Traffic Balancing in a GSM Network that from one of the neighboring BTSs, using a constant handoff threshold value. The selection of the threshold is important to handoff performance. If the threshold is too small, many unnecessary handoffs may be take place. On the contrary, the quality of service (QoS) could be low and calls could be dropped if the threshold is too large. Hence by varying the constant threshold value, it is possible to force mobile to force handoff neighboring BTS or attach on serving BTS depending on the load of the BTS. Step to balance uneven load distribution in cluster using handoff process i. Find top over-congested BTS. ii. Check the status of the neighbours of congested BTS and identify only undercongested BTS iii. Check the number of outgoing handoffs from the congested BTS to the neighbouring BTSs. iv. Identify a higher type of handoff undertaking (due to RSS, Due to path-loss, due to interference, due to TA, due to mobility, etc). v. Introduce handoff the threshold for type of handoff in step 4. Therefore, by adjusting the handoff regions between neighboring BTS, It is possible to cause cell edge users in over-loaded BTS to migrate to lessloaded neighboring BTS. Such an approach is referred to as mobility load balancing, thereby increasing the efficiency of resource utilization [11] Drawbacks of Blocking Relief Methodologies All the above techniques of load balancing are not selfadaptive in the sense that every time congestion occurred in the network either of the above methods has to be manually applied. Once congestion is relieved, those parameters still take effect which in turn affects customer satisfaction. Therefore, the problem of network congestion control remains a critical issue and a high priority, mainly given the growing size, demand, and speed of the networks. Network congestion is becoming a real threat to the growth of existing real time networks (circuit switching). It is a problem that cannot be ignored. In an effort to address the above challenges, this paper presents a design and implementation of a fuzzy multicriteria handoff algorithm based on signal strength, path-loss, traffic load of BTS and signal to noise ratio (SNR). The algorithm balances traffic in all the neighboring BTSs at any time and enhances the 66 performance of the cellular system by selecting the best network segment. This can be achieved by using Fuzzy Logic. A multi-criteria handoff algorithm can provide better performance than a single criterion handoff algorithm due to the extra number of evaluation parameters and the greater potential for achieving the desired balance among different system characteristics. This multicriteria nature of the algorithm allows simultaneous consideration of several significant aspects of the handoff procedure in order to enhance the system performance [12]. This study is organized in three sections, the first section is a design of the algorithm using a fuzzy logic, the second section is simulation of the algorithm using data from two-ray propagation model on MATLAB (R2010b) and the third section is results and conclusions. 2. Proposed Multi-criteria Handoff Algorithm Based on Fuzzy Logic Fuzzy logic reasoning scheme is applied for mapping of non-linear data set to scalar output figure. When the problems are with doubt and ambiguity, to anticipate the correct value among all uncertainties, then fuzzy logic is selected. Fuzzy logic develops computational techniques that can carry out reasoning and problem solving tasks that requires human intelligence [12]. Thus, fuzzy logic is used to choose the optimal serving base station amongst given neighboring base stations for handoff decision founded on the multiple parameters as crisp inputs and give the best possible answer to choose the best base station. There are four input parameters considered in this study: Received Signal Strength (RSS), Path-loss, Signal to Noise Ratio (SNR) and Traffic Load of the BTS. The only output parameter of the fuzzy inference system is handoff decision. In the proposed algorithm the range of received signal is taken to be -50dbm to - 100dbm, the path loss varies from 20db to 80db, the SNR ranges from the lowest 10db to highest 30db. The range of the base station traffic load range from 0 to 100% Fuzzification In the first step of the handoff process, the model would all collect the four input parameters and feed into a fuzzifier. The fuzzifier transforms real time measurements into fuzzy sets. In order to improve the reliability and robustness of the system, Gaussian

5 Journal of Sustainable Research in Engineering Vol. 1 (2), 2014 membership functions (MFs) are used as an alternative process is necessary or not. to triangular MFS. For instance, if RSS is considered in crisp set, it can only be weak or strong. It cannot be both at a time. However, in a fuzzy set the signal can be considered as weak signal and medium at same time with graded membership. The membership values are obtained by mapping the values obtained for particular parameter into a membership function Fuzzy Inference The second step of the handoff process involves feeding the fuzzy sets into an inference engine, where a set of fuzzy IF-THEN rules are applied to obtain fuzzy decision sets. These sets are mapped to the corresponding Gaussian membership functions. Since there are four fuzzy inputs and each of them has three subsets so there are 34=81 rules. Fuzzy rules can be defined as a set of possible scenarios. For simple understanding, the set (No Handoff, Wait, Be Careful, and Handoff) are used to represent the fuzzy set of output handoff decision, the range of the decision matrix is from 0 to 1, where 0 is no handoff and 1 is exactly handoff Defuzzification Finally, the output fuzzy decision sets are aggregated into a single fuzzy set and passed to the defuzzifier to be converted into a precise quantity during the last stage of the handoff decision. The centroid of area method is elected to defuzzify for changing the fuzzy value into the crisp set [12]. Fig. 2. Membership of RSS Fig. 3. Membership of the path-loss 3. Simulation Using MATLAB For system simulation, Mam-dani Fuzzy Inference system is proposed due to the fact that it is well suited for human input and the non-linear nature of wireless. Fuzzy inference gathers the values of all input parameters and then evaluates them according to the fuzzy interference rules base. The composed and aggregated output of rules evaluation is defuzzified using the centroid of area method and crisp output is obtained [12]. Figure 3 to 6 show the fuzzy input variables for RSS, path loss, the load of the BTS and SNR respectively. Each of the variables has three subsets. To improve the reliability and robustness of the system, Gaussian membership functions (MFs) are used as an alternative to the traditional triangular MFS [15]. Since there are four fuzzy inputs and each of them has three subsets then there are 3 4 = 81 rules. The fuzzy IF THEN rules provide knowledge to the system and decide whether a handoff Fig. 5. Membership function of input SNR Fig. 4. Membership of the traffic load of the BTS 67

6 S. Girma et al., Fuzzy Logic Based Traffic Balancing in a GSM Network Fig. 5. Membership of handoff decision The fuzzy set values for the output decision variable handoff decisions are no handoff, wait handoff, be careful handoff and handoff. The universe of discourse for the variable handoff is defined from 0 to 1. Where 0 is no handoff and 1 is definite handoff 3.1. Rule Base Evaluation for FIS A handoff initiation technique in the proposed algorithm utilizes fuzzy logic with Multiple Objective Decision Making (MODM) approach to select the best network segment. All the available BTSs including the current serving BTS are then ranked. The selection of the target network is done using four fuzzy systems that utilizes the RSS, Path-loss, Interference and traffic load of the serving and the neighboring BTS. Based on those parameters the fuzzy logic will rank both the serving and the neighboring BTSs. If a better neighbor BTS is found, a handoff is requested. In figures 8-10, illustrate three examples of the rank of BTS for three different cases. Case one when all the four parameters are excellent, case two when all those parameter are average and case three when all the four parameters are worst: The lower the rule base evaluation the better for BTS to be chosen. Fig. 7. Rule base when all four parameters are average Fig. 8. Rule base when all four parameters are worst Fig. 6. Rule base when all four parameters are excellent 68

7 Journal of Sustainable Research in Engineering Vol. 1 (2), Handoff Necessity and Estimation The final weight handoff is obtained as the output from the Fuzzy inference system (FIS). The output of FIS is compared against a threshold value to determine if a handoff from the serving BTS is required. This threshold value can be adjusted according to the needs of the service provider. If the margin is too small, numerous unwanted handoffs may take place. However, when the margin is too large calls could be dropped and quality of service could be low. Thus, a balanced value for this threshold is required and need to be found from the simulation. Figure 11(a)-(d) shows handoff index with varying traffic load of the BTS, SNR, Path-loss and RSS respectively.for example, the higher load of the BTS the higher the probability of the MS to handoff from the current point of attachement to the neigboring BTS. Performing simulation and observing the need for handoffs based on the four inputs parameter, we propsed 0.55 to be optimum value for threshold value. Fig. 9. Handoff necessity with individual parameter 3.3. Conventional RSS Based Handoff Algorithm and Mobility In this algorithm, the RSSs of the different neighboring are measured over time and the BTSs with the strongest signal strength is selected to carry out a handoff if it satisfies the minimum threshold set by the network provider [12]. Additional handoff initiation strategies have been defined based on the comparison between the current RSS and that of the candidate RSS in neighboring BTS [12], [13]. i. Conventional RSS: handoff takes place if the neighbouring BTS s RSS is higher than the current RSS (RSSj > RSSi). ii. RSS plus hysteresis: handoff takes place if the neighbouring BTS s RSS is higher than the current RSS with a pre-defined hysteresis margin H. (RSSj > RSSi + H). Figure 12 shows the RSS from two neighboring BTS. It is assumed that the RSS averaged over time, so the fluctuations due to multipath nature of the radio environment can be eliminated [14]. Figure 12, also shows a MS moving from BTS I to BTS J. The RSS of BTS I decrease as MS moves away from the BTS and from BTS j increases as it approaches. With 69

8 S. Girma et al., Fuzzy Logic Based Traffic Balancing in a GSM Network conventional algorithm, looking at the variation of RSS from either BTS; it is possible to tell, 4500m is optimum area where handoff can take place. However, the conventional algorithm with hysteresis allows a Mobile Station (MS) to make handoff decision only if the RSS received from the neighboring BTS is sufficiently stronger than the current one by the specified hysteresis margin, provided a certain minimum signal level is assured. 100 RSS FROM I RSS FROM J RSS FROM J+10db RSS FROM J-10db RSS recieved in dbm 9001 Distance in meter between BTS I and J Fig.12. Conventional RSS algorithm with mobility Load balancing can be achieved, with RSS based handoff algorithm, by handing-off BTS edge users in over-loaded cells to migrate to under-loaded neighboring BTS. But this type of load balancing is manual in the sense that every time load unbalance occurred in the cluster, optimization engineers have to tune the hysteresis to solve load unbalance. 4. Results and Discussion Figure 13 shows the received signal strength from BTS i decrease as MS moves from the BTS i to ward BTS j, while the received signal strength from BTS j increases. In the middle between these BTSs, MS gets an ideal boundary where the received signal strength is equal. This is handover region where conventional handoff takes place to keep the connection. The conventional handoff algorithm cannot handoff the MS from BTS i to BTS j unless on the ideal boundary, even if the base station i is fully congested and base station j has no users on the entire coverage area. However, with the proposed algorithm, it is possible to handoff from BTS i to BTS j when the load of BTS i has high load and BTS j has medium load or low load. For example from Figure 13, when a MS receives Signal strength of -70dbm from BTS i, it will receive- 90dbm from BTS j. The path loss will be 40db and 60db respectively for BTS i and j, for transmission of -30dbm transmitter power. At any particular time, it is assumed, a MS receives SNR of 20db for both BTSs. For the handoff threshold of 0.55, MS can handoff from BTS i to j for traffic load of above 75% on BTS i only when the traffic load of BTS j is less than 30% (Figure 14), though there is 20db difference between the received signal from base station i and j. 70

9 Journal of Sustainable Research in Engineering Vol. 1 (2), 2014 Fig. 10. Received signal strength from base station i and j Fig. 11. Comparison of handoff for varying load between base station i and j On the ideal boundary, when a MS receives signal strength of -80dbm from either side, the path loss will be 50db for BTS for -30dbm transmitter power. As Shown on Figure 15, a MS can handoff from BTS i to j for traffic load of above 75% on BTS i if and only if BTS j has less than75% of traffic load This shows how the algorithm distribute traffic among all the neighboring BTSs thus making better resource utilization and improving the performance of the overall network 71

10 S. Girma et al., Fuzzy Logic Based Traffic Balancing in a GSM Network Fig. 12. Comparison of handoff for varying load between base station i and j 4.1. Proposed Handoff Algorithm and Mobility As shown on figure 16, the fastest handoff takes place at about 1700m when BTS i have high load and BTS j has low load. Therefore, the coverage area of a BTS i can virtually be shrunk inwards and MS in that region served by the neighbouring BTS j. This technique is called dynamic mobility load balancing. The logic behind this method is to adjust the handoff regions by biasing the handoff region depending on the load of the BTSs, causing cell-edge users in loaded BTS to migrate to less loaded overlapping cells, thereby releasing some traffic channels being occupied by edge user so that the call can proceed. With this process a new call blocking can be solved in case the neighbouring BTS has low load and satisfying the minimum RSS required. For the same load on either side, the handoff takes Place on the ideal boundary as in conventional handoff.. 72

11 Journal of Sustainable Research in Engineering Vol. 1 (2), 2014 low load on BTS I medium load on BTS I High load on BTS I Low load on BTS J medium load on BTS J High load on BTS J Handoff index(handoff probability) Distance in meter between BTS I and J Fig. 13. Proposed handoff algorithm with mobility 73

12 S. Girma et al., Fuzzy Logic Based Traffic Balancing in a GSM Network 5. Conclusion and Future Works In cellular environment, the selection of a BTS that can fulfill end-user satisfaction while keeping ongoing call is very important task, a wrong selection of BTS may lead to call drop, network over-congestion and wastage of scarce resource. Normally a selection of a network segment is done through handoff process. Traditional network selection was done by single-criteria like RSS, Power Budget, interference, distance and among others. A single-criterion is not an intelligent enough as it does not take into account other important parameters like level congestion on the BTS. This paper was on design and analysis of an intelligent multi-criteria handoff algorithm that enable to select the best network segment based on multiple parameters. The main objective of this work was to develop an algorithm that balances traffic load among all BTSs in cluster through handoff process. We showed that it is possible to balance traffic load of cellular network by handing off some ongoing calls on cell edge in over-loaded BTS to migrate to overlapping under-loaded BTS, such that the coverage area of loaded BTS virtually shrink towards BTS center of a loaded sector. In case of low load scenarios, the coverage area of a BTS is presumed to be virtually widened to cover up to the partial serving area of neighboring BTS. This was achieved using fuzzy logic. The proposed scheme was based on FLC and membership function that are biased in nature. Therefore, a research needs to be done in finding out the different types of membership functions that can result in optimal handoff performance. One alternative is to utilize an Adaptive- network based Fuzzy Inference System (ANFIS) where they system can construct an input-output relation based on human knowledge. A second alternative is to include suitable learning such as neural network. The proposed algorithm has been tested using data from wireless propagation models based on straight line mobility between two BTSs. Further improvement may be achieved by using real data from live wireless network. References [1] Moshe Sidi and David Starobinski, New call blocking versus handoff blocking in cellular networks, Wireless Networks, pp 15 27, [2] K.R Sudhindra and V. Srindhar, An overview of Congestion Relief methodology in GSM Network, IEEE, [3] K.Raymond, R. Arnott, R. R. Trivisonno and M. Kubota, On Mobility Load Balancing for LTE Systems, IEEE,2010. [4] P. Mu nuz, R. Barco, I. de la Bandera, M. Toril and S. Luna- Ram ırez, Optimization of a Fuzzy Logic Controller for Handoff-based Load Balancing, IEEE,2011. [5] K.A. Akpado, Oguejiofor O.S, Ezeagwu C.O and Okolibe A.U, Investigating the Impacts of BTS Antenna Height, Tilt and Transmitter Power on Network Coverage, International Journal of Engineering Science Invention ISSN (Online): , ISSN (Print): , 2013 [6] Goldsmith, wireless communication, Cambridge university press, [7] A.K. Ali., H. S. Hassanein, and H. T. Mouftah, Directional Cell Breathing Based Reactive Congestion Control in WCDMA Cellular Networks, in Proc. of IEEE Symposium on Computers and Communications,2007 [8] Ajay R Mishra, Advanced cellular network planning and optimization 2G/2.5G/ 3G evolution to 4G, John wiley & Sons, ltd, [9] S. Srinivas, A. Sahu and S.K. Jena, Efficient load balancing in Cloud Computing using Fuzzy Logic, IOSR Journal of Engineering (IOSRJEN), ISSN: Volume 2, PP 65-71, 2012 [10] Chandrasekhar and P. K.Behera, Use of Adaptive Resonance Theory for Vertical Handoff Decision in Heterogeneous Wireless Environment, International Journal of Recent Trends in Engineering, Vol 2, No. 3, November 2009 [11] R.Nasri.R and Z. Altman, Handoff Adaptation for Dynamic Load Balancing in 3GPP Long Term Evolution Systems, Proc. Of International Conference on Advances in Mobile Computing &Multimedia (MoMM.), 2007 [12] G.P.Pollini, Trends in handoff design, IEEE Communication Magazine, vol. 34, no. 3, pp ,1996. [13] Ahmed H, B. Liang and A. Saleh, Signal Threshold Adaptation for Vertical Handoff in Heterogeneous Wireless Networks, ACM/Springer Mobile Networks and Applications (MONET), 2005 [14] Liton Chandra Paul, Handoff/handover Mechanism for Mobility Improvement in the Wireless Communication, Global journal of research in engineering electrical and electronics engineering, [15] Ibrahim A. Hameed, Using Gaussian membership functions for improving the reliability and robustness of students evaluation systems, Elsevier,