Design and Performance Evaluation of Cost Function Based Vertical Handoff

The main objectives of a handoff procedure are, first, to minimize the number of link transfers and second, to minimize the handoff processing delay by correct choice of target BS/AP with speedy execution. This minimizes the probability of connection interruptions and reduces the switching load. If the handoff is not fast enough, the quality of the service experiences degradations. A handoff should be evaluated as to its impact on the mobile to network connection. I.INTRODUCTION In cellular networks such as GSM, [2] a call is seamlessly handed over from one cell to another using hard handoff without the loss of voice data. This is managed by networks based handoff control mechanisms that detect when a user is in a handoff zone between cells and redirect the voice data at the appropriate moment to the mobile node via the cell that the mobile node has just entered. In 4G networks a handoff between different networks is required. A handoff between different networks is referred to as a Vertical handoff (VHO). Handoff management process Many literatures describe the handoff process in three phases. 1. Handoff Information Gathering: 2. Handoff Decision: 3. Handoff Execution: II. CLASSIFICATION OF VERTICAL HANDOFF For VHO, there are two additional and useful classifications to understand why VHO mechanisms are different from traditional horizontal handoff (HHO) mechanisms such as signal strength-based. The first classification is: upward and downward. An upward VHO occurs from a network with small coverage and high data rate to a network with wider coverage and lower date rate. On the other hand, a downward VHO occurs in the opposite direction. As an example for this classification let‟s consider the case of two of the most important current wireless technologies: 3G cellular networks and WLANs. The WLAN system can be considered as the small coverage network with high data rate while the 3G cellular system is the one with wider coverage and lower data rate. The trend in the literature has been to perform downward VHOs whenever possible.[8] The second classification is: imperative and alternative. An imperative VHO occurs due to low signal from the BS or AP. In other words, it can be considered as an HHO. The execution of an imperative VHO has to be fast in order to keep on-going connections. On the other hand, a VHO initiated to provide the user with better performance (e.g., more bandwidth or lower access cost) is considered to be an alternative VHO. This VHO can occur when a user connected to a 3G cellular network goes inside the coverage of a WLAN, even if the signal of the connection to the 3G cellular networks does not lose any signal strength, the user may consider the connection to the WLAN a better option. III. COST FUNCTION-BASED STRATEGIES (CFBS) Vertical handoff decision cost function is a measurement of the benefit obtained by handing over to a particular network. It is evaluated for each network n that covers the service area of a user. It is a sum of weighted functions of specific parameters. The general form of the cost function fn of wireless network n is fn=ΣsΣiws,i.p n s,i P n s,i is the cost in the i th parameter to carry out service s on network n; ws, i: the weight (importance) assigned to using the i th parameter to perform services (with Σiwi=1). The parameters used are bandwidth Bn that network n can offer, power consumption Pn of using the network device for n and monetary cost Cn of n. The cost of using a network n at a certain time, with N (i) as the normalization function of parameter i is defined as: fn= wb .N(1/Bn)+wp .N(Pn)+wc .N(Cn) The network that is consistently calculated to have the lowest cost is chosen as the target network. Therefore, this cost function-based policy model estimates dynamic network conditions and includes a stability period (a waiting period before handoffs) to ensure that a handoff is worthwhile for each mobile. The proposed policy-enabled handoff system allows users to express policies on what is the best network and when to handoff. To achieve flexibility, the system separates the decision making scheme from the handoff mechanism (routing table manipulation and sending location updates). To achieve seamlessness, the system considers user involvement (for policy specification) with minimal user interaction (for Sunita Sharma , Dr. S.K.Bodhe / International Journal of Engineering Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com Vol. 2, Issue 3, May-Jun 2012, pp. 257-262 258 | P a g e automation). To improve system stability in the handoff mechanism, load balancing solution is proposed avoiding the handoff synchronization problem (simultaneous decision by many mobiles). For that, we have implemented a performance agent that collects the information on the current bandwidth usage at base stations, and periodically announces this information to its coverage. Since, all data traffic goes through base station they have the most accurate information on current bandwidth usage and the available bandwidth in the network. They solve the problem through a randomized stability period. We use the utility function (higher utility = target network), to evaluate the reachable wireless networks discovered (bandwidth and movement speed as factors) and to quantify the QoS (Quality of Service) provided by the wireless network on the mobile terminal. We introduce two adaptive handoff decision methods adjusting the stability period, according to the network resources and the running applications on the mobile terminal. In the proposed handoff scheme, the handoff decision method is preceded by a system discovery method (Algo-1). [10] The latter is based on an adaptive interface activating method that adjusts the interface activating interval relying on the distance between the mobile terminal and the base station. For that, an ideal coverage concept (i.e., the real coverage in a wireless overlay network) is introduced in which mobile terminal‟s position information and a Location Service Server can assist mobile terminal in deciding when to activate its interfaces (Algo-2). Thus, the system discovery method can balance the power consumption and system discovery time. The proposed vertical handoff decision is based on a policybased networking architecture (i.e., IETF framework). All the described decision strategies were evaluated on two types of networks: WLAN and WWAN such as GSM or GPRS. In the simulations, the evaluated heterogeneous wireless networks consist of a single GSM network, 100 WLANs where WLANs are randomly deployed [11] . The topology covers an area 3000 m in length and 3000 m in width with 10 base stations. The number of mobile nodes ranges from 10 to 70 in the area of 100×100 m, and the mobility of random way point is adopted for each mobile node with random direction and random velocity from 1 to 25 m/s. The transmission range of GSM covers the whole area, where that of each WLAN is 100 m 2 . The bandwidths of GSM, WLAN, are 384 kb/s, and 54 Mb/s respectively. The simulation considers two classes of traffic, i.e., constant bit rate (CBR) and variable bit rate (VBR). The CBR traffic is assumed to arrive at the heterogeneous network to a Poisson distribution with arrival rate λ. The average holding time of the CBR traffic is exponentially distributed (μ), and its mean is normalized to unity. On the other hand, the VBR traffic is assumed to arrive based on a self similar model. The RSS received by a mobile node is different when it uses different wireless networks.Several useful parameters for the simulations are summarized in Table I. Table1: Simulation parameters for Vertical Handoff Evaluation Simulation Parameter Values Area of Coverage for Simulation 3000(m)×3000(m) Number of WLAN Access Points 100 Number of Base Stations 10 Transmission Range of WLAN 100 m Transmission Range of GSM 1000 m Path Loss Exponent (WLAN) 4.5 Path Loss Exponent (GSM) 2.8 Arrival Rate CBR variable Arrival Rate Self Similar Traffic variable Data Rate : WLAN 1.6 Mbps Data Rate: GSM 384 Kbps Mobility model (Random way Point) Velocity (1-25 m/sec) IV. RESULTS FOR CBSF VERTICAL ALGORITHMS In case of CBSF we have evaluated the performance in terms of number of handoffs, handoff delay and dropping probability. The blocking probability is not evaluated since this is a heterogeneous network, because it is assumed that two different networks have already accepted the call/request from the mobile terminal. The RSS based algorithms are derivatives of the conventional handoff algorithms in homogeneous networks. In case of heterogeneous network we applied the same logic only difference is the propagation characteristics and the threshold is different for two networks. Figure 2 shows the performance of vertical handoffs in discussion in terms of number of handoff requests generated. From the graphs we observe that for handoff algorithm based on RSS threshold performs similar to the handoff algorithm in homogeneous network. In this case also we observe the „ping-pong‟ effect.[6] To improve the performance we have tested the network for handoff algorithm based on RSS with hysteresis, the results of simulation indicate that number of handoff requests have reduced by a factor of 10 for traffic of 100 packets per second. For algorithm-1 the numbers of requests are 2 times less as compared to RSS with Threshold based algorithm. The simulation results indicate that the algorithm-2 improves the performance by a factor of 1.6 as compared to algorithm-1. Sunita Sharma , Dr. S.K.Bodhe / International Journal of Engineering Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com Vol. 2, Issue 3, May-Jun 2012, pp. 257-262 259 | P a g e Figure1:Number of vertical handoffs requested for variable traffic (CBR) In case of heterogeneous networks it important to evaluate the handoff process in terms of time taken by the mobile terminal to change of its link from one type of network to another. We call this as „handoff delay‟. Figure .3 shows the performance of the network under test for variable traffic (CBR). It is observed that handoff algorithm-2 strategy results in maximum delay for handoff execution; this is due to time taken by both the networks to evaluate handoff requirement and passing on the information of requirement and availability of resources and confirming the handoff request. Further it is observed that the variability of the delay is very high with respect to the traffic variations (σ = 452 with mean delay of 64 ms). The delay of handoff algorithm-1 is less as compared to algorithm-1 (σ = 244 with mean delay of 45 ms). Figure:2 Handoff delay for vertical handoff execution for variable traffic (CBR). 10 -1 10 0 10 1 10 1 10 2 10 3 10 4 10 5 Traffic Packets/Second N u m b e r o f v e rt ic a l H a n d o ff s Vertical handoff in the Test network