Optimizing Vertical Handover Decision Making in Heterogeneous Wireless Networks

Abstract

Ever-increasing demands of users and the development of modern communication technologies have led to the evolution of 4th Generation (4G) heterogeneous wireless networks. The integration of wireless networks of different characteristics and the demands of: user, mobile device, applications and service providers result in issues such as seamless mobility management, security, administration, billing, etc. The key issue among these challenges is the handover process of mobility management for seamless communication of mobile devices in heterogeneous wireless networks with maximized users’ satisfaction. Always Best Connected (ABC) services anywhere at anytime is one of the key objectives of 4G in integrating IEEE and cellular technologies. This thesis mainly addresses the Vertical Handover Decision (VHD) making in heterogeneous wireless networks for seamless communication of mobile devices. The dependency of VHD on multiple attributes in heterogeneous wireless networks demands an optimized handover process in terms of minimized complexity with improved reliability and flexibility. Several existing methods like fuzzy logic, neural networks, game theory and Multiple Attribute Decision Making (MADM) have been used for VHD. However there are still open issues such as, complexity, reliability and flexibility in these methods. MADM is one such method which supports multiple attributes based decision with minimum complexity for multiple criteria dependent VHD in heterogeneous wireless networks. The main problem with the MADM method is unreliable network selection and the rank reversal problem due to its dependency on attributes normalization and weight calculation methods. Hence, this thesis presents an optimized MADM method referred to as Simplified and Improved Multiple Attributes Alternate Ranking (SI-MAAR) for overcoming the limitations of classical MADM methods. Thus, SI-MAAR method is optimized in terms of minimized computational complexity and improved network selection reliability with the elimination of rank reversal problem. With MATLAB simulations, the analytical model of SI-MAAR method is demonstrated for 100% reliable VHD with the 0% rank reversal problem in heterogeneous wireless networks. iFurther, many of the classical MADM methods used in VHD in heterogeneous wireless networks depend on attributes weight computation techniques such as, Entropy, Variance, Analytical Hierarchy Process (AHP) etc. Expectations of users and applications during VHD in heterogeneous wireless networks is subjective in nature. AHP is one such popular method which supports computation of subjective attributes weight. The main problem with AHP is computation of reciprocal matrix through the involvement of the decision maker which will result in unreliable attributes weight and further to unreliable network selection in VHD. Hence, this thesis also presents an optimized AHP method referred to as Simplified and Improved Analytical Hierarchy Process (SI-AHP) to overcome unreliability in attributes weight computation. Thus, SI-AHP method is optimized in terms of minimum involvement of the decision maker resulting in reduced attributes weight computational complexity with the improved reliability. With MATLAB simulations, SI-AHP method is demonstrated for 100% reliable attributes weight computation used for VHD in heterogeneous wireless networks. In this thesis, SI-MAAR and SI-AHP methods are numerically analysed using MATLAB simulations and results demonstrate that SI-MAAR and SI-AHP methods are outperforming classical MADM and AHP methods respectively. Similarly, simulations using network simulators and further validation by testbed-based approaches are also required for justifying the proposed analytical solutions. Among the available open source network simulators such as NS2, NS3, OMNET++ and J-Sim, simulation of heterogeneous wireless networks is supported only in NS2’s distribution provided by National Institute of Science and Technology (NIST). The major problems with the NIST’s NS2 distribution are: (i) support for only one mobile node simulations (ii) minimal support for VHD and (iii) non-availability of configuration and result analysis tools such as TCP Performance Evaluation suite for simulations of heterogeneous wireless networks. Hence, this thesis also presents NS2 based Evaluation Suite for User Datagram Protocol applications referred to as “ES-UDP” for configuration, simulations and results analysis of multiple mobile nodes in heterogeneous wireless networks. Thus, ES-UDP tool provides both text and graphical results of handover, packets sent and received, throughput, packet delay and jitter of heterogeneous wireless networks simulations. iiOn the other hand, real time experimentation is subject to testbed’s deployment complexity, cost and time. The other major challenge of testbed experimentations is Linux kernel support for heterogeneous wireless networks. Although for handover execution, network layer protocol-Mobile IPv6 is supported by Linux kernel, the major issues are lack of testbed deployment informations, high cost, and nonavailability of testing checkpoints and debugging procedures. Thus, this thesis also presents a cost-effective testbed of Mobile IPv6 for handover execution in homogeneous and heterogeneous wireless networks. Further, this testbed can be used for VHD by deploying proposed solutions: SI-MAAR and SI-AHP in the Linux kernel. To summarize, the main contributions of the thesis are, improving the network selection reliability and the elimination of rank reversal problem of classical MADM methods used in VHD of heterogeneous wireless networks with “SI-MAAR”. Simplifying the attributes weight computation and improving the attributes weight reliability of AHP with “SI-AHP”. Ease of NS2’s configuration, simulation and results analysis of multiple mobile nodes experimentations in heterogeneous wireless networks with “ES-UDP” tool. Finally, simple and cost-effective Mobile IPv6 testbed for handover execution and further to handover decision in heterogeneous wireless networks