Design And Simulation Of Markov-based Vertical Handoff Decision Schemes

Future wireless communication systems will contain all kinds of wireless access networks, which are called heterogeneous networks. This kind of networks can provide users with different access technologies, different data rates, different services, which can satisfy users' requirements better. Vertical handoff is one key part of radio resource management techniques in heterogeneous networks and can ensure QoS (Quality of Service) across different networks. Therefore, efficient vertical handoff algorithms are extremely important.This thesis mainly studies the handoff decision making, which is a key step of a vertical handoff process, and proposes two vertical handoff decision algorithms based on Markov. In order to overcome the shortcomings of the existing single-attribute decision algorithms, we propose a novel algorithm. The proposal is based on RSS (Received Signal Strength) and Markov mobility model. We investigate lifetime-based handoff module which can keep users stay longer in the preferred network. A dynamic handoff threshold module is also proposed in this algorithm. The system can adjust the handoff threshold based on users'velocity, and then improve system performance.The second proposal is based on the Multiple Attribute MDP (Markov Decision Process). The decision making problem is designed as a MDP model for the objective of maximizing the totally expected reward in every connection. The network resources are captured by a link reward function, which is composed of multiple QoS factors. The G1 and entropy methods are applied to determine the weights of the QoS factors. The value iteration algorithm is used to reach a stationary policy.In order to make performance evaluation, we construct simulation environment by MATLAB and NS-2 to make comparisons with other existing algorithms. The numerical results demonstrate that:the first single attribute algorithm performs better in numbers of handoff, available bandwidth, decision delay, WLAN utilization; the second one has a higher expected total reward and a lower average numbers of handoff.