QoS demands splitting for RAT selection in heterogeneous networks

Heterogeneous network is becoming more and more crowded with a huge increment of users and their diverse requirements today. Case exists when a single network cannot meet a user´s QoS requirement. To this end, we propose the concept of "QoS demands splitting". When no single channel can satisfy the QoS requirement, user can connect to another network simultaneously to share his load. Our system model is originated from the QoS satisfaction game, but with split demand. To approach the Nash Equilibrium, we propose a satisfaction-guaranteed better reply algorithm. The convergence property of the satisfaction-guaranteed better reply algorithm is proved by using quasi-potential function. Specifically, when given a complete interference graph, the preference-based matching algorithm is proposed to obtain a more stable equilibrium. In this case, users can approach a unique pure Nash equilibrium. Simulation results illustrate the efficiency of the satisfaction-guaranteed better reply algorithm and preference-based matching algorithm. Additionally, it shows that more crowded the system is, more improvements in the number of satisfied users and social welfare will be achieved.