Novel RAT selection mechanism based on Hopfield neural networks

The concept of wireless heterogeneous networks is based on the coexistence and interoperability of different types of radio access technologies in a unified wireless heterogeneous platform. A key process in the wireless heterogeneous networks is the initiation and execution of vertical handovers. Variety of mechanisms are proposed in the literature that are tailored to optimize the choice of radio access network in the process of vertical handover initialization. This paper proposes a novel mechanism for radio access technology (RAT) selection, the Hopfield neural network Rat selection Mechanism (HRM), that utilizes the Hopfield neural networks as a powerful decision making tool. The main advantage of the proposed mechanism lays in its ability of Artificial Intelligence (AI) reasoning and the use of multi-parameter decision by exploiting the options of IEEE 802.21 protocol. This approach allows improvements in terms of service retainability, decreased packet loss, jitter, etc. The performance of the novel mechanism is compared with two other types of RAT selection mechanisms (RSS based, and Multi-criteria policy based mechanism) throughout extensive simulations. The results clearly show the benefits of the proposed new mechanism and its justified usage in wireless heterogeneous systems.