| Abstrakt: |
Existing and future wireless networks are expected to have multiple Radio Access Technologies (RAT), primarily due to legacy reasons and introduction of newer technologies. However, new technologies, e.g., Self Organizing Networks (SON), tend to focus only on future generations networks, ignoring the existing benefits provided by legacy systems. The motivation behind this work is that, with minimal changes newer functionalities can be extended to legacy systems as well. This paper proposes an Software Defined Networking assisted SON architecture for multi-RAT cellular networks. Using this architecture, a multi-RAT Automatic Neighbour Relation functionality of SON is constructed, and the same can be used to predict mobility of User Equipments (UEs) to a cell. This mobility aspect is modeled using random graphs, and a Mixed Integer Non-Linear Programming problem is formulated, to maximize probability of mobility of UEs to a cell. It is observed that the problem is non-convex and also NP-Hard. Subsequently, a Particle Swarm Optimization (PSO) approach is developed to solve this problem. Results show that PSO output matches with that of exhaustive search for finding the maximum value of UE mobility probability to a cell, thus validating the solution approach. Results also reveal, compatibility of QoS parameters (required by UEs) among source and target cells, increases mobility probability. However, higher arrival rate of UEs towards a cell decreases the same, within target cell capacity limits. Subsequently, implications of UE mobility probability on core SON functionalities, such as load balancing and mobility optimization, are studied. [ABSTRACT FROM AUTHOR] |
Full text from SpringerLink