AI-Driven Handover Management Scheme for HetNets at the Edge

Bringing artificial intelligence (AI) to operate on the devices which are resource constrained and are requiring high computational resources so that they should not overload the cloud and manage at the edge is called AI on the Edge or Edge AI or Edge Intelligence. Artificial Intelligence employs machine learning, deep learning and other techniques for solving complex decision-making problems or for data transmission and computation. One of such large-scale complex systems is Heterogenous Network (HetNet) which consists of varied access technologies with different properties. And when a device moves across such diverse environment, providing seamless continuity in session and call becomes an issue. Hence, in this paper Edge AI has been employed where an edge node is intelligent enough to make an optimal handoff decision to avoid ping-pong and unnecessary handoffs. In this light, ‘Edge based Multi Decision Making Algorithm (EMDMA)' is proposed. It is an edge based multi-criterion-based decision-making model for network selection during handoff implemented at the edge of the network where a user's device is in mobility. The performance of EMDMA is compared with the traditional algorithms TOPSIS and FTOPSIS. After simulation, average number of handoffs for proposed algorithm is reduced to 12% and the probability of handoffs failure has also been reduced to 0.06.