Single Linkage Weighted Steepest Gradient Adaboost Cluster-Based D2D in 5G Networks

Abstract

Efficiency of data transmissions with minimum latency levels and better resource utilization is a challenging issue in 5 G device-to-device (D 2D) environments. A novel technique referred to as single linkage steepest gradient gentle AdaBoost cluster-based device (SLSGAC) is introduced to improve device-to-device communications with minimum latency. The proposed technique uses the ensemble clustering approach to group mobile devices by constructing a set of weak clusters, based on the Minkowski single linkage clustering technique. In the weak clustering process, residual energy, bandwidth and SINR are estimated, and mobile devices are grouped based on the Minkowski distance measure. Results of the weak clustering process are combined to provide the final ensemble’s clustering output by applying the steepest gradient function to minimize the error rate. For each cluster, a head is selected from among the group members to improve the data transmission rate and minimize latency. Simulations are conducted comparing the proposed technique with the existing methods based on such metrics as energy efficiency, data delivery ratio, packet loss rate, throughput and latency.