Abstract—The Internet of Things (IoT) is a disruptive discipline which has gained great attention lately. However, attaining energy efficiency is still challenging due to the resource constraint
nature of IoT devices. This paper designs a novel energy-aware two-tier model for edge-assisted IoT. In the proposed two-tier model, the first tier (IoT tier) is segregated into multiple clusters. Voronoi Cell-based Correlation Cluster Formation (VC3F) algorithm is proposed for cluster formation. In each cluster, the optimal Cluster Head (CH) is selected by a Multipart CH Score (M Score) value. The CH eliminates redundant data using a novel Graphical Representation-assisted Similarity Rank (GR- SimRank) scheme. The redundant-free data is transmitted to the edge-tier through an optimal route selected by a Multi-Objective Flower Pollination (MO-FPO) algorithm. In the edge tier, a novel Fast Fully Connected Neural Network (F2CNN) is proposed with the Lempel-Ziv-Welch (LZW) compression algorithm. The F2CNN-LZW compression works upon the sensitivity level of the data determined by the data validation approach. The efficiency of the proposed two-tier model is evaluated through experiments performed in ns-3.26 simulator. The observations show promising results in energy consumption, throughput, packet delivery ratio, delay and reliability.