An energy-aware traffic offloading approach based on deep learning and optimization in massive MIMO

In the wireless communication, the shortage of bandwidth has motivated the investigation and study of the wireless access technology called massive Multiple-Input Multiple-Output (MIMO). In multi-tier heterogeneous Fifth Generation (5G) networks, energy efficiency is a severe concern as the power utilization of macro base stations' is comparatively higher and proportional to their traffic load. In this paper, a novel African Vulture Shepherd Optimization Algorithm (AVSOA) is established that relies on macro cells and small cell system load information to determine the highly energy-efficient traffic offloading system. The proposed AVSOA model is a combination of the African Vulture Optimization Algorithm (AVOA) and the Shuffled Shepherd Optimization Algorithm (SSOA). The system load is predicted here by exploiting a Deep Quantum Neural Network (DQNN) algorithm to perform the conditional traffic offloading in that every macro-Base Station (BS) conjectures the offloading systems of other macro cells. The experimental evaluation of the adopted model is contrasted with the conventional models considering the energy efficiency, spectral efficiency, throughput, and system load. Finally, the performance analysis of the proposed model achieved better energy efficiency, spectral efficiency, and throughput of 0.250598, 0.184527, and 0.820354 Mbps and a minimum system load of 697.