Fractional squirrel–dolphin echolocation with deep belief network for network‐controlled vertical handoff in disparate and heterogeneous wireless network

Abstract

The innovative characteristics of different heterogeneous wireless networks provide better service. The disparate wireless access networks consist of wireless metropolitan access network (WMAN) and cellular networks. Various techniques are devised for dynamic network selection, but these techniques concentrated on attaining improved performance. To address this issue, a novel method is devised to improve the energy efficiency of disparate heterogeneous wireless networks. Here, the proposed fractional squirrel-dolphin echolocation (FrSqDE) algorithm is devised by integrating squirrel search optimization (SSA) and fractional dolphin echolocation (fractional DE) algorithm. In addition, the fitness function is considered with certain factors that include delay, bit error rate (BER), jitter, packet loss, energy consumption, probability of hot spot, average bit rate (ABR), and received signal strength (RSS). The proposed FrSqDE optimization algorithm is used for training deep belief networks (DBNs) to select optimal weights. Here, the network selection factors are given as input to the proposed FrSqDE-based DBN, wherein optimal decision is made to handle vertical handoff. The proposed FrSqDE-based DBN offered improved performance with a minimal call drop probability of 0.050, minimal delay of 0.013 s, minimal energy consumption of 0.393 J, and maximal throughput of 117.972 kbps.

Assessing the quality of experience in wireless networks for multimedia applications: A comprehensive analysis utilizing deep learning-based techniques

“…In recent years, with the advancements in deep learning, various deep learning models have been applied to QoE prediction. For example, one article utilized deep belief networks to implement the mapping from QoS to QoE [ 20 , 21 ]. The method preprocesses the values and utilizes deep belief networks to fit MOS values.…”