To search, Click below search items.


All Published Papers Search Service


Machine Learning-Based Simplified Early Link Failure Detection Model for Mobile Ad-Hoc Network


Manjunath B Talawar, Dr. D V Ashoka, Dr. R Nagaraja


Vol. 22  No. 4  pp. 491-500


Mobile Ad-Hoc Network (MANET) is a self-configuring, wireless network consisting of a combination of mobile devices connected over a wireless link. Nodes in MANET are connected via multi-hop wireless links and mobility in nature that cause link failures. MANETs are mostly dynamic, with different nodes leading to common link failure problems. Link failures in MANET reduce the network performance and increase network overhead. The detection of link failures and the prediction of reliable links for data transmission always plays a vital role in several research topics of the network community. Therefore, in this paper, the proposed approach detects link failure through a Machine Learning (ML) based simplified early link failure detection model. Initially, the possibility of link failure is identified by using a simplified early link failure detection (SELFD) method, which employs Ad-hoc On-demand Distance Vector (AOMDV). The proposed approach follows both the unsupervised and supervised method of the one-Dimension Deep Auto Encoder (1D-DAE) method and logistic regression for estimating the link failure probability. Finally, the Flamingo Search Algorithm is proposed to fine-tune the hyperparameters of the proposed 1D-DAE method to detect the link failures in routing efficiently. The proposed approach is implemented in the PYTHON platform.The performance of the proposed approach is compared with existing routing methods based on packet delivery ratio, normalized routing load, average end to end delay, throughput, buffer occupancy and bit error rate.


Mobile Ad-Hoc Network (MANET), One Dimensional-Deep Auto Encoder (1D-DAE) method, Flamingo Search Algorithm (FSA), Link Failures, nodes.