A framework for traffic flow survivability in wireless networks prone to multiple failures and attacks

Abstract

Transmitting packets over a wireless network has always been challenging due to failures that have always occurred as a result of many types of wireless connectivity issues. These failures have caused significant outages, and the delayed discovery and diagnostic testing of these failures have exacerbated their impact on servicing, economic damage, and social elements such as technological trust. There has been research on wireless network failures, but little on multiple failures such as node-node, node-link, and link–link failures. The problem of capacity efficiency and fast recovery from multiple failures has also not received attention. This research develops a capacity efficient evolutionary swarm survivability framework, which encompasses enhanced genetic algorithm (EGA) and ant colony system (ACS) survivability models to swiftly resolve node-node, node-link, and link-link failures for improved service quality. The capacity efficient models were tested on such failures at different locations on both small and large wireless networks. The proposed models were able to generate optimal alternative paths, the bandwidth required for fast rerouting, minimized transmission delay, and ensured the rerouting path fitness and good transmission time for rerouting voice, video and multimedia messages. Increasing multiple link failures reveal that as failure increases, the bandwidth used for rerouting and transmission time also increases. This implies that, failure increases bandwidth usage which leads to transmission delay, which in turn slows down message rerouting. The suggested framework performs better than the popular Dijkstra algorithm, proactive, adaptive and reactive models, in terms of throughput, packet delivery ratio (PDR), speed of transmission, transmission delay and running time. According to the simulation results, the capacity efficient ACS has a PDR of 0.89, the Dijkstra model has a PDR of 0.86, the reactive model has a PDR of 0.83, the proactive model has a PDR of 0.83, and the adaptive model has a PDR of 0.81. Another performance evaluation was performed to compare the proposed model's running time to that of other evaluated routing models. The capacity efficient ACS model has a running time of 169.89ms on average, while the adaptive model has a running time of 1837ms and Dijkstra has a running time of 280.62ms. With these results, capacity efficient ACS outperforms other evaluated routing algorithms in terms of PDR and running time. According to the mean throughput determined to evaluate the performance of the following routing algorithms: capacity efficient EGA has a mean throughput of 621.6, Dijkstra has a mean throughput of 619.3, proactive (DSDV) has a mean throughput of 555.9, and reactive (AODV) has a mean throughput of 501.0. Since Dijkstra is more similar to proposed models in terms of performance, capacity efficient EGA was compared to Dijkstra as follows: Dijkstra has a running time of 3.8908ms and EGA has a running time of 3.6968ms. In terms of running time and mean throughput, the capacity efficient EGA also outperforms the other evaluated routing algorithms. The generated alternative paths from these investigations demonstrate that the proposed framework works well in preventing the problem of data loss in transit and ameliorating congestion issue resulting from multiple failures and server overload which manifests when the process hangs. The optimal solution paths will in turn improve business activities through quality data communications for wireless service providers.