Reinforcement Learning-Based Multidimensional Perception and Energy Awareness Optimized Link State Routing for Flying Ad-Hoc Networks

One of the uncrewed aerial vehicles (UAV) in a Flying Ad-hoc network (FANET) can link directly to the infrastructure. At the same time, the other UAVs in the system may have a multi-hop connection in which each node works as a relay and a data collection node. We may not have the support of traditional infrastructure-based networks when natural disasters such as floods or earthquakes strike. This is fatal because trapped people are challenging to find by search and rescue personnel. In such cases, an airborne network of small drones is valuable for giving quick and adequate coverage of the affected area and instant insights to rescue workers. At the same time, such networks face various challenges, and ongoing research and development show promise in making such technology more dependable and effective. This paper presents Multidimensional Perception and Energy Awareness Optimized Link State Routing (MPEAOLSR) for Flying Ad-hoc networks, which is based on Reinforcement Learning (RL). The protocol is a mobile wireless LAN-specific version of the traditional link state algorithm. The protocol largely relies on the idea of multipoint relays (MPRs). During the flooding process, MPRs are chosen to forward broadcast messages. This technique considerably minimizes message overhead associated to a standard flooding system in which each node retransmits each message after receiving the first copy. In RL-MPEAOLSR, only nodes designated as MPRs generate link state information. Furthermore, the RL-MPEAOLSR node can opt to report just links between itself and its MPR selectors, decreasing the number of control messages flooding the network. Because MPRs function well in large and dense networks, the MPEAOLSR protocol is suited for them. The proposed approach outperforms the conventional Energy Awareness Optimised Link State Routing for Flying Ad-Hoc Networks, according to the results. The technique far outperforms current methods in terms of modern bandwidth consumption of 1478.04kpbs, network density of 95.64%, packet delivery ratio of 95.85%, packet loss ratio of 31.94%, delay in transmission time of 6.981 s and accuracy 6.981%.