Secure routing protocol for lossy, congested, and attack-prone Flying Ad Hoc Networks

Abstract

Flying Ad Hoc Networks (FANETs) provide flexible and rapidly deployable wireless communication among unmanned aerial vehicles (UAVs) without relying on fixed infrastructure. These networks are useful in applications such as disaster response, surveillance, environmental monitoring, military communication, and emergency coordination, where timely and reliable packet delivery is important. However, FANET communication is difficult to maintain because UAV nodes move rapidly in three-dimensional space, causing frequent topology changes, link breakage, route instability, queue buildup, packet retransmissions, and energy consumption. In addition to these normal wireless and mobility-related challenges, FANETs are also vulnerable to low-rate Denial-of-Service (LDoS) attacks. Such attacks are difficult to detect because they do not necessarily generate a continuously high traffic volume. Instead, FB-Shrew-like attackers transmit short bursts of relatively small packets at carefully selected intervals, occupying buffer space and forcing legitimate TCP packets to experience delay, retransmission, or loss. This thesis proposes a secure multipath routing protocol called AOMDV-GAD to improve communication resilience in lossy, congested, and attack-prone FANET environments. The proposed protocol integrates lightweight sender-side packet-loss classification with Genetic Algorithm-based route optimization. Traffic characteristics that are unavailable at the source are measured at the receiving node and conveyed to the source through ICMP feedback, after which the source distinguishes packet loss caused by random wireless conditions, ordinary congestion, and low-rate DoS activity. This distinction is important because these three events require different network responses. Random loss should not automatically cause a route to be treated as malicious, congestion should encourage the selection of routes with better queue availability, and LDoS activity should cause routes containing suspected nodes to be avoided. The routing component of AOMDV-GAD extends the multipath capability of Ad hoc On-Demand Multipath Distance Vector (AOMDV) routing. AOMDV first discovers multiple candidate routes between a source and destination. The proposed DoS detection output is then used as a mandatory route-screening condition. Routes containing a DoS-suspected UAV are removed from the candidate route set before fitness evaluation. Only the remaining safe routes are evaluated using queue availability and residual-energy information. The queue factor reduces the selection of heavily loaded intermediate UAVs, while the residual-energy factor reduces repeated dependence on weak or energy-depleted nodes. A Genetic Algorithm is then applied to the safe route population using selection, crossover, mutation, and survivor selection to identify an efficient forwarding route. In this design, route security is separated from route-performance optimization, preventing an attack-affected route from being selected simply because it has favourable congestion or energy values at a particular instant. The proposed protocol was implemented and evaluated using NS-3.35 in a three-dimensional FANET simulation environment. The evaluation considered variations in the number of UAV nodes, percentage of malicious nodes, UAV mobility speed, and simulation duration. AOMDV-GAD was compared with AOMDV, AOMDV-FG, HWSCS-HDL, CLUN-LSR, and JRP-LA using throughput, Packet Delivery Ratio (PDR), end-to-end delay, routing overhead, and energy consumption. The simulation results show that AOMDV-GAD improves packet delivery and throughput while reducing delay, routing overhead, and energy consumption under dense, mobile, and attack-prone conditions. For example, at 150 UAVs, AOMDV-GAD achieved approximately 2.28 Mbps throughput and 83.07% PDR, compared with approximately 1.06 Mbps throughput and 46.23% PDR for AOMDV. Similar improvements were observed when the malicious-node percentage reached 40%, the UAV speed reached 40 m/s, and the simulation duration reached 100 s. The results demonstrate that combining packet-loss classification, DoS-based route screening, congestion-aware evaluation, residual-energy-aware selection, and GA-based route optimization provides a more resilient routing mechanism for highly dynamic FANETs. The proposed approach reduces the probability of repeatedly forwarding packets through compromised, congested, or energy-weak routes and improves the ability of the network to maintain useful communication under adverse conditions.