This paper investigates the use of genetic algorithms for automated selection of parameters in a wireless ad hoc networking system. Experimental results demonstrate that the genetic algorithm can optimize for different classes of operating conditions. We also compare our genetic algorithm optimization against hand-tuning in a complex, realistic scenario and show how the genetic algorithm provides better performance.
The paper presents a method for optimizing the configurable parameters of a proactive link‑state mobile ad hoc network protocol using a genetic algorithm. It explains the protocol’s key parameters—such as heartbeat interval, score thresholds, and routing interrupt periods—and why manual tuning is difficult.
The authors detail the genetic algorithm design, including representation, mutation, crossover, and evaluation using OPNET simulations. Experiments on a DARPA/Army demonstration dataset and synthetic scenarios show that automated tuning reduces dropped packets and delay compared with manual tuning.
A genetic algorithm is employed to automatically select parameter values for the ad hoc networking protocol.
The scoring function combines dropped packets and transmission delay, with dropped packets weighted ten times higher.
A simulation typically takes about 7–8 minutes for the Lakehurst dataset and slightly less for generated scenarios.