A Recursive Random Search Algorithm for Black-box Optimization

This paper proposes a new heuristic search algorithm, Recursive Random Search(RRS), for blackbox optimization problems. Specifically, this algorithm is designed for the dynamical parameter optimization of network protocols which emphasizes on obtaining good solutions within a limited time frame rather than full optimization. The RRS algorithm is based on the initial high-efficiency property of random sampling and attempts to maintain this high-efficiency by constantly “restarting” random sampling with adjusted sample spaces. Due to its basis on random sampling, the RRS algorithm is robust to the effect of random noises in the objective function and it performs especially efficiently when handling the objective functions with negligible parameters. These properties have been demonstrated with the tests on a suite of benchmark functions. The RRS algorithm has been successfully applied to the optimal configuration of several network protocols. One application to a network routing algorithm is presented.