An Immunologically-Inspired Adaptation Mechanism for Evolvable Network Applications

Large-scale network applications are expected to be more autonomous and adaptive to dynamic changes in the net- work to improve user experience, expand applications' opera- tional longevity and reduce maintenance cost. Based on the ob- servation that various biological systems have already met the requirements (i.e., autonomy and adaptability), this paper de- scribes a biologically-inspired framework, called iNet, to design autonomous and adaptive network applications. iNet is designed after the mechanisms behind how the immune system works. iNet models a set of environment conditions (e.g., network traffic and resource availability) as an antigen and a behavior of network applications (e.g., migration and reproduction) as an antibody. iNet allows network applications to autonomously sense its sur- rounding environment conditions (i.e., antigens) and adaptively invoke a behavior (i.e., antibody) suitable for the conditions. The configuration of antibodies evolves via genetic operations (e.g., mutation and crossover). Simulation results show that iNet allows agents to autonomously adapt to changing environment condi- tions by invoking their behaviors suitable for the current envi- ronment condition and evolving their antibody configurations.