Online distributed evolutionary optimization of Time Division Multiple Access protocols

With the advent of cheap, miniaturized electronics, ubiquitous networking has reached an unprecedented level complexity, scale and heterogeneity, becoming core several modern applications such as smart industry, buildings cities. A crucial element for network performance is protocol stack, namely sets rules data formats that determine how nodes in exchange information. great effort been put to devise formal techniques synthesize (offline) protocols, starting from system specifications strict assumptions on environment. However, offline design can be hard apply most applications, either due numerical or fact environment might unknown not available. In these cases, online adaptation potential offer a much more scalable robust solution. Nevertheless, so far only few attempts have done towards automatic design. These approaches, however, typically require central coordinator, need build update model environment, which adds complexity. Here, instead, we envision emergent property network, obtained by environment-driven Distributed Hill Climbing (DHC) algorithm uses node-local reinforcement signals evolve, at runtime without any coordination, scratch, needing We test this approach with 3-state Time Division Multiple Access (TDMA) Medium Control (MAC) observe its emergence networks various scales settings. also show DHC reach different trade-offs terms energy consumption performance.