Traffic patterns in manufacturing machines exhibit strong temporal correlations due to the underlying repetitive nature of their operations. A MAC protocol can potentially learn these patterns and leverage them to efficiently schedule nodes’ transmissions. Recently, with the advent of low power MEM based sensors, wireless sensing in these machines has gained prominence. Communication in control loops must cater to extremely low hard real-time latencies while embracing low-power design principles. In this paper, we present a novel MAC, i-MAC, a wireless MAC protocol that learns to expect and plan for traffic bursts and consequently coordinate node transmissions efficiently.