We present an approach to monitoring power consumption in a distributed computationally constrained micro-grid. Computationally constrained micro-grids typically operate over low-powered computational devices on insecure and unreliable communication networks. Monitoring these networks is important in distinguishing faulty power reports from adversarially manipulated ones. We address this problem with a two-pronged approach, first by characterizing electrical appliances according to consumption behavior models. Second, we propose a delayed snapshot algorithm for collecting aggregated data on power consumption patterns. Our snapshot algorithm is demonstrably efficient in terms of message exchange and offers the added advantage of being resilient to malicious attacks aimed at communication disruption.