Energy disaggregation, also known as non-intrusive load monitoring (NILM), is the process of analyzing energy consumption in a building and identifying individual appliancelevel energy usage. This approach can provide valuable insights into energy consumption patterns and help reduce overall energy usage, costs, and carbon emissions. This paper proposes a new method for tackling the disaggregation problem by using data from low-cost wireless sensor networks. The proposed approach estimates appliance states using a GMM model and uses these states as features to improve energy disaggregation. The performance of the proposed method was evaluated on a real-world dataset called SmartSense deployed in our lab, and the results showed that it significantly improved the accuracy of conventional NILM disaggregation performance.