Tracking food spoilage in the smart home using odour monitoring

Abstract

Use of continuous odour monitoring in a smart home represents a novel sensor modality with the potential to recognize various activities of daily living and identify unsafe conditions for the occupant. In this paper, we focus on food spoilage as one such condition. Using a metal-oxide sensor (MOS) based electronic nose, we measured the odour signatures of two common foods (milk and yogurt) that were stored at 25°C during a week-long period. Feature vectors were constructed using the maximum absolute sensor responses, and their components exhibited a smooth trajectory as the samples aged. Applying principal component analysis (PCA) revealed that the two substances followed distinct trajectories during spoilage. We conclude that an electronic nose can be used to track the spoilage of various foods in a manner that is a) repeatable for a specific food, and b) unique for different foods. Additionally, we found that the sampling protocol used in this work resulted in better repeatability than our previous work in this area. This result demonstrates the potential of using an electronic nose for smart home monitoring.