Authors: Om Barbude, Professor Abhishek P. Bangar, Professor Sachin B. Bhosale, Dr. A. A. Khatri
Abstract: Milk can go wrong at almost any point between the cow and the consumer. Natural variation in composition, bac-terial growth, rough handling and deliberate adulteration all leave their mark on the final product. Laboratory methods such as spectroscopy, chromatography and microbiological testing are still the reference, but most farms and small collection centres cannot run them for every can of milk. Internet of Things (IoT) sensing is often sug-gested as a cheap first check that fills this gap. In this paper we review the verified literature on IoT-based milk quality monitoring. We look at four connected topics: what quality indicators and adulteration involve, which sen-sors (pH, electrical conductivity, temperature, turbidity, optical and gas) are available, how an ESP32-class con-troller and a cloud dashboard can collect and share the readings, and how machine learning can combine several weak signals into one screening decision. The studies we found show that working prototypes are possible, but they say little about long-term field use. Cross-sensitivity, calibration drift, differences between milk batches, unreliable reference labels and cost remain open. A sensor reading can flag a sample, but it cannot name a chemical, so la-boratory confirmation is still needed. We also compare the representative studies, propose a conceptual six-layer architecture with the sensors chosen for a student prototype, and list research gaps in order of priority. No new experimental results are reported.
