Temperature deviations in food cold storage can accelerate spoilage and compromise product safety. Although IoT-based monitoring systems offer continuous automated oversight, their operational reliability under variable network conditions remains insufficiently documented. This study evaluated the Arunika ColdWatch IoT monitoring system deployed at a food cold storage facility in Kota Palu, Indonesia. A 3 × 2 factorial experiment with 10 replicates (60 trials) was conducted from April 2–16, 2026, across three temperature conditions (normal, upper deviation, lower deviation) and two network conditions (stable, disrupted), generating 10,800 total sensor records. Sensor bias ranged from −0.135°C to +0.211°C, with mean absolute error between 0.123°C and 0.233°C. Alert sensitivity was 1.000 under both network conditions; specificity was 0.767 (disrupted) and 0.667 (stable), reflecting false alerts from the highest-bias sensor. Transmission success was 99.1% (stable) versus 94.5% (disrupted; p < 0.001). Notification delivery within the 120-second deadline reached 96.7% under stable conditions but dropped to 0.0% under disruption, with median delay increasing from 5.1 s to 154.6 s (p < 0.001). The system reliably detected all temperature deviations but requires stable network connectivity for actionable notification timeliness. Alert threshold refinement to reduce false-positive rates is also recommended.
Copyrights © 2026