Cold Storage Monitoring System Development iCold Case Study

iCold is the international version of the mobile application developed by us. In contrast with a single sensor probe, this solution is implemented in the form of an entire cold room instrumentation package, consisting of a control panel and electrical cabinet on site, a sensor network up to 15 sensors inside the cold room, a cloud-based management system that carries out equipment health monitoring using AI algorithms, and a mobile app featuring remote monitoring, energy management, intelligent maintenance, and risk alerting.

The technical design of the above product will be reverse engineered using specifications that are available publicly and by analyzing the behavior of the platform.

In instances where there is a particular component whose exact part number is not mentioned in publicly available literature, then the electrical and functional limitations will help determine what silicon family must be in use.

The FLAND system follows a clean four-layer topology:

Layer	Function	Observable evidence
Sensing layer	Distributed temperature, humidity, and equipment-status measurement inside the cold room	15-channel sensor array documented across multiple product descriptions
Control layer	Electrical cabinet with digital display, relay outputs for compressor/defrost/evaporator, parameter setting	Cold-storage electrical control box with digital tube display, −100 °C to +100 °C control range
Connectivity layer	Uplink from the control panel to the cloud platform	OTA-capable, cloud-connected management platform
Application layer	Mobile app and cloud dashboards for monitoring, control, alerts, and AI diagnosis	FLAND iCold app features and platform description

It is intentional to have this separation: the control panel is in charge of deterministic refrigeration control (responsible for the compressor cycling, defrost timing, alarm threshold) while the cloud adds on the analytics, remote visibility, and multi-site management.

Thus, the app does not serve as a replacement for the control loop it rather serves as an extension of it.

3.1 Sensor Topology and Measurement Chain

The 15-channel sensor array is the major hardware part of the FLAND device. It is clearly indicated in all manuals that a 15-channel array is to be used for cold storage monitoring and hot warning temperature fluctuations. It is not a localized sensor, but a distributed sensing network allowing temperature distribution detection for finding hot spots near the door, evaporator coils, and product stacks.

The sensors are NTC thermistors with 10 kΩ at 25 °C and a B-value of 3435 K. It is a cheap and well-known sensor for cold storage applications.

The measurement range is stated as −49 °C to +199 °C, with the temperature resolution of 0.1 °C / 1 °C and the accuracy of ±1 °C at temperatures from −20 °C to +50 °C which enables the equipment to be used within the most typical temperatures for cold and frozen storage. NTC sensors can be linearized with the help of the controller firmware according to the Steinhart–Hart or B-parameter equations, or using a specialized signal-conditioning frontend equipment.

There is also a sensor fault warning in case there is a problem with the sensor operation and the system can distinguish between a faulty sensor and an actual temperature excursion.

The set of sensors is not limited to just temperature. The documentation for the platform mentions temperature, humidity, equipment status, and energy use; in addition, humidity measurements are cited along with temperature ones in the description of deployments.

3.2 Control Panel and MCU Class

Edge controller is an electrical control box for cold storage with a digital tube display and control range from -100°C to +100°C.

The controller is supplied with power from 110-480 V AC (configurable on site) and produces output relays for the compressor, defrost heaters, evaporator fans, and alarm circuits.

In some configurations of FLAND cold room, the automatic electric controller is a PLC (Programmable Logic Controller) or a refrigeration controller board.

The exact MCU part number is not published. The functional and electrical requirements, however, constrain the silicon class tightly:

Multi-channel analogue input

Fifteen NTC channels will need either a large ADC multiplexer in the ADC or an external one connected to a mid-range ADC. A minimum 12-bit ADC would be needed to discriminate 0.1 °C temperature in the cold room environment.

Relays and digital IO

The relay driver circuitry needed to drive the compressor, defrost cycle, fan control, alarm signal, and door open condition will require multiple relays and digital opto-couplers for IO.

Real-time deterministic control

Defrost timing, compressor minimum off-time, and delay logic for alarms would be done in a real-time control loop in a firmware supervisor, not in a generic Linux application.

Display and user interface

The digital display and setpoint adjustment interface implies a menu driven firmware supervisor with non-volatile memory.

Communication

The OTX functionality and remote monitoring requires a UART for Wi-Fi/Cellular Modem and enough flash for firmware images.

These demands imply that the system is likely to be composed of an industrial 8/16/32-bit microcontroller or a small PLC-type controller with onboard multi-channel ADC, relay drivers, UART, and non-volatile storage. In the Chinese refrigeration control industry, this category is usually implemented by MCUs from companies like STM32 from STMicroelectronics, NXP, or local manufacturers, coupled with an ESP8266/ESP32-type Wi-Fi chip or 4G Cat.1 transceiver for cloud uplink.

This type of design places emphasis on long-term availability, predictable execution, and isolation more than computational performance, same principles as those in use in European cold-room controllers.

3.3 Communications and OTA

In this case, the platform will provide support for OTA (Over-The-Air) upgrade of the refrigeration system firmware. This is a key manufacturing capability, which implies that the edge device has a bootloader and an image check & rollback feature, and that the cloud can update control parameters or patch the firmware without visiting the facility.

Uplink communication will most probably be via Wi-Fi (802.11 b/g/n, 2.4 GHz) for indoor cold rooms, and 4G Cat.1 for remote/mobile cold storage, with the modem connected to the controller MCU via UART using AT commands, or MQTT/TLS protocol stack.

The app's real-time monitoring and remote parameter-setting functions require a persistent or periodically polled bidirectional session between the cloud and the edge device.

The cloud layer is explicitly positioned as an IoT + big data + cloud computing + AI platform. Its functional decomposition is visible in the app's feature map:

4.1 Telemetry Ingestion and Device Management

The system receives measurements in time-series form, signals for equipment status, and energy consumption statistics from the 15-sensor network.

Device management features remote configuration capabilities, meaning that the parameters such as temperature set points, defrost cycle times, and alarm thresholds can be adjusted from the app, implying that the cloud has a command line to the edge device, rather than just a data reading connection.

The multi-device lists and advanced filters have been introduced in the app's release history.

4.2 Alarm Engine and Four-Channel Notification

The alarm system is implemented in the form of a four-way alert system, which includes SMS, WeChat, push notifications from mobile application, and email notifications.

This redundancy is required by the cold chain protocol, as it will ensure that SMS alerts will go through to the people who do not have access to smartphones, WeChat will cover the dominant Chinese social media ecosystem, push notifications will provide all necessary information, and the email will create an audit trail.

4.3 AI-Powered Equipment Health Diagnosis

Unique selling proposition of the solution is the artificial intelligence-based equipment health analysis. The description of the product on Fulande's website mentions a solution that detects any deviations in the cold room, makes intelligent analysis and early prediction of possible risks and remote diagnostic of faults. Functionally speaking, it implies the following correlation performed by the cloud:

• Deterioration of the compressor or refrigerant leakage - indicated by declining rates of cooling or changes in the cycle duration.
• Broken defrost system - indicated by unusual temperatures after defrosting.
• Issues related to door management - indicated by recurring temperature increases that cannot be attributed to machine faults.
• Sensor problems - revealed by inconsistencies across channels and comparison with redundant channels.

The ECO energy-efficient control function operates next to the AI layer: the system modifies its defrost and set point strategies according to the load patterns, aiming at lower operating costs and maintaining product temperature.

5.1 Architecture and Platform

FLAND iCold app (package com.fland.coldstorageforeign) is available on Google Play and App Store, where there is an additional version of the app for the domestic market for Chinese users. The app is available in Simplified Chinese, Traditional Chinese, English, and Arabic languages. For iOS, it should be noted that iOS version 15.6 and above are required. The size of the app is 176-183 MB which correlates with React Native/Flutter.

The functions of the mobile application correspond to the functionality of the platform as follows:

Device list and advanced filtering

Multi-device filtering, which is a recently developed function.

Real-time dashboard

Temperature, humidity, device status, and energy usage, along with historic data lists and curve analysis.

Remote control

Parameter setting and refrigeration/defrosting states controlling.

Alarm box

Mobile application endpoint for the notification system which consists of four channels.

Risk warnings and intelligent maintenance

Proactive notifications and maintenance scheduling.

Untended vehicle control

The recently added function.

5.2 Internationalisation and Data Posture

The coldstorageforeign library being separately packaged suggests that this product is intentionally internationalized into English and Arabic versions along with the Chinese version to cater to markets in the Middle East and Southeast Asia, in addition to those using Chinese.

The mobile application is essentially a thin, cloud-based client and there is no trace of an offline-first local database or BLE-provisioning feature. All the heavy lifting – i.e., sensor data gathering, logic for controlling the equipment, alarm analysis – is done by the edge controller, while the application serves the role of the visibility and command interface.

Advantages. The FLAND device has the advantage of having the cold room considered as a system that is instrumented, instead of just having one sensing point.

With fifteen channels of NTC sensors, this system has more spatial coverage than any single-probe logging devices, and the separation between the deterministic control panel and cloud analytics helps to maintain integrity of control while also supporting remote management.

Four-channel alarm systems suit the Chinese market, and the inclusion of OTA update function makes it a feature that is present in high-end products but not in cheaper ones.

The diagnostic AI layer solves a real problem, the distinction of equipment degradation and normal defrost process and the ECO mode of control helps reduce operating costs.

Constraints and Open Issues. Although the accuracy of NTC to ±1 °C in the range of −20 °C to +50 °C is acceptable for commercial refrigerated storage but it is insufficient for achieving ±0.5 °C of traceable calibration required for controlled pharmaceutical cold chain; the latter should have calibrated sensor pathway and uncertainty budget.

It is a good example of how the cold chain intelligence system can be realized by a refrigeration equipment producer: an NTC sensor network of 15 channels, a control cabinet containing relay output and local alarms, devices for connectivity providing OTA feature, a cloud platform with diagnostics and energy management with the help of AI, and a mobile application developed for operators in catering, food industry, and cold chain transportation.

It was not possible to identify the exact type of MCU, but it is clear from the stated requirements analogue acquisition on multiple channels, deterministic relay control, display and menu management, cloud communication via UART, and OTA with roll-back functionality that the MCU falls into the category of industrial microcontrollers / compact PLCs.

The principal engineering lesson is one of layered responsibility: the edge guarantees control and local alarm integrity, the cloud adds analytics and multi-site visibility, and the app delivers that intelligence to the operator without becoming a single point of failure for the cold room itself.

Live App: Google Play Store