ISSN 2738-0971 | eISSN 2738-1013

Bojan Jovanović

The Academy of Applied Studies of Kosovo and Metohija, Dositeja Obradovića bb, 38218 Leposavić, Serbia

Articles

Open Access Original Scientific Paper

A MODULAR IOT PLATFORM FOR NEXT-GENERATION SMART HOMES: ARCHITECTURE, REAL-TIME CONTROL, AND EDGE AI READINESS

Rapid technological progress and increasingly pressing needs for energy efficiency, safety, and personalised comfort have driven the development of intelligent systems for residential automation. This paper presents the design and implementation of a modular IoT smart-home system based on a microcontroller architecture with real-time data processing. The developed prototype integrates sensor modules for detecting temperature, humidity, air quality, illuminance, vibration, precipitation, and flame, as well as actuators for automated control of windows, doors, lighting, ventilation, and alarm mechanisms. The system is connected to a mobile application that enables monitoring and interactive control in real time, and users can define scenarios such as “night mode” or “away mode”. Special emphasis in the design is placed on the system’s modularity, its energy optimisation, and the ability to adapt behaviour based on historical data and user habits. The system’s functionality was tested on a physical model and in real conditions, establishing that it reacts within a time window of 1–3 seconds from the moment a change in environmental parameters is detected. The obtained results indicate significant potential for integrating microcontrollers, an IoT platform, and adaptive control algorithms in the domain of smart buildings and future concepts of urban automation. The paper also opens up avenues for further development with integrated machine-learning and artificial-intelligence algorithms aimed at achieving fully autonomous control of the residential environment. This iteration includes a fully functional physical prototype and application, while the predictive AI part is evaluated offline via simulation/emulation based on recorded logs, without on-device inference.

Open Access Original Scientific Paper

HYBRID BLE–PDR LOCALIZATION SYSTEM FOR SMART RETAIL ENVIRONMENTS

Accurate indoor user localization is a key component in the development of cashier-free smart stores, enabling advanced customer experiences, security monitoring, and behavioral flow analysis. This paper presents a hybrid localization approach that combines inertial motion tracking (Pedestrian Dead Reckoning – PDR) with Bluetooth Low Energy (BLE) tag signals. Unlike systems that require dedicated infrastructure, the proposed solution uses existing BLE electronic shelf labels (ESLs) as reference points. Their identification signals are used to correct the PDR drift, thereby reducing the cumulative error typical of purely inertial methods. The mobile application continuously measures the Received Signal Strength Indicator (RSSI) from BLE tags and applies threshold-based position corrections, while the PDR module, based on the Scarlett step-length model, maintains continuous tracking between tags. The system additionally integrates map-based spatial constraints that eliminate physically impossible paths through a particle-filter mechanism. Experimental evaluation in a retail environment demonstrated an average error of 0.4 m for linear movement and 1.3 m for a complex circular trajectory, confirming meter-level localization accuracy without the need for cloud processing. All computations are performed locally on the user’s device, ensuring privacy protection and enabling real-time movement analysis and context-aware retail interaction.