MFA Multi‑Factor Authentication has become one of the crucial security methods to fight against cyber threats. This paper discusses the MFA techniques and draws inspiration from the previous existing papers, reports, and cyber security frameworks which focus on the various methods of authentication like biometrics, security keys, etc. The traditional methods are based on the static authentic layers, which are vulnerable to threats. This paper proposes a Dynamic Threat Level Detection System to continuously monitor user behaviour and change the authentication levels based on the irregularities detected. The threat level ranges from 0 to 5. The level increases in the response to suspicious activities. A Split One‑Time Password (OTP) Mechanism is integrated into the proposed work where the six‑digit OTP is divided into two and is sent separately to the mail and the user's phone number which prevents the attackers from intercepting the OTP. At higher levels, a fake password interface is designed where the user must enter the hidden original window to enter the password; this process is only known to the user. An extra security method, i.e., the Song‑Based Authentication method is introduced at the highest level where the users select a particular segment of a preloaded song. If the user passes all the levels, a biometric verification is performed, and if any wrong input is detected, the system shuts down immediately to prevent access to unauthorised people. This method enhances the resistance to phishing and brute force methods. The research has further integrated AI for anomaly detection and blockchain‑based process authentication which provide a defensive mechanism against fast‑improving cyber threats.
Automatic power‑operated lift doors are a major source of lift malfunctions, yet practical predictive maintenance solutions remain limited. This study proposes and validates a non‑intrusive, real‑time monitoring framework that integrates vibration signature analysis with Internet of Things (IoT) and Artificial Intelligence (AI) technologies for predictive maintenance of lift entrance systems. Two 3‑axis accelerometers were installed on the motor‑driven car door panels of a 30‑year‑old freight lift to continuously collect data over five months. Acceleration signals were analysed using time‑domain feature extraction and frequency‑domain Fourier Transform to identify characteristic mechanical frequencies. A physics-informed energy module numerically integrated acceleration to estimate door velocity and compute kinetic energy, enabling real‑time safety threshold evaluation. For automated anomaly detection, a Convolutional Neural Network (CNN)‑based DetectNet v2 framework with a ResNet‑18 backbone was trained to classify six operational conditions, including normal and collision events. The system achieved a mean Average Precision of 63.4% under real operational constraints. Unlike previous laboratory‑based studies, this field‑deployed, cloud‑integrated architecture combines vibration analysis, energy monitoring and deep learning in a unified predictive maintenance framework, demonstrating scalable and non‑intrusive monitoring for lift door systems.
An Intelligent Crowd Diversion System (ICDS) has been developed, utilising Artificial Intelligence (AI) and Machine Learning (ML) technologies to assist in crowd management at Kai Tak (KAT) and Sung Wong Toi (SUW) stations during mega events. The ICDS integrates real-time CCTV data and tailored AI-based video analytics to provide accurate people counting and crowd control metrics. The ICDS predicts real-time platform waiting times for dispersal based on statistics, including previous waiting times, crowd densities, and a unique formula for reliable measurement. Real-time updates are displayed on LCD panels for passenger guidance. A dashboard offers an overview of station busyness status, empowering station managers to make real-time adjustments and provide information for upcoming events during non-event days. The incorporation of AI technologies enables predictive measures for managing sudden increases in crowd volumes during largescale events. Data exchange with MTR Data Studio allows enhancing the comprehensiveness of databases and increasing the accuracy of predictions. Estimated waiting times benefit by keeping passengers informed with timely information, ensuring a smooth flow of passengers and improving the mobility efficiency. With the implementation of innovative crowd diversion, MTR can replace traditional manual crowd control management with the ICDS and become a modern railway system in the world.
The Hongkong Electric Co. Ltd. (HK Electric) has a long history of integrating Geographic Information System (GIS) technology into its operations, starting with its first computer mapping system in the 1980s. Over the years, HK Electric has transitioned from a Computer-Aided Design and Drafting (CADD)-based system to an Automated Mapping and Facilities Management (AM/FM) system. In 2019, HK Electric initiated a significant transformation of its GIS platform to create robust and future-proof infrastructure. The implementation began in February 2021 and was completed by February 2024, following data verification and user acceptance tests. Key challenges included transitioning to a standardised commercial off-the-shelf (COTS) product, managing a lengthy migration process, ensuring data accuracy, and addressing performance issues. Solutions involved emphasising the importance of data models, automating data transformation, and leveraging the intrinsic design and data structure of the product. The project also focused on streamlining business processes and managing changes effectively. This paper shares the project team's experiences and solutions to benefit other companies embarking on similar journeys.
The Electrical and Mechanical Services Department (EMSD) is advancing its digital transformation from Building Information Modelling (BIM) to BIM-Asset Management (BIM-AM) to create a Digital Twin for building Operation and Maintenance (O&M). Since 2015, the organisation has been developing and customising its BIM-AM System for enhancing its O&M. In 2019, it launched the Government-Wide IoT Network (GWIN), dedicated IoT infrastructure for near real-time monitoring of Electrical and Mechanical (E&M) assets. To streamline O&M workflows, the organisation has also explored technologies like the Cave Automatic Virtual Environment (CAVE) and smart glasses, utilising Virtual Reality (VR) and Augmented Reality (AR). Since 2020, the organisation has initiated a government-wide asset digitalisation exercise, covering over 450 government venues to support smart building initiatives. Additionally, the organisation has aimed to establish the first on-premises Common Data Environment (CDE) compliant with ISO 19650 for the Government of the HKSAR, alongside the development of an Asset Information Management Platform (AIMP) and Virtual Desktop Infrastructure (VDI) to facilitate BIM-AM adoption. This paper will discuss the integration of these technologies in creating a Digital Twin for effective building O&M.
