Degree Name

Doctor of Philosophy


School of Information Systems and Technology


Public transport contributes greatly to people’s mobility. Intelligent Transportation Systems (ITS) are used to provide safer and better transport services to ensure people’s mobility. Digital ecosystems, analogous to natural ecosystems, are self-evolving and self-sustaining.

A small scale ITS system that bears the characteristics of a digital ecosystem was designed and developed in this PhD study.

Based on the Internet Engineering Task Force presence model, the system was designed with three discrete but inter-connected components: an on-board device installed on a distributed network of shuttle buses; a central server; and a mobile application that allows users to access the system on the go. Using a combination of different communication protocols (e.g. HTTP, WSDL, SOAP, SSL), the system allows effective communication among the three components and ensures secured communication when needed. Using the Service Oriented Architecture (SOA) principle, the system software functions are designed to provide a stratum of services to the system, with the central layer of services designed to serve the other Connected Mobility Digital Ecosystem applications.

Passengers’ travel patterns, shuttle bus travelling time on road, dwelling time at bus stops, and schedule adherence data was gathered and analysed. A historical data based prediction model was developed using an algorithm based on the historical travel data combined with the real-time bus travelling information. It incorporates variables of realtime bus travel speed, route distance, historical travelling time, and historical dwelling time at each stop before reaching a particular bus stop. The prediction accuracy of travel time by using historical data based model and linear regression model is evaluated using the Root Mean Square Error (RMSE) method in comparison with the prescribed bus timetable. The evaluation results show clearly that the model based predictions of travel time from both linear regression and historical data based models show a significant advantage over using timetable. Furthermore, the prediction accuracy from the historical data based model consistently outperformed that using linear regression model in all the conditions, because the historical data based model has taken into account the dwelling time to improve the prediction accuracy while the linear regression model does not include the effect of dwelling time in its model.

The system collects data from its distributed sensors (i.e. network of shuttle buses equipped with the on-board devices, and the passengers who use the app through smartphones), performs data analysis and predicts bus arrival time. While passengers utilise the system to better plan their travel, the system identifies the repeatable passenger usage patterns, which allows the dynamic scheduling of buses that optimises throughput. This process continues and forms an on-going digital ecosystem cycle, which allows better system performance and management. This system behaviour embodies the CMDE’s self-evolving, self-sustaining and mutual benefit design philosophy.

An extensive evaluation study was conducted to assess the system’s usability and usefulness. Results showed that passengers highly value the system, with high ratings on the information sufficiency, accuracy, usefulness, relevance, and ease of use. The app’s main function, prediction of bus arrival time, attracted the most use. Most passengers believed the app provided them with the benefits of being able to plan ahead for a trip and elimination of the likelihood to miss a bus.