رسائل دكتوراة
English
2022
A NEW CLUSTERING APPROACH FOR IMPROVING DATA DISSEMINATION IN VANET BASED ON MARKOV PROCESS
سرى جاسم محمد عبد الأمير, سعد طالب حسون الجبوري
كلية تكنولوجيا المعلومات-جامعة بابل · العراق
In the last few years, Vehicular Ad hoc Network (VANET) has opened wide new research directions. It represents the core of the Intelligent Transportation System (ITS). VANET helps in providing users with safety and infotainment. However, there are many issues in VANET, such as efficient data dissemination, network scalability, and stable network topology.
Firstly, to study VANET's behaviour and evaluate its performance, the clustering technique must be the first step to controlling its changeable topology and building a meaningful group of vehicles. A highway scenario is assumed in this dissertation. In the proposed clustering approach, the moving vehicles are organized into non-overlapping clusters. Specific criteria are suggested to select the cluster head. Generally, there are two strategies for cluster formation: distributed (Ad hoc) and centralized. In this dissertation’s proposed system, the centre-based clustering strategy is selected to manage and control the clustering process with the assistance of the Road Side Units (RSUs).
Secondly, the most important issue is to predict the future movement to build a reliable system and enhance both the communication in VANET and the Intelligent Transportation System. The efficiency of the clustering algorithm depends on the mathematical approach called the stochastic process. The joining and leaving of vehicles to the clusters can be predicted mathematically using a Markov process. But in this study, practical implementation is applied to practically fit the mathematical approach of the Markov concept. The number of vehicles in each cluster was observed at different sequential times to monitor the leaving and joining vehicles.
Finally, in VANET, information can be spread among vehicles that contain communication equipment. Each vehicle plays the role of the sender, router, or receiver. Warning messages can propagate among vehicles. Therefore, the process of modelling the data dissemination in VANET is also implemented in three types: broadcast, multicast, and unicast, and evaluated based on the proposed clustering approach in different scenarios.
The experimental results show that the proposed clustering algorithm’s performance was better than the Centre-Based Secure Clustering (CBSC) in terms of cluster stability and the number of created clusters, and the metrics indicate improvement in results with increasing network size. In data dissemination models, the sparse scenario with unicast data dissemination achieves better performance metrics than broadcast and multicast. The proposed data dissemination model’s performance was better than the Clustering and Probabilistic Broadcasting (CPB) algorithm in terms of delay and packet delivery ratio.
