One of the key performance-limiting factors in achieving the Quality of Services (QoSs) in Vehicle-to-Infrastructure (V2I) communications over a heterogeneous network is the increased number of unnecessary handovers. In heterogeneous network environments, handovers consume a lot of network resources and consequently increase end-to-end network latency. Other factors affecting the handover performance are the traffic type and network loading. In this thesis, an efficient network selection method for handover is investigated. In the study, a multi-tier, multi-RATs heterogeneous network is considered. We first propose a novel algorithm for selecting the most suitable candidate network. We then propose a method for reducing the number of unnecessary handovers by considering the distance between the vehicle and the target Base Station (BS). The short-list of the potential target networks includes only those lying in the direction of movement and are determined based on the geo-location of both vehicle and the candidate network BSs. Certainly, the better network performance is achieved by reducing the scanning time of the candidate networks.

The surveyed literature has shown that the network load at the target network has a significant impact on the handover performance due to an increased handover drop probability. For example, a macro-cell BS could be regarded as a good candidate network for V2I communication due to the reasonable sojourn time, however, in the case, the BS is overloaded all handover requests are dropped. The integration of mobility and network information can further improve handover performance. A network selection method that applies the knowledge of both mobility and network load information is proposed for further improving handover performance. The better performance is achieved by optimizing the parameters such as cell dwelling time, load index, and Received Signal Strength (RSS) values. The type of traffic such as safety-related applications that require stringent QoS requirements and the network usage-cost also influence the network selection for handover. The concept of multi-criteria decision making which combines cell QoS metrics and user budget is investigated and applied for developing an application- aware network selection for handover method. Five application profiles are proposed based on the QoS metrics and network usage monetary cost. It was observed that the QoS of V2I safety-related applications could be guaranteed once they are allocated to the appropriate application profile.

The thesis includes three main contributions. First, the scanning time of all available networks in the direction of movement is reduced by geo-locating both the vehicle and candidate network base stations. Second, by getting the knowledge of network information and vehicle mobility we further improved the system performance. Third, the overall network performance/utilization is improved by considering traffic type and network usage-cost. The thesis concludes with network design guidelines and deployment strategies for various practical network scenarios.