The mass-market adoption of high-end mobile devices and increasing amount of video traffic has led the mobile operators to adopt various solutions to help them cope with the explosion of mobile broadband data traffic, while ensuring high Quality of Service (QoS) levels to their services. Deploying small-cell base stations within the existing macro-cellular networks and offloading traffic from the large macro-cells to the small cells is seen as a promising solution to increase capacity and improve network performance at low cost. Parallel use of diverse technologies is also employed. The result is a heterogeneous network environment (HetNets), part of the next generation network deployments. In this context, this thesis makes a step forward towards the “Always Best Experience” paradigm, which considers mobile users seamlessly roaming in the HetNets environment. Supporting ubiquitous connectivity and enabling very good quality of rich mobile services anywhere and anytime is highly challenging, mostly due to the heterogeneity of the selection criteria, such as: application requirements (e.g., voice, video, data, etc.); different device types and with various capabilities (e.g., smartphones, netbooks, laptops, etc.); multiple overlapping networks using diverse technologies (e.g., Wireless Local Area Networks (IEEE 802.11), Cellular Networks Long Term Evolution (LTE), etc.) and different user preferences. In fact, the mobile users are facing a complex decision when they need to dynamically select the best value network to connect to in order to get the “Always Best Experience”. This thesis presents three major contributions to solve the problem described above: 1) The Location-based Network Prediction mechanism in heterogeneous wireless networks (LNP) provides a shortlist of best available networks to the mobile user based on his location, history record and routing plan; 2) Reputation-oriented Access Network Selection mechanism (RANS) selects the best reputation network from the available networks for the mobile user based on the best trade-off between QoS, energy consumptions and monetary cost. The network reputation is defined based on previous user-network interaction, and consequent user experience with the network. 3) Network Reputation-based Quality Optimization of Video Delivery in heterogeneous networks (NRQOVD) makes use of a reputation mechanism to enhance the video content quality via multipath delivery or delivery adaptation.