Dynamic adaptive content delivery using genetic algorithm

Abstract

In this thesis a new framework for dynamic adaptive content delivery is presented, which is suitable for diversified mobile devices. The proposed framework can dynamically adapt itself for diversified web contents available at the numerous content delivery sites around the globe. Our approach differs form previous works as it is not only based on adapting single type of content in static predefined way, but also capable to adapt multiple types of content dynamically on population changes. Every type of content is different from the others by different attributes they have and even different attribute values. The adaptive content delivery problem considered here is an NP hard problem with exponential time complexity. We introduce Genetic Algorithm for the dynamic learning at the initial phase and at the time when the environment changes due to introduction of new clients. In the proposed framework the Dynamic Content Adaptation has been established by using Genetic Algorithm to identify the Majority Supported Capability Set at the leaming engine in the learning phase using the information from client historical base. The current client environment can be easily identified using the client historical base information and the change in the client environment can also be identified in real-time. We show that Dynamic Adaptive Content Delivery (DACD) can minimize the limitations of existing content adaptation techniques and also add new scope to the current research directions. The framework is verified using real telecom network data with help of WURFL repository. Results indicate that the DACD framework can efficiently identify the MSCS which can deliver content that closely matches the capability of the population and reduces the variety of content significantly. The proposed framework has been compared with the existing research on content adaptation. It is found that the solution of the proposed framework performs better in terms of real-time content adaptation capability and maximization of server resource utilization.