E-learning has become an important service offered over the Internet. Lately many users are accessing learning content via wireless networks and using mobile devices. Most content is rich media-based and often puts significant pressure on the existing wireless networks in order to support high quality of delivery. In this context, offering a solution for improving user quality of experience when multimedia content is delivered over wireless networks is already a challenging task. Additionally, to support this for mobile e-learning over wireless LANs becomes even more difficult. If we want to increase the end-used perceived quality, we have to take into account the users’ individual set of characteristics. The fact that users have subjective opinions on the quality of a multimedia application can be used to increase their QoE by setting a minimum quality threshold below which the connection is considered to be undesired. Like this, the use of precious radio resources can be optimized in order to simultaneously satisfy an increased number of users. In this thesis a new user-oriented adaptive algorithm based on QOAS was designed and developed in order to address the user satisfaction problem. Simulations have been carried out with different adaptation schemes to compare the performances and benefits of the DQOAS mechanism. The simulation results are showing that using a dynamic stream granularity with a minimum threshold for the transmission rate, improves the overall quality of the multimedia delivery process, increasing the total number of satisfied users and the link utilization The good results obtained by the algorithm in IEEE 802.11 wireless environment, motivated the research about the utility of the newly proposed algorithm in another wireless environment, LTE. The study shows that DQOAS algorithm can obtain good results in terms of application perceived quality, when the considered application generates multiple streams. These results can be improved by using a new QoS parameters mapping scheme able to modify the streams’ priority and thus allowing the algorithms decisions to not be overridden by the systems’ scheduler.