Design And Simulated Implementation Of Mobility Prediction-Based Resource Reservation Mechanism In NGI

With the rapid development of the network technologies and applications, Internet has penetrated into people’s work, study and other aspects of life, and people’s requirements of communication are growing fast. It has become an urgent demand for users to enjoy the network service with high quality at any time and any place in NGI (Next Generation Internet).Therefore, NGI must support ABC (Always Best Connected) and provide QoS (Quality of Service) guarantee for all users. In NGI, handoff caused by users’movement may lead to interruptions of users’communication and lower users’QoS. In order to ensure users’ communication QoS, two kinds of mobility prediction-based resource reservation mechanism are designed in this thesis.Both of the resource reservation mechanisms are divided into three stages:mobility prediction, resource reservation and QoS negotiation. At the first stage, the prediction operations will be started periodically after the user has connected the current access network, if the necessary reservation condition is met, the most likely target access networks will be predicted for the user. The main work at the second stage is to reserve resource from the identified target access networks. According to the heterogeneity of the NGI network model, two types of reservation are considered in this thesis, which are horizontal reservation and vertical reservation. The third stage is optional, if most of the resource reservation procedures are failure, the procedure of QoS negotiation must be deployed. The two kinds of resource reservation mechanism are different between the prediction methods of reservation target access networks. One mechanism firstly records the user’s historical mobility information (speed, direction) using different queues with different time granularity, and then an area which covers all spots that the user is going to move to with a high possibility is predicted combining with the user’s immediate mobility information (position, speed, direction), and the target access networks are identified. The other mechanism uses user’s access networks sequences instead of user’s mobility trajectory information combining with temporal information, and user’s most likely reservation target access networks are predicted using the user’s current mobility trajectory having one kind of regularity.In order to verify the feasibility and effectiveness of the two kinds of resource reservation mechanism, simulated implementation is done and the performances are analyzed based on NS2 (Network Simulator version2). The simulation results show that the two kinds of resource reservation mechanism are feasible and effective. They can effectively deal with the inaccuracy of network information and the fuzziness of user’s requirement, guarantee the users’QoS requirements with more adaptability, improve access networks resources utilization and reduce the call blocking probabilities and session dropping probabilities of the access networks.