Mobility management in IP-Based Networks
Mobile communication networks experience a tremendous development clearly evident from the wide variety of new applications way beyond classical phone services. The tremendous success of the Internet along with the demand for always-on connectivity has triggered the development of All-IP mobile communication networks. Deploying these networks requires, however, overcoming many challenges. One of the main challenges is how to manage the mobility between cells connecting through an IP core in a way that satisfies real-time requirements. This challenge is the focus of this dissertation. This dissertation delivers an in-depth analysis of the mobility management issue in IP-based mobile communication networks. The advantages and disadvantages of various concepts for mobility management in different layers of the TCP/IP protocol stack are investigated. In addition, a classification and brief description of well-known mobility approaches for each layer are provided. The analysis concludes that network layer mobility management solutions seem to be best suited to satisfy the requirements of future All-IP networks. The dissertation, therefore, provides a comprehensive review of network layer mobility management protocols along with a discussion of their pros and cons. Analyses of previous work in this area show that the proposed techniques attempt to improve the performance by making constraints either on access networks (e.g. requiring a hierarchical topology, introducing of intermediate nodes, etc.) or mobile terminals (e.g. undertaking many measurements, location tracking, etc.). Therefore, a new technique is required that completes handoffs quickly without affecting the end-to-end performance of ongoing applications. In addition, it should place restrictions neither on access networks nor on mobiles. To meet these requirements, a new solution named Mobile IP Fast Authentication protocol (MIFA) is proposed. MIFA provides seamless mobility and advances the state of the art. It utilizes the fact that mobiles movements are limited to a small set of neighboring subnets. Thus, contacting these neighbors and providing them in advance with sufficient data related to the mobiles enable them to fast re-authenticate the mobiles after the handoff. The dissertation specifies the proposal for both IPv4 and IPv6. The specification of MIFA considers including many error recovery mechanisms to cover the most likely failures. Security considerations are studied carefully as well. MIFA does not make any restrictions on the network topology. It makes use of layer 2 information to optimize the performance and works well even if such information is not available.In order to analyze our new proposal in comparison to a wide range of well-known mobility management protocols, this dissertation proposes a generic mathematical model that supports the evaluation of figures such as average handoff latency, average number of dropped packets, location update cost and packet delivery cost. The generic model considers dropped control messages and takes different network topologies and mobility scenarios into account. This dissertation also validates the generic mathematical model by comparing its results to simulation results as well as results of real testbeds under the same assumptions. The validation proves that the generic model delivers an accurate evaluation of the performance in low-loaded networks. The accuracy of the model remains acceptable even under high loads. The validation also shows that simulation results lie in a range of 23 %, while results of real testbeds lie in a range of 30 % of the generic model?s results. To simplify the analysis using the generic mathematical model, 4 new tools are developed in the scope of this work. They automate the parameterization of mobility protocols, network topologies and mobility scenarios. This dissertation also evaluates the new proposal in comparison to well-known approaches (e.g. Mobile IP, Handoff-Aware Wireless Access Internet Infrastructure (HAWAII), etc.) by means of the generic mathematical model as well as simulation studies modeled in the Network Simulator 2. The evaluation shows that MIFA is a very fast protocol. It outperforms all studied protocols with respect to the handoff latency and number of dropped packets per handoff. MIFA is suitable for low as well as high speeds. Moreover, there is no significant impact of the network topology on its performance. A main advantage of MIFA is its robustness against the dropping of control messages. It remains able to achieve seamless handoffs even if a dropping occurs. The performance improvement is achieved, however, at the cost of introducing new control messages mainly to distribute data concerning mobile terminals to neighbor subnets. This results in more location update cost than that resulting from the other mobility management protocols studied. Due to excluding any constraints on the network topology, MIFA generates the same packet delivery cost as Mobile IP and less than other protocols.An additional focus of this dissertation is the development of an adaptive eLearning environment that personalizes eLearning contents conveying the topics of this dissertation depending on users? characteristics. The goal is to allow researchers to quickly become involved in research on mobility management, while learners such as students are able to gain information on the topics without excess detail. Analyses of existing eLearning environments show a lack of adaptivity support. Existing environments focus mainly on adapting either the navigation or the presentation of contents depending on one or more selected users? characteristics. There is no environment that supports both simultaneously. In addition, many user characteristics are disregarded during the adaptivity process. Thus, there is a need to develop a new adaptive eLearning environment able to eliminate these drawbacks. This dissertation, therefore, designs a new Metadata-driven Adaptive eLearning Environment (MAeLE). MAeLE generates personalized eLearning courses along with building an adequate navigation at run-time. Adaptivity depends mainly on providing contents with their describing metadata, which are stored in a separate database, thus enabling reusing of eLearning contents. The relation between the metadata that describe contents and those describing learners are defined accurately, which enables a dynamic building of personalized courses at run-time. A prototype for MAeLE is provided in this dissertation as well.