Architectural considerations for managing mobility

Ever since the first communication systems were deployed, requirements for mobility have been one of the most important causes of architectural changes. For instance, it has taken a long period of research and standardisation to arrive at the deployment of mobile telephony, the architecture of which is very different from that of the traditional/fixed telephony. Similarly, in the Internet, the proposed architecture of Mobile-IP adopts different methods of IP address allocation and routing, among other changes. On the other hand, the mobile system architectures introduced by the different communities, e.g. telephony, Internet, data management, nomadic computing, and wireless communications, all differ significantly. They provide different kinds of mobility, they adopt different approaches to managing mobility and each of them has certain architectural problems. For example, the problem of triangular routing in Mobile IP is an important drawback of its architecture. Existing mobile system architectures are restricted by their legacy. Also, their scope is confined to particular aspects of mobility. For these two reasons, these architectures are not generic, and they cannot be extended to support additional kinds of mobility. In this thesis, we firstly discuss the problems of existing mobile system architectures. Then, we abstract mobile systems to a generic level and we define the common denominator of mobile systems in a framework of terms. Using the terminology of this framework we examine the problem of managing mobility on an abstract level and we arrive at architectural requirements for mobile systems deployment, which are incorporated into the framework. These requirements for mobility are investigated from two different ODP viewpoints: the information viewpoint and the computational viewpoint. From the information viewpoint we justify the necessity for unique IDs for identifying mobile entities and we discuss the nature and the scope of these unique IDs. From the computational viewpoint we outline a number of entities that a system should feature in order to support mobility. The computational viewpoint requirements are specified following a design pattern template. The hypothesis that motivated this thesis is that a technology independent framework for mobile systems is feasible. Also, that it can be generic enough to represent a variety of mobile systems and that at the same time it can be of practical value. Finally, there is a claim that such a generic and technology independent framework for mobile systems can help to consider many aspects of mobility and therefore assist to building coherent systems that avoid pitfalls common in existing mobile system architectures. In order to investigate the validity of this hypothesis a framework of terms for mobile systems is specified. Using the terminology that this framework provides architectural requirements for mobility are investigated from the information and the computational ODP viewpoints. To test their generality, the terms of this framework are mapped to a range of real systems such as Mobile IP and GSM. A system for the novel area of role mobility that respected the specified requirements of the framework was deployed in a TINA environment in order to investigate the practical value of this framework and the requirements that are expressed in its terms. The design and implementation of the role mobility system is presented in this thesis. Additional to this, this framework was used to investigate some well-known problems of existing mobile systems like GSM and Mobile IP. In this way, the practical value of the framework and the architectural requirements that where expressed in its terms are validated further and the claim that this framework can assist in building coherent systems is substantiated. This thesis, therefore, is intended to provide an insight into the problem of mobility and further directions to carry this research forward are suggested.