The Networking Area of the Telecommunication Group at the New University of Lisbon Paulo Pinto and Luis Bernardo Faculdade de Ciências e Tecnologia, 2829-516 Caparica, Portugal Phone: +351-212948545, Fax: +351-212948532, e-mail: {pfp,lflb}@uninova.pt Abstract — This paper describes the main activities of one out of three areas of the Telecommunication Group at the New University of Lisbon – the Networking Area. Our area of intervention is the middleware level of distributed systems. Highly populated systems are the object of research. Mobility and wireless systems is the other main area. The focus in this second area is on management of mobility at middleware and application levels. However, in the effort to build an entire architecture we have proposed a new interworking architecture for future wireless systems. The paper begins with our view of the main driving ideas that will be important in the future systems in our area of intervention. The second part of the paper describes our current research and some near future directions. I. INTRODUCTION The Telecommunication Group of the New University of Lisbon and UNINOVA [1] is divided in three main areas: networking; signal processing; and propagation and radiation. This paper provides an overview of the activities of the networking area of the Group. II. A THREAD OF ACTION The current lines of research follow the two main activities the senior members had in the recent past. One of them is Highly Populated Systems, and the other is Internet Technologies. Each line has a follow up but the intersection of both is also a challenging scenario. Prior to a description of the activities, it might be interesting to understand the motivation that led us to our current work. A. Highly Populated Systems If we consider a system with millions of active users and thousands of applications servers, we reach very quickly the conclusion that most of the algorithms we use today will not work [2]. Network protocols and service (and application) structures will also fail. Examples are very easy to find: • Each time a minor percentage of the users decide to perform a certain task there is congestion. Imagine accessing an on-line journal web site after a catastrophe, of placing a bet in the lotto just a few minutes before the deadline. • Servers can become so crowded that new approaches have to be taken to solve the problem (for instance, disappear to force users to locate another server). • The amount of information could be so large that managing a hierarchical structure to hold it can be prohibitive (both in terms of placing new information or updating the existent one). We have found that with such systems the concept of mobility just appeared naturally for several reasons [3]: either because new servers have to be placed in the system to raise the response power to users; or because servers cease to exist because they were so overcrowded that it was better for users to find another one. Another relevant aspect is the management of information. In very large systems, the amount of information can be so large that it becomes impossible to have absolute knowledge of the system for certain issues. The amount of information and the update requirements can be prohibitive. If we cannot have absolute information, then some kind of proximity must exist. Information starts to be managed by some metrics (distance, age, etc.) and the larger the value of the metric the less users know about it. For instance, when I am driving towards the Lisbon city center it is irrelevant to me how many free places are there in one of Tokyo’s parking lots. This problem of information management is even worse if we let servers to move (or to disappear and appear elsewhere) because updates are hard to perform. Very large information systems nowadays, such as DNS (Domain Name System), work on quasi-static information and partition the information using pathnames. Highly populated systems can only be feasible (meaning efficient) if we consider a certain level of uncertainty in the algorithms and service structures [4]. This line of research fits nicely in the current problems of ad-hoc and sensor networks. B. Internet Technologies One of the successes of the Internet (although many can call it a fault) was to ignore the way users are connected to the Internet. The physical and logical layers were always a dark subject. The important issue was the ability to send and receive IP packets… A parallel can be drawn with the recent research towards wireless 4G systems. One of the characteristics of 4G systems is the possibility of using one of several radio access networks (RAN) to access the core and the services. Probably the secret in the future 4G architecture might be the