A Decentralized Session Management Framework for Heterogeneous Ad-Hoc and Fixed Networks

Abstract

Wireless technologies are continuously evolving. Second generation cellular networks have gained worldwide acceptance. Wireless LANs are commonly deployed in corporations or university campuses, and their diffusion in public hotspots is growing. Third generation cellular systems are yet to affirm everywhere; still, there is an impressive amount of research ongoing for deploying beyond 3G systems. These new wireless technologies combine the characteristics of WLAN based and cellular networks to provide increased bandwidth. The common direction where all the efforts in wireless technologies are headed is towards an IP-based communication.

Telephony services have been the killer application for cellular systems; their evolution to packet-switched networks is a natural path. Effective IP telephony signaling protocols, such as the Session Initiation Protocol (SIP) and the H 323 protocol are needed to establish IP-based telephony sessions. However, IP telephony is just one service example of IP-based communication. IP-based multimedia sessions are expected to become popular and offer a wider range of communication capabilities than pure telephony.

In order to conjoin the advances of the future wireless technologies with the potential of IP-based multimedia communication, the next step would be to obtain ubiquitous communication capabilities. According to this vision, people must be able to communicate also when no support from an infrastructured network is available, needed or desired. In order to achieve ubiquitous communication, end devices must integrate all the capabilities necessary for IP-based distributed and decentralized communication. Such capabilities are currently missing. For example, it is not possible to utilize native IP telephony signaling protocols in a totally decentralized way.

This dissertation presents a solution for deploying the SIP protocol in a decentralized fashion without support of infrastructure servers. The proposed solution is mainly designed to fit the needs of decentralized mobile environments, and can be applied to small scale ad-hoc networks or also bigger networks with hundreds of nodes. A framework allowing discovery of SIP users in ad-hoc networks and the establishment of SIP sessions among them, in a fully distributed and secure way, is described and evaluated. Security support allows ad-hoc users to authenticate the sender of a message, and to verify the integrity of a received message.

The distributed session management framework has been extended in order to achieve interoperability with the Internet, and the native Internet applications. With limited extensions to the SIP protocol, we have designed and experimentally validated a SIP gateway allowing SIP signaling between ad-hoc networks with private addressing space and native SIP applications in the Internet. The design is completed by an application level relay that permits instant messaging sessions to be established in heterogeneous environments. The resulting framework constitutes a flexible and effective approach for the pervasive deployment of real time applications.

The invention of the phone has radically changed the way people communicate, as it allowed persons to get in contact instantly no matter of their location. However, phone communication has been confined for decades to a fixed location, be it one's own house or a phone boot. The widespread affirmation of cellular technologies has had for fixed telephony a similar impact that the invention of the phone has had on communications years before. With mobile phones, people are enabled to talk with each other anytime and anywhere.

Internet has also revolutionized the way people communicate. E-mails have soon become one of the Internet killer applications. Later on, instant messaging, popularly known as chatting, has gained huge consensus among net surfers. Only recently, the use of the Internet for voice communication is becoming mainstream, and the so called Voice over IP (VoIP) applications (Skype is probably the most famous for the masses) are becoming common use.

Despite its popularity, Internet still suffers from the inherent limitations that affected early telephony: it is fixed. The usage of Internet on the move still does not constitute the easiest and most satisfactory user experience, due to capabilities and limitations of the access technology, terminals, services and applications.

Efforts for mobilizing the Internet are ongoing both in the industrial and in the academic worlds, but several bricks are needed to build the wall of mobile Internet. This dissertation provides one of these bricks, describing a solution that allows the deployment of multimedia applications (chat, VoIP, gaming) in mobile environments. In other words, this dissertation gives solutions for facilitating ubiquitous Internet-based communication, anytime and anywhere.

The vision that we want to become true is that Internet must become mobile in the same way as fixed telephony has become mobile thanks to the cellular technology. More than this, we do not want that users are limited by the presence of an infrastructure to communicate with each other. In order to achieve this, we present solutions to deploy Internet-based services and applications in environments where no support from servers is available. In other words, we enable direct device-to-device, user-to-user Internet communication.

Our contribution is mainly focused on the steps needed to establish the communication, the so called session establishment or signaling phase. We have validated our signaling framework by building a chat application that utilizes its features and works in server-less environments.

The custom server-less solution does not prohibit to connect at the same time with the Internet, so that one can engage in a chess game using direct communication with a person in the proximity while having a chat in progress with a friend using standard Internet services. The challenge that we had to face is that Internet services and applications are usually built implying support from a centralized server. In order to deploy direct user-to-user Internet services, while maintaining interoperability with mainstream services, we had to enhance native Internet services to work without infrastructure support, without sacrificing interoperability with standard Internet applications.

To conclude, we have placed our brick on the still yet to be completed wall of mobile Internet. Our hope is that one day, thanks also to this brick, everybody will be able to enjoy Internet-based applications as easily as now it is possible to use mobile telephony services.