Haptic Virtual Environment Performance Over Policy-based IP Networks: A case study zyxw Rima Tfaily Souayed, Dominique Gaiti zyxwvu Universitk de Technologies de Troyes I2 Rue Marie Curie, BP 2060 Guy Pujolle Universitk Pierre et Marie Curie 8 Rue du Capitaine Scott 7.5015 Paris, France Guy.pujolle Olip6.R zyx IO01 0 Troyes Cedex, France Qiitm. zyxwvutsr 7 zyxwvutsrqp yaiily, Dominique. Gaiti);@utt,fv Wai Yu, Alan Marshall Queen zyxwvu ’s University of Bevast School of Electrical zyxwvu & Electronic Eng. Stranmillis Road BT9 5AH- Belfast, UK A Iax.MarshaIlG3ee. qub.ac. uk K YwZlqub.ac.uk, Abstract Thispaper reports on the performance of Distributed Haptic Krtual Environment (DHPE) applications deployed over policy-based IP networks. Our previous work reported that a best eflort service in IP network is inadequate for satis&ing the Quality of Service (QoS) requirements of such applications. The current work tries tofind out whether applying differentiated Services (D@Serv, satisfies DHVE QoS requirements or at least improves them. This work also determines which policies should be applied on the network. A set of experiments was conducted over a network connecting two departments of the Queen’s University of Bevast. The results are presented in this paper and compared with our previous work results. Keywords Haptic Virtual Environment, PHANTOM, Force Feedback, QoS, DlfPjerv, IP Networks. I. Introduction Distributed Haptic Virtual Environments (DHVEs) [2] are systems that provide a user an illusion of sharing the same mediated space with other people who may be geographically distributed zyxwvuts all over the world. These environments support haptic devices that reinforce, in addition to video and audio, the user’s feeling of telepresence.DHVEs include training and simulation of telemedicine, teleroboticd teleopemtion, etc. Unlike non-interactive me&a (audio/ video media) which changes its state over time without correspondence to external events, DHVEs zyx are classified as interactive media where the system state changes in response to external events to the media or over time zyxwv [8]. This leads to more complex mechanisms for loss detection and reliability than those for non- interactive media. DHVE applications also have more critical networking requirements in terms of packet loss, delay, jitter and bandwidth. They may also require multicasting capabilities over the network, as well as a certain level of security according to the transmitted data type. Failure in satisfying these requirements may lead to destabilization of a control system, desynchronization of the environment between participants, or possibly to a disaster when it concerns teleoperation for example. Multicasting and security issues are part of our future work. The standard Intemet Protocol (IP) provides by default a best effort data delivery whch maintains simplicity in the core network and requires more complexity in the end-hosts. As the network becomes loaded, the service level degrades, and applications such as DHVE cannot adapt to inconsistent service levels. In order to provide an adequate and consistent service to such sensitive applications, qualitative/quantitative services must be implemented to distinguish M ic with strict requirements from other traffic. In other words, sensitive traffic must be prioritized. The most common 0-7803-8108-4/03/$17.00 0 2003 IEEE 71