Randomized Forwarding Approach for Reliability Support in Distributed Mobile System

Chunhung Richard LIN  

IEICE TRANSACTIONS on Communications   Vol.E87-B    No.3    pp.705-716
Publication Date: 2004/03/01
Online ISSN: 
Print ISSN: 0916-8516
Type of Manuscript: PAPER
Category: Network
multicast,  mobile networks,  reliable,  mobile IP,  

Full Text: PDF>>
Buy this Article

IP multicast is an efficient means of sending to a group, but the packets are sent unreliably. Mobility complicates the problem because many multicast protocols are inefficient when faced with frequent membership or location change. In this paper, we propose a new protocol to additionally achieve fault recovery of multicast applications in IP internetwork with mobile participants. Unlike many studies which use the basic unicast routing capability of Mobile IP as the foundation, our protocol is built on top of the existing static hosts IP unicast and multicast forwarding services to avoid triangle routing which always occurs in Mobile IP. Relying only on the existing multicast service model and reconstructing the delivery tree every time a multicast member and/or source move is not always a good solution. By applying the ideas of bi-directional tunneled multicast, our protocol attempts to hide host mobility from all other members of the group. Therefore, the multicast distribution tree will not be updated for the sake of member location change. Furthermore, our protocol has near shortest delivery paths like remote subscription protocol. Exploiting the randomized forwarding service called randomcast in the repair process for packet losses, our protocol achieves local recovery and improves robustness. Additionally, our system structure can minimize the request implosion and duplicate replies. Simulation results show that our protocol has the distinct performance advantages in local recovery and robustness by using randomcast. Our protocol can also adapt to the fluctuation of both host movement and the number of mobile members (i.e., having mobility and scalability properties).