Multicast Networking Research

The recent success of multicast applications such as Internet teleconferencing, distributed interactive simulation, and data dissemination applications illustrate the tremendous potential of applications built upon wide-area multicast communication services. Our ongoing research is focussed in three key areas of multicast networks: reliable multicast protocols, call admission in multicast networks, and multicast flow control.

Reliable Multicast Protocols

While some multicast applications (such as video and voice) do not require reliable data transfer, others (such as shared whiteboards and data dissemination) do. The requirement of reliable data transfer for this last set of applications poses a difficult challenge to network designers - how to design and implement a reliable multicast protocol that can handle 100s or 1000s of participants. We are currently pursuing several related efforts in the area of reliable multicast protocols.

Our work in [Tows97:Comparison] compared the throughput of a generic sender-oriented relaible multicast protocol and two receiver-oriented protocols. While our analyses demonstrated the general superiority of a receiver-based approach, it also demonstrated that the superiority of one approach over another depended subtly on the assumptions made about the multicast session (e.g., whether communication was 1-many or many-many). Our work in [Yama97:Delay] extended that work by examining the delay performance of the three protocols.
Different forms of structuring the error recovery process process were examined in [Kase97:Scalable,Kase98:Comparison] with the former demonstrating the considerable performance gains that can be achieved by dedicating a small number of multicast channel solely for error recovery use. In the latter work [Kase98:Comparison], we compared server-based and receiver-based multicast error recovery, demonstrating that a server-based approach yields both higher protocol throughput and lower network bandwidth usage.
In [NBT], an FEC-based approach to multicast error recovery was combined with ARQ to achieve scalable reliable multicast with a large numbers of users.
Finally, in [Yajn96] we measured and analyzed the packet loss process of MBone multicast sessions; of particular concern to us there was the spatial and temporal correlation in packet loss.

Multicast Flow Control

In our flow control work in [Bhat98:Efficient] we examined the use of multiple multicast groups with a different purpose in mind - striping a file's data across the multiple multicast channels at different rates in order to accommodate the differing receive capabilities of a heterogeneous receiver population. Our goal there was to minimize various measures of the amount of time needed to reliably transfer the file to all receivers.

Multicast Call Admission

[Firoiu 96] addresses the multicast admission control issue, presenting a general framework for admission control and resource reservation for multicast sessions. Within this framework, efficient and practical algorithms that aim to efficiently utilize network resources are developed. The problem of admission control is decomposed into several subproblems that include: the division of end-to-end QoS requirements into local QoS requirements, the mapping of local QoS requirements into resource requirements, and the reclamation of resources allocated in excess.

References

[Bhat98:Efficient]: S. Bhattacharyya, J. Kurose, D. Towsley, R. Nagarajan, ``Efficient Multicast Flow Control using Multiple Multicast Groups,'' to appear in IEEE Infocom98, (San Francisco, April 1998). The technical Report 97-15, version of this paper is available.
[Firoiu 96]: V. Firoiu, D. Towsley, "Call Admission and Resource Reservation for Multicast Sessions," IEEE INFOCOM'96 (San Francisco). See also: V. Firoiu, D. Towsley, "Call Admission and Resource Reservation for Multicast Sessions," Univ. of Massachusetts Techincal Report UM-CS-1995-017, revised 1996.
[Kase98:Comparison]: S. Kasera, J. Kurose, D. Towsley, ``A Comparison of Server-Based and Receiver-Based Local Recovery Approaches for Scalable Reliable Multicast,'' to appear in IEEE Infocom98, (San Francisco, April 1998). (a prepublication copy is not yet available on-line)
[Kase97:Scalable]: S. Kasera, J. Kurose, D. Towsley, ``Scalable Reliable Multicast Using Multiple Multicast Groups,'' 1997 ACM Sigmetrics Conference, April 1997.
[NBT]: J. Nonnenmacher, E. Biersack, D. Towsley. "Parity-Based Loss Recovery for Reliable Multicast Transmission," Technical Report 97-17, Dept. of Computer Science, Univ. of Massachusetts, March 1997. A shorter version of this paper appear in Proc. 1997 ACM SigComm Conference, (Cannes, France, Sept. 1997).
[Tows97:Comparison]: D. Towsley, J. Kurose, S. Pingali, ``A Comparison of Sender-Initiated and Receiver-Initiated Reliable Multicast Protocols," IEEE Journal on Selected Areas in Communications, April 1997.
[Yajn96]: M. Yajnik, J. Kurose, D. Towsley, ``Packet Loss Correlation in the MBone Multicast Network,'' IEEE Global Internet Conf. (London, Nov. 1996). See also: M. Yajnik, J. Kurose, D. Towsley, "Packet Loss Correlation in the MBone Multicast Network" Technical Report UM-CS-96-32, Computer Science Department, University of Massachusetts, 1996.
[Yama97:Delay]: M. Yamamoto, J. Kurose, D. Towsley, H. Ikeda, ``A Delay Analysis of Sender-Initiated and Receiver-Initiated Reliable Multicast Protocols,'' Proc. IEEE Infocom97, (Kobe Japan, April 97).

Last modified: 12/97