Download 붙임 3 - CCRG

(붙임 3)
Proposal for Collaborative Research
1. Title of Research Project:
ANALYSIS AND DESIGN OF MAC AND NETWORK-LEVEL SCHEMES FOR RELIABLE GROUP
COMMUNICATION IN AD-HOC NETWORKS
2. Duration of Project
From
May
2002
to
May 2004
(
24
Months)
3. Project Manager
Name
J.J. Garcia-Luna-Aceves
Date of Birth
Affiliation
& Position
Professor, University of California at Santa Cruz
Present
Computer Engineering Department
Address
University of California, Santa Cruz CA 95064
TEL
831 4594153 FAX
831 4594829
Degree &
Specialty
Ph.D., internetworking and wireless networks
Oct 20, 1955
E-mail address
[email protected]
Main Scientific publications in the last five years(to a Maximum of five)
The following are selected publications; a complete list of publications is available at:
http://www.cse.ucsc.edu/research/ccrg/publications.html
1. L. Bao and J.J. Garcia-Luna-Aceves, "A New Approach to Channel Access Scheduling for Ad
Hoc Networks," Proc. ACM MobiCom 2001-- Seventh Annual International Conference on Mobile
Computing and networking, July 16-21, 2001, Rome, Italy.
2. J.J. Garcia-Luna-Aceves and A. Tzamaloukas, "Reversing The Collision-Avoidance Handshake
in Wireless Networks ", Proc. ACM/IEEE Mobicom 99, Seattle, Washington, August 15--20, 1999.
3 M. Spohn and J.J. Garcia-Luna-Aceves, " Neighborhood Aware Source Routing", Proc. ACM
MobiHoc 2001, October 4--5, 2001, Long Beach, California.
4. E.L. Madruga and J.J. Garcia-Luna-Aceves , "Scalable Multicasting: The Core Assisted Mesh
Protocol", accepted for publication in ACM/Baltzer Mobile Networks and Applications Journal,
Special Issue on Management of Mobility in Distributed Systems, First Quarter 2001.
5. M. Mosko and J.J. Garcia-Luna-Aceves," Performance of Group Communications over Ad-Hoc
Networks", Proc. The Seventh IEEE Symposium on Computers and Communications (ISCC 02),
Taormina/Giardini Naxos, Italy, July 1 - 4, 2002.
* Please mention the year of the publication, the name and volume of the journal and main/co author
4. Applying Budget
1st year
ETRI
Counterpart
2nd year
ETRI
Counterpart
5. Members of Research Team
Prof. J.J. Garcia-Luna-Aceves, Principal Investigator
Marc Mosko, Graduate Research Assistant
Hari Rangarajan, Graduate Research Assistant
Long Yang, Graduate Research Assistant
Dr. Donkyun Kim, Visiting Scientist
(post-doc appointment)
6.
Research Plan
6. 1. Summary
Group communication over multihop wireless networks is critical to many
applications in the commercial and military sectors.
Data broadcast and data
multicast have many uses in multimedia applications, data distribution
services, and as tools for other communications protocols. More recently, the
availability of less expensive sensors has prompted renewed interest in group
communication within the context of sensor networks.
Multicast and broadcast over multihop wireless networks presents significant
problems for lost data, flow control and error correction schemes.
The loss
of a single packet may affect many recipients due to data replication.
Because data travel simultaneously through multiple network paths, there is
no single flow, so standard end-to-end approaches such as the Transmission
Control protocol (TCP) do not directly apply.
For reliable data delivery,
packet acknowledgments or loss negative acknowledgments may flood a sender or
disrupt other data packets causing further loss.
A fundamental problem particular to multi-hop wireless networks is the
hidden-terminal problem.
Two radio nodes, for instance, may not have clear
reception of each other, but both may be in contact with a third node.
The
third node is the focus of hidden-terminal collisions when the other two
nodes simultaneously transmit. Solutions to the hidden-terminal problem
follow two lines. Contention-based solutions use some form of randomized
channel reservation scheme.
The reservations still contend for the channel
and may cause collisions of the reservation packets, but larger data packets
should not collide.
Schedule-based solutions devise a time-synchronous order
to transmissions and avoid data packet collisions.
Some protocols have
features of both. Group multicast or all-node broadcast puts strains on a
weak aspect of unscheduled ad-hoc protocols.
Collision avoidance protocols,
such as IEEE 802.11 DCF use unicast request-to-send (RTS) and clear-to-send
(CTS) messages to reserve radio time.
side-band carriers.
Busy-tone protocols, such as BTMA, use
Such mechanisms do not work for group communications,
because multiple nodes would respond or busy tones would collide.
These
protocols become pure CSMA for broadcasts, which offers no hidden-terminal
collision protection for multi-hop wireless networks.
Group communications covers several forms of packet delivery mechanisms,
including link-layer broadcast, group-address packets, and IP multicast.
Packet losses significantly affect group communications because of correlated
losses and difficulties with reliability mechanisms.
Correlated losses occur
from the inability of switching nodes to replicate packets. Downstream nodes,
possibly many, suffer correlated loss when one upstream switching node loses
a packet.
Standard unicast reliability mechanics, such as end-to-end or hop-
by-hop acknowledgments, involve significant overhead or prohibitive network
load for communications in a large group.
We propose to study reliable group communication over multihop wireless
networks. Although several proposals exist for supporting unreliable
multicast over multihop wireless networks, very limited work has been
reported on effective approaches for reliable group communication in ad hoc
networks, and the impact of the MAC layer has been ignored until very
recently.
We will investigate modifications to MAC protocols for more
efficient support of reliable multicasting and broadcasting. We will also
investigate network-level schemes for reliable mutlicasting and broadcasting
in mobile ad hoc networks, and the interaction between mechanisms at the MAC
layer and the network layer for reliable group communication. A particular
problem that needs attention is the effect that changes in the neighborhood
of a node (i.e., additions and deletions of neighbors) has on its ability to
disseminate data reliably to a group of other nodes in the network.
6. 2. Development Approach
Our approach will take advantage of our recent work on scheduled channel
access and reliable broadcasting over contention-based MAC protocols.
More specifically, we will carry out the following three main tasks. The
first task aims at developing a network-level scheme for information
dissemination on demand; the other two tasks aim at developing schemes for
the transmission of information to multiple receivers efficiently. Together,
these schemes should improve the throughput and delay characteristics of a
mutlihop ad hoc network substantially in the presence of multipoint data
traffic.

Reliable information dissemination with a publish-subscribe paradigm:
We will develop and simulate in ns2 a protocol for reliable information
dissemination using a publish-subscribe approach. The protocol will
take advantage of extension to a traditional collision-avoidance
handshake and the use of “content tables” to allow network nodes to
disseminate data on demand. This task is an extrapolation of the SPIN
protocol, introduced by Heinzelman et al (W. Heinzelman et al.,
“Adaptive Protocols for Information Dissemination in Wireless Sensor
Networks,” Proc. MobiCom 99) and directed diffusion, advocated by
Intanagonwiwat et al (C. Intanagonwiwat et al., “Directed Diffusion: A
Scalable and Robust Communication Paradigm for Sensor Networks,” Proc.
MobiCom 2000) and will take into account the effects of the MAC layer
in information dissemination, which has been neglected in the main
proposals to date.

Scheduling of multicast transmissions over contention-based MAC
protocols: We will develop extensions to collision-avoidance protocols,
such that the transmissions of nodes originating or forwarding data
reliably are paced in order to reduce or eliminate collisions of data
packets. Our approach is based on work done to date by Mosko and
Garcia-Luna-Aceves on reliable broadcasting in ad hoc networks. We will
develop heuristics to establish an implicit or explicit transmission
schedule based on knowledge of the 2-hop neighborhood of a node, and
additional information if needed.

Scheduling of multicast transmissions over conflict-free MAC protocols:
We will develop extension to our conflict-free MAC protocols to
schedule the transmission of multicast packets in a way that minimizes
or reduces substantially the number of times a given relay receives
replicas of any given packet. Current proposals for conflict-free
channel access do not take into account the need to minimize
unnecessary replicas of broadcast or multicast packets received by a
given relay.
6. 3. Project Milestones (Classified by period)
The projected milestones are the following:
Year 1:

Draft description of the first design of the proposed protocols within
3 months of the start of the project.

Report
addressing
the
correctness
of
the
proposed
protocols,
when
applicable, within 6 months of the start of the project.

Report describing results of simulations in ns2 of the proposed
protocol within 8 months of the start of the project.

Manuscript(s) describing the design, correctness, and performance of
the proposed protocol within 8 months from the start of the project.

Annual technical report describing the results achieved during the first
year of performance.
Year 2:

Report addressing the design and correctness of the proposed
protocols, when applicable, within 18 months of the start of the
project.

Report describing the design, correctness, and performance of the
proposed protocol within 20 months from the start of the
project.

Annual technical report describing the results achieved during the
second year of performance.
The ns2 code that we will be written will be public domain, and therefore
ETRI can have access to it.
Signature of Project Manager