|
|
|
DISTRIBUTED SYSTEMS |
|
MULTIMEDIA NETWORKS |
|
NETWORK OPTIMIZATION/ANALYSIS/SIMULATION |
|
SENSOR NETWORKS |
|
WIRED NETWORKS/INTERNET |
|
WIRELESS NETWORKS |
|
|
DETAILED PROJECT DESCRIPTIONS |
|
DISTRIBUTED SYSTEMS |
- DiffGen
(PI: Dr. Indranil Gupta)
A wide variety of scientists and engineers use differential equations to express
their results and ideas. A subclass of differential equation systems has been shown
to be translatable (and equivalent) to useful distributed protocols such as epidemics,
endemics, etc. As such, these protocols can be systematically converted into
distributed protocols with equivalent stochastic behavior as the original equation system.
The resulting protocols are also probabilistically scalable and reliable, utilize low bandwidths
at each process and have low convergence times. The DiffGen toolkit will allow designers of
distributed protocols to to automatically generate compilable and runnable C code given an input
system of differential equations. We expect the toolkit will be useful to researchers and
designers using natural analogies to design distributed systems by cutting short the protocol
design life cycle. The system is written in Java and the protocol code outputted is in C.
The toolkit is designed to run on Windows and Unix.
- Folklore
(PI: Dr. Indranil Gupta)
A distributed file system that realizes "forever storage" of files
even in the presence of highly malicious and aggressive attackers. The core
protocol in Folklore is an endemic replication model derived from the survival
of endemic diseases in human populations. The "derivation" of the protocol
from the natural phenomenon is done systematically using our methodology to
translate differential equations into distributed protocols. This appears as
a paper in StoDiS 2005, and is also under review elsewhere.
- MON (Management Overlay Networks)
(PI: Dr. Indranil Gupta
and Dr. Klara Nahrstedt)
Develops on-demand Management Overlay Networks for PlanetLab-like testbeds,
so that instant queries can be carried out cheaply (without using heavy-weight DHTs)
but very reliably. This appears as a Usenix WORLDS 2005 paper, and future plans
include developing support for automated .mon scripts, detailed queries,
medium-term queries, and exploring applications such as distributed logging
and automatic experiment restarts.
This project studies effective management framework of large-scale overlay P2P
networks using light-weighted membership functions, fast on-demand overlay network
construction algorithms to reach maximum number of peer nodes, and distributed
programming environment with a push/pull-enabled script language to allow for
fast pulling of status information from the various peer nodes as well as for
fast pushing of different management information and software updates
to the various peer nodes.
- Overhaul
(PI: Dr. Indranil Gupta)
A complete solution to combat flash crowds. Overhaul combines stochastic real-time
detection at the web-server and detection-enabled cooperative caching to effectively
mask out flash crowds at the server, and without losing latency at the client.
It is an effective hybrid between a centralized and a p2p solution, a methodology
we believe will be popular for systems of the future. The original Overhaul was a
WCW 04 paper, but the current version is quite different (and improved) and is
currently under review elsewhere.
- peerCounter
(PI: Dr. Indranil Gupta)
peerCounter is a project that aims to address the problem of estimating the size of large and dynamic peer-to-peer networks.
It consists of an API and an application, both implemented in Java.
- Proactive Protocol Composition Language
(PI: Dr. Indranil Gupta)
It is essential to develop methodologies for the design of distributed protocols in order to
reduce the complexity of distributed systems, and for maturing the field. We have developed a
composition methodology for designing a range of proactive distributed protocols, with applicability
to large-scale distributed groups in both the Internet (e.g., peer to peer systems) and wireless
sensor networks. The methodology consists of several basic building blocks and different composition
techniques which can be used to combine the blocks into solutions for distributed computing problems.
The compositions preserve properties of the original components (reliability, scalability, liveness).
Proactive Protocol Composition Language (PPCL) is a simple language that can be used to specify the
compositions described in the methodology. The PPCL compiler uses the PPCL specifications to compose
existing (C language) source code of building block implementations in order to automate the above
design methodology. Given appropriate source code, PPCL can be used to generate protocols already
existing in literature, as well as to generate new, undiscovered protocols.
^Top
|
MULTIMEDIA NETWORKS |
-
Real-Time Streaming of Multimedia Data in IP Networks with Loss Concealment
(PI: Dr. Benjamin Wah)
This research is on the design of protocols and coding methods for the concealment
of losses that occur during real-time transmissions of audio, imagery, and video
data over unreliable IP networks, such as the Internet and wireless networks.
Since multimedia transmissions may tolerate some degree of loss, this research
develops schemes that involve trade-offs in their real-time requirements,
tolerance to loss, and perceptual quality. By studying proper coding of
transmitted data, protocols to schedule transmissions and feedbacks, and
reconstruction schemes to recover lost data, we can apply the developed
results to emerging multimedia-ready 3G and 4G cellular networks.
- Tele-immersive Environments for Everybody (TEEVE)
(PI: Dr. Klara Nahrstedt)
This project investigates semantic protocols to deliver 3D multi-camera content
over Internet networks to multi-display receivers. The challenging questions
that we aim to answer are: (a) what are the next generation system architectures
and high-performance protocols to build tele-immersive 3D environments using
COTS components, (b) what are the 3D compression algorithms for these large video
streams (one 3D frame from one 3D camera is 300Kbytes), (c) what are the adaptive
control algorithms, mathematical models and stable solutions that are required
for multi-camera, multi-display TEEVE environments.
^Top
|
NETWORK OPTIMIZATION/ANALYSIS/SIMULATION |
- Network Modeling and Simulation
(PI: Dr. P. R. Kumar)
Sponsored, in part, by enabling techniques for large-scale network simulation and
emulation, supported by DARPA Network Modeling and Simulation (NMS) program, NSF
Next Generation Software (NGS) program, and an unrestricted gift from Cisco, Inc.
This project focuses on device-enabling techniques for simulating large-scale networks
with high fidelity and in real-time. A component-based, compositional
network simulation and emulation environment, J-sim
has been developed as part of the deliverables for the project.
^Top
|
SENSOR NETWORKS |
- Computation in Sensor Networks and maximizing lifetime of sensors
(PI: Dr. P. R. Kumar)
In a sensor network, one is only interested in collecting from a sink node a
relevant function of the sensor measurements. We aim to study the maximum rate
at which functions of sensor measurements can be computed and communicated to
the sink node. We are also looking at how block computation and function structure
can be exploited to provide greater computational throughput and tackling distributed
computation problems in networks subject to noise. Another focus is to find
communication schemes which maximize the functional lifetime of the sensors.
- Sensor Networks
(PI: Dr. R. Srikant)
A key difference between ad hoc sensor networks and ad hoc wireless networks
in that sensor networks are usually designed for specific applications.
Thus, significant resource gains can be obtained in sensor networks by
designing algorithms for in-network processing of information before transmitting
it to a fusion center. We are developing distributed algorithms for such in-network
processing for specific applications.
- Smart Attire
(PI: Dr. Tarek Abdelzaher)
Clothing embedded with sensing, computing, and communication capabilities will be
the next generation of fashion, that will improve lifestyle of our society by providing
various innovative personal monitoring services.
- EnviroTrack
(PI: Dr. Tarek Abdelzaher)
As an object-based distributed middleware system, EnviroTrack raises the level of programming abstraction by
providing a convenient and powerful interface to the application developer geared towards tracking the physical environment.
EnviroTrack is novel in its seamless integration of objects that live in physical time and space into the computational environment
of the application. It contains run-time mechanisms that efficiently abstract groups of sensors by logical objects which maintain
aggregate environmental state. Such objects may be logically attached to moving entities in the physical environment, in order to
monitor the state of the tracked entity.
- Data-centric wireless sensor networks
(PI: Dr. Jennifer Hou)
Sponsored, in part, by the NSF ANI program and the
MURI/ONR program, this project focuses on two synergistic thrusts.
Along the theoretical thrust, we carry out an analysis of
fundamental performance limits with respect to coverage, connectivity,
lifetime scalability, and critical power for connectivity. Along the
protocol design and systems prototyping thrust, we design algorithms
and protocols that are capable of approaching these performance
limits. In particular, we have implemented on a Berkeley motes
testbed a proof-of-concept prototype for acoustic tracking
applications.
^Top
|
WIRED NETWORKS/INTERNET |
- Computer Security and Privacy
(PI: Dr. Nikita Borisov)
Nikita Borisov's research interests lie in the area of computer security and privacy,
especially as applied to large-scale distributed systems. My current focus
is on anonymity: analysis of existing anonymous systems and the design of
new peer-to-peer anonymous networks. In this research he takes a systems approach,
with rigorous, quantitative metrics of anonymity and analysis with help from simulations.
He is also interested in other privacy enhancing technologies, such as techniques
for privacy-preserving data aggregation, as well as research into building reliable
and secure peer-to-peer networks.
- Multiclass Scheduling and Congestion Control in Communication Networks
(PI: Dr. Benjamin Wah)
- Congestion Control and Pricing for a Heterogeneous Internet
(PI: Dr. R. Srikant)
We consider the problem of allocating resources fairly in the Internet through
congestion management and pricing in the Internet. We develop algorithms for end
users and Internet users which require minimal information exchange while
optimally utilizing the network resources.
- Resilient Internet Routing and Service Routing
(PI: Dr. Klara Nahrstedt)
This project investigates vulnerabilities in Internet routing, especially
in the BGP domain. We are developing a resilient routing framework at the IP
and overlay layers to greatly minimize the impact of the failures and vulnerabilities.
Our framework incorporates the development of models, metrics and algorithms that
would assist us in the delivery of reliability, improved network availability and survivability.
- Orpheus: A Compose-able, Value-added Service Overlay Network
(PI: Dr. Klara Nahrstedt)
This project investigates a QoS-aware service and content composition over overlay
Internet networks for applications such as IPTV, Tele-immersive Environments and other.
The challenges in this project are (a) automated service composition where the
application designer provides a high-level application service graph and our
composition system, called Orpheus, maps the service graph onto the physical
overlay infrastructure in a QoS-aware fashion, (b) automated service routing where
we investigate functional and data routing to find the best mappings onto the underlying
overlay and IP networks from a high-level application service graph description.
This work relies on our prior on Spidernet, QoS talk and QoS-Compiler where we started
to investigate service networks in smaller scale (smart rooms) and investigate the
creation of service graphs and their mapping onto smart room infrastructures.
^Top
|
WIRELESS NETWORKS |
- Mobility and Security for Networks
(PI: Dr. Yih-Chun Hu)
Yi-Chu Hu's general research interests are in security and systems, with emphasis on the areas
of secure systems and mobile communications. He has published papers in the areas of
secure Internet routing, secure routing in wireless ad hoc networks, security and
anonymity in peer-to-peer networks, efficient cryptographic mechanisms for routing security,
and the design and evaluation of ad hoc network routing protocols.
- IT Convergence Laboratory
(PI: Dr. P. R. Kumar)
The convergence of communication and computation over the past two decades has
given us the Internet. We believe that the next phase of the information technology
revolution will be the convergence of control with communication, and computation.
This will provide the ability for large numbers of sensors, actuators, and
computational units, all interconnected wirelessly or over wires, to interact
with the physical environment. A key issue for these systems is, we believe, to
determine what are the appropriate abstractions and architecture to enable this
to occur. Also key to proliferation will be services provided by middleware,
as well as methodologies for design. These issues are investigated both experimentally
and theoretically in the Convergence Lab.
- Capacity, coverage and connectivity of wireless networks
(PI: Dr. P. R. Kumar)
The wireless medium is inherently unreliable. We investigate the effect of fading on the
transport capacity of wireless networks. Further, in random networks, we need to find how
many neighbors we need for each node so that the whole network can be connected in a
multihop fashion. We are looking into the requirements for a network to have theta-coverage
and asymptotic connectivity.
- Congestion control and Routing in wireless networks
(PI: Dr. P. R. Kumar)
The fundamental difference is that wireless networks are of a broadcast nature. There is no
strict notion of a ÒlinkÓ, since transmissions from nearby nodes interfere with each other.
We are attempting to develop a transport protocol which would take into account these
problems and yet effectively control congestion. We are also looking at routing in wireless networks.
In particular we are looking at constructing flow-avoiding routing protocols which
automatically load balances flows.
- PERM: Practical End-host Residential Multihoming
(PI: Dr. Haiyun Luo)
PERM is the Practical End-host collaborative Residential Multihoming framework. 802.11 networks
have spread rapidly in the residential area, and it is common for neighbors to receive signals
from each other's home wireless networks. PERM allows residents to leverage such an opportunity
to improve their last-mile Internet connectivity, at no additional cost, by pooling their
Internet accesses together. PERM is practical in that it does not rely on support from
the network infrastructure in terms of advanced naming scheme or proxy, the remote host
in terms of new transport protocols, or the end-user in terms of explicit application
feedback or configuration. Instead, PERM employs automated on-line analysis of the user's
networking behaviors, and exploits the identified patterns to achieve high-performance
scheduling at the flow level. PERM is also highly usable for normal residential users.
It preserves a user's privacy and security, and mitigates the free-riding problem.
We have implemented PERM for Linux clients and the open-source Linksys wireless router.
- SELECT: Self-learning Wireless Networks
(PI: Dr. Haiyun Luo)
Abstract: SELECT is an effective and efficient self-learning collision avoidance strategy
to address the open hidden/exposed receiver problem in wireless networks. SELECT is based
on the observation that carrier sense with received signal strength (RSS) measurements
at the sender and the receiver are strongly correlated. A SELECT-enabled sender exploits
such correlation using automated on-line learning algorithm, and makes informed judgment
of the channel availability at the intended receiver. SELECT achieves collision avoidance
at packet-level time granularity, involves zero communication overhead, requires no
hardware support beyond what is vailable in off-the-shelf 802.11 devices, and easily
integrates with the 802.11 DCF.
- WINE: Wireless Interlay Architecture
(PI: Dr. Haiyun Luo)
Abstract: The 802.11 PHY/MAC is designed for single hop wireless LANs and the attempts to
extend this protocol to multihop wireless meshes have not proved as successful. Notable
problems include poor TCP performance, unfairness and self-interference across multiple hops.
Nevertheless, there is a strong desire to leverage the prevalence of cheap 802.11 devices
into large-scale community wireless mesh. We propose WINE, a wireless interlay network
which sits atop 802.11 MAC but below the network layer. WINE is deployed on a carefully
selected set of mesh switches, which coordinate among themselves to limit interference
over multiple hops and provide an abstract link mechanism to upper layers.
- Real-Time QoS provisioning for ad-hoc wireless communication
(PI: Dr. Marco
Caccamo)
Today's embedded computers are increasingly mobile and ubiquitous, are capable of interacting
with the environment, and can communicate with one another over possibly vast and pervasive
networks. While most of the wireless networks are not expected to be capable of controlling
their own motion, new technological possibilities are emerging to provide small embedded
devices with the means to propel themselves, with an energy expenditure that is comparable
to the energy budget of communication and computation. A Controlled-Mobility Wireless Network (CMWN)
is defined as a network of embedded devices endowed with computation, communication and
motion capabilities. The purpose of this project is the development of a new conceptual
framework for the design, development and operation of efficient and reliable networks
with such characteristics.
- Routing/transport issues in wireless networks
(PI: Dr. Nitin Vaidya)
Routing and transport layer protocols for wireless
networks, including multi-hop mesh networks, sensor networks,
and ad hoc networks.
- Pulsar
(PI: Dr. Robin Kravets)
Cross-Layer Approach to Energy Conservation in Mobile Ad Hoc Networks. The increasing demand
for long-lived mobile computing devices has brought power conservation to the forefront of
research. Improving lifetime is particularly important for mobile ad hoc and sensor networks
where devices are expected to be deployed for long periods of time without the potential
for recharging their batteries. To support energy-efficient communication, it is necessary
to consider energy consumption at multiple layers in the network protocol stack. While many
techniques have been studied in isolation, any change to communication at one of the layers
impacts the other and so may impact energy consumption and communication quality. The Pulsar
framework provides a hierarchical approach to energy conservation that focuses on the
interrelationships between layers.
- Medium Access Control, Routing and Congestion Control for Wireless Networks
(PI: Dr. R. Srikant)
We are exploring the role of queue length information in efficiently using resources in
single-hop and multi-hop wireless networks. This problem is especially difficult in
multi-hop networks since each node only has access to local topology and traffic information.
A key goal is to design distributed algorithms that achieve a guaranteed fraction of the
performance achievable by a centralized algorithm.
- Security and Trustworthiness in Wireless and Wireline Peer-to-Peer Networks
(PI: Dr. R. Srikant)
We use game theory to study the impact of selfish behavior in peer-to-peer networks
and design incentives and punishments to encourage socially-optimal behavior.
We are also studying the use of peer-to-peer file dissemination as a means to
combat the spread of worms in the Internet.
- Loss Concealments for Real-Time Low Bit-Rate Voice Transmissions Over Wireless Networks
(PI: Dr. Benjamin Wah)
In this research, we are studying a novel end-to-end multiple-description coding (MDC)
method for concealing packet losses in transmitting low-bit-rate coded speech, without
special protocol and hardware supports from the underlying network. Based on correlations
of coding parameters observed, we investigate a new coder-dependent parameter-based MDC.
Specifically, for low-bit-rate coding, we have observed high inter-frame correlations
in linear predictors and low correlations in excitation parameters. Hence, we generate
multiple descriptions at senders by interleaving linear predictors and reconstruct
lost ones at receivers by linear interpolations. The hard-to-reconstruct excitations
are replicated to multiple descriptions. The design is done in a way that does not
increase the bandwidth of transmission and that adapts its number of descriptions
to network loss conditions. We evaluate the reconstruction quality of three linear
predictor representations: Reflection Coefficient, Log Area Ratio, and Line Spectral Pair (LSP).
Based on the results, we study enhancements on excitation quality. Finally, we integrate
our concealment schemes into the FS CELP, ITU G.723.1, and FS MELP coders and test their
performance under different loss scenarios.
- First responder System: A Mission-Critical Communication Architecture with Integrated QoS/QoP
(PI: Dr. Klara Nahrstedt)
This project investigates scalable and secure overlay communication architecture in
first responder systems to allow for end-to-end high quality data transmission and
trustworthy delivery of information in incident areas. The challenges in these
networks are (a) heterogeneous networks along the end-to-end path, ranging from
ad-hoc wireless networks to Internet2 where the command and control happens,
(b) communication across multiple agencies and within agency along different
command structures, and (c) different user roles, security and performance requirements
and demands in case of accidents to be responded to.
- Secure and Timely Data/Resource Aggregation in PowerGrid Systems
(PI: Dr. Klara Nahrstedt)
This project is part of a large center-like cyber-trust grant that investigates
trustworthy information distribution in PowerGrid systems. One important part of the
overall PowerGrid system are the SCADA networks that collect information from small
intelligent devices and report the power status to command centers for further
decision making in power distribution. We investigate problems of trustworthy
aggregation as information are moved from thousands of devices towards the command center(s).
We are studying aggregation architectures, algorithms and protocols on smart gateways which are
connected via wireless 802.11 Ethernet with the intelligent IED monitoring devices.
The smart gateways will assist in secure and timely collection of information and provide
the alarm detection system in case of problems as well as informed aggregated information
as needed to the control center(s).
- Dyanmic ad-hoc wireless networks (DAWN)
(PI: Dr. Jennifer Hou)
Supported by the MURI/ARO program, we aim in this project
to tackle issues that work toward a better definition/characterization
of wireless links and its implication for higher-layer protocol design
and optimization. We would like to (i) understand how, and to what
extent, wireless links are affected by PHY/MAC attributes and other
environmental factors, (ii) characterize the behavior of wireless
links in such a way that they become amenable to rigorous analysis and
reasoning, and (iii) identify control knobs in the MAC/PHY layers with
which the network capacity can be optimized.
- An open, dependable and evolvable software infrastructure for assisted-living
(PI: Dr. Jennifer Hou)
Supported by the NSF Computer Systems Research program,
our project aims to design, implement, and evaluate (i) an
assisted-living supportive software infrastructure that allows
disparate technologies, software components, and wireless devices of
different protocol families to work together in a low cost,
dependable, and secure fashion with predictable properties; (ii)
interface that abstracts this software infrastructure and provides
various services to facilitate elder people with their independent
and/or assisted living needs.
- Towards building a
performance-predicable wireless mesh network
(PI: Dr. Jennifer Hou)
Supported by the NSF Network Technologies and Systems program,
our project seeks (i) A thorough understanding of how, and to what
extent, wireless links are affected by PHY/MAC attributes and other
environmental factors, (ii) characterization of the behavior of
wireless links in such a way that they become amenable to rigorous
analysis and reasoning, and (iii) identification of control knobs in the MAC/PHY layers with which the network capacity can be optimized.
^Top
|
|
 |
|
|
|
