IWSCN 2010
Keynote Speakers
The
Changing Performance Goals in Internet Video Streaming
Carsten
Griwodz, Simula Research Laboratory
AS, Norway
Internet video streaming over the Internet has been investigated for decades
and it has reached a mass market with commercially backed offerings such as
YouTube and PPLive. Why should anyone still use research resources to look into
improving video transport over the Internet? We discuss the development of
performance criteria over time and put them into the perspective of today's
wired and wireless networks. We present the questions of appropriate visual
performance that are currently dominating ideas for improving streaming
infrastructure and performance evaluation, and present the demands of streaming
that goes beyond "classical" video streaming.
Dependable
Protocols for Wireless Sensor Networks
Levente
Buttyán, Budapest University of Technology and Economics, Hungary
Wireless sensor networks are often envisioned to be
used in applications where their dependability is highly important. Examples
include the use of wireless sensor networks for monitoring critical
infrastructures, such as electricity distribution networks and drinking water
supply systems. On the one hand, sensor networks can be relatively easily
deployed at large scale, and as they are normally built from low cost devices,
they can provide the monitoring service in a cost efficient manner. On the
other hand, the usefulness of the wireless sensor network in critical
infrastructure monitoring is primarily determined by its dependability: a
monitoring system that fails to report a faulty condition or an attack prevents
the operator from fixing problems before their consequences affect the
operation of the critical infrastructure.
In the WSAN4CIP Project (http://www.wsan4cip.eu/), we work on the
application of wireless sensor networking technology in critical infrastructure
protection, and for this reason, we develop mechanisms that increase the
dependability of wireless sensor networks. Overall, our goal by improving the
dependability of the monitoring sensor network itself is to ensure that the
failures of and the attacks on it have a minimal impact on the monitored
infrastructure. Failures may happen and attacks may be targeted at any layer of
the sensor network architecture starting from the node hardware and operating
system, through the networking protocol stack, up to the middleware and service
layers. Therefore, our approach in the WSAN4CIP Project is to address the
problem of dependability at all layers of the architecture.
In this talk, I first give a brief overview of the
WSAN4CIP Project, and then I focus on two specific problems addressed in the
project. More specifically, I present our work on (i) a dependable distributed
storage scheme for wireless sensor networks and (ii) a private cluster head
election protocol.
(i) In sensor networks, there are usually multiple,
distributed sources that generate data that, in some cases, must be stored
efficiently in multiple storage nodes, each having constrained communication,
computation, and storage capabilities. Using the principles of network coding
and storing encoded data instead of raw data can help to implement a reliable
storage service in an efficient manner. However, coding based approaches has a
potential problem in hostile environments, where an adversary may attack the
storage nodes. In particular, the problem that we are interested in is the so
called pollution attack, whereby the adversary modifies some of the stored
encoded data, which results in erroneous decoding of a large part of the
original data upon retrieval. We propose algorithms to detect and recover from
such attacks. In contrast to existing approaches to solve this problem, our
approach is not based on adding cryptographic checksums or signatures to the
encoded packets; rather, we take advantage of the inherent redundancy in such
distributed storage systems.
(ii) Clustering is a useful mechanism in wireless
sensor networks that helps to cope with scalability problems and, if combined
with in-network data aggregation, may increase the energy efficiency of the
network. At the same time, by assigning a special role to the cluster head
nodes, clustering makes the network more vulnerable to attacks. In particular,
disabling a cluster head by physical destruction or jamming may render the
entire cluster inoperable temporarily until the problem is detected and a new
cluster head is elected. Hence, the cluster head nodes may be attractive
targets of attacks, and one would like to make it difficult for an adversary to
identify them. The adversary can try to identify the cluster head nodes in
various ways, including the observation of the cluster head election process
itself and the analysis of the traffic patterns after the termination of the
cluster head election. Here, we focus on the former problem, which we call the
private cluster head election problem. This problem has been neglected so far,
and as a consequence, existing cluster head election protocols leak too much
information making the identification of the elected cluster head nodes easy
even for a passive external observer. We propose a private cluster head
election protocol for wireless sensor networks that is designed to hide the identity
of the elected cluster head nodes from an adversary that can observe the execution
of the protocol.