Brief Update: July 17
In the past few days, Mahdi has been
lecturing and informing us on the Hierarchical
Node ID Distribution process and chunk factors. Also, he informed us that Persea has a way of not only testing the system against Insertion attacks and back pointer attacks but also of using a silent or active attacker. Thus far we have been simulating with an active attacker but now we will test with a silent attacker as well(my simulations).
An active attacker will return an
incorrect result where as a silent attacker will just
ignore the query or drop the request.
With a silent attacker, no incorrect
result is returned so the look up request still has
a chance of a correct result being returned as well as the DHTs of the source remaining uncorrupted. Thus we can see that usually the active attacker is a more malicious threat, but also that it is harder to accomplish and requires more control/bandwidth.
Our prediction of the silent attacker
simulations is that the probability of a successful
result will not be influenced by the number of silent attackers. This prediction has thus far been correct with only one data point/simulation left to finish. |
|
July 6, 2012
After we received the Persea code,
Mahdi explained more attacks
such as sybil, insertion and
reflection attacks.
Persea starts with bootstrap nodes
and a basic level of security
based on a trusted network of friends
that expands only to
other nodes that must be trusted by
one node in the network.
This requirement to be invited
differentiates Persea from a regular
Kademlia network where no mechanism
checks for trusted/honest nodes.
Another level of success in Persea
that enables effecient lookups is Persea's
Replication process. Within a
network, a key/value pair is saved
in more than one location equally
spaced within the network so
if a lookup encounters an attacker
node it still leaves a chance of success
from the other nodes in the network
that also hold that key/value information.
Last week, we ran lookups of 10000
and measured the success of the lookups without
initializing the Persea system or
more complex attackers.
This week on Monday, Michael and I
were assigned parameters that would
not only enable the Persea system but
also initialize sybil attacks, and in my
simulations Node insertion and
Reflection attacks. Based on our data, we can
compare how successful our lookups
are with Persea on the defense as well as
how well it holds up against more
complex attacks.
|
June 17 - 19, 2012
June 19, 2012
I arrived yesterday and moved into
Kalpana Chawla Hall on the University of
Texas at Arlington campus.
Coming from Wesleyan the first
impression of the University was based on its
size. 400 vs. 32000. After a few hours, my roommate and I got settled into our room and prepared for our first day participating in the NSF funded REU in Hazard Mitigation under the Civil Engineering Deparment. Orientation:
Monday and Tuesday included a
whirlwind of introductions to the other
professors and the other 10 people in our REU group.
Day 1 was basic REU information and a
quick presentation on each project.
Day 2 was all about ethic personal,
professional, research and student.
Overall experience:
Excited - to be on this huge campus
and being offered this extraordinary
opportunity
Nervous - Let's just be honest being
in a whole new place is crazy.
Questions thus far:
What does peer to peer look like in
action? And what is it really?
Why is the gym so far away?
|
