M. Stanojevic et al.: Can You Fool Me?...
Looks at automatic detection of protocol gullibility, where gullibility 
is defined as a posibility that a subset of the participants can violate 
desirable properties of the protocol by disobeying the protocol in some 
fashion.
So gullibility is not bugs or security breach in terms of authority, 
integrity or privacy. It is not an attack by an external adversary, 
rather a manipulation by a legitimate participant.
The problem is defined as  a game between angelic (always respects 
protocol)  vs. demonic (can do anything) components. Hence, protocol is 
gullible if there exists a strategy for a demonic component to violate 
protocol property.
The challenges in automatic detection (insted of manual protocol 
analysis as done in the past) are primarily:
1. Practicality of search over a demonic component space of strategies, 
which is large (think of just flipping a bit or two of a valid packet). 
This challenge can be addressed by using space reduction techniques, 
e.g. considering only packet header and not payload.
2. Variability in nework conditions: some protocol properties are 
exploitable only under certain conditions, such as ECN bit is use only 
under congestion.
3. Determining whether a strategy has been successful, because of 
nondeterminism of the angelic component: need to compare with reference 
behavior and use statistical tests.

Prototype checker has been implemented and tested on a simple example of 
ECN protocol, showing feasilbility of automatic checking.

M. Buettner et al.: Smart Dust with RFID
Looks at integration of RFID and Wireless Sensor Networks to produce 
RFID Sensor Networks (RSN) whose nodes are called WISPs (Wireless 
Identification and Sensing Platforms). The idea is te extend RFID to 
include sensing, and get a a nde that is small, inexpensive, can 
directly communicate (not P2P), and has no battery limitation.
Questions surroundung this concept include:
1. Feasibilty: can WISPs  be built? And yes they can, in fact Intel has 
an evolution of WISPs already.
2. Applications: can RSNs do anything useful? Yes again, with the 
following examples:
- "Milk": long-lived sensing as a cold-chain monitoring
-"Planes" - use them as strain gages, to check structural integrity
-"Brains" - implanatable medical devices
-"Elders" - long-term elder care, monitoring of people's activities
Open questions not yet looked at:
3. Dealing with intermittent power: how to program for this system? 
Power is a huge issue and we need strategies to mitigate it.
4. Dealing with asymmetric protocols. RFID uses slotted ALOHA now but 
more efficient protocols ar eneeded when we have changing sensed data etc.
Audience was interested in manufacturing and cost issues (realistic that 
WISP gets cheaper than RFID), environmental impact of using large number 
of WISPs everywhere (to be possibly addressed by "retiring" rather than 
throwing them away),
using omni-directional vs. directional antenae, total power needed to 
"ligh up" the space where WISPs are deployed, is Moore's Law indeed 
going to improve usefullness of WISPs, whether nodes will have enough 
power to complete the needed task and return the result (probably, since 
less power is needed to send than to store to flash memory), scheduling 
for readers, possibility of wireless recharge...

R. Mahajan et al.: Eat All You Can
Paper advocates that aggressive use of resources can be better than 
strictly following the efficiency principle, eg. in wireless 
communication. The criterion for this method is whether the marginal 
cost can be driven below marginal benefit.
Proposed is a buffet principle, "eat all you can for fixed price". For 
example, buffet FEC (using available spectrum for extra FEC bits) could 
be practical with embedding bits in separate packets, sending them at 
lower priority and keeping the FEC packets small.
Identified challenges are in applying the principle so that it does not 
detract from productive work (use priorities or utility-driven 
approach), quantifying marginal cost and benefit (could be tricky due to 
side-effects), but precise accounting may not be needed in some cases. 
It is argued that buffet is good for non-conservable resources (i.e. 
ones that you either use or lose).
Audience questions included: correctly estimating available resources, 
while acclaiming the general priciple there is concern that network 
components are engineered to work well only at relatively low capacity, 
but not nowhere near 99.999% load, difficulty in imagining the possible 
side-effects of buffet. Also mentioned is a suggestion to consider "use 
it at low cost or lose it" approach. From control theory, stability and 
responsiveness must be maintained, so this appears as good for a 
well-controlled system but questionable if deplyed out in the wild. 
Finally, environmentalists again raised an issue of who's controlling 
the power usage, since use of resources aggressively often implies high 
power consumption.