Incentives - Peer-to-Peer Computing: Applications, Architecture, Protocols, and Challenges - page 111

can avoid collusion between two neighboring devices in the forwarding auction
game.
Buchegger and Le Boudec [Buchegger and Boudec, 2005] observed that
economic incentives such as payment approaches can entice selfish users to
help in routing and forwarding data but may not be able to handle other types
of misbehaviors such as packet dropping, modification, fabrication, or timing
problems. Thus, they proposed to use a reputation system in which every
user provides “opinion” data to the network based on observing the behaviors
of neighboring devices. After a user device has gathered such opinions (both
from itself or from others, i.e., second-hand information), it can carry out a
Bayesian estimation so as to classify the neighboring devices as malicious or
normal. A neighboring device that is identified as a malicious user is then
isolated from the network by rejecting its routing and forwarding requests.
Felegyhazi et al. [Felegyhazi et al., 2006] reported an interesting game
theoretic analysis of the forwarding problem in ad hoc networks. Instead of
using a payment-based strategy, the model employs a purely utility concept
in that a device's utility is equal to its payoff when it acts as a data source
(i.e., the sender of a multihop tra c), minus the cost when it acts as an
intermediate device (i.e., a forwarder of other sender's tra c) in any time-
slot. Here, both the payoff and cost are defined in terms of data throughput.
Thus, an important assumption in this model is that only the sender has a
positive payoff, while all the intermediate devices enjoy no payoff but just
incur forwarding costs. Specifically, the destination device (i.e., the receiver
of the multihop tra c) also enjoys no payoff. This may not conform to a
realistic situation. Simulations were done to estimate the probability that the
conditions for a cooperative equilibrium hold in randomly generated network
scenarios.
Table 5.3 gives a qualitative comparison of various data forwarding ap-
proaches in wireless ad hoc networks. In general, some form of payment is
required. However, as the devices in a wireless ad hoc network are not under a
centralized authority's control, it is very di cult to enforce a secure payment
clearance mechanism. Auction schemes are interesting but are also di cult to
implement in practice because a highly trusted communication infrastructure
is required for exchanging bidding information. Yet this is a paradoxical re-
quirement as the communication among wireless peers is itself the ultimate
goal in data forwarding. Similarly, a reputation-based approach, while not dif-
ficult to implement in practice, could also lead to a paradoxical situation in
the sense that the reputation values may not be trustworthy.
5.3.2 Wireless Information Sharing Systems
Wolfson et al. [Wolfson et al., 2004] investigated an interesting opportunis-
tic wireless information exchange problem in which a moving vehicle transmits
the information it has collected to encountered vehicles, thereby obtaining
other information from those vehicles in exchange. The incentive mechanisms