We study the link congestion problem in contention-based access
networks.  These networks are characterized by the use of a random
access protocol for channel access. We assume adaptive users or
applications in the sense that their traffic flows are elastic. We
further assume that individuals' service requirements can be
characterized by demand functions that are price-sensitive. A
price-based rate control strategy, where the price indicates the
current congestion state, is considered to control channel congestion.

The effectiveness of price-based rate control is studied in the
context of the classic slotted ALOHA model with an infinite
population. The price as the new state variable is dynamically updated
based on control parameters and the ternary channel feedback. Our
results show that under this model stabilization of the ALOHA channel
can be achieved. In particular, by using drift analysis, we prove that
the associated Markov chain is positive recurrent. The resulting
steady state probability distribution thus characterizes an operating
point for the model. Moreover, a desired operating point as such could
be selected by proper choice of the control parameters. We also show
that service differentiation is realized at the operating point. From
a control perspective, where demand functions become predetermined
system parameters, such a price-based scheme offers a simple mechanism
to provide quality of service in a best-effort contention-based
network.