In this paper, we study the delay performance of
CSMA policies in wireless networks, where the delay is defined
as the average time that a silent wireless link needs to wait until it
accesses the channel for packet transmission. It is well-known that
CSMA policies can incur an access delay that may be correlated
over time and may grow exponentially with the network size. This
discourages practical implementation of CSMA policies in even
mid-sized networks. In this paper, we provide a new perspective
on the delay performance of CSMA policies. We present recently
developed results for two important interference models and show
how CSMA policies can be used to ensure an access delay that
is memoryless over time or that does not grow exponentially with
the network size. The two interference models that we consider
are primary interference and the “lattice interference graph”.
Our results suggest that CSMA policies can achieve a delay
performance, as well as a delay-throughput trade-off, that makes
them viable to be used in practice.