To Sense or Not To Sense: A Comparative Study of CSMA With Aloha

A fundamental difference between the two most representative random-access schemes, Aloha and carrier sense multiple access (CSMA), is sensing. There has been a common belief that the access efficiency can always be improved by the use of carrier sensing, which is indeed based on an implicit assumption that the packet length is much larger than the sensing time. For machine-to-machine (M2M) communications featured with short packets, the benefit of sensing may not overweigh the cost any more. It is therefore of paramount importance to identify the conditions for CSMA to outperform Aloha. In this paper, the sum rate performance of CSMA networks with two representative receiver structures, i.e., the collision model and the capture model, is characterized and optimized, based on which a comparative study of the optimal sum rate performance between Aloha and CSMA is conducted to establish criteria for beneficial sensing. The analysis shows that the maximum sum rates of CSMA with both receiver structures logarithmically increase with the mean received SNR
$\rho $
at the high SNR region, and the rate gain of the capture model over the collision model is significant only when
$\rho $
is small. The critical threshold for the ratio of the sensing time to the packet length for beneficial sensing is characterized under various scenarios, and found to be close to zero at the low SNR region when the capture model is adopted, indicating that the packet length needs to be extremely large for CSMA to outperform Aloha in that case. The analysis sheds important light on the access design of M2M communications, and suggests that Aloha could be a more favorable option when short packets are sent by a massive amount of low-power machine-type devices.