Effect of buffering on the throughput of ALOHA

Interest in ALOHA has waned because its throughput is low in comparison to other more complex MAC protocols. Yet those protocols suffer substantial performance degradation in networks with long relative propagation delay. Abramson's well accepted analysis of ALOHA estimates a maximum throughput of 18.6% assuming an infinite number of nodes and a Poisson distribution of transmitted packets. Previous work extended the analysis to allow for a finite number of nodes. However, in modern use buffered transmission is an obvious implementation feature that is not addressed by published analysis. Here we derive a method to calculate the throughput of ALOHA for both buffered and unbuffered cases with a finite number of nodes. We calculate theoretical values for two different collision cases and thereby determine the relationship between offered load and transmitted load. Ultimately, we show by analysis and simulation that the maximum throughput for a low number of active nodes approaches to 32.22% with buffering and 27.37% for unbuffered ALOHA. Therefore, ALOHA is even more attractive than commonly understood as a MAC for networks with large relative propagation delay, such as underwater acoustic networks.