Performance evaluation of OFDMA and aggregation downlink stateless service disciplines in Wi-Fi networks

Aggregation and Orthogonal Frequency Division Multiple Access (OFDMA) are two fundamental features allowing access points and stations to benefit from the high physical transmission rates of the recent WiFi standards. They aim to pool frames with a unique overhead (mainly the physical header and the acknowledgment) to mitigate its impact on the throughput. Natural implementations of these features lead to nonFIFO (First In, First Out) service disciplines that may generate unfairness between frames. In this paper, we evaluate three stateless service disciplines that require a low level of resources (computation and memory). We prove that when OFDMA does not introduce additional delay, one of these greedy algorithms minimizes the overhead. Whereas this service discipline maximizes the system capacity, it generates strong unfairness between the stations. Instead, the two other algorithms may offer a good trade-off between capacity and fairness. These algorithms are evaluated through a theoretical framework and simulations that replay actual Wi-Fi traces.