Trying to use a wireless access point in a crowded venue can be very frustrating. The bandwidth is limited, at least in the short term; therefore, the fight for the available bandwidth with your fellow users has to be resolved in some equitable way. Should you all be given an equal share, maximizing fairness, or should those with higher connection speeds be prioritized, maximizing throughput?

The authors propose a new packet scheduling algorithm for access points that dedicates longer time slices to higher speed devices while still iterating through all users in a round-robin fashion to ensure some level of fairness. At each iteration point, the algorithm dynamically adjusts the time window dedicated to that client. The time window is optimized over a discrete number of durations to find the time interval that will maximize the sum of a throughput efficiency computation and the Jain’s fairness index using two mixing parameters.

Unfortunately, the two mixing parameters are left for the implementer to choose with only some empirical results as guidelines. One redeeming aspect of the research is that the authors implemented the algorithm on a commercial wireless router to ensure that the exhaustive search algorithm could actually be computed in real-time using currently available hardware.