A comparison between opportunistic and fair resource allocation scheduling for LTE

The increasing demand for data traffic through high capability mobile devices and the trade-off between revenues and costs for upgrading the communication infrastructures strongly push the network operators to make the best use of the available network resources, which renews and reinforces the interest on resource management strategies. In this respect, a key challenge consists in finding a proper balance between an opportunistic approach, which aims at maximizing the total amount of carried traffic, and a fairness-based approach that, instead, aims at providing some type of fairness among the users. In this paper, we compare a few resource allocation policies for the downlink channel of LTE networks in terms of aggregate capacity, user fairness, and scheduler complexity. Two schemes are based on the Hungarian method, which was designed to associate exactly one resource to each user, while minimizing the overall cost of the assignment. This method hence provides a nice tradeoff between resource exploitation and user fairness in terms of number of allocated resources. We then propose practical implementations of the Hungarian method that work even when the number of users differs from the number of available resources, and users have heterogenous channel conditions and offered traffic. In addition, we consider a third scheduler that follows a back-pressure approach, which which assignes resources to users with longer backlog in an opportunistic manner, thus pursuing both aggregate throughput maximization and fairness among users with equal traffic demand.