Wireless scheduling subject to slice constraints

Speaker : Matthew Andrews
Nokia US
Date: 02/03/2022
Time: 3:00 pm - 4:00 pm
Location: Zoom + LINCS


We present methods for achieving slice guarantees in cellular networks by adjusting the MAC scheduling weights. A network slice is a portion of a network that is carved out for a particular set of users or a particular type of traffic. One slice provided by a national operator could be used to support a Mobile Virtual Network Operator (MVNO), another could support Ultra-Reliable Low Latency Communication for vehicular applications, and a third could support an enterprise application such as an advanced wireless enabled factory.

Our goal is to optimize network utility (e.g. Proportional Fairness) subject to a variety of slice constraints such as the minimum aggregate rate provided to a slice, the minimum total resources provided to a slice, or a guarantee on the fairness of service within a slice. We show how such constraints can be achieved using multiplicative or additive adjustments to the standard Proportional Fair scheduler weights and we analyze the differences between the two approaches. We also discuss how the interaction between the constraints can make the situation significantly more complex than the analogous problem with single-user minimum rate constraints only. For example, with minimum resource constraints the optimum solution can lie in the interior of the achievable rate region.

This is joint work with Sem Borst, Siegfried Klein, Hans Kroener and Silvio Mandelli.