Evaluating Inter-Operator Cooperation Scenarios to Save Radio Access Network Energy

Abstract

Reducing energy consumption is crucial not only to reduce OPEX but also to reduce the human debt to our planet. Over the past few years, most service providers (SPs) have actively tackled this issue, particularly targeting periods of low activity. Indeed, having fewer customers during these periods allows SPs to downsize or shut down part of their infrastructure. But this is not always optimal. Despite multiple energy-efficient optimizations, a mobile national operator (MNO) still need to maintain significant radio access network (RAN) infrastructure active at night. Could MNOs do better by cooperating with each other in such a way that an MNO can redirect its subscribers to a partner MNO, thus allowing its entire infrastructure to be temporarily deactivated while switching roles with the partner during a subsequent drop in activity period? To answer this question, we investigated a novel collaborative framework based on multi-agent reinforcement learning (MARL) allowing for negotiations between SPs as well as trustful reports from a distributed ledger technology (DLT) to evaluate the amount of energy saved. We leveraged it to experiment three different sets of rules (free, recommended, or imposed) regulating the negotiation between multiple SPs (3, 4, 8, or 10). Based on the observation of four cooperation metrics (efficiency, safety, incentive-compatibility, and fairness), the simulations showed that the imposed set of rules proved to be the best mode.