Joint Sleep Mode Activation and Load Balancing with Dynamic Cell Load: A Combinatorial Bandit Approach
Wajahat Bashir Gilkar, Gourab Ghatak
TL;DR
The paper tackles energy efficiency in ultra-dense HetNets by jointly optimizing sleep-mode activation of gNB-SCs and CRE-based load balancing under QoS guarantees. It introduces a combinatorial CMAB framework using CUCB with an $(\alpha,\beta)$-approximation oracle to select SC ON/OFF patterns, accounting for non-monotonic rewards caused by dynamic load. A derivative-free Powell method tunes CRE parameters for the selected super-arm to minimize total load while enforcing $\rho_M \le 1$ and $\rho_i \le 1$. Numerical results show the proposed approach reduces power consumption compared to ALL-ON and VFA-SARSA, while maintaining QoS and scaling to larger UE counts, and it is suitable for deployment as O-RAN near-real-time xApps.
Abstract
We propose a combinatorial bandit formulation to opportunistically trigger sleep modes in gNode-B (gNB) small cells (SCs), followed by a cell range expansion (CRE)-based load balancing procedure. This is implemented by ensuring that the fifth generation (5G) quality of service identifier (5QI)-requirements of user equipments (UEs) are maintained. The key challenge is the fact that while deactivating a given SC gNB reduces its own consumption, it may increase the load on neighboring gNBs and the macro gNB (coverage cell), impacting the overall energy efficiency. This phenomenon is accurately characterized by modeling the dynamic cell load that jointly takes into account the location of the UEs, their relative locations to all the SCs, and their data demands. We experimentally show that the proposed combinatorial upper confidence bound (CUCB) followed by the load balancer outperforms not only the naive strategies like arbitrarily keeping all the SCs on, but also other state-of-the-art reinforcement learning solutions. The proposed algorithm can be implemented as open-radio access network (O-RAN) near-real-time (NRT) RAN intelligent controller (RIC) xApps.