Virtualized resource sharing in cloud radio access networks: An auction approach

As mobile terminals proliferate and mobile Internet traffic explodes, demand for efficient and high capacity cellular access escalates. Two fundamental techniques are being implemented to help meet such a demand. The first is virtualization based, centralized cloud processing. Baseband signals are sampled and transmitted through front-haul links to a mobile cloud, for processing by mobile base station instances deployed in an on-demand fashion. User and channel information are aggregated to the cloud, facilitating optimized decision making. The second is the separation of infrastructure ownership from service provisioning in cellular networks. The "Tower" company now specializes in deployment and maintenance of the cloud-radio access network (C-RAN) infrastructure. Mobile operators focus instead on their sole business of wireless service provisioning. Mobile operators lease C-RAN resources that include spectrum resources at remote radio heads, front-haul bandwidth and mobile base station instances. This work proposes a natural auction approach for inter-operator resource sharing, where each operator bids a capacitated sub-network of the C-RAN. Drawing from the theories of Maximal-in-Range auctions and efficient graph algorithms, we design and test a C-RAN resource auction that is truthful, polynomial-time computable, and achieves close-to-optimal social welfare.