Approaches to adaptively reduce processing effort for LTE Cloud-RAN systems

Cloud-RAN is a novel architecture for LTE networks, where antennas with only limited processing capabilities are deployed in the field and all baseband and higher layer processing of the base stations is pooled in a central office. It has been shown that the centralization leads to multiplexing gains for signal processing hardware. When a system is able to efficiently cope with overload of the compute resources, significant savings are possible. This allows to install less resources (saving Capital Expenditure, CAPEX) and to switch off more resources during low-load periods (saving Operational Expenditure, OPEX). In this publication, the resource allocation of a system of LTE base stations is formulated as an optimization problem. The formulation includes the reduction of interference by not using radio resources for transmission and the flexibility of using different MIMO modes. The results of optimization runs are evaluated. They show that about 50% of the compute resources can be saved without impacting the system performance. In overload situations, only 20% of the resources are sufficient to deliver 87% of the system performance. The most efficient approach to reduce processing effort is to adapt the MIMO mode and to reduce the number of virtual transmit antennas.