A QoS-Demand-Aware Computing Resource Management Scheme in Cloud-RAN

This article focuses on computing resource allocation in Cloud Radio Access Networks. A game-based optimization algorithm was developed to distribute the computing resources among BaseBand Units (BBUs) in a BBU-pool whereby resources utilization is maximized. The model allocates computing resources on-demand, based on the instantaneous requests of BBUs, using a game-theory bargaining approach. In the case that the available resources are not sufficient to fulfil all instantiation requests, BBUs are prioritized to ensure the adequate Quality of Service, low-priority ones being always guaranteed a minimum computing resource to avoid them to crash. The performance of the proposed model is observed over time, concerning resource usage, BBU fulfilment level and efficiency. Simulations in a group of cells with a mixture of heterogeneous services in tidal traffic conditions show that resources allocated to BBUs are consistent with the priority of ongoing services and in line with real-time demand. Results also show that improving the average fulfilment level from 98% to 100% requires doubling the available resources at the cost of the average resources usage being cut in half.