Towards Elastic SDR Architectures Using Dynamic Task Management

SDR platforms integrating several types and numbers of processing elements in System-on-Chips become an attractive solution for baseband processing in wireless systems. In order to cope with the diversity of protocol applications and the heterogeneity of multi-core architectures, a hierarchical approach for workload distribution is proposed in this paper. Specifically, a system-level scheduler is employed to map applications to multiple processing clusters, complemented with a cluster-level scheduler - the CoreManager - for dynamic resource allocation and configuration as well as for task and data scheduling. A performance analysis of the proposed approach is presented, which shows the advantages of dynamic scheduling against a static approach for variable workloads in the LTE-Advanced uplink multi-user scenarios.