Scalable 5G Signal Processing on Multiprocessor System: A Clustering Approach

Abstract

5G supports the variety of new services with different requirements for throughput, latency and reliability. Multicore computing platforms are used to meet the various requirements while allowing scalability and flexibility in the implementation of the base stations. The challenge in this regards is the efficient distribution and processing of signal processing tasks on parallel processors. Moreover, with increasing of the application complexity, the management and synchronization overhead increases disproportionately, which limits the increase in performance and system efficiency. To cope with this problem the application granularity reduction using task clustering was proposed recently and demonstrated impressive performance improvement. Unfortunately, no practical clustering algorithm have been studied in this regards. Our motivation is to study and design well suited clustering algorithms to these needs. More particularly, we modify Clustering And Scheduling System II(CASSII) algorithm in order to gain higher speed-ups and show the performance improvement in regards to original algorithm and not clustered graphs.