A new data flow analysis model for TDM

This paper proposes a new data ow model for analyzing the worst-case temporal behavior of resource arbitration through Time Division Multiplexing (TDM). TDM arbitration allows resource sharing amongst the tasks of concurrent applications, where each application may have its own end-to-end hard real time requirements, such as minimum throughput and maximum latency. Current data flow modeling techniques for the temporal analysis of TDM arbitration over-estimate the worst-case temporal behavior of tasks. This causes unnecessary over-reservation of resources to the application, leading to under-utilization of system re- sources and unnecessary rejection of additional applications. We propose a conservative data ow model that accurately estimates the worst-case temporal behavior of TDM arbitration. Unlike existing models, we do not make restrictive assumptions on the characteristics of TDM, nor on the amount of resources reserved. This enables optimized resource allocation for TDM arbitration. We present a new model that closely mimicks the worst-case temporal behavior of TDM arbitration. We formally prove that this model is conservative with respect to the worst-case behavior of TDM arbitration, and we prove that it is strictly more accurate than the state-of-the-art. Quantitatively, we show that our new model leads to a 20% improvement of resource allocation, in a case study of a wireless LAN radio down-link