Scheduling Analysis of TDMA-Constrained Tasks: Illustration with Software Radio Protocols

In this paper a new task model is proposed for scheduling analysis of dependent tasks in radio stations that embed a TDMA communication protocol. TDMA is a channel access protocol that allows several stations to communicate in a same network, by dividing time into several time slots. Tasks handling the TDMA radio protocol are scheduled in a manner to be compliant with the TDMA configuration: task parameters such as execution times, deadlines and release times are constrained by TDMA slots. The periodic task model, commonly used in scheduling analysis, is inefficient for the accurate specification of such systems, resulting in pessimistic scheduling analysis results. To encompass this issue, this paper proposes a new task model called Dependent General Multiframe (DGMF). This model extends the existing GMF model with precedence dependency and shared resource synchronization. We show how to perform scheduling analysis with DGMF by transforming it into a transaction model and using a schedulability test we proposed. In this paper we experiment on "software radio protocols" from Thales Communications & Security, which are representative of the system we want to analyze. Experimental results show an improvement of system schedulability using the proposed analysis technique, compared to existing ones (GMF and periodic tasks). The new task model thus provides a technique to model and analyze TDMA systems with less pessimistic results.