Modern automotive embedded systems are composed of multiple real-time tasks communicating by means of shared variables. The effect of an initial event is typically propagated to an actuation signal through sequences of tasks writing/reading shared variables, creating an effect chain. The responsiveness, performance and stability of the control algorithms of an automotive application typically depend on the propagation delays of selected effect chains. Depending on the communication model adopted, the propagation delay of an effect chain may significantly vary, as may be the resulting overhead and memory footprint. In this work, the trade-offs between three communication models that are widely adopted for industrial automotive systems, namely Explicit, Implicit, and Logical Execution Time (LET), are explored. In particular, a formal analysis of the LET model is provided for real-time systems composed of periodic tasks with harmonic and non-harmonic periods, analytically characterizing the control performance of LET effect chains. It is shown that by introducing tasks offsets, the real-time performance of non-harmonic tasks may improve, getting closer to the constant end-to-end latency experienced in the harmonic case. Further, a heuristic algorithm is introduced that obtains a set of offsets that might reduce end-to-end latencies, improving LET communication determinism. The results are compared against an industrial case study consisting of an automotive engine control system. On the other hand, real-time servers have been widely explored in the scheduling literature to predictably execute aperiodic activities, as well as to allow hierarchical scheduling settings. As they facilitate timing isolation between different software components, there is a renewed interest for the adoption of fixed priority real-time servers in the automotive domain, as a way to implement more efficient reservation mechanisms than TDMA-based methods. In this thesis, the extended Polling Periodic, the Deferrable and Sporadic Servers are the main focus. Despite their popularity, only sufficient schedulability conditions exist for real-time systems scheduled with any of these servers. Thus, a formal characterization of an exact response time analysis for fixed priority systems based on the aforementioned servers in a multi-level scheduling setting under preemptive scheduling is developed. In addition to that, an experimental characterization of the schedulabilty improvement that can be obtained with respect to existing sufficient schedulability tests is provided, proving the effectiveness of the proposed exact analysis. Additionally, an overhead-aware schedulability analysis for hierarchical fixed priority preemptive systems is given, and a practical server parameterization heuristic technique, that preserves the least possible utilization as well as enhances the aggregated WCRT of the tasks in a hierarchical scheduling setting, is exposed. Moreover, the temporal isolation offered by the above-mentioned servers is further exploit to serve tasks originally scheduled in the background so that they can meet their deadlines. Hence, a method to implement a Deferrable Server (DS) on top of ETAS RTA-OS, a ubiquitous AUTOSAR-compliant OS, is presented and a heuristic to select its parameters is proposed. Lastly, the effectiveness of the parametrization is proven by applying the technique to an industrial case study consisting of an automotive engine control system.
I moderni sistemi embedded automobilistici sono composti da diversi task con constraint di tempo reale, questi task comunicano per mezzo di variabili condivise. In questo senso, l'effetto di un evento iniziale è tipicamente propagato ad un segnale di attuazione attraverso sequenze di task che scrivono e leggono variabili condivise, creando un ‘effect chain’. La reattività, le prestazioni e la stabilità degli algoritmi di controllo di un'applicazione automotive dipendono tipicamente dai ritardi di propagazione degli effect chain selezionati. A seconda del modello di comunicazione adottato, il ritardo di propagazione di effect chain può variare in modo significativo, come può essere l'overhead risultante e il footprint di memoria. In questo lavoro, esploriamo i compromessi tra tre modelli di comunicazione che sono ampiamente adottati per i sistemi automobilistici industriali, vale a dire, Explicit, Implicit, e Logical Execution Time (LET). In particolare, viene fornita un'analisi formale del modello LET per sistemi in tempo reale composti di task periodici con periodi armonici e non armonici, caratterizzando analiticamente le prestazioni di controllo degli effect chain usando l’approccio LET. Si dimostra che introducendo le compensazioni dei task, le prestazioni in tempo reale dei task non armonici possono migliorare, avvicinandosi alla costante latenza end-to-end sperimentata nel caso delle armoniche. Inoltre, viene presentato un algoritmo euristico che ottiene un insieme di offset che potrebbe ridurre le latenze end-to-end, migliorando il determinismo della comunicazione LET. I risultati sono confrontati con un caso di studio industriale che consiste in un sistema di controllo del motore di una macchina. D'altra parte, i server in tempo reale sono stati ampiamente esplorati nella letteratura per eseguire prevedibilmente task aperiodiche, così come per consentire impostazioni di programmazione gerarchiche. Poiché facilitano l'isolamento temporale tra i diversi componenti software, c'è un rinnovato interesse per l'adozione di server in tempo reale a priorità fissa nel settore automobilistico, come modo per implementare meccanismi di prenotazione più efficienti rispetto ai metodi basati su TDMA. In questa tesi, il Polling Periodico esteso, i Deferrable server e Sporadic server sono il focus principale. Nonostante la loro popolarità, esistono solo condizioni di schedulabilità sufficienti per i sistemi in tempo reale programmati con uno qualsiasi di questi server. Pertanto, viene sviluppata una caratterizzazione formale di un'esatta analisi dei tempi di risposta per i sistemi a priorità fissa basati sui suddetti server in un'impostazione di schedulazione multilivello sotto pianificazione preventiva. Inoltre, viene fornita una caratterizzazione sperimentale del miglioramento della schedulabilità che può essere ottenuta rispetto ai test di schedulabilità esistenti, dimostrando l'efficacia dell'analisi esatta proposta. In questo lavoro, approfittiamo del loro isolamento temporale per svolgere compiti originariamente programmati in background, in modo che possano rispettare le loro scadenze. Pertanto, presentiamo un metodo per implementare un Deferrable server (DS) in aggiunta all'ETAS RTA-OS, un sistema operativo che segue lo standard AUTOSAR, e proponiamo una euristica per selezionarne i parametri. Dimostriamo poi l'efficacia della parametrizzazione applicando la tecnica ad un caso di studio industriale costituito da un sistema di controllo del motore automobilistico.
Analisi di schedulabilità real-time per sistemi embedded per automotive
MARTINEZ GARCIA, Jorge Luis
2021
Abstract
Modern automotive embedded systems are composed of multiple real-time tasks communicating by means of shared variables. The effect of an initial event is typically propagated to an actuation signal through sequences of tasks writing/reading shared variables, creating an effect chain. The responsiveness, performance and stability of the control algorithms of an automotive application typically depend on the propagation delays of selected effect chains. Depending on the communication model adopted, the propagation delay of an effect chain may significantly vary, as may be the resulting overhead and memory footprint. In this work, the trade-offs between three communication models that are widely adopted for industrial automotive systems, namely Explicit, Implicit, and Logical Execution Time (LET), are explored. In particular, a formal analysis of the LET model is provided for real-time systems composed of periodic tasks with harmonic and non-harmonic periods, analytically characterizing the control performance of LET effect chains. It is shown that by introducing tasks offsets, the real-time performance of non-harmonic tasks may improve, getting closer to the constant end-to-end latency experienced in the harmonic case. Further, a heuristic algorithm is introduced that obtains a set of offsets that might reduce end-to-end latencies, improving LET communication determinism. The results are compared against an industrial case study consisting of an automotive engine control system. On the other hand, real-time servers have been widely explored in the scheduling literature to predictably execute aperiodic activities, as well as to allow hierarchical scheduling settings. As they facilitate timing isolation between different software components, there is a renewed interest for the adoption of fixed priority real-time servers in the automotive domain, as a way to implement more efficient reservation mechanisms than TDMA-based methods. In this thesis, the extended Polling Periodic, the Deferrable and Sporadic Servers are the main focus. Despite their popularity, only sufficient schedulability conditions exist for real-time systems scheduled with any of these servers. Thus, a formal characterization of an exact response time analysis for fixed priority systems based on the aforementioned servers in a multi-level scheduling setting under preemptive scheduling is developed. In addition to that, an experimental characterization of the schedulabilty improvement that can be obtained with respect to existing sufficient schedulability tests is provided, proving the effectiveness of the proposed exact analysis. Additionally, an overhead-aware schedulability analysis for hierarchical fixed priority preemptive systems is given, and a practical server parameterization heuristic technique, that preserves the least possible utilization as well as enhances the aggregated WCRT of the tasks in a hierarchical scheduling setting, is exposed. Moreover, the temporal isolation offered by the above-mentioned servers is further exploit to serve tasks originally scheduled in the background so that they can meet their deadlines. Hence, a method to implement a Deferrable Server (DS) on top of ETAS RTA-OS, a ubiquitous AUTOSAR-compliant OS, is presented and a heuristic to select its parameters is proposed. Lastly, the effectiveness of the parametrization is proven by applying the technique to an industrial case study consisting of an automotive engine control system.File | Dimensione | Formato | |
---|---|---|---|
Martinez-diss.pdf
accesso aperto
Dimensione
4.04 MB
Formato
Adobe PDF
|
4.04 MB | Adobe PDF | Visualizza/Apri |
I documenti in UNITESI sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/20.500.14242/78862
URN:NBN:IT:UNIMORE-78862