To fully utilize multi-processors, new tools are required to manage software complexity. We present a novel technique that enables automating hierarchical process network transformations to derive optimized parallel applications. Designers leverage a library of process constructors and data-parallel algorithmic skeletons, utilizing the well-defined semantics of a restricted set of operators. This carefully chosen set addresses both temporal and spatial aspects of computation, enabling the automated identification of various parallel patterns. We utilize an augmented version of a meta-modeling framework grounded in system graphs and trait hierarchies to generate an intermediate representation (IR) of the system model to simplify automatic transformations and evaluations. Our augmentation allows for capturing skeletons and hierarchical networks. By meticulously selecting the underlying framework, we alleviate the need for tool integration in our design flow. We validate our approach through a proof-of-concept implementation, where our automated tool applied 193 transformations to fully parallelize an image processing application.

Design space exploration (DSE) is a key activity in embedded design processes, where a mapping between applications and platforms that meets the process design requirements must be found. Finding such mappings is very challenging due to the complexity of modern embedded platforms and applications. DSE tools aid in this challenge by potentially covering sections of the design space that could be unintuitive to designers, leading to more optimised designs. Despite this potential benefit, DSE tools remain relatively niche in the embedded industry. A significant obstacle hindering their wider adoption is integrating such tools into embedded design processes. We present two contributions that address this integration issue. First, we present the design space identification (DSI) approach for systematically constructing DSE solutions that are modular and tuneable. Modularity means that DSE solutions can be reused to construct other DSE solutions, while tuneability means that the most specific DSE solution is chosen for the target DSE problem. Moreover, DSI enables transparent cooperation between exploration algorithms. Second, we present IDeSyDe, an extensible DSE framework for DSE solutions based on DSI. IDeSyDe allows extensions to be developed in different programming languages in a manner compliant with the DSI approach. We showcase the relevance of these contributions through five different case studies. The case study evaluations showed that non-exploration DSI procedures create overheads, which are marginal compared to the exploration algorithms. Empirically, most evaluations average 2% of the total DSE request. More importantly, the case studies have shown that IDeSyDe indeed provides a modular and incremental framework for constructing DSE solutions. In particular, the last case study required minimal extensions over the previous case studies so that support for a new application type was added to IDeSyDe.

The design of modern embedded systems is increasingly complex. Different applications must share a heterogeneous embedded platform while satisfying demanding design requirements. The financial cost and engineering effort to implement such modern embedded systems are proportional to this increasing complexity. Model-driven-engineering (MDE) approaches mitigate this complexity by using well-defined models as the central elements of the design process. Namely, these models can be used to automate design process activities, such as design space exploration (DSE).

This thesis brings two contributions within this context. First, a novel meta-modelling approach for MDE with its implementation ForSyDe IO. Second, the design space identification (DSI) approach with its implementation IDeSyDe. ForSyDe IO is a language-agnostic MDE framework that promotes cooperative development through its underlying common model and capabilities. IDeSyDe is a modular and tuneable MDE DSE framework that enables the composable construction of DSE solutions via DSI. Of greater practical value, DSI and its implementation IDeSyDe seamlessly combine different DSE solutions to improve the overall exploration performance.

To ensure these contributions have immediate practical value, this thesis also presents four different MDE DSE scenarios originating from industrial cooperation incorporated into IDeSyDe. The industrial cooperation includes periodic workloads from avionics and automotive contexts and digital signal processing applications from academic contexts. ForSyDe IO was used to express each case study’s applications, platforms and requirements, which shows how it can be incrementally adapted for different scenarios. At the same time, IDeSyDe is used to construct DSE solutions for each case study in a fashion that displays IDeSyDe’s modularity, tuneability and capabilityfor synergizing different DSE solutions.

The case studies show qualitatively how the contributions, DSI in particular, aid in providing a cooperative and modular environment for developing MDE DSE solutions. At the same time, the numeric results of case studies show quantitatively that the overhead of the DSI automated proceduresis negligible compared to the overall DSE process and that the transparent combination of explorers improves the overall exploration performance without additional development effort.

A challenge in designing resource-constrained embedded systems for digital signal processing (DSP) is their complexity due to their vast design spaces, where only a fraction of implementations are feasible or optimal. A crucial tool to aid in this challenge is automated design space exploration (DSE). However, no exact, multi-objective, and preference-free DSE approach exists for DSP applications on resource-constrained embedded platforms. We propose a novel DSE solution with these ideal characteristics to perform DSE of analyzable DSP applications for tile-based multiprocessing embedded platforms. Our proposal harmonizes the exactness of constraint programming (CP) and the exploration efficiency of genetic algorithms (GA). Through this synergy, no single-objective reduction strategy or a priori objective preferences is required. We evaluate the proposal through state-of-the-art single-objective case studies and multi-objective case studies inspired by these. The evaluations show that our proposal improves the single-objective state-of-the-art and finds high-quality approximate Pareto-frontiers for the multi-objective case study. Therefore, our proposal is a more performant single-objective DSE solution than the state-of-the-art, and it is the first exact, multi-objective, and preference-free DSE approach for the problem addressed.

    Future avionic systems will be increasingly automated. The size and complexity of the avionics functions in these systems will increase likewise. The degree of attainable automation directly depends on the avionics system's computing power and the efficiency of available tools that map the overall functionality onto the target heterogeneous platform architecture. In safety-critical scenarios, these automation tools must also provide safety guarantees that aid or drive the certification processes.

    In line with this automation goal, We propose a novel design space exploration technique for the mapping functionality on IMA platforms.    The design space exploration technique returns mappings of the functionality onto the platform that are safe and increasingly resource-efficient.    A safe mapping is one where the functional and extra-functional requirements are met.    A resource-efficient mapping is one where fewer processing elements are used to achieve a safe mapping.    More importantly, the proposed technique can return computational proof that no safe mapping is likely possible. This proof is key for safety-critical contexts.

    To demonstrate the suitability of our technique for avionics systems design scenarios, we investigate its use with an industrial avionics case based on the ones from the PANORAMA ITEA3 project. The case study includes two avionics functionalities,    one control functionality, and one streaming-like functionality. The platform is hierarchical and heterogeneous, with elements oriented for higher safety and elements oriented for higher performance.    The avionics case-study evaluation shows that our novel design space exploration technique's abstractions and assumptions adequately represent avionics design scenarios directly or through a systematic overestimation.

    The technique is openly available within the design space exploration tool IDeSyDe. Therefore, designers can immediately benefit from the optimality and safety guarantees given by our novel design space exploration technique in their avionics design process.

Model-driven engineering (MDE) addresses the complexity of modern-day embedded system design. Multiple MDE frameworks are often integrated into a design process to use each MDE framework's state-of-the-art tools for increased productivity. However, this integration requires substantial development effort. In this paper, we propose an MDE, framework based on a formalism of system graphs and trait hierarchies for programming-language-agnostic integration between tools within our framework and with tools of other MDE frameworks. Implementing our framework for each programming language is a one-time development effort. We evaluate our proposal in an MDE design process by developing a Java supporting library and an AMALTHEA connector. Then we perform an MDE, industrial avionics case study with both. The evaluation shows that our framework facilitates the integration of different tools and the independent development of different system parts. Therefore, our framework is a reliable MDE, framework that lowers the effort of integrating tools to benefit from their combined state-of-the-art.

Nowadays, it is a reality to launch, operate, and utilize small satellites at an affordable cost. However, bandwidth constraint is still an important challenge. For instance, multispectral and hyperspectral sensors generate a significant amount of data subjected to communication channel impairments, which is addressed mainly by source and channel coding aiming at an effective transmission. This paper targets a significant further bandwidth reduction by proposing an on-the-fly analysis technique on the satellite to decide which information is effectively useful for specific target applications, before coding and transmitting. The challenge would be detecting clouds and vessels having the measurements of red-band, green-band, blue-band, and near infrared band, aiming at sufficient probability of detection, avoiding false alarms. Furthermore, the embedded platform constraints must be satisfied. Experiments for typical scenarios of summer and winter days in Stockholm, Sweden, are conducted using data from the Mimir’s Well, the Saab AI-based data fusion system. Results show that non-relevant content can be identified and discarded, pointing out that for the cloudy scenarios evaluated, up to 73.1% percent of image content can be suppressed without compromising the useful information into the image. For the water regions in the scenarios containing vessels, results indicate that a stringent amount of data can be discarded (up to 98.5%) when transmitting only the regions of interest (ROI). 

New methodologies are needed for the development of avionics systems to meet today’s software explosion in complexity and related cost due to the increased functionality in the aircraft. Current design flows for software-intensive systems do not have a clear path from the functional specification to the final implementation and cannot provide real-time guarantees. The situation will become even more difficult because, in the future, more and more applications will share the same computation nodes and the network in a distributed hierarchical network-based system. In order to overcome the present situation, a novel methodology for a correct-by-construction design of safety-critical embedded avionics systems has been created and formulated within the Vinnova NFFP7 project CORRECT. Correct-by-construction design is a radical departure from current design practice, with the potential to decrease the verification costs for future systems significantly. The paper presents the underlying foundation of the methodology, its carefully selected ingredients, and discuss available results and existing tool support. The methodology is based on a disciplined system modelling environment grounded on a sound formal foundation, a design space exploration technique, and a clear path to hardware and software synthesis. An industrial case study investigates the potential of the methodology.

Design space exploration (DSE) is a key activity in embedded system design methodologies and can be supported by well-defined models of computation (MoCs) and predictable platform architectures. The original design model, covering the application models, platform models and design constraints needs to be converted into a form analyzable by computer-aided decision procedures such as mathematical programming or genetic algorithms. This conversion is the process of design space identification (DSI), which becomes very challenging if the design domain comprises several MoCs and platforms. For a systematic solution to this problem, separation of concerns between the design domain and decision domain is of key importance. We propose in this paper a systematic DSI scheme that is (a) composable, as it enables the stepwise and simultaneous extension of both design and decision domain, and (b) tuneable, because it also enables different DSE solving techniques given the same design model. We exemplify this DSI scheme by an illustrative example that demonstrates the mechanisms for composition and tuning. Additionally, we show how different compositions can lead to the same decision model as an important property of this DSI scheme.

The size and complexity of automotive software systems are steadily increasing. Software functions are subject to different requirements and belong to different functional domains of the car. Meanwhile, streaming applications have become increasingly relevant in emerging application areas such as Advanced Driving Assistance Systems. Among models for streaming applications, the Synchronous Data Flow model is well-known for its analysable properties. This work presents transformation rules that allow transforming applications described by the Synchronous Data Flow model to an automotive component model. The proposed transformation rules are implemented in form of a software plugin for an automotive tool suite that allows for timing analysis, code synthesis and deployment to a Real-Time Operating System. To demonstrate the applicability of the proposed approach, a case study of a Kalman filter that is part of a simplified cruise control application is presented. An abstract Synchronous Data Flow model of the filter is transformed into a component that is deployed on an Electronic Control Unit with hard timing guarantees.