Fátima Mattiello-Francisco

InRob: An approach for testing interoperability and robustness of real-time embedded software

Advances in digital technologies have contributed for significant reduction in accidents caused by hardware failures. However, the growing complexity of functions performed by embedded software has increased the number of accidents caused by software faults in critical systems. Moreover, due to the highly competitive market, software intensive subsystems are usually developed by different suppliers. Often these subsystems are required to interact with each other in order to provide a collaborative service. Testing approaches for subsystems integration support verification of the quality of service, focusing on the subsystems interfaces. The increasing complexity and tight coupling of real-time subsystems make integration testing unmanageable. The ad-hoc approach for testing is becoming less effective and more expensive. This article presents an integration testing approach denominated InRob, designed to verify the interoperability and robustness related to timing constraints of real-time embedded software. InRob guides the construction of services, based on formal models, aiming at the specifications of interoperability and robustness of test cases related to delays and time-outs of the messages exchanged in the interfaces of interconnected subsystems. The proposed formalism supports automatic test cases generation by verifying the relevant properties in the service behavioral model. As timing constraints are critical properties of aerospace systems, the feasibility of InRob is showed in the integration testing process of a telescope onboard in a satellite. The process is instantiated with existing testing tools and the case study is the software embedded in the telescope.

Designing fault injection experiments using state-based model to test a space software

Software for space applications requires significant testing. This paper presents an evaluation of the CoFI testing methodology as applied to actual space software, where deterministic fault cases derived from state-based models were executed using the software-implemented fault injection technique. Different models were used to represent the behavior of embedded software in a real satellite computer under the presence of both normal inputs and external faults in communication, processor, and memory. CoFI methodology was used for model construction, the Condado tool for test derivation, and the QSEE-TAS tool for test execution. In total, 8,620% of 471 fault cases detected errors in the software; this is a very large number, and more so considering that the software had already been tested by the company which developed it before being subject the CoFI methodology.

An Independent Software Verification and Validation Process for Space Applications

This paper presents an Independent Software Verification and Validation process that applies reviews for verification and a systematic testing methodology to guide validation. This process was applied to a pilot project named Quality Software Embedded in Space Missions (QSEE) at INPE and pointed very good results. The main feature of the process is that it uses a particular testing methodology named CoFI and an automatic test cases generation tool based in state-models. These features allowed systematizing validation activities which were carried on by a team not involved with the software development. The main activities of the process, the results in terms of the errors found not only through the reviews but also through the tests are presented. Lessons learned including drawbacks and benefits are discussed as well.

A Brazilian Software Industry Experience in Using ECSS for Space Application Software Development

This paper presents the tailoring of ECSS software product assurance requirements aiming at the development of scientific satellite payload embedded software by a Brazilian software supplier. The software item, named SWPDC, developed by DBA Engenharia de Sistemas LTDA within Software Factory context, is part of an ongoing research project, named Quality of Space Application Embedded Software - QSEE, developed by National Institute for Space Research — INPE, with FINEP financial support. Among other aspects, QSEE project allowed to evaluate the adherence of a Software Factory processes to INPE’s embedded software development process requirements. Although not familiar with space domain, the high maturity level of such supplier, CMMI-3 formally evaluated, facilitates the Software Factory to comply with the requirements imposed by the custumer. Following the software verification and validation processes recommended by ECSS standards, an Independent Verification and Validation - IVV approach was used by INPE in order to delegate the software acceptance activities to a third party team. ECSS standard tailored form contributions along the execution of the project and the benefits provided to the supplier in terms of process improvements are also presented herein.

Extended interoperability models for timed system robustness testing

Time incompatibilities in the interactions of timed subsystems on board satellites can cause deadlocks and data loss. Verification of time requirements is one of the major challenges in integration testing of complex timed systems. This paper proposes a new approach to enrich interoperability formal models with timing deviations in order to derivate robustness test cases. The combination of such an approach with architectural aspects of the test system improves the testing process, adding controllability and observability. Experiments demonstrate the generation of effective interoperability and robustness test cases for the integration phase of a subsystem embedded in a space telescope system.

Dependability Verification of Nanosatellite Embedded Software Supported by a Reusable Test System

The use of CubeSats has increased tremendously over the 15 years since the standard creation because the low cost and reduced project development cycle. However, one of the most concerns in reducing a project delivery time is the collateral effect in test process, resulting in failures in the mission operation. This paper proposes the combined use of the Model-Driven Engineering (MDE) and Model-Base Testing (MBT) approaches in the Validation and Verification (V&V) process of a nanosatellite mission, focusing on an evolved way to measure the dependability requirements of the interoperable on-board software. The proposal counts on a reusable Test System (TS) based on Arduinos that are integrated in the engineering model of the Cubesat architecture via I2C bus. From the behavioral models of both the on-board computer and the satellite payloads, source code can be generated in order to be embedded in the Arduinos, prototyping in the TS the expected behavior of the interactions between the specified subsystems. These models can also be useful to derive test suites following a MBT approach. Thus the TS can support the execution of different test cases at different stages of development of the software intensive subsystems. The proposed V&V process is discussed in the context of a particular nanosatellite named NanosatC-Br2 under development at INPE.