Research

Principal Investigators:  

Mark Lawford, Vera Pantelic, Victor Bandur  

Description: 

The project aims to evaluate various aspects of centralized automotive E/E (Electric/Electronic) architectures, including hardware platforms, communication protocols (CAN, CAN FD, Automotive Ethernet), functional safety, and software quality. A physical prototype of an automotive electric powertrain is being built as part of a project with a large automotive company. 

Potential for HQPs involvement:  

HQPs can be involved in this project, and they will have the opportunity to work with a special-purpose hypervisor to enable the running of several virtual machines on an embedded platform; write a minimal working hypervisor from scratch; work with industrial Simulink models of both controllers and plants on a Hardware-in-the-Loop (HiL) testing platform; and port existing industrial software to new hardware platforms. Furthermore, the resulting prototype of the powertrain E/E architecture will be evaluated for different communication protocols, including automotive Ethernet, and various schemes for the allocation of software components to available hardware units. 

Principal Investigators:

Richard Paige, Alan Wassyng

Description:

This project focuses on the software engineering of different kinds of health care applications, including personalized health care applications running on mobile devices (e.g., smart watches), as well as novel ultrasound treatment devices.

Potential for HQPs Involvement:

HQPs can participate in various components of work including:

• Automating the development of parts of a smart watch application via the use of code generation and/or user interface mockups.
• Integration of machine learning-based models with code generators to automate personalization of user interfaces for mobile health applications.
• Security assessment of smart watch applications for health care, where personalized data is collected and analyzed: how can we make end-users aware of the security risks associated with the use of the app, e.g., through a novel user interface?
• Use of model transformation and/or machine translation to enable harmonization of electronic health records (building on top of Google Cloud infrastructure).
• Enhancements to smart watch applications to support sign language users.
• Safety assurance of novel ultrasound devices: we are constructing an assurance argument that a novel device is acceptably safe to deploy, and this argument needs to be refined, rigorously specified, and then validated with the domain experts building the device.
• Model-based engineering of software components for an ultrasound device, emphasizing code generation of configuration data, as well as traceability between high-level requirements and the generated code.

Principal Investigator:

Sebastien Mosser

Description

This project focuses on enriching the jPipe language (https://github.com/ace-design/jpipe). So far, jPipe is a textual language used to model Justification diagrams, as a way to document different kinds of systems. A Justification diagram assembles evidence (e.g., “tests are ready”) through strategies (e.g., “execute test suite”) to reach conclusions (e.g., “quality assessed through testing”). We recently experimented with Continuous Integration/Continuous Delivery platforms (e.g., GitHub Action) to operationalize such justifications and execute them as CI/CD pipelines. We applied these ideas to pipelines and evaluated the reproducibility of machine learning notebooks. The jPipe environment was initially started as a summer internship and has since grown into a more fully-fledged environment with interest from industrial partners such as TELUS and Airbus. This was recently presented at CONFLANG last October (https://www.youtube.com/watch?v=NbEkN8OOQFQ&t=5210s).

Potential for HQPs Involvement:

HQPs can participate in various activities, including exploring the capacity of automating what was done manually in our initial experiments, by providing a way to transform a justification diagram into an executable GitHub action pipeline automatically.