About

Background. We are beginning to feel a disconnect between research and industrial practice on topics related to the design and implementation of complex embedded systems such as those common to automotive or avionics systems. As an example, some research results on the topic of mixed-critical systems assume that it is acceptable to abruptly drop (or kill) tasks in a running system in case the criticality level changes.

Whereas such a simplification does help in formulating elegant solutions, nonetheless, abruptly dropping tasks in practice can lead to unacceptable system-instability availability issues. On the other hand, we have seen that the industry is not able to appropriately transfer known and difficult problems to the research groups. Appropriate abstraction and formalization while avoiding trivialization of a challenging industrial problem in a way that protects the intellectual properties of a company(s) remains one of the biggest hurdle in an academic-industrial collaboration. Consequently, the research community is forced to make (sometimes unjustified) assumptions in order to complete the statement of the problem, leading to methods and tools which cannot be easily applied to solve the corresponding industrial problem, and thus, the embedded systems industry is not able to satisfactorily leverage the innovation taking place in the universities or other research groups.

The Objective of the Workshop. To respond to this challenge, the first workshop on Collaboration of Academia and Industry for Real World Embedded Systems, co-located with ESWeek 2016 in Pittsburgh, USA, intends to to bring together experts from academia and the industry in a collaboration for the design and implementation of industrial strength  embedded systems.

This workshop will focus on the following aspects of such a collaboration through talks and an interactive session:

  A. Challenges in the transfer of appropriate and formalized problem statements from industry to academia; and

  B. Challenges in the transfer of methods and tools developed by the academia back to the industry, and how to establish quality control over methods and tools, e.g., using existing or new benchmarks.

A. Challenges in the transfer of transfer of appropriate and formalized problem statements

The first hurdle is the construction of appropriately abstracted, but not trivialized problem statements, thereby allowing research partners to reason about core challenges contained in the problem without getting mired in implementation details. Too often, it has been observed that the specification of the problem is incomplete resulting in a solution based on unjustified assumptions. As a result, such solutions are usually unfit for adoption in the industry. As a concrete example, some researchers working on mixed criticality topics often assume that it is acceptable to simply and abruptly drop tasks once the criticality level changes. In practice, this is not acceptable, and often results in system stability and availability issues.

B. Challenges in the transfer of methods and tools to the industry and associated quality controls

The second hurdle is the incomplete, or often completely missing expectation(s) on the quality of solutions and tools developed for solving an agreed problem. The resulting methods and tools cannot successfully enter production since these do not scale, or rely on information (or assumptions) which may not be available in actual practice. Consequently, the innovations from the academia are unable to make the full impact on the industry. It may be beneficial to require that methods and tools developed in the course of research are tested for properties (e.g., scalability) using standardized benchmarks. During the workshop, we will discuss whether the existing benchmarks are sufficient, or new ones required, along with the associated specifications.

Format of the Workshop. Talks, and an interactive session. See the workshop program.

Comments are closed