Dr Bernd Fischer

Dr Bernd Fischer joined ECS in January 2006 as a Senior Lecturer in the Dependable Systems and Software Engineering group.

Bernd Fischer spent much of his career writing code for NASA. Now in his new role at ECS he is working on new code generator validation techniques which could save the space agency large sums of money.

As a student in high school Bernd got a home computer and became very keen to learn it for real. When he finished at High School in 1985, he went on to do a degree at Technical University Brunswick, Germany, followed by a PhD in Computer Science.

His PhD was entitled Deduction-based software component retrieval and involved him designing an environment which provided a way for users to write programme specifications which enabled them to find components proven to fit their needs.

In 1998, Bernd moved to California to join NASA Ames Research Center where he began working on programme synthesis. He devised a system which allows users to write high-level descriptions of the problem they are trying to solve from which code can be generated automatically to suit the user's needs.

'This is a very useful method,' Bernd commented. 'It can be used in engineering – for example, it can be used to translate differential equations or statistical models into code.'

He joined ECS in 2006 because he wanted to be back in Europe and to work in academia again. 'ECS is a good fit to my interests and I learned that it is one of the top schools in the country,' he said. 'British friends advised me not to join any school rated below five as a good school with good students is very important to me.'

It is this work on code generation which has led to Bernd working on his current project for NASA. Together with collaborators still working at NASA, he has come up with a way to slim line the generation of code that can be used in the navigation of airplanes and satellites.

'There is a lot of effort and money being spent on code generation,' Bernd commented. 'But at the moment, people don't trust code generators because they are still unreliable and can produce code that could also contain undetected errors.'

Bernd is developing a way to validate that the produced code is free from errors without the need for it to be tested a second time which is current industry practice.

This is good news for the aerospace industry and could also benefit automotive manufacturers.

'Reliable code generation is a real necessity at the moment,' commented Bernd. 'It will save time and money and increase confidence.'